Optogenetic Disruption of Memory-Driv~, Oculomotor Beha~the
Non-Human Pnmate
~
by
.
~ASS~~"~~~~~;~~~,~JT~2~,---~:~""!"'§!"""'(I1~-U~TE~.
I ~~,~ -;-~2014 ]
Leah C. Acker
LiBRARJES
S.M. Electrical Engineering and Computer Science
Massachusetts Institute of Technology, 2009
SUBMITTED TO THE HARVARO-MlT PROGRAM IN HEALTH SCIENCES AND TECHNOLOGY
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PlflLOSOPHY
AT THE
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
September 2014
© 2014 Leah C Acker.. All rights reserved.
The author hereby grants to MIT permission to reproduce and to distribute publicly paper and electronic
copies of this thesis document in whole or in part in~edium now known or hereafter created.
Signature redacted
Sign.ature of Autltor ..........~ __~...................................................... "..................... .
.
Harvard-MIT Pro2I'am in Health Sciences and Technology
Signature redacted
May23,2014
Certified by ...... .
Robert Desimone
~oris ~y ProfeSsor of Neuroscience
.
. S ig natu re redacted
Certified .by.. .. ... .... .. .
. ......................... .
Ed Boyden
Associate Professor and AT&tT Chair
Sig natu re redacted
C/ ...................................................
Accepted by.......... .
Thesis Supervisor
ThesisSupervisor
..
····E~~N B~~;iID:PhD
Professor of Computational Neuroscience and Health Sciences and Technology
Director, Harvard-MIT Program in Health Sciences and Technology
Optogenetic Disruption of Memory-Driven, Oculomotor Behavior in the Non-Human Primate
by
Leah C. Acker
Submitted to the Harvard-MIT Division of Health Sciences and Technology on May 23, 2014
in Partial Fulfillment of the Requirements for the Degree of
Doctor of Philosophy in Biomedical Engineering
Abstract
Pharmacological inactivation studies have shown that the frontal eye field (FEF) is critical for
executing saccades to remembered locations. FEF neurons increase their firing rate during the
three epochs of the memory-guided saccade task: visual stimulus presentation, the delay interval,
and motor preparation. It is unclear, though, whether FEF activity during each epoch is necessary
for memory-guided saccade execution. To address this question, techniques for millisecondprecise optical inactivation of the primate brain were invented. A red-shifted halorhodopsin
(Jaws) and a novel large-volume tissue illuminator were used in two rhesus macaques to
inactivate part of the FEF at different times during the memory-guided saccade task. Neuronal
recordings showed that the inactivated tissue volume (i.e., the volume where the firing rate of >
80% of neurons decreased by >80%) spanned several cubic millimeters, which is consistent with
histological findings. When the target was in the inactivated receptive field, error rates (i.e.,
failures to execute memory-guided saccades to the proper target location) increased in both
monkeys with inactivation during either the target, delay, or motor period. This implies that FEF
neuronal activity contributes to performance throughout the memory-guided saccade task.
2
Acknowledgments
First, I would like to thank Robert Desimone. Bob taught me how to ask hard questions, how to
design experiments, how to interpret results, and how to present the findings-in short, he taught
me how to think deeply and clearly about science. Just as importantly, Bob welcomed me into a
lab that feels like a family, where he is a kind of father-figure and where the "siblings" remain
close throughout their careers.
I would like to thank Ed Boyden for his support, help, and constructive feedback. Ed helped me
to obtain whatever resources I needed for my thesis, and he challenged me to think broadly about
research and the impact of my work.
I would like to thank Roger Mark, my committee chair, for all of the kind, honest, and practical
advice he has given me as a housemaster, professor, and mentor. Clearly, I have had the honor
of knowing him for many years in many different circumstances.
I would like to thank Dr. Ann Graybiel for serving on my committee with such enthusiasm and
dedication. Ann has long been an inspiration to me, but as we have gotten to know each other
better, I have discovered a kindred spirit, a mentor, and a friend.
I am thankful for the support of my parents, my brother, and especially, my husband, Jared.
I am grateful to my colleagues in the Desimone and Boyden Labs, from whom I have learned so
much. Specifically, I want to acknowledge: Amy Chuong, whose opsin, Jaws, was critical for this
study; Erica Pino, who assisted with the rodent experiments and fiber fabrication; Narcisse Bichot
and Huihui Zhou, who patiently taught me much of what I know about primate
neurophysiology; Fumiaki Yoshida, from whom I learned to be a good surgeon; Bob Schafer and
Bruno Gomes, my delightful officemates who taught me a lot about statistics and made it a joy to
work in 46-6121; Tim Buschman, Grant Mulliken, Reza Rajimehr, and Ying Zhang, who
generously shared their wisdom regarding everything from data analysis to experimental design
to career planning; the Desimone Lab technicians Ellen DeGennaro, Matt Heard, and Jonathan
Winkle-without their support this work would not have been possible; and, finally, Azriel
3
Ghadooshahy, Harbi Sohal, Min Xu, Daniel Baldauf, Yasaman Bagherzadeh Bioki, Rogier
Landman, Ben Lu, Justin Kinney, Annabelle Singer, Anthony Zorzos, Brian Allen, Jake Bernstein,
Aimei Yang, Brian Chow, Ian Wickersham, Xue Han, Christian Wentz, Nathan Klapoetke, Suhasa
Kodandaramaiah, and Giovanni Talei Franzesi for their friendship, support, feedback, and
encouragement.
I also would like to thank my 61 floor "neighbors": Jeetu Sharma, Robert Ajemian, Margo Cantor,
Henry Hall, Christine Keller, and Jannifer Lee. I am grateful to Kostas Tomadakis for fabricating
many of my designs over the years, and for the veterinary and animal care staff here at MIT, who
make our research possible.
Finally, despite my best effort above, it is impossible for me to properly thank everyone who has
encouraged, supported, and inspired me here at MIT, in general, and at the McGovern Institute
for Brain Research at MIT, in particular.
I have been humbled by the interest that has been shown in my work and by the generosity with
which so many here have shared their time, equipment, expertise, and insight. The McGovern
Institute is a truly special place, and it has been an honor to work here not just among colleagues,
but among friends.
4
Table of contents
Acknowledgments.....................................
.
3
................
Table of contents .................................-...-.---.------------------------------..................... 5
List of Figures..........................----......--------------------.---...-.........-----Chapter 1: Introduction ..
.......................
....---- 00-...--....---------------------------------
1.1 Frontal Eye Field ......................--.............----.-..-...............................................-
11
1.1.1 Anatomy .........................................................................................................
11
1.1.2 Frontal Eye Field Inactivation ................................-...........................................
15
1.1.3 Significance of studying the frontal eye field.......................................17
1.1.4 The pre-motor theory of attention....................................................18
1.1.5 Local FEF
18
-- -----------...............................................
circuitry ..................................................................-------
1.1.6 FEF in the context of larger oculomotor and attentional circuits....................................................... 18
19
-----------------------------...............................................
1.2 Optogenetics .........................................
21
1.2.1 Expressing the opsin in a large enough fraction of neurons................................................................
1.2.2 Opsin expression in rhesus macaques..............................................................21
1.3 Illumination and challenges with current light propagation models......................................22
1.4 "Backward" effects: activation of neurons with inactivating opsins..................23
24
1.5 Other challenges with primate optogenetics .................................
1.5.1 Absence of behavioral change.........................................................
1.5.2 Inconsistent behavioral changes .............................................................
1.6 Motivation for the present study ..............................
24
................
----
.... 25
26
....................
Chapter 2: Visible light propagation in the living brain: implications for optogenetics........27
28
2.1 Introduction .........................................................---------------...............................................
2.2 Materials and Methods........................................................................30
2.2.1 Anim al Care...........................................................
.................---
----- ----
2.2.2 Isometric light measurements ..........................................................................-----2.2.3 Illum inator calibration ....................................................--....................
---
30
----.............................................
... 31
.....-------------------........................... 33
2.2.4 Aligner placement and isometric probe insertion ................................
2.2.5 In vivo light measurements..............................................................................35
5
... 34
2.2.6 Determining absorption and reduced scattering coefficients ...............................................................
36
2.2.7 Diffusion model of light propagation in tissue .....................................................................................
37
2.2.8 Deriving absorption and reduced scattering coefficients from light measurements............................ 37
2.2.9 Geometric limit of light propagation in tissue......................................................................................
38
2.2.10 Kubelka-Munk model...............................................................................................................................39
2.2.11 Internal geometric model ........................................................................................................................
39
2.3 Results ...................................................................................................................................
40
2.4 Discussion ..............................................................................................................................
45
2.4.1 Comparison of measured absorption coefficients and those estimated based on the hemoglobin content
of the b rain .........................................................................................................................................................
45
2.4.2 Comparison of absorption coefficients reported with values in the literature ....................................
46
2.4.3 Kubelka-Munk model.................................................................................................................................47
2.4.4 Implications for the selection of opsins................................................................................................
48
Chapter 3: Illuminatorfor optogeneticmodulation oflarge brain volumes........................50
3.1: Introduction ..........................................................................................................................
50
3.2 Materials and Methods...........................................................................................................54
3.2.1 Illuminator construction ............................................................................................................................
54
3.2.2 Interfacing the illuminator with a laser ...............................................................................................
57
3.2.3 Illuminator profiling and calibration....................................................................................................
58
3.2.4 Subjects and surgical procedures .........................................................................................................
61
3.3 Results ...................................................................................................................................
3.3.1 Optical artifact in local field potential ..................................................................................................
63
63
3.3.2 Electrophysiology.......................................................................................................................................65
3.4 Discussion ..............................................................................................................................
67
3.4.1 Potential challenges...................................................................................................................................67
3.4.2 Extensions and further applications ....................................................................................................
68
Chapter 4: Optogenetic modulation of memory-guided saccades reveals that visual, delayperiod and motor-related FEF firingactivitycontributesto behavior..................................70
4.1 Introduction ...........................................................................................................................
4.2 Methods.................................................................................................................................72
4 .2.1 Su bjects......................................................................................................................................................72
6
70
- -... 73
-------------------...............................--
4.2.2 Surgical procedures ...........................................................--........
73
4.2.3 Localizing injection targets in behavior monkeys .................................................................................
.
4.2.5 Behavioral testing ..............................................................-............
.... 76
......
4.2.4 Virus injections in behavior primates .................................................................
----------------------..........................
81
4.2.6 Simultaneous recording and optical inactivation .................................................................................
....... 82
4.2.7 Classification of behavioral errors ..........................................................................
82
4.2.8 Electrophysiological analysis .......................................................................
...
4.2.9 Quantification of rebound ................................................................
.---.
--------.................... 83
84
4.2.10 Viral vector screening ....................................................................................................------................
4 .2.11 H istology .................................................................
.......... ...
----------------------------------------------....-------... 85
-------..---------------------------------------------...-.
4.3 Results .....................................................-.---........--.
78
88
4.3.1 Histology.....................................................------------------------------------------------...............................................88
4.3.2 FEF was identified using micro-stimulation .........................................................................................
89
4.3.3 Spatial specificity of neuronal responses..............................................................................................
90
4.3.4 Neurons with visual, delay, and motor activity were recorded...........................................................
92
4.3.5 FEF neurons were optogenetically inactivated in both monkeys .........................................................
94
4.3.6 Overall population level firing-rate analysis .................................................................................
94
4.3.7 Silencing of individual neurons............................................................................
.. ... 95
.......
4.3.8 All subtypes of neurons were silenced in both monkeys ....................................................................
4.3.9 Rebound activity .................................................................................
... . ---...--------------------.............. 102
102
4.3.10 Behavioral results ......................................................................................------.....................................
103
4.4 Discussion ....................................................................................--------------........................
.....
4.4.1 Significance of the frontal eye field .............................................................................
4.4.2 Technical and scientific significance ......................................................................................
Appendixes ..................................................................-----..-...
96
103
104
.................-....................... 108
Appendix 1: Shutter characterization plots.................................................................................-108
Appendix 2: Supplemental micro-stimulation and neuronal recording data. ................... 110
Appendix 3: Viral vector selection..........................................................................
............ 110
Appendix 4: Behavioral data analysis..................................................................................
Appendix 5: List of injected viral vectors. ............
........
.............
.....................-----.-- 113
Appendix 6: Fluorescence probe leads to isometric response ......................................................
7
112
115
Appendix 6.1: Geometric estimations of isometric probe shadowing .............................................................
116
Appendix 6.2: Supplemental methods for ruby sphere calibration .................................................................
118
Appendix 6.3: Internal geometric model..........................................................................................................123
Appendix 7: Heating measurements in tissue ..............................................................................
125
Appendix 7.1: Possible negative effects of heating in the brain ......................................................................
125
Appendix 7.2: Neuronal firing rate increases with increased brain temperature ............................................
125
Appendix 7.3: Heating may play a role in driving behavior in primate visual cortex .......................................
125
Appendix 7.4: BOLD fMRI activation results from heating alone .....................................................................
126
Appendix 7.5: Results of temperature measurements in vivo .........................................................................
127
Works Cited ............................................
......................................................................
8
130
List of Figures
Figure 1: The Frontal Eye Field in Humans and Macaque Monkeys............................................12
14
Figure 2: Response profiles of FEF neurons. ...............................................................................
Figure 3: Light-sensitive proteins, opsins, can modulate neuronal firing...................................20
Figure 4: Absorption coefficients of hemoglobin as a function of wavelength..........................30
Figure 5: Isometric light probe .............................................................................
33
.
--.. . . . . . . . 35
Figure 6: Custom aligner ..........................................................................
------------
...
36
Figure 7: Experimental setup for in vivo light measurements......................................................
Figure 8: Average light decrease with distance from illuminator.....................................................41
42
Figure 9: Light fall off for all anim als ......................................................................................
Figure 10: Blue light propagation.......................................................................................43
Figure 11: Green light propagation..............................................................................................44
45
Figure 12: Red light propagation ........................................................................
Figure 13: Comparison of large-volume illuminator and standard optical fiber.......................53
56
Figure 14: Etched tip of the large-volume illuminator..................................................................
Figure 15: Optical setup interfacing the large-volume illuminator with a laser........................58
Figure 16: Setup of integrating sphere and illuminator that allow for measurements of the light
emission profile along the length of the illuminator.......................................................................60
Figure 17: Locations of the illuminator and recording sites in cortex. ........................................
62
Figure 18: Local field potential recordings of representative neurons.........................................64
Figure 19: LFP and instantaneous spike values for Monkey L measured using a multi-contact u.
probe...............................................................................
-- ..... -
....------------------............................ 66
75
Figure 20: Circuit used for electrical micro-stimulation ................................................................
Figure 21: Task used during testing......................................................................................
.......... 78
Figure 22: Injection fields and target presentation locations for both monkeys.........................79
Figure 23: AAV8-Jaws-GFP injection of 1 uL / site at two sites in a histology primate. ............ 89
Figure 24: The end points of saccades .............................................................................................
Figure 25: Firing profiles ...............................................................................................................
9
90
. --91
Figure 26: A neuron with only a visual response ...........................................................................
92
Figure 27: Two neurons with visual, delay, and motor-related activity. .....................................
93
Figure 28: Distribution of response types among task-responsive FEF neurons.......................94
Figure 29: Decrease in firing rate with illumination for all FEF neurons in both monkeys. ......... 95
Figure 30: Raster plot of optogenetically-inactivated FEF neuron..............................................
96
Figure 31: Firing rate profile of visually-responsive neurons during target presentation. ......
97
Figure 32: Firing rate decrease in visually responsive neurons in both monkeys......................98
Figure 33: Firing rate profile of delay - responsive neurons during the delay period ..............
99
Figure 34: Firing rate decrease in delay-responsive neurons in both monkeys............................100
Figure 35: Firing rate profile of motor-related neurons during the motor period........................101
Figure 36: Firing rate decrease in motor-related neurons in both monkeys..................................101
Figure 37: Error rates for both monkeys at different silencing times..............................................102
Figure 38: Characterization of shutter opening .................................................................................
108
Figure 39: Characterization of shutter closure...................................................................................109
Figure 40: Examples of micro-stimulation evoked saccades............................................................110
Figure 41: Ruby absorption coefficients for red and green light.....................................................115
Figure 42: Demonstration of angle independence of isometric light probe...................................116
Figure 43: Beam profiler used for absolute light power measurements.................118
Figure 44: Water bath for isometric probe calibration. .....................................................................
119
Figure 45: Beam profile of planar illuminator....................................................................................122
Figure 46: Internal geom etric..............
.
--.......................................
..............................................
123
Figure 47: Illustration of how rays can internally reflect and yield a uniform beam in the center.
................................................................
..-...... . . . .................................................................................
124
Figure 48: In vivo temperature measurements with the primate specific illuminator for a single
laser p u lse ................................................................................................................................................
128
Figure 49: Additive heating with frequently spaced laser pulses ...................................................
129
10
Chapter 1: Introduction
Abstract
This thesis uses a novel neuro-modulation technique, optogenetics, to reversibly inactivate the
primate frontal eye field (FEF) at different times during the memory-guided saccade task. The
scientific contribution of this thesis is a better understanding of the time-dependent contributions
of FEF to oculomotor behavior and to short-term memory. The technical contributions include
adapting optogenetics to a scientific study in non-human primates and providing methods that
allow other groups working with other species (e.g., squirrel monkeys, ferrets, cats, etc.) to adapt
optogenetics to their specific needs. The following chapter has two parts, consistent with the dual
scientific-technological character of this thesis. The first section describes the frontal eye field and
significant questions about this area that can be answered optogenetically. The second section
reviews optogenetics and the challenges with its use in non-human primate models.
1.1 Frontal Eye Field
By definition, the FEF is the part of the primate prefrontal cortex in which low-current electrical
stimulation (as low as 10 tA, <50 paA threshold) reliably evokes saccadic eye movements (Bruce,
Goldberg et al. 1985, Murphey and Maunsell 2008). FEF plays a major role in planning and
executing saccades and in controlling visual attention. The anatomy of FEF and the firing
responses of its neurons indicate that FEF may be critical for sensory, motor, and working
memory functions, among others; however, it can be difficult to determine how brain regions like
FEF, contribute to larger neuronal circuits. For example, we still cannot pinpoint why inactivating
FEF prevents subjects from making memory-guided saccades, a simple task that is extensively
used in both primate cognitive neuroscience and clinical neurology.
1.1.1 Anatomy
FEF, which includes Brodmann's areas 8Ac and 45 and is located on the rostral bank of the arcuate
sulcus near dorsal prefrontal and ventral premotor cortex. Figure 1 shows the locations of FEF in
the human and macaque monkey brains. The anatomical relationship of FEF and the other eye
11
fields of the primate frontal cortex are preserved between macaques and humans (Amiez and
Petrides 2009), and FEF generally is believed to be present only in higher species although this is
debated (Bahring, Meier et al. 1994).
Spa
6
-
Ew ?
Human
Macaque
Monkey
-f-OV
Figure 1: The Frontal Eye Field in Humans and Macaque Monkeys
Human (left) and macaque frontal eye field (PEP) is highlighted above. Notice the similarity in brain
structure between the two species. FEF includes areas 8Ac and area 45 on the rostral bank of the arcuate
sulcus (AS) in macaque. FEF is not present in the mouse brain (not shown). Adapted from (Wise,
Boussaoud et al. 1997).
1.1.1.1 Cytoarchitecture
At the cellular level, FEF has characteristics of both motor and prefrontal cortex. For example,
FEF contains many more large Layer V pyramidal neurons, which are typically seen in primary
motor areas, than other nearby cortical areas contain (Stanton, Deng et al. 1989). Conversely, even
though primary motor cortex lacks granular Layer IV neurons, frontal cortical regions and FEF
both contain these granular Layer IV neurons (Stanton, Deng et al. 1989). The cytoarchitecture of
FEF hints at its functional responses.
12
1.1.1.2 Receptive fields
FEF neurons exhibit spatially selective visual responses, or receptive fields, similar to neurons in
primary visual cortex. However, FEF receptive fields are larger and less defined (Mohler,
Goldberg et al. 1973) than those in primary visual cortex. Populations of FEF neurons also exhibit
so-called motor fields -spatially tuning of firing rate in preparation for a saccade to a particular
part of the visual field (Bizzi 1968). Visual receptive fields and motor fields generally overlap
within FEF, but they are not necessarily a perfect match, particularly at the level of single neurons
(Bruce and Goldberg 1985, Bruce, Goldberg et al. 1985). This thesis studies the visual receptive
and motor fields in FEF, which may help us to better understand the interplay between FEF's
sensory and motor functions.
1.1.1.3 Firing rate characteristics
When Bruce and Goldberg (1985) recorded from single units in macaque FEF during visuallyand memory- guided saccades, they identified three kinds of pre-saccadic activity (visual, motor,
and delay-period / anticipatory activity). Forty percent of neurons had visual activity but no
motor activity. Twenty percent had little or no visual response, but a discharge prior to
movement, and the remaining 40% of neurons had both visual- and motor-related activity. One
in five visuomotor and motor neurons showed anticipatory motor activity, but only 2% of visual
neurons showed anticipatory activity (Bruce and Goldberg 1985).
Hanes et al. (1998) also
observed these three classes of neurons in FEF (Hanes, Patterson et al. 1998). The discharge rates
during a memory-guided saccade task are shown in Figure 2. A detailed description of the
memory-guided saccade task can be found in Chapter 4, Section 4.2.5.
13
A ATF
100
Trigger
B
100.
C
100.
200 400 -4W -200
0
0
Time from
Time from
target (mu) saccade (ms)
D
FT
u
100
C#
~
1
0 200
-200 0
Time from
Time from
fixation (mg) maccado (mS)
Figure 2: Response profiles of FEF neurons.
A) a neuron with a primarilyvisual response;B) a neuron with a visual response and delay period activity;
C) a neuron with a slight visual response and delay period activity, but primarily a motor response; d) a
very rare motor suppression cell. From Hanes et al., 1998 (Hanes, Patterson et al. 1998). The F bar
corresponds to "fixation." The T bar corresponds to "target" presentation, and the word "trigger"
corresponds to the end of "fixation,"which cued the monkey to preparea saccade.
14
1.1.2 Frontal Eye Field Inactivation
Inactivation demonstrates FEF's necessary role in mediating behavior. Conventional techniques
for inactivation (i.e., ablation, pharmacological inactivation, and cooling) are either permanent or
act over timescales much longer than the actions of the neural circuit (e.g., hours v. milliseconds).
Optogenetic inactivation occurs at the millisecond timescale, which is appropriate for dissecting
neural circuit function. Earlier studies using cruder inactivation methods, however, have
informed the pioneering experiments with optogenetic inactivation in primate FEF as described
in Chapter 4.
1.1.2.1 Ablation
FEF ablation has modest, and generally temporary, effects on saccade behavior; however, saccade
defects become more profound as larger volumes of FEF are affected (Schiller and Tehovnik
2003). For example, three weeks after lesioning the left FEF in a rhesus macaque, Schiller and
Chou (1998) demonstrated that the latency of visually-guided saccades to right-sided targets
increased by 45 ms, the latency of saccades to left-sided targets decreased by 13 ms, and saccadic
peak velocities decreased for right-sided saccades from 534*/s to 459*/s (Schiller and Chou 1998).
After four months, though, both the latency and differences decreased, and after a year, no deficit
remained (Schiller and Chou 1998).
In contrast to the post-ablation compensation seen in saccade tasks, Schiller and Chou (2000)
found that an FEF lesion caused a permanent bias in a two-target, temporal asynchrony task.
Prior to ablation, the macaque would make a saccade to whichever of the two targets appeared
first; however, after unilateral ablation of left FEF, the macaque showed a lasting bias for leftsided targets, even if the right target appeared more than 80 ms before the left target did (Schiller
and Chou 2000).
15
1.1.2.2 Pharmacological inactivation
Because the brain can compensate for FEF ablation with time, reversible pharmacological
inactivators, such as Lidocaine or Muscimol produce more dramatic behavioral defects than
ablation does.
Dias and Segraves (1996) hypothesize, "Acute inactivation of the FEF of
monkeys.. .produced much more severe oculomotor impairment [than ablation did]. This
difference is probably due to the acute nature of the Muscimol effect, which does not allow time
for reorganization of the control of eye movements before testing begins." One would expect even
less reorganization with optogenetics than with pharmacological inactivation because the
duration of silencing is even shorter with optogenetics.
1.1.2.2.1 Muscimol inactivation
Dias et al., (1995) found that injections of Muscimol in FEF prevented monkeys from making
either visually or memory-guided saccades to targets represented by neurons at the center of the
injection site. They also reported longer saccade latencies to targets adjacent to the area
represented by the injection site. For memory-guided saccades, Dias et al., (1995) reported that
saccade latencies to targets in the opposite direction of inactivated receptive field decreased and
that the saccades often were initiated prior to the "go-cue" -something
we have termed a
"premature saccade." Later Dias and Segraves (1999) reported that monkeys initiated fewer
saccades to the retinotopic representation of the inactivated site. Further memory-guided
saccades to targets in the inactivated site become even less accurate and had decreased peak
velocities with longer delay periods (Dias and Segraves 1999). The primates also were less
accurate in their fixation to the central cue (Dias, Kiesau et al. 1995, Dias and Segraves 1996, Dias
and Segraves 1999).
1.1.2.2.2 Lidocaine inactivation
Sommer and Tehovnik (1997) reported similar deficits in memory-guided saccades when they
pharmacologically inactivated part of FEF with Lidocaine (Sommer and Tehovnik 1997).
Specifically, they found that saccade latency, error rate, and "no saccade" frequency increased for
16
targets represented by the inactivated receptive field (Sommer and Tehovnik 1997). Further,
Sommer and Tehovnik (1997) reported an increased frequency of premature saccades to target
locations opposite of the inactivated receptive field. Unlike Dias et al., (1995), Sommer and
Tehovnik (1997) did not find a consistent effect on saccade peak velocity. In order to estimate the
minimum volume of tissue that must be silenced optogenetically to disrupt memory-guided
saccades, we consider the Tehovnik and Sommer (1997) studies of Lidocaine in macaque FEF,
which are reviewed in more detail in Chapter 2.
1.1.2.3 Significance of pharmacological inactivation
Overall, the Lidocaine and Muscimol results could point to deficits in motor activity, perception,
working memory, impulse control, saccade initiation, and/or fixation control. The firing activity
of FEF neurons implies that the area responds differently during different times in the task, but
since pharmacological inactivation encompasses the entire task, we cannot separate different
aspects of FEF inactivation from the observed behavior deficits. However, with optogenetic
silencing during specific epochs of the memory-guided saccade task, we were better able to tease
apart the role of FEF in generating memory-guided saccades, as described in Chapter 4.
1.1.3 Significance of studying the frontal eye field
Even, in a straightforward, classic, extensively-used task like the memory-guided saccade,
conventional inactivation techniques do not allow us to pinpoint why a particular brain region,
(i.e., FEF) is critical for behavior. Therefore, this thesis uses optogenetics to answer the following
questions about the role of FEF in memory-guided saccades.
Does FEF allow the subject:
* To "see" the target through its visual response?
STo recall the target through the delay period activity?
To execute the saccade?
*All of the above?
17
In addition to the basic questions on which this thesis focuses, the results presented in Chapter 4
provide a better understanding of FEF, which is critical to the field of cognitive neuroscience
because FEF lies in the center of many open questions, some of which are described below.
1.1.4 The pre-motor theory of attention
The pre-motor theory of attention suggests that latent saccade commands give rise to covert
spatial attention (i.e., attention to a part of the visual field outside of direct gaze) even if the eye
movement is never executed (Thompson, Biscoe et al. 2005). While the pre-motor theory remains
controversial, most agree that a better understanding of FEF, which plays a role in generating eye
movements (Bruce and Goldberg 1985, Bruce, Goldberg et al. 1985, Hanes, Patterson et al. 1998,
Tehovnik, Sommer et al. 2000) and which is involved in spatial attention allocation (Moore and
Armstrong 2003, Moore and Fallah 2004), could help improve understanding of the extent to
which saccadic control and attentional control share common neuronal mechanisms.
1.1.5 Local FEF circuitry
Moreover, the local FEF circuitry, with its diverse neuronal subtypes, is poorly understood. We
know that, in addition to the visual-, delay-, and motor-related firing rate changes described
above, FEF neurons can be selective for everything from visual features (Zhou and Desimone
2011) to decision signals (Ding and Gold 2012) to auditory responses (Schall, Morel et al. 1995,
Kirchner, Barbeau et al. 2009). However, we do not know how or if FEF neurons modulate one
another's responses. Understanding FEF better and developing the optogenetics tools to probe
functionally heterogeneous brain regions, such as FEF, provide new tools for understanding local
microcircuits.
1.1.6 FEF in the context of larger oculomotor and attentional circuits
FEF is part of the larger oculomotor and visual attention circuits, and it shares functional
similarities with connected brain regions, such as the superior colliculus (SC) and the lateral
interparietal area (LIP). The interplay among these areas is an area of active research. McPeek et
18
al. (2003) showed that sub-threshold electrical micro-stimulation of the superior colliculus
(McPeek, Han et al. 2003) and FEF (McPeek and Takahashi 2006) resulted in eye movements
curved toward the area of the visual field corresponding to the location of micro-stimulation,
implying shared control of motor planning. Schafer and Moore (2011) suggested that FEF plays a
stronger role in top-down attention while Buschman and Miller (2007) pointed to LIP playing a
stronger role in bottom-up attention (Buschman and Miller 2007, Schafer and Moore 2011). In
Chapter 4, 1 transiently remove FEF from the oculomotor circuit during motor planning or during
the delay period of a memory-guided saccade, which provides some clues about how much
shared functions, such as motor preparation, depend on FEF.
1.2 Optogenetics
In addition to its scientific contribution, this work makes two key technological contributions.
First, this thesis adapts optogenetics, which modulates neuronal firing on the time-scale of the
brain, to a scientific study in a primate. Second, it details tools and methods so that researchers
who work with other species can tailor optogenetic parameters for their specific experiments.
Optogenetics genetically incorporates light sensitive proteins, called opsins, into neurons and
subsequently modulates neuronal firing rate with light. Neuronal firing can be driven with
excitatory opsins (such as Channel-rhodopsin 2, ChR2) or suppressed by inhibitory opsins (such
as ArchT or eNpHR), by illuminating with a particular color of light (Figure 3). In addition, it is
possible to target opsins to genetically-defined cell populations (Boyden, Zhang et al. 2005,
Aravanis, Wang et al. 2007, Gradinaru, Thompson et al. 2007, Han and Boyden 2007, Gradinaru,
Thompson et al. 2008, Zhang, Prigge et al. 2008, Zhao, Cunha et al. 2008, Chow, Han et al. 2010,
Gradinaru, Zhang et al. 2010, Gunaydin, Yizhar et al. 2010, Han, Chow et al. 2011, Yizhar, Fenno
et al. 2011, Chow, Han et al. 2012).
19
B
A
archaerhodopsins and
bacterorhodopsins
(e.g., Arch, Mac, BR)
C
halorhodopsins
(e.g., Halo/NpHR)
channelirhodopsins
(e.g., ChR2)
H+, Na+, K+, Ca 2
Ci1i
H.
D
E
100% neural sen
in cortlcal neurons of awake mio
mediated by Arch
y
-A
I
-4
-
I
F
Halo expreselng neuron
in vitro, quieted by yellow light
1
5
1
5
10
15
20
15
W
two different ChR2-expressing
neurons in vitro, responding to
the samne train of blue lit pulse
LLUlUI
I
1~ml
W
10
50nVI1
a
iIJmiL1UJaOD M
Is
Figure 3: Light-sensitive proteins, opsins. can modulate neuronal firing
A) Archaerhodopsin, Bacteriorhodopsins and B) Halorhodopsin are membrane proteins that can be
incorporatedinto neurons to silencefiring, D) and E). C) Channelrhodopsinsare also membrane proteins,
but illuminating these proteins can drive neuronalfiring, as shown in F). Adapted from (Boyden 2011).
Optogenetic manipulations of rodent brain circuits have provided insight into functional
physiology (Adamantidis, Zhang et al. 2007, Zhang, Holbro et al. 2008, Airan, Thompson et al.
2009, Cardin, Carlen et al. 2009, Claridge-Chang, Roorda et al. 2009, Sohal, Zhang et al. 2009, Tsai,
Zhang et al. 2009, Cardin, Carlen et al. 2010, Carter, Yizhar et al. 2010, Ciocchi, Herry et al. 2010,
Haubensak, Kunwar et al. 2010, Johansen, Hamanaka et al. 2010, Lee, Durand et al. 2010,
Llewellyn, Thompson et al. 2010, Witten, Lin et al. 2010, Carter and de Lecea 2011) and disease
states (Bi, Cui et al. 2006, Alilain, Li et al. 2008, Claridge-Chang, Roorda et al. 2009, Gradinaru,
Mogri et al. 2009, Tonnesen, Sorensen et al. 2009, Zhang, Ivanova et al. 2009, Covington, Lobo et
al. 2010, Kravitz, Freeze et al. 2010, Tomita, Sugano et al. 2010, Witten, Lin et al. 2010); however,
the optogenetics have not yet been applied to scientific questions in primates. Because the
macaque brain is 200x larger than the mouse brain and contains almost 100x as many neurons
20
(Herculano-Houzel 2009), two key challenges for primate optogenetics are 1) expressing the opsin
in a large enough fraction of neurons over a large enough volume of tissue to impact behavior
and 2) illuminating a large enough cortical volume to reliably change behavior with minimal light
delivery.
1.2.1 Expressing the opsin in a large enough fraction of neurons
While transgenic mice expressing opsins are the model of choice for rodent optogenetics studies,
transgenic, opsin-expressing animals do not exist in another other species (except rat) (Tomita,
Sugano et al. 2009, Tomita, Sugano et al. 2010). Transgenic macaque models are rare, (Chan 2009,
Chan and Yang 2009, Niu, Yu et al. 2010), and none of those reported to date would be viable for
primate optogenetics. Further, in utero electroporation (Gradinaru, Thompson et al. 2007, Huber,
Petreanu et al. 2008) can introduce opsins into embryos, but this is not practical for primate
research given the long gestation period (164 days v. 19 days in mice), low number of offspring
per pregnancy (1 v. 7.4 / litter for C57 black mice (Little and Pearsons 1940)), and lengthy time to
full maturity (-8 years for male rhesus macaques (Dixson and Nevison 1997) v. 6-8 weeks for
mice). While there are three standard ways of incorporating opsins in whole animalstransgenics, in utero electroporation, and viral vectors-only viral vectors are currently practical
and feasible in adult macaques.
1.2.2 Opsin expression in rhesus macaques
Opsins have been introduced into the primate brain using lentivirus (Han, Qian et al. 2009,
Diester, Kaufman et al. 2011, Han, Chow et al. 2011), AAV1 (Jazayeri, Lindbloom-Brown et al.
2012), AAV5 (Diester, Kaufman et al. 2011, Gerits, Farivar et al. 2012) and AAV8 (Cavanaugh,
Monosov et al. 2012) with both ubiquitous promoters (Han, Qian et al. 2009, Han, Chow et al.
2011, Cavanaugh, Monosov et al. 2012, Gerits, Farivar et al. 2012) and with neuron-specific
promoters (Diester, Kaufman et al. 2011, Jazayeri, Lindbloom-Brown et al. 2012). Yet, a particular
virus and promoter combination often does not yield consistent neuronal expression levels across
species, brain regions, or research groups. For example, a marmoset study of AAV8 found
expression of a reporter gene in up to 91% of neurons (Masamizu, Okada et al. 2010), while
21
Cavanaugh et al. (2012) reported transfection of <40% of cells with AAV8 using a ubiquitous CAG
promoter and a GFP-tagged inhibitory opsin (ArchT) in the macaque superior colliculus
(Cavanaugh, Joiner et al. 2012). Even within the same species, Han et al. (2009) reported that more
than half of neurons in macaque cortex expressed ChR2 with a lentiviral construct, but when
Gerits et al. (2012) injected the same virus in the same brain area, they found the virus to be
ineffective (Gerits, Farivar et al. 2012).
With so few reports of opsin expression in primates and disagreement among those reports, there
was no obvious choice of virus. Therefore, I systematically screened several viral vectors in the
rhesus monkey brain in order to determine the best vector for this particular project and for the
general benefit of the field of primate optogenetics. Specifically, I evaluated whether lentivirus or
AAV gave better expression, which serotype of AAV gave the best expression, and whether any
of the viruses appeared to damage neurons over time. These results are summarized in Chapter
4 and detailed reports are available upon request.
1.3 Illumination and challenges with current light propagation models
In addition to better characterization of viral vectors, the optogenetics community lacks (yet
desperately needs) systematic measurements of light propagation in the living brain. Without
these measurements, optogenetics researchers cannot determine the light power needed to
illuminate a given tissue volume. The solution has been to err on the side of over-illumination
but increasing the input light power is more likely to result in heating, which 1) could damage
tissue and 2) may confound results by increasing neuronal firing rates or driving increases in the
fMRI BOLD signal. The potential for tissue heating is described in detail in Appendix 7.
Measurements from ex vivo tissue specimens (Aravanis, Wang et al. 2007, Huber, Petreanu et al.
2008) inform popular irradiance predictors used within the optogenetics community (e.g.,
(Yizhar, Fenno et al. 2011) and optogenetics.org) while scattering and absorption coefficients
taken from ex vivo tissue samples (Yaroslavsky, Schulze et al. 2002) underlie the Monte Carlo
models of light propagation published in papers introducing new opsins (Chow, Han et al. 2010)
or new illuminators (Bernstein, Han et al. 2008). Unfortunately, models based on in vitro or ex vivo
22
measurements do not reflect in vivo light propagation because in vitro and ex vivo tissue samples
differ geometrically from living tissue and contain less blood than living tissue does. Current light
propagation models do not show significant wavelength-dependent light propagation
differences, but one would expect substantial wavelength-dependent light propagation
differences in vivo since all wavelengths of visible light expect for red are absorbed much more
readily by the hemoglobin than red light is (Eggert and Blazek 1987, Robles, Chowdhury et al.
2010).
Chapter 2 of this thesis describes techniques for in vivo light propagation measurements and
presents the results of these measurements in the living brain. These measurements allowed us
to design a primate-specific illuminator to deliver light to the desired volume of FEF, which is
described in Chapter 4.
1.4 "Backward" effects: activation of neurons with inactivating opsins
All of the previous studies that used inhibitory opsins in primate cortex reported a subpopulation
of neurons that were activated by the illumination rather than inhibited. Han et al. (2011)
expressed an inhibitory opsin (ArchT) in the parietal cortex (area 7a) of one macaque and the
visual cortex (area V1) of another macaque (Han, Chow et al. 2011). The primates were awake
and freely viewing during the recordings, in which Han et al. (2011) reported a significantly
reduced firing rate in 60% of neurons (45 / 74) relative to baseline and a significantly increased
firing rate in 10% (7 / 74) of neurons during illumination (Han, Chow et al. 2011). Diester et al.
(2012) found no change in movements elicited by electrical stimulation with simultaneous
illumination in an eNpHR2.0-injected area of motor cortex despite eNpHR2.0 expression in 70%
of neurons, based on histology. In contrast to the histology, Diester et al. (2012) reported that only
38% of neurons (55/144) were silenced in this area of the cortex and -10% of neurons (15/144)
increased their firing rate in response to illumination (Diester, Kaufman et al. 2011).
Han et al. (2011) and Diester et al. (2012) both hypothesized that these heterogeneous neuronal
responses resulted from inactivating inhibitory neurons. Results from Ohayon et al., (2013) seem
to be consistent with this hypothesis. Ohayon et al., (2013) used a lower light power density (190
23
mW/mm2 ) than Diester et al, (2012) or Han et al., (2011), yet Ohayon et al., (2013) still reported
increased firing with inhibitory opsins (ArchT and eNpHR2.0). It is possible, though, that the
"backward" effect in Ohayon et al. (2013) may have resulted from heating (even at the lower light
power), from an interaction between light delivered to these sites and adjacent locations that were
injected with ChR2, or possibly from the network effects and inactivation of interneurons that
Han et al., (2011) and Diester et al, (2012) proposed.
Regardless of the source of these heterogeneous effects, activating some cells while
simultaneously inhibiting others could negate the overall effects of modulation, particularly if the
goal is to modulate behavior. Thus, this work tackles this potential problem both by achieving
high levels of neuronal expression across all layers of cortex (limiting potential network effects)
and by carefully controlling for possible effects of heating.
1.5 Other challenges with primate optogenetics
The challenges discovered in previous primate optogenetics studies (e.g., varying virus
expression levels, inverse modulation, and little behavioral modulation) raised questions about
whether optogenetics could be meaningfully applied to a scientific study in non-human primates.
Chapter 4 of this thesis presents results that affirm the viability of optogenetics in scientific
studies using non-human primates. Specifically, Chapter 4 shows consistent, optogeneticallydriven behavioral changes in two monkeys. Below, there is a brief review the groundwork that
others have laid for primate optogenetics and the challenges that they have identified.
1.5.1 Absence of behavioral change
To date, there have been nine published reports of optogenetics in rhesus macaques (Han, Qian
et al. 2009, Diester, Kaufman et al. 2011, Han, Chow et al. 2011, Cavanaugh, Monosov et al. 2012,
Gerits, Farivar et al. 2012, Jazayeri, Lindbloom-Brown et al. 2012, Ohayon, Grimaldi et al. 2013,
Ruiz, Lustig et al. 2013, Dai, Brooks et al. 2014). Behavioral modulation or subthreshold "optical
micro-stimulation" has been reported in primates, but only with excitatory opsins (Gerits, Farivar
24
et al. 2012, Jazayeri, Lindbloom-Brown et al. 2012, Ohayon, Grimaldi et al. 2013, Dai, Brooks et al.
2014) or by targeting structures in the brainstem (Cavanaugh, Monosov et al. 2012).
Three studies reported optogenetic inactivation in primate cortex, but reported behavioral
changes. Diester et al. (2012) injected viruses containing ChR2 in somatosensory cortex in two
macaques and in motor and pre-motor cortex of a single macaque. Despite expression in >65%
of neurons in motor cortex, no behavior changes were evoked (Diester, Kaufman et al. 2011). Han
et al. (2011) reported inactivation but no behavioral results. Ohayon et al, (2013) did not report
any behavioral changes with ArchT or eNphR2.0 (inhibitory opsins) in FEF.
1.5.2 Inconsistent behavioral changes
Cavanaugh et al. (2012) expressed ArchT in the superior colliculus of three rhesus macaques and
studied the effect of inactivation during visually-guided saccades in two of the primates and
during free viewing in the third. Despite large light power densities (650 - 1600 mW/mm 2 ), there
were only modest effects. Visually-guided saccades showed a shift in saccade end points (7.3%
and 4.6% of saccade magnitude in the two primates, respectively), a reduced peak velocity
(decreases of 79*/s and 9.6*/s, in the two monkeys, respectively), and increased latencies (7 ms
and 7.9 ms). Cavanaugh et al. (2012) reported that neuronal response decreased 68.2% on average
in one macaque and 27.4% in the second, but it is not clear whether there were inverse effects (i.e.,
excitation with illumination) included in this average. Regardless, Cavanaugh et al. (2012)
suggest that effects were much stronger in one primate than the other due differences in the extent
of neuronal silencing, but they do not provide insight into what methodological improvements
could be made to ensure consistent silencing across primates (Cavanaugh, Monosov et al. 2012).
In the Gertis et al. (2012) study, rhesus macaques injected with ChR2 in FEF performed a visuallyguided saccade task during an fMRI. The group reported that saccade latencies were shorter to
ipsilateral targets in cases of optical stimulation than in non-stimulation trials, but in one rhesus
macaque (M2), the difference between stimulation and non-stimulation trials was present for both
ipsilateral and contralateral targets, implying an attentional increase in subject M2 coincident
with the illumination alone. Further, the authors suggested that the other monkey (Ml) may have
25
sustained unilateral FEF damage, which alone may explain an ipsilateral, but not contralateral,
effect. Better methodological controls (e.g., confirming whether the primate sustained a lesion
prior to testing, controlling for light leakage during illumination, running sham illumination in
non-injected areas, etc.) may have been sufficient to explain inconsistent behavioral results.
Ohayon et al., (2013) reported a few evoked saccades with optical stimulation using ChR2 in FEF,
but the effect was not replicable across sessions, monkeys, or even nearby locations within FEF.
1.6 Motivation for the present study
FEF plays a critical role in executing memory-guided saccades. Using optogenetics, we can
determine which of these firing rate increases-visual, delay, or motor-are behavior-critical by
inactivating FEF on the millisecond time scale of the neural circuit.
In order to apply optogenetics to a scientific study in the primate, I had to overcome specific
technical challenges. Chapter 2 presents measurements of light propagation and heating in the
living brain. Chapter 3 describes a specialized illuminator that delivers light over a large volume
of the frontal eye field without overheating the cortical tissue. Chapter 4 describes the
identification of viral vectors sufficient for primate physiology and details techniques for
monitoring viral vector expression in real-time in the living brain.
Finally, Chapter 4 also details the successful manipulation of memory-guided saccades in the
rhesus macaque through optogenetic FEF inactivation. From these results, I determine that FEF
critically contributes to saccades through its visual activity, its delay-period activity, and its motor
activity because silencing FEF at any of these times during the task increases the saccade error
rate.
26
Chapter 2: Visible light propagation in the living brain: implications for optogenetics
Abstract
The distribution of red, green, and blue light was measured in the anesthetized rodent brain using
a novel approach designed for in vivo, visible light measurements. 12 A 1.5 mm-diameter
illuminator was placed on the surface of the cortex. Fluence rates were measured at depths of 0.5
to 2.5 mm using an isometric light probe. Mathematical models based on diffusion theory were
applied to these measurements, and the absorption and effective coefficients in the living brain
were derived for red, green, and blue light. The in vivo absorption coefficients for green and blue
light are substantially higher than the in vitro and ex vivo coefficients typically used in the
literature to estimate light propagation in optogenetics studies; however, the values presented
here closely match the absorption coefficients that one would expect based on the amount of
blood in the living brain and the absorption coefficient of oxygenated blood. These results imply
a substantial advantage for red-light sensitive opsins in terms of lower light power requirements,
decreased risk of heating, and larger potential illumination volumes.
Keywords:
optogenetics, tissue optical properties, absorption, scattering, brain optical properties, isometric,
Glossary:
KM-Kubelka-Munk, a model of light loss as a function of distance.
NA - numerical aperture
1Yellow light was not tested because a suitable laser at the light powers needed for these experiments and
the heating experiments (>100 mW total light power after coupling) was not commercially available.
Further, at the time of this study, evidence to suggest that neuronal silencing with ArchT (green-light
responsive) was substantially better than eNpHR (yellow-light responsive) had already been published,
e.g., Han et al., 2011 and Chow et al., 2010. Thus, the true comparison was whether to use green-light
sensitive ArchT or a new red-light sensitive halorhodopsin, Jaws.
2 Blue light was tested, even though no blue-light sensitive opsin was considered for the primate study,
because ChR2 (a blue light sensitive excitatory opsin) is used more than any other opsin. Thus, information
about blue light is generally valuable to the field.
27
2.1 Introduction
While optogenetics is widely used in vivo, current models of visible light propagation use tissue
properties derived from in vitro or ex vivo specimens (Aravanis, Wang et al. 2007, Huber, Petreanu
et al. 2008, Yizhar, Fenno et al. 2011). These measurements inform popular irradiance predictors
used within the optogenetics community (e.g., the "Stanford" predictor (Yizhar, Fenno et al. 2011)
and optogenetics.org), which estimate light propagation based exclusively on scattering and
geometry, ignoring absorption. Even when groups have attempted to include absorption
coefficients via Monte Carlo models, e.g., (Bernstein, Han et al. 2008, Chow, Han et al. 2010), they
had resort to ex vivo-derived coefficients (Yaroslavsky, Schulze et al. 2002) because in vivo
absorption and scattering coefficients for non-red, visible light have not been determined.
In vivo preparations contain flowing, oxygenated blood unlike either ex vivo or in vitro
preparations. The hemoglobin in blood may prove critical to understanding the optical properties
of the brain because hemoglobin attenuates visible light more than other tissue components.
Further, hemoglobin does not just scatter light, it absorbs light in a wavelength-dependent
manner. One would expect red light to propagate much further in vivo than any other color of
visible light because, as Figure 4 shows, all wavelengths of visible light expect for red are
absorbed much more readily by the hemoglobin than red light is (Eggert and Blazek 1987, Robles,
Chowdhury et al. 2010).
It is likely more important to account for light absorption than any other optical property of tissue.
Unlike scattering and geometric factors, which spread light over larger volumes, absorption
converts light energy to heat. Temperature impacts neuronal firing and excessive illumination
can damage tissue, as is reviewed in Appendix 7.
In vivo light propagation measurements are feasible, as the photodynamic therapy community
has demonstrated, but these measurements are only practical with current techniques for longer
wavelengths that propagate far through tissue (such as those in the deep red and infra-red range).
The parts of diffusion theory used to estimate absorption and scattering in these experiments
require the light source to approximate an isotropic point source. This assumption can only be
28
made when the light measurement probe is far from the light source. These distant measurements
cannot be made with visible light of the wavelengths typically used in optogenetics because that
light does not propagate that far in living, blood-perfused tissue.
Here, techniques for in vivo light measurements used in the photodynamic therapy community
are adapted to visible wavelengths in a novel experimental preparation. Effective and absorption
coefficients were determined in the living mouse brain for red (635 nm), green (532 rum) and blue
(473 nm) light. These values can be used directly in models of light propagation (e.g., Monte Carlo
models) or heating (via the bioheat equation).
Recently, the optogenetics community has expanded to include many different species of animal
models (Gradinaru, Zhang et al. 2010, Gerits, Farivar et al. 2012, Roberts, Gobes et al. 2012),
wavelengths of light (Zhang, Prigge et al. 2008, Chow, Han et al. 2010, Gunaydin, Yizhar et al.
2010), illumination sources (Zorzos, Boyden et al. 2010, Anikeeva, Andalman et al. 2012, Stirman,
Crane et al. 2012), and tissue targets-both in the brain and in other parts of the body (Arrenberg,
Stainier et al. 2010, Cheng, Zhang et al. 2012). The techniques presented here are flexible enough
to be adapted to any of these recent developments and to continue to grow with the field.
Experimenters can use these techniques to directly measure light propagation in their specific
experimental preparations, reducing the risk of heating, over-illumination, and tissue damage
regardless of the wavelength, illuminator or animal model.
29
400
HbO
2
Hb deoxy
350
300
_U
250
U
1200
8
1
.a
4 15
IVV
I
-
50
'
45 0
0~'
500
------------
"
"
-
0
550
600
Wavelength (nm)
650
700
Figure 4: Absorption coefficients of hemoglobin as a function of wavelength
Based on data provided courtesy of Scott Prahl, Oregon Medical Laser Center
2.2 Materials and Methods
2.2.1 Animal Care
Sixteen male C57BL/6J mice (Taconic) aged 8-16 weeks were used for light power measurements
(n = 5 for 473 nm; n = 6 for 532 nm; and n = 5 for 635 nm). All procedures were approved by the
MIT Committee on Animal Care and were in accordance with the NIH Guide for Care and Use
of Laboratory Animals. Prior to surgery, analgesics (buprenorphine, 0.1 mg/kg, IP; Meloxicam, 2
mg/kg, IP) were administered. An anti-inflammatory corticosteroid (Dexamethasone, 1 mg/kg,
IP) was administered to prevent edema. Under isofluorane anesthesia (1-4% isoflurane in
oxygen), the top of the head was shaved and a depilatory (Nair) was then applied to ensure
complete fur removal. The field was scrubbed 3x with betadine and isopropyl alcohol prior to
opening. A 3 mm (medial-lateral) x 2.5 mm (rostral-caudal) craniotomy was made in the parietal
30
bone. For consistency, the top left corner of the craniotomy was always 0.5 mm caudal and 0.5
mm lateral to bregma. To keep the cortical surface free from blood, a thin coat of transparent
silicone (Kwik-Sil, WPI) was placed in a ring around the bony margins of the craniotomy, as
needed, to prevent bleeding. The craniotomy was kept moist with saline.
Once the craniotomy surgery was complete, pentobarbital (50mg/kg, IP) was administered, and
3
isoflurane anesthesia was removed 3-5 minutes later, depending on the depth of anesthesia. Mice
remained anesthetized in the stereotax and on the heating pad during testing. Reflexes were
tested regularly to ensure adequate depth of anesthesia. Supplementary pentobarbital (25 mg/kg,
IP) was administered as needed. Euthanasia was performed at the conclusion of testing. If
bleeding on to the surface of the craniotomy was detected either pre- or post-experiment, the
animal was excluded from this study.
2.2.2 Isometric light measurements
Because opsin-expressing cells can be modulated by photons approaching from all directions, an
isometric light probe is needed to measure how much light is reaching a particular point in the
brain. Isometric light measurements also allow tissue optical properties to be derived from
diffusion theory. The ruby-tipped probes used in this study are isometric because fluorescence is
an inherently isometric process. The incident angle of the excitatory blue, green, or red photon
has no bearing on the emission angle of the measured ruby-colored photon (Skinner 1964).
Appendix 6 describes fluorescence and isometric theory in more detail.
3 Isofluorane
was used to quickly induce anesthesia in the test mice and maintained during surgery. It was
not continued during testing, though, because isofluorane increases cerebral perfusion which can lead to
bleeding or cerebral edema (Sicard, Shen et al. 2003), both of which would complicate the light
measurements. Cerebral blood flow with pentobarbital is close to what is seen in the awake animal
(Goldman and Sapirstein 1973).
31
2.2.2.1 Isometric probe construction
Isometric, ruby-tipped light probes similar to those presented by Bays et al. (Bays, Wagnieres et
al. 1997) were constructed (Figure 5a). A 300 pm diameter spherical ruby ball lens (NT46-223,
Edmund Optics) was centrally affixed to the flat-cleaved, polished end of a 400 tm diameter
multimode optical fiber (ThorLabs, BFH48-400, NA = 0.48) with optically-transparent, UV
curable adhesive (NT37-322, Edmund Optics). An SMA connector on the other end of the
multimode optical fiber allowed the probe to interface with the spectrometer.
2.2.2.2 Spectrometer set up
The fluorescence emitted from the isometric light probe was measured using a CCD spectrometer
(HR2000, Ocean Optics) and recorded using SpectraSuite software (Ocean Optics). Prior to
testing, a dark spectrum measurement was taken and subtracted out of all subsequent
measurements. A non-linearity correction was applied using the default linearity coefficients of
the spectrometer. The integration duration of the spectrometer was adjusted to prevent saturation
in the ruby wavelength range (690 nm to 695 nm). Integration times ranged from 3 ms to 3 s with
most integration times in the 0.1 to 1 s range. All recorded spectra were saved in their raw form
without averaging. Figure 5b shows an example spectrum with peaks representing both the
incident light and the ruby fluorescence. Only the ruby peak was used in analysis, and all photon
counts were converted to fluence rates with units of photons/second in offline processing.
2.2.2.3 Isometric probe calibration
The probe was calibrated in water by applying collimated light. Prior to calibration the beam of
the collimated light was profiled with a beam profiler calibrated for absolute light power
measurements (BC106-Vis, ThorLabs). The linear relationship between incident light power
density at the test wavelength and the emitted fluorescence at the peak ruby emission wavelength
(-695 nm, defined as the 0.47 nm-wide bin with peak amplitude between 693 nm and 697 nm)
was determined for each probe. A linear relationship between the applied light power density
and the ruby photon emission rate was generated as shown in Figure 5c. Two small corrections
32
were applied to the slope prior to using it for testing. The first accounted for the difference in the
refractive indices of the brain, n = 1.37, (Binding, Ben Arous et al. 2011) and water, n = 1.33. The
second accounted for the difference in the angle between the light source during (900) and testing
(720). See Appendix 6 for detailed probe calibration methods and theoretical explanation of the
calibration process.
Output from photon counter
Ruby sphere
3am dlamtr
Sample calibration curve relating
photon count to light power
-
4000
:~3500-
4
3.5
3000
2000
r
3
Applied light
wavelength
5r
o 1000
(
a)
b)
y5.57710E-05
2
1500-
Ruby
wavelength
500
400
500
800
700
800
Wavelength(nm)
900
RS9.95746E-01-
0
1000
0
20000 40000 60000
80000 100000
c)
Figure 5: Isometric light probe
Isometric light probe (a), spectrometer output from probe (b), and calibrationcurve (c)
2.2.3 Illuminator calibration
The light power output from the planar illuminator was measured prior to each testing session.
For red (635 nm) and green (532 nm) light tests, an analog, fiber-coupled DPSS laser (Shanghai
Laser Optics Company) was connected to a TTL-controlled optical shutter (OzOptics, SH-200-33470/700-M-0-0-SP) via a 200 tm diameter (numerical aperture = 0.22) multimode FC/PC
terminated fiber. For 473 nm light, a fiber-coupled diode laser (Vortran) was used. The output
from the shutter was FC coupled to the planar illuminator-a 1.5 mm diameter (numeric aperture
= 0.5)
plastic optical fiber (CK-60, Industrial Fiber Optics, Inc.). The planar illuminator, which was
fixed in the aligner, as shown in Figure 6, was lowered into an integrating sphere (S142C,
ThorLabs) with attached power meter (PM100D, ThorLabs).
33
For the red and green lasers, a 15 minute warm up was required for stable output. Afterward, the
laser power was manually adjusted to achieve total light powers from the planar illuminator of
0.44, 0.88, 1.77, 3.53, and 8.83 mW. When these quantities are divided by the area of the planar
2
,
illuminator (1.77 mm2 ), they yield light power densities of 0.25, 0.5, 1, 2, and 5mW/mm
respectively. For blue light tests, total light powers of 1.17, 1.57, 2.35, 4.03, and 8.83 mW were
,
used. These correspond to, light power densities of 0.66, 0.885, 1.33, 2.275, and 5 mW/mm 2
respectively. 4
2.2.4 Aligner placement and isometric probe insertion
To ensure alignment between the planar illuminator and isometric probe, a custom aligner
(Figure 6) was fabricated. The aligner was attached to a stereotactic arm and lowered on to the
surface of cortex using a micromanipulator (Siskiyou). 5 The isometric probe was mounted to
stereotactic holder (Kopf) and lowered at a 28' angle through the custom aligner. Ruby fluence
4 The blue light powers differ from those of red and green because the integrating sphere used initially consistently
misreported low powers of 473nm light. This was identified after the first three blue-light tests. The first integrating
sphere reported light powers of 0.44, 0.88, 1.77, 3.53, and 8.83 mW yielding light power densities of 0.25, 0.5, 1, 2, and
5mW/mm 2, but measurement with a second, new integration sphere reported light powers of 1.17, 1.57, 2.35, 4.03, and
8.83 mW yielding light power densities of 0.66, 0.885, 1.33, 2.275, and 5 mW/mm 2, respectively. Because the difference
in the reported values between the two integrating spheres was consistent and replicable, data from the three mice
already tested was included and subsequent mice were tested with the powers that had already been used with blue
light. In these subsequent tests, the new integrating sphere reported the values listed above and the old integrating
sphere reported the incorrect values, as expected. Calibrations were repeated after testing as well. The actual light
powers used do not matter for the final normalized result.
s Despite prophylactic dexamethasone administration, some mice had slight cerebral edema. To address the variability
of tissue height relative to the skull, a low intensity of light was coupled into the planar illuminator during lowering.
This created a well-defined spot on the surface of the brain that changed color when the illuminator was fully in contact
with the cortex (similar to the change in the color of green laser pointer when one shines a green laser pointer on his
thumb). This change in color allowed experimenters to ensure that the planar illuminator was fully on the surface of
the cortex.
In a few cases, the laser light became visible to the experimenter during the shallowest testing depth, presumably
because the dexamethasone had taken effect only after the aligner had been placed. In those cases, the aligner could
not be lowered further as this would risk damaging the isometric probe. Thus, an adjustment was made to the depths
if they were with 0.1 mm of a 0.5 mm increment and if the isometric probe monitoring confirmed that the brain had
sunken between aligner placement and the first measurements.
34
was monitored in real time during lowering and retraction using the spectrometer and
SpectraSuite software.
Planar
illumina
1.5mm
diameter
The ruby
sphere probe
will be
lowered into
cortex across
the beam path
of the planar
\
illuminator.
Planar illuminator
Figure 6: Custom aligner
Custom alignerviewed from the top (left) andfrom the side (right).
2.2.5 In vivo light measurements
Fluence rates were measured in 0.5 mm increments from 0.5 mm to 2.5 mm deep to the surface
of cortex. At each depth, continuous light pulses were applied at power densities of 0.25, 0.5, 1, 2,
and 5 mW/mm2 via the planar illuminator. Note that for blue light these light power densities
were slightly different as described in the calibration section above. Integration times ranged
from 3 - 3000 ms, with higher powers requiring shorter integration times. During a single pulse,
the spectral output from the light probe was recorded at least 25 times for each applied light
power density. The pulse durations ranged from 5-20 s, depending on the maximum integration
time possible without saturating the spectrometer measurement (i.e., without exceeding the
maximum photon count at the ruby wavelength). 6 Figure 7 shows a diagram with the
6
The total pulse length necessary to achieve 25 repeats was a bit longer than the sum of the 25 integration periods due
to transfer time between the spectrometer and the computer running the SpectraSuite software.
35
experimental set up for in vivo light measurements. Note that the hardware set up is nearly
identical to that used for calibration. The probe was not retracted until the end of testing to
minimize bleeding. Offline analysis and plotting were performed in MATLAB.
Photon count Information
from spectrometer
-
Laserandshu
mo
eter
Loser
S
*
enablel
power,
signal S
I.1ser
Cofto
fiberprb
fiber
al
Light
measurement
optical
optical
Lae
Light
probe
Aesthetized
mouse
Iflum,
.-
Figure 7: Experimental setup for in vivo light measurements
2.2.6 Determining absorption and reduced scattering coefficients
In the photodynamic therapy literature, in vivo light propagation measurements of red and
infrared light are made several millimeters away from a small, narrow light source. This
paradigm allows for absorption and scattering coefficients to be estimated using simple diffusion
theory equations for isotropic point sources. For visible light of sub-red wavelengths, it is not
36
feasible to use point source estimates because too little light reaches these distant points to get
accurate measurements. Thus, this work uses a wide, collimated-beam estimate to determine
scattering and absorption coefficients. For this work, a planar illuminator was used to
approximate a wide, collimated-beam source (Appendix 6).
2.2.7 Diffusion model of light propagation in tissue
According to diffusion theory, in the limit where
Ita,
the absorption coefficient, is much larger
than ts', the reduced scattering coefficient, a wide-beam collimated irradiance incident on a thick
slab with refractive index matched boundaries has the following fluence rate, PDt(z):
(Pt(z)
(1)
= (5/(1 + 2*a/peff))*exp(-peff *z)-2*exp(-pt*z)
where z is the vertical distance from the illumination source to the point measured, teff is the
effective attenuation coefficient, and tt is the total attenuation coefficient.
Assuming
ia
/
[s <<1, no sources deep within tissue, and a semi-infinite medium with isometric
scattering, the effective and total attenuation coefficients can be related to one another as follows:
leff~
(2)
sqrt(3*Ia* lit)
Equation 2 can be rearranged to solve for the total attenuation coefficient,
ut = je/
.tt:
(3)
(3* p.)
Equation 3 can be substituted into Equation 1 to yield Equation 4
Pt(z) =(5/(1 + 2*p/i /jeff))*exp(-eff *z)-2*exp(-pe# *z/3/p,)
(4)
A full derivation of all of the equations shown above can be found in (Star 2011).
2.2.8 Deriving absorption and reduced scattering coefficients from light measurements
Two manually calibrated spectrometers (HR2000, Ocean Optics) were used in this study, and
their peak ruby wavelengths differed slightly, leading to different ruby wavelength ranges. The
37
ruby wavelength range was 695.7 to 698nm for the older device and 692-695 nm for the newer
spectrometer. All tests with a given ruby sphere probe were performed using the same
spectrometer that was used for calibration without disconnection. In SpectraSuite, photons were
measured across a range of "ruby" wavelengths.
Offline, in MATLAB, the photon count was normalized by the integration time to yield a fluence
rate (photons/ second). Fluence rates for each condition in each mouse were averaged across all
trials at every wavelength (omitting the first trial due to possible software lag). The calibration
curve was used to convert the mean fluence rates (photons/second) into light powers (mW). The
light power reaching a given depth was divided by the applied light power to yield a normalized
fluence rate (or fraction of light power remaining). The normalized fluence rates for all light
powers at a given depth in a given mouse were averaged to yield mean normalized fluence as a
function of distance from the illuminator for each mouse.
The normalized fluence functions for every mouse were fit to Equation 4 using a least-squares
approach. All available depths were included in the fit for each mouse. All fits had an R2 value >
0.95. From those fits,
aand
Leff were determined for every mouse. Those absorption and effective
coefficients were then averaged across all mice of a given color to yield the estimated absorption
and effective coefficients for a given color.
2.2.9 Geometric limit of light propagation in tissue
Light power density in tissue decreases as a function of distance from the optical fiber due to
geometry. Light clearly spreads out as it exits the tip of a multi-mode optical fiber so it makes
sense that there would be some geometric limit to how much light could reach a given depth in
tissue even in the absence of scattering or absorption.
The Kubelka-Munk (KM) model (Kubelka 1948) has been proposed as the geometric limit for light
reaching a given point in tissue, both in the light dosimetry literature (Eggert and Blazek 1987,
Star, Marijnissen et al. 1987) and in the optogenetics literature (Aravanis, Wang et al. 2007, Yizhar,
38
Fenno et al. 2011). This study puts forth an alternative geometric limit: the internal geometric
model.
2.2.10 Kubelka-Munk model
The most commonly used estimate for geometric light decrease is the Kubelka-Munk model
(Equations 5 and 6).
I (z)
_
I (z=O)
p2
(z+p)
(5)
2
where
(6)
p r-
Here I(z) is the flux at depth z, I (z =0) is the initial flux, r is the radius of the optical fiber, n is the
refractive index of the medium, and NA is the numerical aperture of the optical fiber. Arvanis et
al., 2007 expanded the KM model to include a wavelength dependent scattering factor, S, as
shown in Equation 7. This has now become the basis of the Stanford predictor.
I (z)
I;(z=o)
_
-
p2
(sz+1)(z+ p) 2
2.2.11 Internal geometric model
In a multi-modal optical fiber, many modes of light are coupled into a single optical fiber. The
modes exit the fiber at some point on the surface of the tip and at some angle relative to the fiber
surface. This model of geometric loss was developed to estimate the light due to light spread as
different modes of light exited the tip of the fiber at different locations and angles. For a light
mode to spread maximally away from the multi-mode optical fiber, it would have to exit the fiber
tip at the farthest distance possible from the center of the fiber (r = rfiber) along the trajectory of the
maximum divergence angle, 0&iv, which is calculated in Equation 8.
(8)
Odv = sin- 1 NAfiber
ntissue
39
where ntissue is the refractive index of the tissue and NAfiber is the numerical aperture of the fiber.
This maximally spreading mode crosses the center of the optical fiber at a distance, zfiber above the
tip.
Zfiber is
calculated in Equation 9.
Zfiber
-
(9)
rfiber
tan(Odiv)
At a distance zfiber away from the tip of the fiber, all of the modes crossing the exact center will be
emitted at some angle ranging from - Odiv to Ouv. None of those modes will be internally reflected
because they all have a trajectory angle, 0 5 1 Odiv I. For points off of the center axis at distance
zfiber
from the tip, some of the modes crossing through those points will be internally reflected
while others would be emitted from the tip (Appendix 6). The launch angle and point along the
fiber can be calculated in each case. The code for this function is available upon request. By
integrating across the longitudinal axis of the fiber, a profile of modes at the tip of the fiber can
be generated with each mode having a launch angle, and a launch distance from the center of the
fiber. Those profiles at the tip can be fit to a model which projects where they will go at various
distances from the tip of the illuminator.
2.3 Results
Using the methods described above, normalized fluence rates were determined as a function of
distance from the illuminator for red, green, and blue light. Figure 8 shows that red light
propagates much further in tissue than either green or blue light does and that very little of the
applied green and blue light remains at distances 1 mm or more from the light source.
Curves based on Equation 4 were fit to the average normalized fluence rate for each mouse
(Figure 9). All fits had R-squared values > 0.95, typically > 0.99. For red (635 nm) light, the mean
best fit for the absorption coefficient was found at
ta =
2.14 cm-1 (standard error = 0.15 cm1). For
the effective coefficient, the best fit was Ideff = 6.61 cm 1 (standard error = 0.53 cm-1). For green (532
nm) light, the mean best fit was found at Ida = 13.7 cm- 1 (standard error = 1.33 cm-1 ) and 4eff = 41.1
40
cm-1 (standard error = 4.04 cm-1 ). For blue (473 nm) light, the mean best fit was found at
lta
= 10.6
cm-1 (standard error = 0.486 cm-') and e = 31.8 cm-1 (standard error = 1.46 cm-').
Mean light fall off with standard error
--
0.9
-
473 nm
532 nm
635 nm
-
1
0.8
E
0.7
0.6
0.5
0
0.4
LL
0.3
-
0
C.
0.2
0.1
0
0.5
1
1.5
2
2.5
Depths (mm)
Figure 8: Average light decrease with distance from illuminator
Mean normalizedfluence rates representing the fraction of applied light power reaching a given depth for
blue (473 nm), green (532 nm) and red (635 nm) light as a function of distancefrom the illuminatorwith
standard error bars
41
Light fall off with distance
1
473 nm
532nm
0.9
- - -
0.7
a
- -%
t%
0.6
\~%%
4-.
0.5
0.
-
%%
A
'4*4'4"b
OA
0
%
*4t*f*
C
0.3
X4"W
f
0.2
0.1
0
)
*4
*4
%
.0
635nm
0.8
4,
0.5
1
1.5
2
2.5
Depths (mm)
Figure 9: Light fall off for all animals
Normalizedfluence ratesfor each animal are plotted (markers) and curves for each mouse were fit using
Equation 4. From these curve fits values for the absorptionand effective coefficients at each wavelength
were determined.
Next, the light powers found here for the three colors of light were plotted against two commonly
used estimates for light propagation (Figure 10, 11, and 12). First, the Stanford prediction was
plotted. Recall that the Stanford online calculator of irradiance combines the KM estimate with
scattering estimates to predict light fall with distance (Equation 7). Next, the KM estimation was
plotted to represent the widely-assumed geometric limit which light powers should not exceed
even in the absence of scattering and/or absorption. Finally, the new "internal geometric"
estimate of the geometric limit of light propagation was plotted as an alternative geometric limit,
contrasting with KM.
The in vivo measurements for red light exceed the KM model, which is not supposed to be possible
if KM is truly the geometric limit. The in vivo measurements never exceed the 'internal geometric"
42
estimate, which unlike the KM estimate, accounts for the internal reflections within multi-mode
optical fibers.
For blue light, the in vivo measurements generally match with the Stanford estimate at 473 nm
(Figure 10). For green light, in vivo measurements are lower than predicted by the Stanford
estimate. Some part of this difference between the Stanford estimate and the in vivo data may
result from a slight difference in the wavelength used for the estimate (560 nm) and the actual
measurements (532 nm). For red light, the in vivo and estimate wavelengths are much closer (630
nm for the estimate, 635 nm for the tests), and the difference is even greater.
Measurements and estimates of blue light propagation
x x xxx
x x xxxxxxx
x
xxxxx
.x
X
X
0.8
x
x
x
AA
-I
Measured light powers:
x
x
(in vivo)
0.6
-A
A
Predictions:
A
-------- Stanford
A
A
-r
Current study
-
A
Kubelka Munk
AA
0.4
(geometric limit)
AAAA
x
IAAAAAAAAA&
U-
Internal geometric
(geometric limit)
AAAA
0.2
AAAAAAAA
............
0
0
0.5
1
1.5
2
2.5
Depths (mm)
Figure 10: Blue light propagation
Blue light propagationin this study compared with the widely-used Stanford predictor and two estimates
of the geometric limit (Kubelka Munk and the internal limit presented here).
43
Measurement and estimates of green light propagation
I
1
-x
xx
x
I
x
x
x
III
x xx
x
xx
xXXxxx
x
x
A
Measured light powers:
-
'A
x
x
A
A
x
AA
0.6
Predictions:
A
A
A
A
a
A
AA
AA
OA
A
x
A
A AA
AA
AAA
A
A
Kubelka Munk
(geometric limit)
Internal geometric
(geometric limit)
A
A
U-
Current study
(in vivo)
(532 nm)
xx
Ax
01
--
-
0.8
A
A
A
AA
A
A
A
A
0.2t
OL
0
0.5
1
1.5
2
2.5
3
Depths (mm)
Figure 11: Green light propagation
Green light propagationin this study compared with two estimates of the geometric limit (Kubelka Munk
and the internal limit presented here). Data from the Stanford predictor is omitted because the closest
available value is 560 nm, whereas these measurements were taken at 532 nm.
44
Measurement and estimates of red light propagation
1
x xx xx xxx xx xx xx x x x x~
x
.Ax
X
-A
1E
0.8
S
-- A
A
Measured light powers:
- Current study
X
x
(invivo)
. .Predictions:
A
'.-
-------- Stanford
.
Kubelka Munk
x
(geometric limit)
internal geometric
(geometric limit)
SOA
0
-..
LA.
0.2
-*-**--.*-mea
0I
0
0
0.5
1
1.5
2
2.5
3
Depths (mm)
Figure 12: Red light propagation
Red light propagationin this study compared with models based on ex vivo measurements, the widelyused Stanford predictor and two estimates of the geometric limit (Kubelka Munk and the internal limit
presented here).
2.4 Discussion
2.4.1 Comparison of measured absorption coefficients and those estimated based on
the hemoglobin content of the brain
Since, to our knowledge, there are no other in vivo measurements of light propagation at 473 nm
and 532 nm, it may prove useful to consider the physiology of the cortex in the context of the
green and blue light absorption estimates. At 473 and 532 nm, the only major source of absorption
in tissue is hemoglobin. In particular, most absorption is due to oxygenated hemoglobin because
oxygen saturations are typically >95% in healthy individuals (Koeppen and Stanton 2009). Within
the cranium, blood and CSF each comprise 10% of the total volume while the brain makes up the
45
other 80% (Rengachary 2005). Assuming that 3-5% of the intracranial blood is in large vessels
(e.g., the Circle of Willis) and that the rest is in tiny vessels throughout the brain tissue itself, it is
reasonable to expect that the blue and green absorption coefficients in brain tissue would be 57% of the absorption coefficients for oxygenated hemoglobin in whole blood. In Figure 4, the
absorption coefficient for oxygenated hemoglobin
235 cnr1 at 532 nm. Thus, at 473 nm, our value of
aH
_oxy = 165.35 cm.- at 473 nm and
a = 10.6
IaHb-oxy =
cm1 (standard error = 0.486 cm-1) is
6.4% of the absorption coefficient of oxygenated hemoglobin. At 532 nm, our value of
4a
= 13.7
cm- 1 (standard error = 1.33 cm- 1) is 5.8% of the absorption coefficient for oxygenated hemoglobin
at that wavelength. The absorption coefficients determined here are consistent with physiology
of the brain, and imply that absorption is the main source of light loss for green and blue light in
the living brain.
2.4.2 Comparison of absorption coefficients reported with values in the literature
Even though, we know of no other in vivo reports of absorption coefficients at 532 nm and 473 nm
blue light comparisons with in vitro and ex vivo data may provide insight into how close these
two types of measurements are. At 473 nm, this paper reports
Via
= 10.6 cm1 (standard error =
0.486 cm'). Gottschalk (Gottschalk 1992) reported Ia= 9 cm-1 at 456 nm for in vitro for gray matter
(reprinted in (Mobley 2003)). Mesradi et al., 2013 reported Va Z 4.5 cnrl for fresh slices of rat brain
and 4a = 3 cm1 for frozen slice of rat brain (Mesradi, Genoux et al. 2013). Yaroslavsky et al., 2002
reported Via = 0.7
cm1 and
Ia
= 1.4 cm-1 for gray and white matter, respectively, based on
measurements taken in post mortem human tissue (Yaroslavsky, Schulze et al. 2002). The
Yaroslavsky et al. values have been used widely in the optogenetics literature (Aravanis, Wang
et al. 2007, Bernstein, Han et al. 2008, Gutierrez, Mark et al. 2011, Kahn, Desai et al. 2011, Yizhar,
Fenno et al. 2011, Han 2012, Jing, Fabien et al. 2012, Osawa, Iwasaki et al. 2013) despite being an
order magnitude lower than the in vivo measurements presented here.
At 532 nm, this paper reports Va = 13.7 cm-1 (standard error = 1.33 cm-1). Again this value is similar
to what Gottschalk (Gottschalk 1992) reported for in vitro gray matter at 514 nm (ha = 11.7 cm-1).
The Yaroslavsky et al., values are even lower for green light (reported at 510 nm) than they were
46
for blue with
ta =
0.4 cmr 1 and
ta =
1 cm-1 for gray and white matter, respectively. Mesradi et al.,
2013 report comparable values for green light as well as for blue (Mesradi, Genoux et al. 2013). In
vitro measurements of human gray matter taken 24 hours post mortem yielded
Jta
= 19.5 cm- 1 at
514 nm; however, that paper notes that a separate sample from this patient's brain contained
melanoma (Sterenborg, Saarnak et al. 1996). Thus, the tissue yielding the measurement was not
from a healthy brain, and it is even possible that some melanin expressing cells were present in
the sample that was taken as "gray matter."
In contrast to the lower two wavelengths, the literature offers some, limited in vivo light
measurements of red (635 nm) light. Our value of
mice is a bit higher than was reported in rats
Jta=
La =
2.14 cm- 1 (standard error = 0.15 cn- 1) in
0.57 cm-' at 632 nm (Angell-Petersen, Hirschberg
et al. 2007); however, this difference may result from our measurements being taken in gray
matter while the other measurements were likely taken in gray matter and the thalamus in order
to keep the probe far enough from the source to use the point-source approximation in diffusion
theory. Overall values from extracted tissue have a wide range of absorption coefficients from
ha
= 0.20 to 4.5 cm-1 (Wilson, Patterson et al. 1986, Flock, Wilson et al. 1987, F. 1995, Sterenborg,
Saarnak et al. 1996, Yaroslavsky, Schulze et al. 2002, Angell-Petersen, Hirschberg et al. 2007,
Mesradi, Genoux et al. 2013). Our value for red light falls squarely within that range.
2.4.3 Kubelka-Munk model
Even though the Kubelka-Munk (KM) model has been applied to biological tissues and in
estimates of light propagation for optogenetics, its primary and more appropriate application is
in the paint industry (Star, 2011). This model assumes completely diffuse light (Kubelka, 1948),
which is appropriate for the ambient room lighting incident on a painted wall, but not for the
output of a laser-coupled optical fiber in tissue. Further, the Kubelka-Munk model essentially
assumes a one dimensional process in which scattering decreases flux with distance. This differs
from transport theory, where the three dimensional radiance decreases due to geometric spread,
scattering, and absorption. In addition, the KM model assumes that the radius of the fiber, r, is
much smaller than the distance between the optical fiber and the light source, z. In optogenetics,
47
researchers often use optical fibers with diameters on the order of 0.1 mm, and they couple their
electrodes and optical fibers so that they can record from targets only a few hundred microns
from the light source. Thus, the KM model assumption that r << z does not hold for in vivo
optogenetics experiments, like the ones in this study.
The KM model seems to have become popular for use with optogenetics for three reasons: 1) it is
simple mathematically, 2) when the radius of the optical fiber is much smaller than the
measurement distance the source begins to resemble the isotropic, diffuse, uniform source
assumed in the KM model, and 3) the in vitro and ex vivo preparations used for light measurement
contain little blood, and, thus, have little absorption due to hemoglobin. Often these ex vivo
preparations are purposefully exsanguinated, but even if they are merely excised, much of the
blood is lost in the excision process and any remaining blood would contain only deoxygenated
hemoglobin. Thus, in these ex vivo preparations, scattering is the dominant cause of light loss
while in an in vivo preparation, absorption is the dominant cause of light loss.
The fact that our measured red light values exceed the maximum predicted values possible (in
the absence of absorption and scattering) suggests that the KM model is inappropriately applied
as a geometric limit and that a model such as the internal geometric model presented in Appendix
6.3 would be more appropriate.
2.4.4 Implications for the selection of opsins
Red light propagates much farther in tissue than green light or blue light does. Models based only
on scattering and/or geometric loss fail to demonstrate just how dramatically absorption
attenuates light in the living brain. When models fail to account for absorption, the difference in
red and blue/green light propagation is less apparent. The in vivo measurements presented here
show the dramatic advantage of using red light in the brain rather than other visual light colors.
Now, that the difference is better understood, the choice of opsin for neuroscience studies may
shift to reflect this finding. In particular, new red-light sensitive opsins like Crimson (Klapoetke,
48
Murata et al. 2014) and Jaws (Chuong 2014) are likely even better choices for most neuroscience
studies than ChR2, eNpHR, Arch, or ArchT. Jaws differs from previous inactivating opsins
because it has a red shoulder, meaning that in addition to having a peak sensitivity in the yellow
range (like other halorhodopsins) it is also sensitive to light at higher, red wavelengths, which is
sometimes called a red shoulder. Further, these red-light sensitive opsins are great candidates for
primate neuroscience or even for translational applications. Thus, for this particular study, we
chose to use Jaws, a red-light sensitive opsin that inhibits neuronal firing rather than Arch / ArchT
or eNpHR, which are sensitive to green and greenish-yellow light, respectively.
49
Chapter 3: Illuminator for optogenetic modulation of large brain volumes
Abstract
A large-volume illuminator was developed for optogenetic manipulations in the non-human
primate brain. The light-emitting surface of this illuminator is two orders of magnitude larger
than a conventional optical fiber of equal diameter, which allows for substantially greater light
delivery to brain tissue with no greater penetration damage than a conventional fiber. This
illuminator delivers light in the macaque cortex over a tissue volume of ~ 10 mm3, which is at
least 2 mm in diameter and at least 3 mm in length. When used in conjunction with a red-light
sensitive halorhodopsin, Jaws, neuronal silencing spanning a 3 mm depth was observed 1 mm
away from the light source. This illuminator may prove to be an enabling technology for
widespread optogenetics use in non-human primate studies.
Keywords:
primate optogenetics, large-volume illumination, halorhodopsin, channelrhodopsin, illuminator
Glossary:
Jaws-a red-shifted halorhodopsin
LFP - local field potential
3.1: Introduction
While optogenetics is widely used in the rodent neuroscience community (Boyden 2011,
Bernstein, Garrity et al. 2012), this technique has been used only a handful of times in non-human
primates, primarily in proof-of-concept studies (Han, Qian et al. 2009, Diester, Kaufman et al.
2011, Han, Chow et al. 2011, Cavanaugh, Monosov et al. 2012, Gerits, Farivar et al. 2012, Jazayeri,
Lindbloom-Brown et al. 2012, Dai, Brooks et al. 2014). The size of the macaque brain is a challenge
for optogenetics. Because the macaque brain is 200x larger than the mouse brain, the tissue
volume comprising a region of interest in the macaque brain may be significantly larger than an
analogous tissue volume in a rodent (Herculano-Houzel 2009). For example, macaque tracer
studies have found single axons that span distances longer than the entire mouse brain (Amir,
50
Harel et al. 1993), and electrophysiological studies show that the point image spread in macaques
is up to 10 mm in certain brain areas (such as 7a) and at least a few millimeters in most other
visual areas (Hubel and Wiesel 1974, Dow, Snyder et al. 1981, Gattass, Gross et al. 1981, Van
Essen, Newsome et al. 1984, Tootell, Switkes et al. 1988, Andersen, Asanuma et al. 1990,
Boussaoud, Desimone et al. 1991, Felleman and Van Essen 1991). Thus, in order to silence or drive
a neuronal population in the primate brain corresponding to a particular function or
representation using optogenetics, tissue volumes on the order of cubic millimeters may need to
be illuminated.
Previous optogenetics studies in the primate have illuminated cortical tissue in a manner similar
to what Aravanis et al., (2007) presented for rats (Aravanis, Wang et al. 2007, Gradinaru, Zhang
et al. 2010, Diester, Kaufman et al. 2011, Dai, Brooks et al. 2014). In this "dual-pronged optrode,"
a solid-state laser diode is coupled to a flat-cleaved glass or silica multimodal optical fiber, which
is glued to an electrode and inserted into the brain as a single unit (Aravanis, Wang et al. 2007,
Han, Qian et al. 2009, Diester, Kaufman et al. 2011, Han, Chow et al. 2011, Cavanaugh, Monosov
et al. 2012, Gerits, Farivar et al. 2012, Jazayeri, Lindbloom-Brown et al. 2012).
Other groups have identified the need to illuminate larger tissue volumes in the primate brain
and proposed possible solutions. Diester et al., (2011) coupled multiple flat-tipped fibers to a
single electrode (Diester, Kaufman et al. 2011), while Tamura et al., (2012) enclosed several flattipped optical fibers and a tungsten microelectrode together in glass (Tamura, Ohashi et al. 2012).
Unfortunately, these approaches double or triple the diameter of the illuminator that is inserted
into the brain and have been noted to cause damage (Diester, Kaufman et al. 2011), perhaps due
to the large brain volumes that must be displaced for a given degree of illumination achieved.
Because penetration damage is proportional to the diameter of whatever is inserted into the brain
(Sharp, Ortega et al. 2009), the large-volume illuminator described here was designed to increase
the volume of light delivery without increasing the fiber diameter or the surface light power
density.
51
As shown in Figure 13, the large-volume illuminator has the same diameter as a standard optical
fiber (70 -250 microns), but its light emitting surface area is on the order of cubic millimeters,
rather than tens of microns. When placed in a brain phantom (such as the 1" cube of 1.75% agarose
shown in Figure 13) and coupled to lasers with equal total input light power, the large-volume
illuminator delivers light over a larger volume than the traditional fiber does. The surface light
power density of the large-volume illuminator is approximately two orders of magnitude smaller
than the light power density at the tip of the traditional fiber because its light emitting surface
area is approximately 5 mm 2 compared with 0.05 mm 2 for the traditional fiber. Even so, the large
volume illuminator distributes light more broadly. This broad distribution allows both light and
heat to spread more evenly through the tissue without creating a "hot spot" near the fiber tip.
Beyond its tip geometry, the large-volume illuminator differs from conventional optical fibers
because it is made of a flexible, biocompatible plastic rather than glass or silica. Further, it is made
of a single material without any glue, doping, or surface asymmetry. As Ozden et al., (2013) noted,
the axial asymmetry of "dual-pronged optrodes" induces cortical damage and increases the risk
of optical fiber breakage in the brain (Ozden, Wang et al. 2013). While coupling the illuminator
and electrode is still feasible, this illuminator delivers light to a large to enough volume that
optogenetic silencing may be recorded with a separately inserted electrode placed a millimeter
or more away from the illuminator. Thus, this fiber has a reduced risk of breaking both because
it is made of plastic and because it is not necessary to couple it to an electrode.
While this illuminator allows for broad tissue illumination regardless of wavelength, it is most
powerful when used with red light, which is less readily absorbed by tissue than any other color
of visible light. The illuminator presented here delivered light over a 3 mm depth of macaque
cortex (>10 mm 3) and silenced neurons expressing the red light-sensitive halorhodopsin, Jaws,
(Chuong 2014) over the same distance.
52
Figure 13: Comparison of large-volume illuminator and standard optical fiber.
The large-volume illuminator(left) and a standardopticalfiber with equal diameter and equal total input
light powers are shown in a block of 1.75% agarviewed from the top (top) and side (bottom).
53
3.2 Materials and Methods
3.2.1 Illuminator construction
Illuminator construction began with etching the light emitting surface of the fiber, near the tip.
Figure 14a illustrates the result of tip etching by comparing the cross section of the tip of a
traditional optical fiber with the cross section of the tip of the large volume illuminator. A 22gauge wire stripper (StripMaster) was used to remove the polyethylene jacket 15- 20 cm from the
end of a 250 tm diameter plastic optical fiber (Industrial Fiber Optics). A table vise clamp
(Wilton) was locked onto the stripped end of the fiber while the experimenter held the jacketed
end taut and parallel to the floor (perpendicular to the vise). The stripped section of the fiber was
thinned to a diameter of 60-100 pm using the lower setting of a dual temperature (570 / 1000'F)
heat gun (Milwaukee). While still holding the fiber taut, the experimenter removed heat and
allowed the fiber to cool, which prevents the fiber from curling. Once the fiber cooled, the
experimenter gripped the fiber about an inch to either side of its thinnest point and pulled the
fiber apart to create a tapered tip. While this pulling technique requires some practice,
particularly in regulating the heat to prevent the fiber from melting or curling, several members
of the lab have mastered the technique in less than a day.
Next, the tapered tip of the fiber was examined under a 4x dissection microscope (VistaVision).
Figure 14b shows the tip of the illuminator grossly and Figure 14c shows the tip under a
microscope. Fibers with forked or curled tips were discarded. To etch the tip, the length of desired
light emission (typically 3-7 mm) was measured up from tip and lab tape (VWR) was applied
above that level. This left only the desired length of light emission exposed and protected the rest
of the fiber during etching. The exposed tip of the fiber was then uniformly etched on all sides
using a 5 tm silicon carbide lapping sheet (ThorLabs) followed by a 3 .tm aluminum oxide
lapping sheet (ThorLabs).
At the other end of the fiber, the 22-gauge wire stripper was used to remove about 5 mm of the
jacket. This end was cut flat using a hot knife (Industrial Fiber Optics) and polished to a smooth
finish using successively smaller fiber polishing sheets ranging from 5 tm to 0.3 tm (ThorLabs).
54
A fiber microscope (ThorLabs) was used to ensure a smooth and uniform polish. The flat end of
the polished fiber was inserted into a stainless steel ferrule with 260 urm inner diameter (Precision
Fiber Products) until it was flush with the coupling end of the ferrule. The fiber was then secured
in the ferrule with plastic epoxy (Industrial Fiber Optics). Excess epoxy was removed after 12 to
24 hours of curing. A dust cap (ThorLabs) was placed on the ferrule to protect the fiber. Figure
14d summarizes the fabrication process.
55
Standard fiber
IlII uminator
Cladding
Cladding
Core
b)
a)
c)
I
d)
Figure 14: Etched tip of the large-volume illuminator
(A) Cartoon cross section showing how both the core and cladding are etched to spread light broadly out
of the tip. (B) The tip of the illuminatortapersfrom a diameter of 250 ym to 60-100 pm. (C) Cladding is
removed along the last 5 mm of the penetrating illuminator. When a laser is applied to the far end of this
illuminator, light is emitted over the 5 mm long etched tip. (D) This flow chart describes the fabrication
processfor etched large-volume illuminators.
56
3.2.2 Interfacing the illuminator with a laser
A diode-pumped solid state (DPSS) laser (Shanghai Laser Optics Co.) was coupled to a
mechanical shutter (Oz Optics) via a 200 lim diameter glass optical fiber (Thor Labs). A PC
running MonkeyLogic software (University of Chicago) and a DS8000 digital stimulator (World
Precision Instruments, Inc.) controlled the shutter opening and laser output. A 200 tm diameter
glass optical fiber (ThorLabs) coupled the output of the shutter to the illuminator with a ceramic
mating sleeve (Precision Fiber Products). This setup (Figure 15) was used for both the calibration
and in vivo testing.
57
Laserandshutter
controlsgn
2 Laser
senable
TTL
Laser
DPSSopen
power
10speca
sign
tt r or
opbicaoptical
'Ollr
fier
fber
Mating
Illuminator
a)
Ceramic
Patch cable with FC connector
mating
Illuminator
b)
Figure 15: Optical setup interfacing the large-volume illuminator with a laser.
(A) The full optical setup is shown here in block diagramformat. (B) A close-up of the opticalfiber mating
sleeve / illuminatorinterface.
3.2.3 Illuminator profiling and calibration
To ensure uniform light emission along the etched fiber tip, a lowering test was performed.
Specifically, the illuminator was lowered with a micromanipulator (Siskiyou) into an integrating
sphere (ThorLabs) in 500
mrincrements. Using light absorbing foil (ThorLabs) and alumuminumn
58
foil (Reynolds), custom shielding, with non-reflective material on one side and reflective material
on the other, was created. The custom shielding placed over the top of the integrating sphere, and
the large volume illuminator passed through a 26-gauge hole in the shielding. No ambient light
or light from parts of the fiber above the sphere could enter the integrating sphere because the
diameter of the hole in the shielding matched the diameter of the penetrating illuminator. The
total light power output was measured along the length of the illuminator and plotted as a
function of how far the illuminator had been lowered into the integrating sphere (Figure 16).
Some novel tip geometries (e.g., diffuser tips) introduce an optical mismatch between the optical
fiber and a secondary material, which decreases total light power output and causes some light
energy to be dissipated as heat. To confirm that this illuminator does not lose light energy due to
its geometry, light power measurements from a flat cleaved optical fiber and the illuminator were
compared under identical input conditions. After initial tests showed identical total light power
output for the current illuminator and flat-cleaved fibers, this testing was not continued.
59
Cladded
optical
fiber
Cadded
optical
fiber
/
.--
\
'4
4$
fiber
I
In'
fiber
Iris
Integratin
sphere
a)
..........
... ........................................
Illumination profile of representative illuminator
with standard error bars
Light absorbing
coating
0.8
ht
0.6 F
0
S0.4
0.2
0
b)
c)
1
2
3
4
5
Length of illuminator in integrating sphere (mm)
Figure 16: Setup of integrating sphere and illuminator that allow for measurements of the light emission
profile along the length of the illuminator.
(A) External view of profiling set up. (B) Cross-sectional view of illuminator profiling set up. (C)
Representative illuminatorprofile with the fraction of total light emission measured as a function of the
length of the illuminatorthat is in the integratingsphere to be measured.
60
6
3.2.4 Subjects and surgical procedures
Two male rhesus macaques weighing 13 kg and 16 kg were tested. The animals were cared for in
accordance with the National Institutes of Health Guide for the Care and Use of Laboratory
Animals and the guidelines of the MIT Animal Care and Use Committee. All surgical procedures
were carried out under anesthesia. Antibiotics and analgesics were administered postoperatively, as needed. In each monkey, a recording well (19 mm inner diameter, Crist
Instruments, Co.) over the frontal eye field and a titanium head post were implanted surgically
under aseptic conditions. After electrophysiological and micro-stimulation mapping, a viral
vector containing Jaws (AAV8-hSyn-Jaws-GFP) was stereotactically injected into cortices of both
monkeys. Illumination testing and electrophysiological recording began six weeks after virus
injection.
3.2.4.1 In vivo illumination
During all testing, the animal was seated in an ergonomically-designed chair facing a computer
screen with its head held. Two micromanipulators (NAN Instruments, Inc.) were mounted on the
recording well. A custom grid was placed in the recording well to ensure that the illuminator
and electrode were lowered in parallel one millimeter apart from one another. The grid was
locked into place relative to the implanted recording well such that the virus injected sites could
be accurately targeted months after injection. The illuminator was lowered to the desired depth
through a 25-gauge stainless steel guide tube held by one of the drives, while an electrode was
lowered in parallel 1 mm away through another guide tube using the second drive. In monkey
C, a single-contact, parylene-coated tungsten microelectrode with approximately 1 MQ resistance
(WeSense) was used, while in monkey L, a 16-channel multi-contact electrode (u-probe, Plexon)
was used. The illuminator and u-probe remained in the same location throughout a recording
session, but the single-contact electrode moved along its insertion trajectory, to allow for
recording at different depths along the length of illumination. Prior to testing, the experimenter
visually confirmed that the primate was not exposed to any stray laser light. The entire implant
and optics setup remained shielded with light-absorbing foil (ThorLabs) throughout testing.
61
Figure 17 shows the recording, illumination, and virus injection locations on an MRI image for
both monkeys.
Subject C
Subiect L
a)
c)
b)
Figure 17: Locations of the illuminator and recording sites in cortex.
(A) Coronal MRI imagefrom Monkey C showing the grid through which electrodes and the illuminator
are lowered and the target region of interest (white box). Yellow lines denote the sides of the recording
chamber. (B) Enlarged view of the region of interest with the trajectories of the illuminatorand electrode
shown as well as the virus injection region and the locations of the sample neurons shown in Figure 18
and 19. Note for spatial reference that the electrode and illuminatortrajectories are separatedby 1 mm. (C)
and (D) Same as (A) and (B), respectively, except with the imagingfor Monkey L.
62
3.2.4.2 Recording and analysis
The electrode input was coupled into a pre-amplifier (Plexon) via a headstage (Plexon) and
electrical connector (Onnetics). The pre-amplifier split the data into spike channels (0.25 -8 kHz)
sampled at 40 kHz and local field potential channels (0.7 - 170 Hz) sampled at 1 kHz. Preamplifier output went into a multi-channel acquisition processor, or MAP box, (Plexon) where it
was filtered and acquired using the Rasputin (Plexon) software. Spikes were sorted offline using
principal component analysis and manual waveform shape analysis (Offline Sorter, Plexon).
MATLAB (Mathworks) was used for electrophysiological data analysis and plotting.
3.3 Results
3.3.1 Optical artifact in local field potential
An optical artifact results when metal electrodes are illuminated with light (Ayling, Harrison et
al. 2009, Han, Qian et al. 2009, Cardin, Carlen et al. 2010, Han, Chow et al. 2011, Ozden, Wang et
al. 2013) presumably due to the Becquerel effect, as per Han, 2012, The presence of a light artifact
in the local field potential (LFP) indicates that light is reaching a particular cortical location, and
because the light artifact is even observed in saline, it is an opsin-independent marker of tissue
illumination. However, as others have noted, the light artifact varies across locations and between
recording sessions (Ayling, Harrison et al. 2009, Han, Qian et al. 2009). Here, the optical artifact
is used only as a digital marker that light has reached a particular point in the tissue, not as an
analog readout of how much light has reached that point.
The optical artifact was observed over 3 mm depth of cortex, which coincided with neuronal
silencing in areas that had previously been injected with the halorhodopsin, Jaws. Figure 18
shows the actual LFP deflection that resulted during illumination in both animals.
63
Light artifact at 0.5 mm
Instanteous firing rate at 0.5 mm
A0
I.
C
aIL
-J
A)l)
-500
0
1I91
Tkm (n"s)50
A)lI)
0
iTme (rns)
Light artifact at 1.5 mm
1
0
Instariteous firing rats at 1.5 rm
112C
C
6C
0.
-J
IN
B))
5
1000
Tme (mn)
0 We 0n) O
1000
Instanteous firing rate at 2.5 mm
Light art actat 2.5 m
f120
so
-J
C)I)
-500
6
0
Om)3oTim
C)ii)
too
I
Light artfact at 3.5 mm
0
Tom (m)
300
OW
Instanteous firing rate at 3.5 mm
.30
D)I)
-400
500
s1mm
D)li)
0
.
Is
a.
So0
Time (Msr0
s0
Figure 18: Local field potential recordings of representative neurons.
Local field potential recordingsof representativeneurons at the locations shown in Figure17 (left column)
and instantaneous spike time histograms (right). The red bar denotes when the laser is on. Note that the
virus injections were performed at depths of 1 mm relative to the surface of cortex to 3.5 mm relative to the
surface of cortex in 0.5 mm intervals. All of this data comes from single contact recordingsin Monkey C.
64
3.3.2 Electrophysiology
Illuminating neurons that express a halorhodopsin with light of the appropriate wavelength
suppresses neuronal firing (Han and Boyden 2007, Gradinaru, Thompson et al. 2008, Arrenberg,
Del Bene et al. 2009, Sohal, Zhang et al. 2009, Diester, Kaufman et al. 2011) and often leads to
rebound excitation after illumination ceases (Arrenberg, Del Bene et al. 2009, Tonnesen, Sorensen
et al. 2009, Tsunematsu, Kilduff et al. 2011, Brown, Tan et al. 2012, Madisen, Mao et al. 2012,
Schone and Burdakov 2012, Smith, Virkud et al. 2012). When the previously-injected area was
illuminated, both the expected neuronal suppression during the light pulse and the rebound
excitation post-illumination were observed. This is consistent with light reaching these neurons
and corroborates the LFP findings for the same locations. The left column of Figure 18 shows the
LFP at four locations spaced one millimeter apart from one another along the recording trajectory
for Monkey C shown in Figure 17. The right column of Figure 18 shows the instantaneous firing
rate for four example neurons recorded in Monkey C at the same locations. Figure 19 shows single
unit recordings and LFPs from Monkey L taken using a multi-contact electrode (Plexon, u-probe)
at depths denoted between the 'A' and 'B' points in Figure 17d.
65
Mean LFP Values for lilmitnated Channels
L*SWr
on
an
uwrobe
13
400
a)
4m, 0w
sic
loco
4O0100
Tbm (Ins) rel. to lser onset
kitanteous spke rats for Contact 11
Instenteous spike rat lo Contact 10
30
Lawe
-
Lawan
I20
0. 4 mo
e"IS)
rel
b)
m
Go
to Ia~n" e
m
we
m
C)
Instentuous spike rate for Contact 12
reF
-Law
Tnetms
oalownset
lnstataous spike rate for Contact 13
LAW
o
Lwon
30
20o
Ito
110
- M0
d)
a
200
raw
400 400
ON
1400
12W200
to la assnt t)m%)retDIsw~
a
ma
a m
logo
ima
Instanteous spike rate for Contact 15
Iistenteous spike rate for Contact 14
-cor"
n
Sin
9)
am oes) ra etoWo
Tm, (ts) rettotow
onset
Figure 19: LFP and instantaneous spike values for Monkey L measured using a multi-contact u-probe.
66
(Caption continuedfrom Figure 19 on previous page) Only channels 6-16 are included because channels
1-5 were superficial to the illuminatorand, therefore, not illuminated. (A) Mean LFP values for channels
6-16. All illuminated channels have a light artifactregardlessof whether there was neuron (colored traces)
or no single unit neuron present (gray traces) on the channel. (B) - (F) The instantaneous spike ratesfor
illuminated channels with a single unit neuron present. Note that the neurons on all channel were either
partly or completely silenced during illumination. The spacing between the contacts is 0.2 mm.
3.4 Discussion
This chapter presents a novel illuminator with diameter equal to that of a conventional fiber and
with a light-emitting surface area two orders of magnitude larger than that of a conventional
illuminator. This illuminator allows for dramatically increased illumination without increasing
the penetration diameter. Further, this illuminator can be used in the primate cortex to illuminate
tissue along a length of at least 3 mm and a volume of -10 mm 3.Most importantly, this illuminator
can silence neurons expressing a halorhodopsin (here, Jaws) over this same span, thus, facilitating
optogenetic silencing of tissue volumes relevant in primate neuroscience.
3.4.1 Potential challenges
The main challenge with this illuminator is mastering the fabrication technique. The experimenter
must learn what tension needs to be applied to the fiber during the heating process. Applying too
much tension will cause the fiber to snap while applying too little allows the fiber to curl upon
itself. Additionally, the experimenter must practice applying heat to the fiber in order to stretch
it enough to taper it, but not enough to curl or break it. Even after mastering fiber heating, the
experimenter must check the tip of each fiber to ensure that it is not curled or forked before
placing it into the brain. While the authors have experimented with automated methods of
heating and tapering the fiber (e.g., a pipette puller), none of these have provided results as
consistent as a practiced individual can achieve manually.
A second challenge involves maintaining the mechanical integrity of the illuminator after
fabrication. Unlike glass or silica optical fibers which break when they are mechanically overstressed or bent too sharply, plastic optical fibers curl or kink. If the illuminator presented here is
forcefully advanced toward the brain before the guide tube has penetrated the dura, the tip will
67
curl on itself. It is apparent when this occurs because the fiber will not enter the brain and will
not advance further through the guide tube. This curling destroys the tip of the fiber, but spares
the tissue. If the same force were applied to a glass fiber, it is possible that shards of glass would
be driven into granulation tissue, the dura and/or the brain. Secondly, kinks can develop in plastic
optical fibers due to overly sharp bending. Just as light leaks out of glass fibers when they are
bent or crushed, light will leak out from the kinked region of a damaged plastic optical fiber. Both
plastic optical fiber illuminators (such as this one) and glass fibers should be handled carefully
and examined regularly to ensure that they have not sustained mechanical damage.
3.4.2 Extensions and further applications
While this chapter presents illuminators with etched, light-emitting surfaces 3-5 mm in length,
longer etched tips are possible for studies requiring illumination over longer trajectories and
shorter tips could be manufactured for more focal illumination. Though this study uses 250 im
diameter fibers, other fiber diameters could be used depending on the desired application.
Further, one could use non-uniform etching to create "dead spots" where no light is emitted. This
would allow for light to be delivered selectively to one layer of cortex.
Because acrylic is more biocompatible than glass (Williams 2001) this illuminator could be placed
chronically in the brain with fewer complications than one would expect with traditional optical
fibers. Further, there is some evidence that more flexible electrodes (e.g., Tungsten micro-wires)
cause less tissue damage than rigid electrodes because there is less shear stress and tearing as the
electrode moves relative to the brain (Freire, Morya et al. 2011). Similarly, this more flexible plastic
illuminator may cause less damage chronically than a rigid glass fiber would.
Although the fiber was tested in the cortex, it is reasonable to assume that it could be used
successfully in deeper brain areas, such as the thalamus, amygdala, or subthalamic nuclei. While
neuronal silencing is demonstrated here using Jaws, this illuminator could be used with many
different opsins, including excitatory opsins, such as the recently developed Chrimson molecule
that is 45 nm red-shifted beyond previous opsins (including VChR2, C1V1, and ReaChR), and has
a spectrum similar to that of Jaws (Klapoetke, Murata et al. 2014). Finally, this illuminator
68
geometry has the potential to eventually be used in human applications where there is an even
greater need to illuminate larger brain volumes, should optogenetics be used in a clinical scenario
(Chow and Boyden 2013).
69
Chapter 4: Optogenetic modulation of memory-guided saccades reveals that visual,
delay-period and motor-related FEF firing activity contributes to behavior
Abstract
The previous chapters lay the groundwork for applying optogenetics to behavioral study in nonhuman primates. Light propagation in living tissue was measured to choose the best color for
large volume illumination (red) and, thus, the best opsin (Jaws). An illuminator specifically for
large volumes was developed. In this chapter, virus expression was evaluated while tissue
heating was addressed in Appendix 7. With the technology optimized, optogenetics could finally
be applied to more than just a proof-of-concept study. In this chapter, we present optogenetic
silencing of nearly every task-relevant neuron in a large region of frontal eye field (FEF). Error
rates increase with FEF inactivation during the memory-guided saccade task regardless of when
the inactivation occurred during the task. This demonstrates that visual, delay, and motor-related
firing in FEF all contribute to the proper execution of memory-guided saccades.
Keywords:
primate optogenetics, frontal eye field, saccade, memory-guided, oculomotor, halorhodopsin,
Glossary:
Jaws-a red-shifted halorhodopsin
FEF - frontal eye field
RF -receptive field
4.1 Introduction
The frontal eye field (FEF) is critical for executing saccades to remembered target locations. When
FEF is pharmacologically inactivated, subjects can no longer perform the memory-guided saccade
task (Dias, Kiesau et al. 1995, Dias and Segraves 1996, Sommer and Tehovnik 1997, Dias and
Segraves 1999); however, it is unclear from the firing patterns of FEF neurons why the saccade
cannot be executed. FEF neurons exhibit spatially selective visual responses (i.e., receptive fields),
70
which are larger and less well-defined than those in primary visual cortex (Mohler, Goldberg et
al. 1973). Populations of FEF neurons also exhibit motor fields, which correspond to the spatial
tuning of firing rate in preparation for a saccade to a particular part of the visual field (Bizzi 1968).
Visual receptive fields and motor fields generally overlap within FEF (Bruce and Goldberg 1985,
Bruce, Goldberg et al. 1985). In addition to visual and motor-related activity, FEF neurons often
have sustained, spatially-specific increases in their firing rate throughout the delay period of a
memory-guided saccade task (Bruce and Goldberg 1985, Hanes, Patterson et al. 1998). Taking
these results together, it is unclear if FEF inactivation prevents subjects from executing memoryguided saccades due to a lack of visual activity (inability to see the target), a lack of delay activity
(inability to remember the target location), a lack of motor activity (inability to move one's eyes
to the target), or some combination of these.
This study used temporally-precise optogenetic techniques to inactivate neurons FEF during only
one epoch of the memory-guided saccade task in each trial. This approach is conceptually
straightforward, and it disambiguates the behavioral contributions of FEF neuronal firing at
various times during the memory-guided saccade task. However, this approach is technically
challenging because much larger tissue volumes must be inactivated than are typically
modulated with optogenetics in other species (e.g., rodents), as discussed in Chapter 1.
Acute inactivation of a large tissue volume seems to best allow for observation of FEF-related
saccade deficits. With either ablation or pharmacological inactivation, saccade deficits become
more profound as larger volumes of FEF are affected (Schiller and Tehovnik 2003); however, over
time, saccades return to normal even after ablation (Schiller and Chou 1998). Dias and Segraves
(1996) write, "Acute [pharmacological] inactivation of the FEF of monkeys.. .produced much
more severe oculomotor impairment [as compared with ablation]. This difference is probably due
to the acute nature of the Muscimol effect, which does not allow time for reorganization of the
control of eye movements before testing begins." (Dias and Segraves 1996) Optogenetic
inactivation is orders of magnitude more temporally precise than pharmacological inactivation
(i.e., on the order of milliseconds rather than hours); however, inactivating a large enough fraction
of neurons over a large enough tissue volume is a technical challenge for primate optogenetics.
71
Pharmacological inactivation studies suggest that in order to disrupt memory-guided saccades
nearly every FEF neuron in a tissue volume of -10 mm 3 must be inactivated (i.e., the firing rate
must be reduced by > 80%). When Tehovnik and Sommer (1997) studied the effects of Lidocaine
in macaque FEF, they determined that silencing nearly all 100% of neurons within a 1.5 mm radius
of the injection site (-14 mm3) prevented saccade execution to the effected receptive field
(Sommer and Tehovnik 1997, Tehovnik and Sommer 1997). Martin et al., (1999) and Dias et al.
(1999) both reported that silencing 100% of neurons in a 1 - 1.5 mm (4.2 - 14 mm2 ) sphere of FEF
with either Lidocaine or Muscimol is sufficient to disrupt memory-guided saccades as well.
Inactivating -80% of neurons by cooling part of FEF also can disrupt saccades (Chafee and
Goldman-Rakic 2000). Thus, the technical goal for this study was to reduce the firing rate of at
least 80% of neurons by >80% over at volume of ~10mm 3, as determined using micro-stimulation,
the standard technique for mapping receptive fields in FEF, and confirming this with
electrophysiology.
A novel illuminator (Chapter 3) was invented for this study and paired with a red-light sensitive
halorhodopsin, Jaw, (Chuong 2014). A red-light sensitive opsin was selected rather than
conventional green-light sensitive silencer based on in vivo light propagation measurements
(Chapter 2). Together these allowed for inactivation of volumes of tissue comparable to those
reported in pharmacological inactivation studies.
4.2 Methods
4.2.1 Subjects
Two healthy, male rhesus monkeys (Macaca mulatta) weighing 13-16 kg were used for the
behavioral study. Two male rhesus monkeys weighing (Macaca mulatta) 10-11 kg were used for
histological studies. The histology monkeys were selected on the basis of pre-existing, worsening
medical conditions (e.g., metastatic cancer) for which veterinarians recommended euthanasia
within a few months.
All procedures and animal care were in accordance with the NIH
guidelines and were approved by the Massachusetts Institute of Technology Animal Care and
Use Committee.
72
4.2.2 Surgical procedures
Anatomical MRI images taken prior to surgery were used determine the stereotactic coordinates
for chamber placement in behavior subjects. Both monkeys already had a titanium head post
surgically affixed to the skull. Under Sevoflurane anesthesia, a recording chamber (Crist) was
stereotactically placed over FEF in each monkey based on the each monkey's MRI. After
surgically affixing the chamber to the skull, a craniotomy was made inside the chamber.
4.2.3 Localizing injection targets in behavior monkeys
An ultem grid (custom) with 1 mm hole spacing was placed in the recording chamber during
awake testing. A screw on the side of the chamber fit into a notch on one side of the grid to ensure
that it was oriented consistently across recording sessions. An anatomical MRI was performed
with the grid in the well, which was filled with a surgical lubricant (SurgiLube). This MRI image
was processed offline in software (Amide) and rotated such that estimates of the distance from
the top of the grid to various points in the brain could be made along the grid holes.
An x-y stage (NAN) was placed over the recording chamber and fixed in a consistent orientation.
Micro-drives (NAN) were mounted on the x-y stage and used for micro-stimulation,
electrophysiological recording, and, later, for in vivo illumination. A 220 tm, parylene-coated,
tungsten microelectrode (Nimer Lab) was lowered through a 25 gauge guide tube to MRI-derived
depths spanning the thickness of cortex, based on the MRI image. One grid hole per day was
tested in order to isolate the exact location of the frontal eye field within the craniotomy.
4.2.3.1 Eye tracking
Subjects performed a central fixation task while seated 57 cm in front of a CRT computer monitor
(resolution of 1024 x 768 pixels and refresh rate of 75 Hz). The MATLAB-based MonkeyLogic
software suite (Asaad and Eskandar 2008, Asaad and Eskandar 2008, Asaad, Santhanam et al.
2013) controlled the task and recorded the eye position, which was monitored at 250 Hz using a
video-based eye tracking system (Eye Link II, SR Research).
73
4.2.3.2 Micro-stimulation
Between trials, the experimenter manually-triggered electrical stimulation via a TTL input to a
digital stimulator (DS8000, WPI) and stimulus isolation unit (DSL 100, WPI). The current output
was constantly monitored via an oscilloscope (Tektronix). Figure 20 shows the electrical circuit
used for stimulation. A train of 100 bipolar pulses was applied with a frequency of 250 Hz, and
a total pulse width of 0.2 ms (0.1 ms depolarization followed by 0.1 ms hyperpolarization) via a
220 tm, parylene-coated, tungsten microelectrode (Nimer Lab, WeSense). Applied currents
ranged from 10 - 300 tA. Eye position was monitored during stimulation via an Eye Link II
system and recorded in MonkeyLogic. The frontal eye field was defined as the region of cortex in
the vicinity of the principal and arcuate sulci where fixed-vector, saccades could be evoked with
current < 150 FtA at least 50% of the time, a standard that other groups have employed (Bruce,
Goldberg et al. 1985, Moore and Fallah 2004, Maunsell and Treue 2006, Murphey and Maunsell
2008). Typically, thresholds were lower, though, around 50 tA or less. This study specifically
sought to find areas of FEF in which micro-stimulation evoked saccades with eccentricities of
approximately 10 visual degrees.
74
Manual trigger
10 k
Wl
Oscilloscope output across 10kD
Connects to
ground
(guide tube)
Connects to
electrode
* voltL
500 ps
Figure 20: Circuit used for electrical micro-stimulation
4.2.3.3 Recording frontal eye field neurons
Neurons in FEF have a well-characterized range of firing responses during a memory-guided
saccade task (Bruce and Goldberg 1985), as shown in Figure 2. Neuronal recordings were made
in the vicinity of FEF coordinates identified during micro-stimulation while the monkey
performed a memory guided saccade task. These recordings confirmed that the proposed
75
injection sites identified during micro-stimulation7 and contained neurons with the firing profiles
expected in FEF (e.g., with visual, delay period, and motor activity), and that opsin injection and
subsequent silencing should be able to affect all of the three main neuronal responses.
4.2.3.3.1 Recording set up
Spikes and LFPs were recorded using a multi-channel acquisition processor system, or MAP box
(Plexon) via a single contact, 220 rim, parylene-coated, tungsten microelectrode (Nimer Lab or
WeSense) coupled to a pre-amplifier (Plexon) via a headstage (Plexon) and electrical connector
(Omnetics). The pre-amplifier split the data into spike channels (0.25 -8 kHz) sampled at 40 kHz
and local field potential channels (0.7 - 170 Hz) sampled at 1 kHz. Pre-amplifier output went into
a MAP box, (Plexon) where it was filtered and acquired using the Rasputin (Plexon) software.
Spikes were sorted offline using principle component analysis and manual waveform shape
analysis (Offline Sorter, Plexon). MATLAB (Mathworks) was used for further analysis and
plotting.
4.2.4 Virus injections in behavior primates
The grid holes and depths for virus injections were determined using micro-stimulation and
recording during a memory guided saccade task, as described above. The results of microstimulation are found in section 4.3.2 and shown in Figure 24. Monkey C was injected along
trajectories in two grid holes while Monkey L was injected along trajectories in three grid holes.
The injected locations were determined solely by the physiological map (determined via microstimulation and electrophysiology) in each monkey.
receptive fields and motor fields generally overlap within FEF, but they are not necessarily a perfect
match, particularly at the level of single neurons Bruce, C. J. and M. E. Goldberg (1985). "Primate frontal
eye fields. I. Single neurons discharging before saccades." I Neurophysiol 53(3): 603-635, Bruce, C. J., M. E.
Goldberg, M. C. Bushnell and G. B. Stanton ibid."Primate frontal eye fields. II. Physiological and anatomical
correlates of electrically evoked eye movements." 54: 714-734.. We did not necessary see a perfect match
here either, but the response were to the same side and fit within the range of point we would expect.
7Visual
76
Virus injections were performed under general (Sevoflurane) anesthesia. Dexamethasone was
administered prior to virus injection to prevent brain swelling and to potentially improve virus
uptake into neurons. The grid used during recording and micro-stimulation was placed in the
well during surgery.
4.2.4.1 Virus preparation
The virus (AAV8-hSyn-Jaws-GFP) was thawed on wet ice, diluted 1:10 with sterile phosphate
buffered saline (pH 7.4, Life Technologies), centrifuged and loaded into the syringe as described
in the histology section below (Section 4.2.11). The virus-loaded syringe was placed in the microsyringe injector pump on a stereotactic arm and aligned such that its trajectory moved directly
along the desired grid hole. The depth when the injection needle reached the top of the grid was
recorded so that the depth for injection could be calculated.
4.2.4.2 Injection sites
For Monkey C, cortical trajectories along two adjacent grid holes were injected. Along one
trajectory, 0.8 iL of diluted virus was injected at each of five sites, spaced 0.6 mm apart. Along a
parallel trajectory 1 mm away, 0.8 VL of diluted virus was injected at each of four sites spaced 0.6
mm apart. The deepest site along each trajectory was injected first. The injection needle remained
at each injection site for 2 minutes before it was retracted 0.6 mm to the next injection site. For the
last, most superficial injection site along a given trajectory, the injection needle remained in place
for at least 20 minutes post-injection.
For Monkey L, trajectories of cortex along three grid holes were injected. Along the first trajectory
(G-4), 0.4 1iL were injected at each of 8 sites spaced 0.6 mm apart. The amount of virus per site
was reduced in order to maintain an equivalent total volume of virus across both behavior
monkeys. Along the second trajectory (H-3), 0.4 iL of virus was injected at each of 5 sites spaced
0.6 mm apart. Along the third trajectory (H-2), 0.4 tL were injected at each of 5 sites spaced 0.6
mm apart. As in the Monkey C, the injection need remained in place for at least 20 minutes before
being removed from the brain.
77
4.2.5 Behavioral testing
Each monkey was were seated in front of a CRT monitor which was controlled by the
MonkeyLogic software package, as described above. Prior to each testing session, an eye
calibration was performed using the built-in function in the MonkeyLogic software (Asaad and
Eskandar 2008, Asaad and Eskandar 2008, Asaad, Santhanam et al. 2013). Figure 21 summarizes
the primary behavioral task used in this study.
1) Visual?
2) Delay?
3) Motor?
Dirpto
114MIS
Initial fixation:
300400 ms
Target:
looms
Delay:
500-1000 ms
Saccade:
<Sooms
Figure 21: Task used during testina
4.2.5.1 Memory-guided saccade
Both behavior monkeys were trained to perform a memory-guided saccade task. The overall task
setup is shown in Figure 21. The trial began with central fixation on a white spot (radius 0.50) for
a randomly determined duration of 300-400 ms. Next, while the central fixation spot remained
on the screen, a peripheral stimulus (also a white dot with a 0.5* radius) flashed for 100 ms in one
of four pre-determined locations, all with an eccentricity of 10 visual degrees. Three possible
stimulus locations were located on the side contralateral to the virus injection. The other target
78
was located ipsilaterally. The target locations with both laser and sham trials are shown in Figure
22. Target selection for a given trial was random. Non-laser sham locations were not tested in
Monkey L because when these were used, Monkey L refused to make any saccades to the injected
target location, presumably recognizing that saccades to the injected location were more difficult
to execute than saccades to other sites (due to inactivation). Both monkeys maintained central
fixation during the stimulus flash and for another 500-1100 ms until the central spot was
extinguished. The disappearance of the central spot served as a "go-cue" for the monkey to make
a saccade to the remembered location where the stimulus had flashed. A reward was given for a
saccade that fell within 30 of the flashed stimulus.
Monkey L
Monkey C
Figure 22: Injection fields and target presentation locations for both monkeys.
All possible target locationsare shown above; however, only one target appearedper trial. The central
fixation dot appearedon every trial. Both the target and the fixation dot were shown as white 1 degree
diameter circles. In the diagram above, targets that were presented only with sham illumination are
shown in white; those with both sham and laser illumination are shown in red. All targets had an
eccentricity of 10 degrees relative to the centralfixation point.
4.2.5.2 Laser and sham shutter opening during the task
At one of three possible points in each trial, a shutter (either the laser shutter or an identical
"sham" shutter) opened for 300 ms. In a third of trials, one of the two shutters opened at the same
79
time as the peripheral cue appeared. In a different third of trials, one of the shutters opened 200
ms after the peripheral cue disappeared (during the delay period). In the final third of trials, one
of the shutters opened at the same time as the central fixation spot disappeared. The "sham"
shutter was three times as likely to open as the laser shutter. This both increased subject morale
and allowed a large enough average time between laser trials to prevent tissue heating (as
discussed in Appendix 7).
4.2.5.3 Optical hardware set up
A 635 nm, 500 mW DPSS laser (SLOC) and two identical mechanical shutters (Oz Optics) were
controlled via TTL pulses from a pulse generator (DS8000, WPI) with timing controlled by a
computer running the MonkeyLogic software package.
The laser was coupled to one of the electrically-controlled fiber optics shutters via 200 im
diameter optical fiber (provided with the laser by SLOC). The output of the shutter was coupled
to a different 200 pm diameter optical fiber (ThorLabs). Any part of the optical fiber that extended
into the primate box was shielded with light-absorbing, black electrical tape (3M) to prevent any
light from being visible to the primate through the shielding. The optical fiber was coupled to a
large-volume illuminator (described in Chapter 3) with a ceramic sleeve (Precision Fiber
Products, Inc.). This coupling was located behind the primate's chair and shielded with optically
absorbing foil (ThorLabs) to prevent any light loss at the coupling from becoming visible to the
monkey. The drives and stage mounted on the recording chamber were shielded with optically
absorbing foil as well. The large-volume illuminator was lowered into the brain via guide tube
that penetrated the dura but remained just above the cortex. The experimenter inserted the largevolume illuminator into the brain by hand because the illuminator is too flexible to be lowered
into the brain by either an electro-mechanical or screw drive. The large volume illuminator was
affixed to a micro-drive (NAN), which retracted the fiber retracted at the end of experiments.
While the opening of the shutter was clean (albeit with a slight delay) the physical closing of the
shutter was not immediate and seemed to "bounce" shut, meaning that it would close fully then
slightly reopen, only to close again as shown in Appendix 1. To ensure a laser pulse with crisp
80
onset and offset, the laser was disabled for 150 ms at the end of the pulse. Thus, sharp light offsets
were maintained even if there was some "bounce" in the shutter. Appendix 1 describes the shapes
of the laser pulses in more detail.
All of the hardware that interfaced with the optical fiber was shielded with optically-absorbing,
black-painted foil (ThorLabs), and the setup with visually inspected before every testing session
to make sure that no stray light would be visible to the monkey. Further, as an additional
prevention against visual light cues, a bank of 5 red LEDs (DigiKey) was placed in the testing box
and flashed continuously at 2.5 Hz (50% duty cycle). This prevented the monkey's eyes from
adjusting to darkness and was an extra precaution against any light leakage during testing. The
monkeys were trained with this flashing light prior to any kind of illumination to ensure that it
did not impact their behavior.
4.2.6 Simultaneous recording and optical inactivation
In Monkey C, a single contact electrode was lowered into the cortex with a micro-drive (NAN)
through 25 gauge guide tube that penetrated the dura. In Monkey L, a 16 channel multi-site, linear
electrode (Plexon U-Probe) was used instead of the single contact electrode to allow for better
characterization of neuronal populations with fewer penetrations. The same method of lowering
and driving the electrode was used in both primates. The electrode was always lowered through
a grid hole adjacent and parallel to (i.e., 1 mm away from) the grid hole through which the optical
fiber was lowered.
The output from each electrode was passed through a high-impedance headstage prior to
extracting the spike and LFP components via the pre-amp (Plexon). The spike channel ranged
from 250 Hz to 8 kHz, and was sampled at 40 kHz. The LFP channel is defined as 0.7 Hz to 170
Hz. Spikes were sorted offline in software (Offline Sorter, Plexon Inc.). Data analysis was
performed using MATLAB.
81
4.2.7 Classification of behavioral errors
A saccade error occurred when the monkey failed to make a saccade to the correct target. We
specifically considered trials in which the monkey maintained fixation until the central spot
disappeared. If the monkey failed to move his eyes out of the central fixation window within 500
ms of the go-cue, a "no saccade" error occurred. If the primate attempted to make a saccade, but
did not move his eyes to the correct target location, an "incorrect" error occurred. 8 If the trials did
not end in an error or correct saccade, the monkey either did not initiate the trial or broke fixation
at some point during the task.
4.2.8 Electrophysiological analysis
This section describes the statistical measures and terms that were used to analyze the firing rate
data presented in the subsequent section.
4.2.8.1 Population-level inactivation
For each neuron, the average firing rate (spikes / second) was calculated during the illumination
period in laser trials and during the equivalent time period in control trials for every neuron. A
paired sample student's t-test was performed on the average laser and control firing rates for all
neurons (or subgroup of neurons) for each monkey.
4.2.8.2 Inactivation of individual neurons
The change in firing rate with illumination was assessed in individual neurons as follows. The
number of spikes during the illumination period (or control period) was calculated for each trial
in each neuron. The firing rate was normalized based on the location of the target and the period
of the trial in which the firing rate was measured (i.e., target v. delay v. go-cue periods).
8 These "incorrect" saccades can be further sub-classified into a) saccades to one of the other possible target sites, b)
saccades that approached, but failed to reach, the correct target, and c) saccades that veered off into space, not approach
any of the possible targets.
82
Specifically, the average firing rate for the target location and task period for control trials was
subtracted from the firing rate calculated for during the test period for each trial. This difference
was divided by the standard deviation of the control trial firing rates for this testing period and
target location. The distribution of normalized trial-by-trial firing rates for given neuron without
illumination was compared with the normalized trial-by-trial firing rates for a given neuron
during illumination. An f-statistic was calculated for each comparison to determine whether the
variances were equal at the a = 0.95 level and the appropriate student's t-test was applied to the
distributions at a 95% confidence level.
4.2.8.3 Classification of neuronal subpopulations based on firing rate response profiles
Neurons were classified based on whether their firing rate increased at different task times. The
firing rate of "visually-responsive" neurons increased significantly when a target was presented
in the associated receptive field. Specifically, in visually-responsive neurons, the firing rate in the
period 50 ms to 200 ms after the presentation of the target was significantly higher than the
baseline firing rate, 200 to 50 ms prior to target presentation. The firing rate of "delay-responsive"
increased significantly during the delay period (relative to the baseline). Specifically, delayresponsive neurons had a significant increase in firing rate for the period 350 to 500 ms after the
target offset as compared with the baseline firing rate (again determined from 200 ms to 50 ms
prior to target presentation). A paired sample student's t-test was used once again with a
significance level of 0.95. The firing rate of "Motor-responsive" increased during motor
preparation. Each neuron was tested for a significant increase in firing at its preferred target
location at each task time using a paired sample student's t-test at a 95% confidence level. Motor
responsive neurons had a significant increase in the firing rate during the 100 ms window prior
to the start of the saccade. The rate during this motor preparation period was compared against
the firing rate calculated for the 100 ms window prior to the go-cue.
4.2.9 Quantification of rebound
The number of spikes in the 50 ms period from 20 ms after the end of the laser pulse to 70 ms
after the end of the laser pulse was measured and average across all trials for each condition,
83
neuron, and monkey. This time period was used to exclude silencing while fully encapsulating
the rebound for all neurons. The average number of spikes in this period was calculated for
control trials that used a sham shutter as well. For each neuron, the average number of spikes in
the rebound period across conditions for laser and control trials was taken by weighting the percondition averages by the frequency of each condition and calculating the weighted mean. The
average number of spikes during this period in the laser trials was subtracted from the average
number of spikes during the rebound period in control trials to yield the average increase in the
number of spikes in the rebound period for each neuron.
4.2.10 Viral vector screening
A large screen of opsin-containing viral vectors was performed to identify which virus(es)
maximized neuronal opsin expression without compromising neuronal health. Two injections of
each of 12 different viral vectors were performed in the cortex of monkey SM under aseptic
surgical conditions. All 12 of the vectors were injected in duplicate in distant locations. To prevent
expression overlap, each injection site was separated from all other injection sites by at least 10
mm in all directions.
4.2.10.1 Surgical procedures for viral vector screening
Anesthesia was induced with Telazol, and then continued with Sevoflurane. Prior to surgery,
Dexamethasone and Famotidine were administered. IV fluids (5% Dextrose) were infused
throughout surgery, Antibiotics and painkillers (Buprenorphine) were administered intraoperatively. Craniotomies were performed bilaterally at stereotactic coordinates determined
from a pre-op anatomical MRI. The bone cap of the craniotomy was preserved in antibiotic and
replaced at the end of surgery.
4.2.10.2 Intra-operative injection preparation
Injection syringes (10 1iL gas-tight, Hamilton) were pre-loaded with 5 1 iL of sterile silicone oil
(Sigma). Each syringe was mounted on a UMP3 micro-syringe injector pump (WPI Inc.). The
plunger was depressed until a bubble of silicone oil formed at the tip of the injection needle. Next,
84
aliquots of virus were removed from dry ice, quickly thawed on wet ice, and centrifuged at 40 C
and 5k rmp for at least 5 minutes. Immediately after centrifugation, the injection needle was
lowered into the aliquot and 2.5 tL of thawed, centrifuged virus was drawn up at a rate of 1 1iL
/ minute. The syringes were visually inspected for bubbles during and after loading the virus.
4.2.10.3 Virus injection
The micro-syringe injector pump with virus-loaded syringe was mounted on a stereotactic arm
(Kopf). The tip of the needle was moved to the appropriate location on the surface of the cortex.
The plunger of the syringe was depressed prior to the needle penetrating the cortex, and a small
amount of virus -0.1 - 0.3 tL was forced out. A droplet formed on the tip of the injection needle,
forcing out any bubbles present in the needle prior to injection. Next, the injection needles were
each observed to enter the brain. The coordinate at which the needle penetrated the brain was
recorded.
For each test injection, 2 VL of virus were infused into the brain at either one site (if cortex were
thin in this region) or two sites along the same trajectory. When two sites were injected, the deeper
site was injected first with 1 iL. The needle was retracted 1 mm and the second site was injected
with 1 iL. Virus was injected at a rate of 0.1 iL/minute at all sites. Injection needles remained in
the brain for at least 10 minutes post-injection. The sites were injected four at a time, using four
different stereotactic arms, pumps, and syringes.
4.2.11 Histology
Five weeks after the virus injections, Monkey SM was sedated with Ketamine (10 mg/kg,
subcutaneous) and then transported to the perfusion space where a pentobarbital overdose (>10
mg / kg, IV) was administered. Trans-cardiac perfusion was performed with 2 liters of phosphate
buffered saline (pH 7.4, room temperature) followed by 4 liters of 4% paraformaldehyde
(prepared the day of the perfusion in PBS, room temperature). A gravity perfusion system (1
meter height) and 18 gauge perfusion cannula were used to match the perfusion solution flow
rate to that of normal blood flow. The brain was removed and blocked immediately after fixation
85
and then soaked in 4% paraformaldehyde for ~36 hours. Next, the tissue was blocked and cryoprotected in a glycerol sinking solution (25% glycerol + 0.1% sodium azide in O.1M phosphate
buffer) and kept on a shaker at 4C for 3-4 days. Coronal slices (80 Im) thickness were made on a
sliding microtome and stored in 0.1M phosphate buffer + 0.1% sodium azide at 4C. All tissue
processing occurred at room temperature on a shaker unless otherwise noted.
4.2.11.1 DAB staining and imaging
Every 611 section (every 480 rim) was washed and permeabilized 3 times (10 min / rinse) in PBSTX (0.01M PBS + 0.2% triton X-100). The sections were incubated in 3% hydrogen peroxide in PBSTX for 10 minutes, and then rinsed 3 more times (10 min / rinse) in PBS-TX. The sections were
blocked for 1 hour in blocking solution (10% normal goat serum in PBS-TX) and incubated
overnight at 4*C in a 1:2000 dilution of polyclonal rabbit GFP antibody (Invitrogen) in blocking
solution.
The next day, the sections were rinsed 3 times (10 min / rinse) with PBS-TX and incubated, at
room temperature for 1 hour, in a 1:500 dilution of biotinylated goat anti-rabbit IgG (Vector) in
blocking solution. The sections were subsequently rinsed 3 times (10 min / rinse) in PBS-TX and
incubated in an avidin/biotin (Vectastain ABC) solution for 1 hour. After 3 more (10 min / rinse)
with 0.1M phosphate buffer, the sections were incubated with 3,3'-diaminobenzidine, hydrogen
peroxide, and a nickel solution (DAB Peroxidase Substrate Kit, Vector Labs) until just exposed.
The sections were rinsed 3 more times (10 min / rinse) with 0.1 M phosphate buffer. The sections
were then mounted onto gel-subbed slides and allowed to dry. Once dried, the slices were placed
in serial dehydrations of 70%, 95%, and 100% ethanol each for 2 minutes. Next, the slides
progressed through three washes with Xylenes, each lasting 5 minutes. Finally, the slides were
cover-slipped with Eukitt.
DAB stained sections were visualized under light microscopy. The DAB staining allowed for each
of the virus injection sites to be mapped to a range of tissue sections, based on both the anatomy
of the sections and the presence of DAB label.
86
4.2.11.2 Fluorescent immunohistochemistry
Sections adjacent to those that stained strongly with DAB were selected for fluorescent
immunohistochemistry. Sections were double-stained. An anti-GFP stain was used to assay for
virus expression while an anti-NeuN stain was used to label neurons. Selected sections were
rinsed 3 times (10 min / rinse) with PBS-TX and then blocked for 1 hour in 10% normal goat serum
in PBS-TX. Next, the sections were incubated at 4C overnight in primary antibody. The primary
antibody consisted of both a 1:2000 dilution of polyclonal rabbit anti-GFP (Invitrogen) and a
1:5000 dilution of monoclonal mouse anti-NeuN (Invitrogen) mixed together in PBS-TX + 10%
normal goat serum.
Next, sections were rinsed 3 times (10 min / rinse) with PBS-TX at room temperature. The well
plates were covered in aluminum foil (Reynolds) to block out ambient light. The sections were
incubated in secondary antibody for 3 hours on a shaker at room temperature. The secondary
antibody consisted of 1:250 dilutions of Alexa Fluor 488 goat anti-rabbit (Invitrogen) and Alexa
Fluor 633 goat anti-mouse (Invitrogen) mixed together in PBS-TX + 10% normal goat serum. With
the aluminum foil in place, the sections were rinsed in 0.1 M phosphate buffer 3 times (10 minutes
/ rinse) and mounted on to gel subbed slides under dim light. The slides were left to dry in
complete darkness. They were then cover slipped in dim light using Vectasheild + DAPI and the
edges of each slide were sealed with clear nail polish. The slides were stored in an opaque
container at 4C.
4.2.11.3 Fluorescent imaging and cell counting
Confocal imaging was performed on slides that underwent fluorescence immunohistochemistry.
Images were taken in 3D over a depth of 50 tm in 2 Itm increments with a 20x lens (0.45 mm x
0.45 mm area). Cells with neuronal label (NeuN) or GFP label were counted offline using ImageJ
(NIH). In the counting process, two researchers independently counted cells and digitally noted
their labeling. Afterward, the researchers compared their digital labels to reach a consensus on
cell classification. Using these classifications, the spatial extent of labeling, the fraction of neurons
87
labeled, the specificity of each viral vector for neurons, and the health of cells infected with
different viral vectors was determined.
The anterior-posterior spread of expression along the surface of the cortex was determined by
measuring the number of coronal sections that had expression in cells in the vicinity of an
injection site. The lateral-medial spread of virus was measured based on the area of expression
on a single slide at the center of the injection site. The fraction of labeled neurons at the center of
the injection site was determined by dividing the total number of neurons by the number of
neurons expressing GFP. Further, these counts were broken down by layer of cortex.
To assess for cell death, the number of total neurons per unit area in the injection site was
compared against the total number of neurons per unit area in adjacent sites that was not injected.
If there were fewer neurons in the injection site, it was assumed that some kind of cell death had
occurred. In the case of frank necrosis, tissue may take on a "Swiss cheese" appearance. Other
signs of poor cell health include blebbing or granular inclusions in the soma or along dendrites.
To control for possible variability in surgical technique, injections were performed with the same
techniques and equipment by the same experimenter, and each virus was injected in two distinct
sites.
4.2.11.4 Histology matched to viral vector and brain region
A second histology monkey, R, was injected in the frontal eye field with AAV8-hSyn-Jaws-GFP
in a surgical procedure similar to that described for the behavior monkeys. Tissue was collected
8 weeks after the injection in the manner described above. Tissue processing and analysis also
was the same as described previously.
4.3 Results
4.3.1 Histology
Virus injections with AAV8-Jaws-GFP yielded expression in > 80% neurons across layers II - V
(Figure 23). Virus injection did not lower the number of neurons per unit area, based on
88
comparisons of the neuronal density within the injection site with the neuronal density of
adjacent, un-injected sites.
Figure 23: AAV8-aws-GFP injection of 1 uL / site at two sites in a histology primate.
4.3.2 FEF was identified using micro-stimulation
FEF was functionally identified in both primates at sites with micro-stimulation thresholds
150
VA. The evoked saccades resembled those in the literature, e.g., (Bruce and Goldberg 1985).
Examples of evoked saccades trajectories are shown in Appendix 2. Micro-stimulation was
performed every 0.25 mm along two trajectories (each spanning -4 mm) in monkey C and along
three trajectories (each spanning 3-4 mm) in monkey L. The end points of evoked saccades at
locations spanning all injection trajectories in both primates are shown in Figure 24. Specifically,
each dot in Figure 24 corresponds to the mean end point of saccades evoked at a single site along
the injection trajectories for each primate. The starting point of these saccades was normalized to
be (0,0). Figure 24 illustrates the receptive fields that were injected with virus and subsequently
inactivated in both primates.
89
Evoked saccade endpoints
20
* Monkey C
15 [ RF M onkey C
.
*
Monkey L
I f~ L
F
m 15
05 0
-5 F
I-S
0
RF Monkey L
&10
-15F
-20
-15
-10
-5
0
5
10
Degrees of visual angle
15
20
Figure 24: The end points of saccades
The end points of saccades evoked using micro-stimulationin the injection sites in both monkeys. All end
points are relative to a saccade that startedin the centralfixation location (0, 0) denoted with the cross. The
possible target locations, each with an eccentricity of 10 degrees are shown as black circles. The general
receptivefields (RFs)for both monkeys are circled and shaded in red (monkey C) or blue (monkey L).
4.3.3 Spatial specificity of neuronal responses
The neuronal responses recorded in the two monkeys were specific for the receptive fields that
were determined using micro-stimulation prior to viral vector injection. Figure 25 shows an
example of the firing profile of a neuron recorded during a memory-guided saccade task session
in Monkey C. In this task, the direct left and direct right target locations (y = 0) were used for
Monkey C and Monkey L, respectively, as the "injected" sites. Visual receptive fields and motor
fields generally overlap within FEF, but they are not necessarily a perfect match, particularly at
the level of single neurons (Bruce and Goldberg 1985, Bruce, Goldberg et al. 1985). For Monkey
90
C, the receptive fields tended to be a bit lower than the motor fields. Further, by placing both the
"injected" target location and the "opposite" target location along the horizontal axis, we
illuminated the possible confounding factor of perceptual asymmetries in the upper v. lower
visual field (Thomas and Elias 2011).
bw
I
180 deg (anti-RF)
+
I
0 deg (RF)
-A
E M
5 400 ms,
Target on
-I
IdwAMWOR
Saccade start
Figure 25: Firing profiles
Firingprofiles for targets at the 8 locations shown (all with eccentricity of 10 degrees relative to central
fixation point, denoted by cross). The receptivefield (RF) demonstrates an elevatedfiring rate throughout
the delay period and ending just prior to the start of the saccade.
91
4.3.4 Neurons with visual, delay, and motor activity were recorded
Specifically, we sought to confirm the presence of three classes of FEF neurons: a) neurons with
firing rate increases in response to visual stimulation in the target area, b) neurons with firing rate
elevation during the delay period, and c) neurons with firing rate elevation during saccade
preparation and execution. These are referred to as visual, delay, and motor neurons
subsequently, and the classifications are not mutually exclusive as described in section 4.2.8.3.
Figure 26 shows the response profile of a "visual" neuron while Figure 27 shows two neurons
that would both be classified as having "visual," "delay", and "motor" activity. This illustrates
the variability of response profiles within neuronal subclasses.
Fixatior
Target
I
Target
not in RF
50
Target
40
in RF
30
-20
10
p
'a
:::
y-vvvY~
-0.5
___________________
0
Time from target
presentation (s)
0.5
//
"
~
-0.5
11me from
saccade start (s)
Figure 26: A neuron with only a visual response.
92
0.5
Fixation
Fixatio r
Target
Targe t
t
25
-
20
Target
not In RF
2
Target
InRF
20
5
10
0
5
5
0
Time from target
presentation (s)
0.5
-05
Time from
saccade start (s)
0.5
Target
in IF
,
5not
15
-0.5
E
-0.5
-
0
Time from target
presentation (s)
.S"0.0
Timefrom
saccade start (s)
Targe t
inRF
0.5
Figure 27: Two neurons with visual, delay, and motor-related activity.
In Monkey L, 67 FEF task-modulated neurons were recorded. In Monkey C, 52 task-modulated
units were recorded. In Monkey L, 62 units had an increase in firing associated with visual
stimulus presentation, 60 units had an increase in firing associated with the delay period, and 39
units had a motor-related increase in firing rate. In Monkey C, 39 units had a visually-related
increase in firing rate, 36 units had a delay-related increases in firing rate, and 33 units had a
motor-related increase in firing rate. These ratios are summarized in Figure 28.
93
100
C
0
:
90
Monkey L
80
Monkey C
7070
W
60-
4-..
0
)
50
'J 30Q-
20-
10
0
Visual
Delay
Motor
Figure 28: Distribution of response types among task-responsive FEF neurons.
A given neuron may respond to more than one type of stimulation (e.g., "visual" and "motor"). Percentage
of task-responsive neurons that increase theirfiring rate during each epoch of the memory-guided saccade
task.
4.3.5 FEF neurons were optogenetically inactivated in both monkeys
The firing rate for neurons in the injected region of FEF decreased significantly with illumination
in both monkeys. The decrease in firing rate was significant for all subtypes of neurons (i.e.,
visual, delay, and motor) at all task times (i.e., during the target, delay or go-cue periods)
regardless of whether or not the target was in the receptive field or opposite to it.
4.3.6 Overall population level firing-rate analysis
On the population-level, the decrease in firing rate with illumination was significant at a level of
p < Ie -10 using a paired sample student's t-test for each monkey. Figure 29 shows the distribution
of firing rate decreases for both monkeys in all task-modulated neurons. In both monkeys, the
firing rate in > 2/3 of units decreased by > 80%. Previous pharmacological studies, defined a
94
neuron as "inactivated" if its firing rate decreases by >80%. None of FEF neurons increased their
firing rate during illumination.
70
70
Monkey L
En
60
60
C
50
so
C
40
IV 40
30
30
20
20
10
10
Monkey C
&A
#A
0
IV.
C
0
0
20
40
60
8
0
100
Percent decrease in firing rate
0
8
60
40
Percent decrease in firing rate
20
100
Figure 29: Decrease in firing rate with illumination for all FEF neurons in both monkeys.
4.3.7 Silencing of individual neurons
In addition to population-level analyses, the difference in the distribution of instantaneous firing
rates for laser v. control trials at equivalent trial times were compared for every neuron. This more
rigorous test evaluates what fraction of individual neurons had a significant decrease in firing
rate during illumination based on the distribution of instantaneous firing rates from all trials. A
significance level of a = 0.95 was used throughout. Of the 67 units recorded in Monkey L with
visual, motor, or delay period activity, 60 units (89.6%) had a significant decrease in firing rate.
Of the 52 units recorded in Monkey C with visual, delay or motor activity, 36 units (69.2%) had a
significant decrease in firing rate with illumination. An example of a neuron that was
optogenetically inactivated is shown in Figure 30.
95
I
I
Tarqet in RF
*
S
e
a
*
so
80
70
60
E
C
I-
I-
50
a0
0
- Target outside RF
40,
40 a ' I. .c0
30
-iw-n-. %
20
C
#
a,
*
a-
.0
10
0
-
-400
-200
0
200
400
600
800
1000
1200
Laser on
1400
1600
1800
Time relative to laser pulse onset (ms)
Figure 30: Raster plot of optogenetically-inactivated FEF neuron
Note that the trial number only reflects the relative trial number within each condition, not the absolute
ordering in which these recordings were made. Trials were re-orderedhere to help the reader visualize the
effects remain regardless of whether or not the target is inside the receptivefield or opposite to it.
4.3.8 All subtypes of neurons were silenced in both monkeys
Because all subtypes of neurons were recorded in both monkeys (see Figure 28), neuronal
silencing could be further evaluated within these neuronal subpopulations.
4.3.8.1 Visual neurons
Visually-responsive neurons were inactivated in both monkeys. When the target was in the
receptive field, the population of visually-driven neurons showed a smaller, later increase in
firing rate relative to the non-illuminated trials, as shown in Figure 31.
96
Fixation
Target
Laser
3
Fn
Monkey L
Control trials
Target opposite to RF
=62
2.s F
Control trials
Target In RF
0.5
2
OA
.aser trials
Target In RF
1.51
0.3
0
I
0.2
z
U. F dP%00$W
0
0.1
0
-200
0
200
400
Time from target presentation (s)
000
-200o
0
200
40
600
Time from target presentation (s)
Figure 31: Firing rate profile of visually-responsive neurons during target presentation.
The top panel shows the response profile when the target was in the RF and the bottom panel shows the
response profile when the target was opposite to the RE. There arefewer neurons for Monkey C here than
in Figure 32 because only a subset of neuronsfrom Monkey C were recorded during target inactivation.
The increase in firing rate immediately after illuminationcorresponds to the rebound effect discussed later
in the chapter.
As Figure 32 shows, the firing rate in the vast majority of visually-responsive neurons in both
monkeys decreased by > 80%. The laser-on and laser-off firing rates of the population of visuallyresponsive neurons were evaluated using a paired samples student's t-test. In both monkeys,
there was a significant decrease in the average firing rate of the visually-responsive neuron
population during illumination (Monkey L: n = 62, p < le-12; Monkey C: n = 39, p < le-5).
97
S70-
MSnao-
Monkey L
60.- Visual units
60
11
70
50
so-
Monkey C
Visual units
60-
so-
140-
L 4030
S30-
> 20o
% 2010~
o
20
40
60
80
Percent decrease in firing rate
100
0
20
40
60
so
100
Percent decrease in firing rate
Figure 32: Firing rate decrease in visually responsive neurons in both monkeys.
Further, when individual visually-responsive units were considered, the majority of units in both
monkeys had a significant decrease in the trial-by-trial spike probability. A student's t-test was
used to compare the distribution of spikes/test period for all control trials against the distribution
of spikes/test period for all laser trials for a given unit. Monkey L had a significant decrease in
firing rate with illumination in 55 of 62 units (88.7%) while Monkey C demonstrated a significant
decrease in 28 of 39 units (72.8%).
4.3.8.2 Delay neurons
Neurons with elevated firing rates during the delay period were inactivated in both monkeys.
During delay period illumination, there was some breakthrough firing toward the end of the laser
pulse in delay-responsive neurons. Figure 33 shows the firing profile for delay-responsive
neurons with delay-period illumination for when the target is in the injected receptive field and
opposite to it. The majority of delay-responsive neurons decreased their firing rate by >80% as
shown in Figure 34. As a population, delay-responsive neurons had a significant decrease in
firing rate with illumination (Monkey L: n = 60, p < le-11; Monkey C: n = 36, p < le-5). When
considered individually, the firing rate of most delay-responsive neurons significantly decreased
98
with illumination. Monkey L had a significant decrease in 53 of 60 delay-responsive neurons
(86.7%) while Monkey C had a significant decrease in 28 of 36 delay-responsive neurons (77.8%).
Fixation
Target
Laser
I
3
r
Monkey L
Monkey C
0.9
n -60
2s
Control trials
Target opposite to RF
n= 36
0.8
Control trials
Target in RF
0.7
2
0.6
'a
Laser trials
Target in RF
0.5
1.5 l-
0A
I
0.3
z
;4e
0.s
0
-200 -100 0
Time rel. to
laser start (ms)
k
02
0.1
0
100
200 300
Time rel. to
laser end (ms)
IW"W
I
-200 -100 0
Time rel.to
laser start (ms)
0
- -,
100 200 300
Time rel. to
laser end (ms)
Figure 33: Firing rate profile of delay - responsive neurons during the delay period.
The top panel shows the response profile when the target was in the RF and the bottom panel shows the
response profile when the target was opposite to the RF.
99
S70
ga
60
80
Monkey L
Delay units
II
C
z
C
SO.
a,
'A
40-
C
0
~30.
Eu
Monkey C
70 - Delay units
60
50
--
40
a, 30
'4-
o 20
0
20
4-
10
C
a,U 10
0~
0
20
40
60
8
100
Percent decrease in firing rate
0
0
80
20
40
60
Percent decrease in firing rate
100
Figure 34: Firing rate decrease in delay-responsive neurons in both monkeys.
4.3.8.3 Motor units
Neurons with elevated firing rates during the pre-motor and motor period were inactivated in
both monkeys. Figure 38 shows the firing profile for motor-responsive neurons with go-cue
illumination when the target is in the injected receptive field and when the target was opposite
to the injected receptive field. On average, these neurons eventually elevated their firing rate in
preparation for movement (albeit it later and to a lesser extent than during inactivated trials), but
the overall population of motor-responsive neurons in both monkeys had a significant decrease
in firing rate with illumination (Monkey L: n = 39, p < le-9; Monkey C: n = 33, p < le-7). Figure 39
shows the percent decrease in firing rate for motor neuron populations in both monkeys. When
the distribution of neuronal firing rates across trials was calculated for each individual unit,
Monkey L had a significant decrease in 34 of 39 motor-responsive neurons (87.2%) while Monkey
C had a significant decrease in 25 of 33 motor-responsive neurons (75.7%).
100
-
Fixation
Laser
0.8
3
Monkey L
n- 39
0.71
Monkey C
n= 13
Control trials
Target opposite to RF
2.s
Control trials
Target In RF
0.6
.2!
M.
2
0.5
Laser trials
Target In RF
.A
1.5 1-
02
z
0
0
-200 -100 0
ime rel.to
laser start (ms)
-
0.5
0
-200 -100 0
Timemel.to
100 200 300
Timerel.to
laser end (ms)
laser start (ms)
laser end (ms)
100 200 300
Time l. to
Figure 35: Firing rate profile of motor-related neurons during the motor period.
The top panel shows the response profile when the target was in the RF and the bottom panel shows the
response profile when the target was opposite to the RE. There arefewer neurons for Monkey C here than
in Figure 36 because only a subset of neurons from Monkey C were recorded during pre-saccadic
inactivation.
=80
70
II
Monkey L
Motor units
80
70
Monkey C
Motor units
60-
-60
S50
so
so-
Ih1.40
04.0
0 404.'
0
930
60
920
20[
10
110
U 0
30-
20
40
60
8
0I
100
0
20
40
60
Percent decrease in firing rate
Percent decrease in firing rate
Figure 36: Firing rate decrease in motor-related neurons in both monkeys.
101
100
4.3.9 Rebound activity
Not during, but immediately after illumination, a rebound effect was observed (see Figure 30, 31,
33, and 35). On average less than one additional spike occurred per trial during rebound period
in both monkeys (Monkey L: mean +/- std. dev. = 0.965 +/- 0.878 spikes / pd; Monkey C: mean +/std. dev. = 0.350 +/- 0.825 spikes / pd.). The extent of the rebound was correlated with the extent
of silencing in Monkey L, but not in Monkey C (p < le-10 in Monkey L, p > 0.1 in Monkey C). To
put the rebound effect into perspective, note that a single additional spike over a 50 ms period
yields a transient spike rate increase of 20 Hz. The raster plot in Figure 30 most clearly illustrates
the extent of the rebound.
4.3.10 Behavioral results
Error rates (e.g., failures to execute memory-guided saccades to the proper target location)
increased in both monkeys with optogenetic inactivation during either the target, delay, or motor
period for targets in the inactivated receptive field. Figure 37 shows the rate of errors made after
the go-cue. Data was collected over four testing sessions in each primate. Because errors were
relatively uncommon and relatively few trials with each condition at each target location were
performed on a given day, trials for all sessions were combined at the end of testing and x2
analysis was performed on the overall error counts.
**
**
40A 30-
0
o
**
Monkey C
Monkey L
8
No Illumination
Target
6-
Delay
Go-Cue
4
I
* p <0.05,)etest
** p<0.05/12
(Bonferroni correction)
2-
Opposite
Injected
0
Opposite
Figure 37: Errorrates for both monkeys at different silencing times.
102
Injected
4.4 Discussion
Optogenetic inactivation of visual, delay, and motor-related activity was observed in both
monkeys. Error rates for saccades to targets in the injected receptive field increased with
optogenetic inactivation at during any of the epochs of the memory-guided saccade task. Thus,
neuronal firing in the frontal eye field activity contributes to performance in all three epochs
(visual, delay, and motor preparation) of the memory-guided saccade task.
Overall, 80.7% of neurons (96/119) had a significant decrease in firing rate with illumination.
Silencing is often reported as inhibition of baseline activity compared with baseline activity itself,
but the firing rates (illuminated v. control) here were measured during the same epochs of the
task in which these neurons typically increased their firing rate. Therefore, we report silencing of
task-relevant, driven activity which reflects the extent to which optogenetic inactivation can block
drive activity in an active brain region. When this driven activity was inhibited at any time during
the task, the rate of errors increased specifically for the targets in the inactivated receptive field.
4.4.1 Significance of the frontal eye field
The frontal eye field is a critical region for both oculomotor and visual attention circuits in the
primate brain. This study demonstrates that silencing of the frontal eye field can impact simple
task behavior, and it allows for the possibility of inactivating FEF to study its role in these larger
circuits.
One aspect to consider is how momentary inactivation of FEF with simultaneous recordings in
other brain areas could be used to study the connections between these areas. FEF's close
connections to the superior colliculus (McPeek, Han et al. 2003, McPeek and Takahashi 2006), the
lateral intraparietal area (Buschman and Miller 2007, Schafer and Moore 2011), pre-frontal cortex,
and the other eye fields are well known (Parthasarathy, Schall et al. 1992), but longer-range
connections between FEF and areas such as V4 (Gregoriou, Gotts et al. 2009, Anderson, Kennedy
et al. 2011, Zhou and Desimone 2011), and the pulvinar (Stanton, Bruce et al. 1995) are active areas
103
of research. In future work, synchrony between FEF and these more distant regions could be
disrupted with optogenetics to study global phenomena, such as attention.
4.4.2 Technical and scientific significance
While previous primate optogenetics studies have shown decreased firing rates when
inactivating opsins were expressed in the cortex (Diester, Kaufman et al. 2011, Han, Chow et al.
2011, Ohayon, Grimaldi et al. 2013), none demonstrated a behavioral change with optogenetic
inactivation. The fraction of neurons with a significant decrease in firing rate was 81.5% (97/119)
in this study, compared with 38% (55/144) with eNpHR2.0 (Diester, Kaufman et al. 2011), 46.5%
(45/74) with ArchT (Han, Chow et al. 2011), 68.1% (96/ 141) with ArchT and 52.9% (54/102) with
eNphR3.0 (Ohayon, Grimaldi et al. 2013).
In addition to silencing a larger fraction of neurons, this study is different from earlier work
because none of the neurons in this study increased their firing rate during illumination. In each
of the previous primate optogenetics studies with inactivating opsins, a subpopulation of neurons
(-10-25% of the total neuronal population) significantly increased their firing rate during
illumination (Diester, Kaufman et al. 2011, Han, Chow et al. 2011, Ohayon, Grimaldi et al. 2013).
The coordinated, population-level silencing and homogenous modulation demonstrated in the
current study presumably allowed for optogenetically-driven behavioral changes.
The rebound effect, which was described in Sections 3.3.2 and 4.3.9, is not the same the as the
heterogeneous neuronal effects that other groups reported. With heterogeneous activity, some
neurons are driven while others are simultaneously inhibited. Here, the response of the population
is homogenous in that nearly every neuron is inhibited during illumination and very transiently
driven once illumination ends. The rebound effect is a known consequence of optogenetic
inactivation with halorhodopsins (Arrenberg, Del Bene et al. 2009, Tonnesen, Sorensen et al. 2009,
Tsunematsu, Kilduff et al. 2011, Brown, Tan et al. 2012, Madisen, Mao et al. 2012, Raimondo, Kay
et al. 2012, Schone and Burdakov 2012, Smith, Virkud et al. 2012). Halorhodopsins inactivate
neurons by allowing an influx of chloride ions to lower the membrane potential beyond baseline
and, thus, block the generation of action potentials. Raimondo et al. (2012) suggested that
104
intracellular chloride accumulation may alter the extracellular chloride concentration and drive
an efflux of chloride at the end of illumination, which manifests itself as the rebound effect. The
rebound occurs before the trial is over in two of the test conditions (visual and delay period);
however, it is unlike that the rebound effect impacted behavior. On average, less than one extra
spike per trial was induced by the rebound effect. Compared to micro-stimulation, which used
100 pulses at 250 Hz (inducing >50 spikes) to induce saccades in this study, the single extra spike
is negligible. For trials with illumination during the motor period, illumination typically
continued through the end of the trials because the mean saccade latencies for both monkeys were
40-80 ms shorter than the illumination period. The overall optogenetic modulation during each
task epoch demonstrated that visual, delay, and motor firing all contribute to performance in the
memory guided saccade task. Chuong et al., 2014 showed that the rebound effect with Jaws than
with ArchT, but that light pulse sculpting (i.e., ramping down the intensity of illumination over
hundreds of milliseconds) can ameliorate the amplitude of the post-inhibition rebound. Pulse
sculpting was not used in this study because 1) temporally-precise light pulse onsets and offsets
were needed at specific task times. This could not be achieved with the durations needed to
"ramp down" the light pulse; 2) this study sought to disrupt a particular kind of neuronal signal
during the task (i.e., the visual signal, the delay signal, the motor preparation signal). Silencing
was not strictly necessary as a meaningless rebound spike could also serve to disrupt the signal
without changing the interpretation of the results; and 3) the temporal specificity of the task and
the monitoring allowed for monitoring of rebound-driven behavior changes, and there were no
rebound-driven changes. When rebound spikes occurred, they occurred in a specific time period
after the end of the pulse. This timing allowed us to look for time-locked events that may have
occurred as a result of the rebound as a control. If we used a ramping pulse, we would not be able
to disambiguate spikes that resulted from some kind of rebound (if any) from spikes that resulted
as normal firing resumed. While we did not use sculpted pulses in this initial study, in the future,
others may find that sculpted pulses work well for different studies or that different techniques
for sculpting pulses may be identified.
105
This study was able to achieve homogenous inactivation during illumination because a large,
functionally-relevant region of cortex was inactivated uniformly. First, virus was injected over a
large, functionally determined volume of cortex. The full receptive field was mapped in FEF prior
to injection and the injection sites were selected based on the physiology of each monkey. In
contrast to injection sites selected based on anatomical MRIs or to those in close proximity to sites
where other opsins were injected, a single inhibitory opsin was injected at all chosen sites in a
given primate. We ensured that the whole tissue volume corresponding to our task was targeted.
Other studies relied on MRI images alone or injected excitatory opsins in close proximity to the
silencing opsin injection site. Second, we delivered light evenly over a large volume and
intentionally sought to avoid heating, which can lead to increased firing rates or damage. In order
to achieve this broad illumination, though, a primate-specific illuminator was developed for this
study. Finally, histologically-evaluated viruses with high expression levels and no evidence of
toxicity were used. These viruses were specifically determined for our monkeys, prep, and brain
area.
In previous studies, low expression levels or small tissue volumes were consistently cited as a
possible reason for the small or absent behavioral changes with optogenetic inactivation (e.g.,
Cavanaugh et al., 2012), the sources for these small or absent effects. In contrast, the volume of
tissue inactivated in this study (see Chapter 3) is comparable with that of the pharmacological
experiments (Sommer and Tehovnik 1997, Tehovnik and Sommer 1997), as is the extent of
silencing (Chafee and Goldman-Rakic 2000). Here, -70% of task-modulated neurons had firing
rate reductions of >80%. In the Muscimol literature, >80% of neurons had firing rate reductions of
> 80% (Keating and Gooley 1988, Dias, Kiesau et al. 1995, Dias and Segraves 1996, Sommer and
Tehovnik 1997, Tehovnik and Sommer 1997, Dias and Segraves 1999, Tehovnik, Sommer et al.
2000, Sommer and Wurtz 2002, Schiller and Tehovnik 2003, Keller, Lee et al. 2008).
Pharmacological inactivation may affect memory-guided saccades not only because FEF neurons
are inactivated during the task, but because FEF neurons are inactivated for a long enough time
that reciprocal connections (e.g., among inactivated and non-affected FEF neurons or between
FEF and other structures, such as the superior colliculus or parietal cortex) are disrupted. For
106
example, FEF sends information related to movement, memory, and vision to the superior
colliculus (Sommer and Wurtz 2001) while the superior colliculus reciprocally activates FEF
(Sommer and Wurtz 1998). Disrupting the FEF input to the SC may in turn weaken the SC input
to the FEF over the time course of Muscimol inactivation. With optogenetic inactivation, it is
unlikely that adaptation would occur given how infrequently the FEF is inactivated and the
brevity of the illumination.
Unlike pharmacological inactivation, in which the brain may begin to adapt to this change and in
which the short-time scale compensatory mechanism may not be present, optogenetic
inactivation disrupts the "normal" state. Ultimately, simultaneous recordings in multiple brain
areas during optogenetic and pharmacological inactivation could help to answer this question,
From this study, we learned that the visual, delay, and motor-related increases in FEF neuronal
firing contribute to performance in the memory-guided saccade task.
107
Appendixes
Appendix 1: Shutter characterization plots
Onset
Lght power measured -Aa kntegratIng sphere
10
8
6
E
4
5 ms
2
1.07
1.0705
1.071
1.0715
I 1.072 I 1.0725
1.073
1.0735
1074
1 0745
1.0735
1.074
1-0745
Time Ons)
1.C 75
XlO0
Enable sig"ul to shiter
5
43
2
1rms
0
I --
1.07
-
-I1
0
1.0705
1. 71
1.0715
1.072
1.0725
Time 91s)
Figure 38: Characterization of shutter opening
108
1.073
105
X10
4
Offset
U t power measured va Irtegring sphere
108-
E
4-
2-
-
0
2.06
2,065
2.07
2.075
2.08
2.005
2.08
2.085
Time (ms)
2.09
x10 4
Enable slgna to shutter
4 -N
148 ms
3-
18 MS
1-
0-1
2.06
2.065
2.07
2.075
Time
Figure 39: Characterization of shutter closure.
109
(ms)
2.09
X104
Appendix 2: Supplemental micro-stimulation and neuronal recording data.
Examples of evoked saccades at injection sites used in this study (monkey L)
Eye m
Iid
giuitshole G4 at adupth (5 mm
muM to the eute ofoactia wDt cuneit o!50 iaA
it
Eye moamn ta
relaftt to1 b a
2
lI tde G4 at a depth of6 mm
ate
t with cutrent of 100 uA
1614-
12
0
810
0
8[-
-1
6-2
Eye pieh
Eye----
*
0
-2
0
2
4
Endnt
6
a
locationEnigWao
Degrees e to rftt)
10
12
14
0
2
4
6
a
10
Depues let to rght)
12
14
16
Figure 40: Examples of micro-stimulation evoked saccades.
Appendix 3: Viral vector selection
A wide variety of opsin-containing viruses have been injected in the primate brain, including
lentivirus (Han, Qian et al. 2009, Diester, Kaufman et al. 2011, Han, Chow et al. 2011), AAV1
(Jazayeri, Lindbloom-Brown et al. 2012), AAV5 (Diester, Kaufman et al. 2011, Gerits, Farivar et
al. 2012) and AAV8 (Cavanaugh, Monosov et al. 2012). Ubiquitous (Han, Qian et al. 2009, Han,
Chow et al. 2011, Cavanaugh, Monosov et al. 2012, Gerits, Farivar et al. 2012) and neuron-specific
promoters (Diester, Kaufman et al. 2011, Jazayeri, Lindbloom-Brown et al. 2012) have been used,
but ultimately, virus expression still varies across different groups depending on injection
techniques, target region, and which batch of virus was used. In this study, we identified the
batch of virus that worked for us and used a consistent injection technique (an intra-operative
preparation). Some viruses failed to work for us (e.g., AAV1) while others worked much better
for us than for others. For example, we tested the same lot of AAV8-CAG-ArchT that Cavanaugh
et al. (2012) used. We found expression in -80% of cortical neurons while (Cavanaugh, Joiner et
al. 2012) reported expression in <40% of cells. In another example, Han et al. (2009) reported
110
expression in more than half of neurons with ChR2 in a lentiviral construct while Gerits et al.
(2012) reported that the same construct in the same brain area was ineffective (Gerits, Farivar et
al. 2012).
111
Appendix 4: Behavioral data analysis
Saccades made after the disappearance of the target were analyzed and classified as correct or as
errors as described above. When no saccade was made, no further analysis was performed.
Saccadic latency was defined as the time from the disappearance of the central fixation dot to the
beginning of pursuit. The instantaneous velocity was calculated from 30 ms prior to when the
eyes left the central fixation window until the eyes reached the desired target location. The peak
instantaneous velocity was determined. The acceleration was calculated for all times from the
first instantaneous velocity calculated until the peak velocity. The beginning of pursuit was
defined as the time of peak acceleration. Trials in which the peak velocity failed to reach 85
degrees / second were excluded. If the monkey failed to make a saccade 500 ms after the go-cue,
the saccade was counted as a "no saccade" trial. Saccadic error distance is the vector distance
between the end-point of the saccade (adjusted linearly by the starting point) and the center of
the target location.
112
Appendix 5: List of injected viral vectors.
Two of the injections failed during surgery due to leaky syringes, which were discarded. Table 1
lists the viral vectors tested and the injection sites.
Table 1: List of all viral vectors injected and their injection locations. All vectors had a GFP tag.
Virus
X (mm, lateral to
Y (mm, anterior to ear
midline)
bars)
Side
AAV9-CAG-ArchT
L
20
19
AAV8-CAG-Jaws Kir2.1
R
12
14
AAV1-CAG-ArchT
L
10
12
AAV5-CAG-ArchT
L
20
10
AAV8-CAG- Jaws Kir2.1
R
5
15
AAV1-CAG-ArchT
L
4
12
AAV8-CAG- Jaws Kir2.1
L
30
15
lenti Jaws -ER2
L
20
2
AAV8-CAG- Jaws
R
9
14
FCK-eNpHR3.0
L
10
5
EIAV-ArchT
L
37
0*
AAV8-CamKII-ArchT
L
2
20*
AAV9-CAG-ArchT
R
8
3
113
AAV8-CAG-ArchT
R
19
0*
lenti of Jaws -ER2
L
37
18
AAV8-CAG- Jaws Kir2.1
L
11
8
AAVDJ-CAG-ArchT
R
25
25*
EIAV-ArchT
R
12
25*
FCK-ArchT
L
30
2
AAVDJ-CAG-ArchT
L
3
8
AAV8-CamKIl-ArchT
R
17
0*
AAV8-CAG-Halo57 Kir2.1
R
9
11
114
Appendix 6: Fluorescence probe leads to isometric response
Because an opsin-expressing cell can be activated by light coming from any direction, the light
measurement probe must be as isotropic as possible. Fluorescence is an inherently isotropic
process because the incident angle of the excitatory photon has no bearing on the emission angle
of the emitted photon(Skinner 1964). In our probe, we couple a 0.3 mm diameter ruby sphere to
a 0.4 mm diameter optical fiber using optically transparent adhesive (Figure 5a). The necessary
attachment to the optical fiber only obstructs about 20% of the ruby sphere surface area (see
Appendix 6.1 on the next page). We chose to use a ruby as the fluorescent source in our probe
because rubies absorb photons at wavelengths across most of the visible light range (Cronemeyer
1966). Figure 41 shows the absorptive spectrum of rubies. Bayes et al., 1995 tested the angular
response of a similar probe and found a flat response for the range of unobstructed incident
angles (-130* to 1300), reprinted below in Figure 42 (Bays, Wagnieres et al. 1995).
4
40
agreen
7' P
5?
1.1
33
cm
ared = 0.15 / cm
20
2
=4/
2
a- 9 PINKROT
07I
SLES 6. 7'. AMO7
6.1
I
F25
30 0
1.06 X 10'1CM-3
1 I
I
40
4"
a
:i
s
VAVELMTH UOP
I
we
-o
,
FIG. 2. Absorption coefficient a(cm-') and absorption cross
section w(10-" c0 at 300"K as functions of wavelength ?(9)
for E.,c and EIc for Linde pink-ruby boule B-9 samples3 6, 7
and 7" with an average Cr-concentration of 1.86X 109 m- . The
points designated by circles refer to ae, e, those by triangles to
a,, g,. The transmittance data were measured with the Beckman
B spectrophotometer.
Figure 41: Ruby absorption coefficients for red and green light.
115
Cronemeyer 1966
S,
..a. "SO
0.4,441
- 19 -126
0.
-~60
IIt
6
0
2
4II
0*
I"a*.s.a......l.
I
-IS0
.12
40
0
Oreutation
60
120
1o0
[degreel
Figure 42: Demonstration of angle independence of isometric light probe.
From Bays et al.
Appendix 6.1: Geometric estimations of isometric probe shadowing
Calculations showing how much coupling between the optical fiber and the ruby sphere limits
light acceptance in the optical probe
116
d=0.2mm
8
Glass optical fiber
D = 0.15mm
Ruby sphere
The ruby sphere surface defined by the angle 0 is blocked by the glass optical fiber and cannot accept
light.
0 = arctan(d/D)
Recall that the surface integral of a sphere in spherical coordinates is
sine de d<p
surface
Let fiblocked be the acceptance angle that is blocked by the optical fiber
Let flid.a be the acceptance angle of a ruby sphere that is not attached to a fiber
2 [Tt
01JOsinedeldv
arctan(
2 7t
= -2n
=
sin9de
cos Iartan
21r2ir
ld..i = .1[
0
-cosO)i
1
2uI
+
(
=
)
fiblocked =
. 0 sinede]dp = 4n
Thus, the fraction of light that the attached ruby sphere collects compared with an ideal sphere is
4r- 21-2n 1
(ne.
- flbloekd) / fideol
1+(d)/ h
47T=
2w+ 2n
1+
1+
1+
471
2
-
_+(__r
2
=0.8
The ruby sphere, even when attached to the optical probe, collects 80% of the light Incident for all
directions.
117
Appendix 6.2: Supplemental methods for ruby sphere calibration
Appendix 6.2.1: Beam profiler set up
For each color of visible light (473 nm, 532 nm, and 635 nm), a beam of collimated light was
generated by coupling a DPSS (532 nm and 635 nm) or diode (473 nm, Vortran) laser to a largediameter collimator (F810FC-A, ThorLabs) via a
20 0
tm diameter (NA = 0.22) multimode FC/PC
terminated fiber (ThorLabs). The collimated beam passed through a custom-built iris with lightabsorbing coating. The total power light power and beam uniformity were measured with a beam
profiler (BC106-Vis, ThorLabs) that had been calibrated to take absolute light power
measurements. The set up for this calibration is shown in Figure 43.
Output from beam profiler with
absolute light power values
fibe r
-1
.0.5
0
5
1
X ("04
Figure 43: Beam profiler used for absolute light power measurements.
Appendix 6.2.2: Fluence measurements in water bath
Next, the collimated beam, with measured, uniform light power density, was applied to the
isometric probe via a water bath. Figure 44 shows the experimental set up. A relationship between
118
incident light power density at the test wavelength and the emitted fluorescence at the peak ruby
emission wavelength (-695 nm, defined as the 0.47 nm-wide bin with peak amplitude between
693 nm and 697 nm) was determined for each probe as shown in Figure 5c.
OpticalI
fiber
Water bath
Figure 44: Water bath for isometric probe calibration.
Appendix 6.2.3: Determination of correction factors for calibration
Equation A.6.2.3.1 is determined by plotting the light power density (mW /mM 2 ) on the surface
of the ruby sphere as a function of fluence rate (photons / second), as shown in Figure 5.
cPrs = m*x I C
(A.6.2.3.1)
Where Ors is light power density on the surface of the ruby sphere probe, x is fluence rate (photons
/ second) at the ruby wavelength(s), C is a constant correction factor determined to be 3.08 as
shown below, and m is the probe- and wavelength-specific slope of the linear curve fit during the
isometric probe calibration. Note that the y-intercept for that linear curve fit is set to 0.
The constant correction factor, C, is determined by Equation A.6.2.3.2.
C=
(1 +fH2o-cortex)* Across section I (Asurface *(1 -fiocked))
(A.6.2.3.2)
Where fmo-cortex is the fraction by which calibration in water over estimates photon counts in
cortex,
Across secion
is the cross-sectional area of the ruby sphere (n r 2 ), the area over which
119
collimated light is applied,
Asurface
is the surface area of the ruby sphere over which it
(theoretically) can absorb photons (471 r2), the area over which light is applied in vivo, and fblocked
is the fraction of the ruby sphere probe surface area that is blocked by the attached optical fiber
and cannot absorb photons. Across section, Asurface, and fblocked are all geometric properties of the
isometric probe; however, the derivation of fmo-cortex is a bit more nuanced and, thus, explained
below.
The difference in the refractive indices of water (nm2o = 1.33) and the ruby (nruby = 1.77, provided
by Edmund optics) is slightly larger than the difference in the refractive indices of cortex
(ncortex =
1.37) (Lue, Bewersdorf et al. 2007) and the ruby. Thus, the critical angle in for the ruby in water is
slightly smaller than the critical angle in cortex. The critical angle for total internal reflection of
photons inside the ruby in both air and water was determined using Equations A.6.2.3.3 and
A.6.2.3.4.
OcriticalH2o= arc(n2o / nruby)
(A.6.2.3.3)
(9 critical_cortex = arc(ncortex / nruby)
(A.6.2.3.4)
Then Snell's law was used to derive an expression for the angle of transmission,
Eh, relative to the
incident angle & (Equations A.6.2.3.5 and A.6.2.3.6).
(A.6.2.3.5)
t_H20 = arcsin( nruby sin((9d nH2o)
&-cortex
=
(A.6.2.3.6)
arcsin(nruby * sin(&) I ncortex)
Fresnel's Equations were rearranged to get reflection and transmission coefficients as a function
of the incident angle for both the parallel and perpendicular cases as shown in Equations A.6.2.3.7
- A.6.2.3.10. The terms parallel and perpendicular are relative to the plane of incidence.
t)
R 11 = tan(9itan(ei+at)
(A.6.2.3.7)
R_=
(A.6.2.3.8)
sin(Oi - 9t)
sin(Oi + 9t)
120
T, '-= 2*si_n(Ot)* cos(9i)
(A.6.2.3.9)
= sin(Oi+ Ot)* cos(9i -
et)
T.= 2*sin(Ot)* cos(Oi)
sin(Oi+
(A.6.2.3.10)
et)
The percentages of light reflected and transmitted for a given incident angle were determined
using the conservation of energy and then integrated over all possible incident angles to yield an
overall percentage of photons emitted by the ruby sphere and then transmitted out of the ruby
sphere back to the medium. In Equation A.6.2.3.11 below, the first term represents the fraction of
fluoresced photons that are reflected back into the ruby sphere when emitted angles are less than
the critical angle. The second term represents the angles over which 100% of photons remain in
the ruby sphere.
A=Oc
A6=90*
f r(0)dO + Jd0
-=0*
(A.6.2.3.11)
A-=&
These equations were solved empirically in MATLAB using dA = 10. It was determined
empirically that water yields a photon count 3.8% higher than expected in cortex. This procedure
was repeated for measurements and estimations in air v. water as well to confirm the correction.
Theoretically, air should give a photon count 29.3% higher than expected in water and the
measurements were 28.2 +/- 2.6% (standard error) higher in water than in air, which empirically
confirms the validity of this theoretical correction.
Appendix 6.2.4: Planar illuminator as a wide-beam collimated source
A planar illuminator was used to approximate a collimated wide beam source. Use of an actual
collimator was not feasible because in order to get a collimator with uniform light power output
over a 1.5 mm diameter area, the collimator itself would need to be much larger than the mouse's
head and only the very center of the collimated beam would be uniform enough for testing. Thus,
an iris with a very small aperture and large physical footprint would be needed between the
collimator and the brain. Even though this is almost exactly the preparation used for ruby sphere
121
calibration, this was not physically feasible in the in vivo preparation. A beam profile of the
output from the planar illuminator showed a uniform output over a centrally located one mm
square area in the center of the beam even at a distance 8 mm away (Figure 45). Further, geometric
decrease models accounting for internal reflection of multi-model coupled rays show that the
total light power directly under the planar illuminator remains at 100% up to 2.5 mm away from
the illuminator surface (Figure 46).
-1.5
0.12
-1
-0.5
0.1
0
E
1
0.06
1.5
0.04
2
2.5
3
-25
-2
-1.5
-1
-0.5
0
x (Mm)
0.5
Figure 45: Beam profile of planar illuminator
Beam profile of planarilluminatorat 8 mm away.
122
1
1.5
Light power
density
(mW/mm
2
)
E
0.0
0.5
Fraction of light reaching a gien depth based on geometric decrease only
1
0.5
0.9
1
0.8
E 1.5
0.7
2
0.6
2.5
0.5
3
0.4
3.5
0.3
4
0.2
4.5
0.1
E
TL
CS
5
-4
2
3
-1
0
1
-2
-3
Distance from center of planar illuminator (mm)
4
0
Figure 46: Internal geometric
Geometric decrease model for planarilluminatorover a distance of 8 mm.
Appendix 6.3: Internal geometric model
Models of geometric light power decrease with distance were generated in MATLAB. These
models used the numerical aperture of the planar illuminator and its diameter to calculate a
decrease in light power density due to the beam "spreading out" as it exited the illuminator. They
showed that the light power in the center of the beam directly under the planar illuminator does
not begin to decrease until about 2.5 mm below the illuminator surface. In addition, beam profiles
of the planar illuminator showed a uniform light power density across its surface and this
uniformity in the middle of the beam was even present 8 mm away from the illuminator surface.
123
rfilber
rfZfibr
rfa,
Zfiber
Figure 47: Illustration of how rays can internally reflect and yield a uniform beam in the center.
124
Appendix 7: Heating measurements in tissue
Appendix 7.1: Possible negative effects of heating in the brain
Cooling the brain after a traumatic injury reduces brain metabolism and limits the extent of injury
(Busto, Dietrich et al. 1987, Brengelmann 1993, Cabanac 1993, Brown, Bae et al. 2007). While
cooling the brain impairs task performance in healthy subjects by reducing neuronal firing rate
(Horel, Voytko et al. 1984, Gregory Keating and Gooley 1988, Keating and Gooley 1988, Quintana,
Fuster et al. 1989, Chafee and Goldman-Rakic 2000), locally increasing brain temperature also can
disrupt saccade behavior (Bajada, Mastaglia et al. 1980). In the extreme, raising the temperature
of the brain by >50 C (e.g., during the course of a methamphetamine overdose or an extreme fever)
may cause permanent brain damage and at times, death (Kiyatkin 2004, Kiyatkin and Brown 2004,
Kiyatkin 2005, Kiyatkin 2007).
Appendix 7.2: Neuronal firing rate increases with increased brain temperature
In vitro studies have shown that heating increases neuronal firing rate and the probability of
neurotransmitter release (Volgushev, Kudryashov et al. 2004). Some populations of neurons
increase their firing rate in response to small (>20 C) increases in baseline temperature (Burgoon
and Boulant 2001, Guatteo, Chung et al. 2005). Further, the brain endogenously contains transient
receptor potential (TRP) channels that increase neuronal firing as the brain temperature increases
(Xu, Ramsey et al. 2002, Story, Peier et al. 2003, Moran, Xu et al. 2004, Bernstein, Garrity et al.
2012).
Appendix 7.3: Heating may play a role in driving behavior in primate visual cortex
Synaptic transmission in the visual cortex is temperature dependent, according to in vitro studies
which showed increased excitability of rat visual cortical neurons with increased temperature
(Hardingham and Larkman 1998). Jazeyeri et al. (2012) introduced ChR2 into the primary visual
cortex (V1) of a rhesus macaque and reliably shifted the primate's free gaze during the inter-trial
interval with illumination. With the given total light powers (20 - 50 mW) and assuming a
125
standard optical fiber (100 - 200 pm diameter), total light power densities used in this study were
-
likely high, between 1500 - 6500 mW/mm2 . Even though Jazeyeri et al. (2012) used short (100
250 ms) pulses with a low duty cycle (5-10%), no temperature measurements. Instead, the control
for possible heating was to place the fiber in a different un-injected site. Similarly, Dai et al., (2014)
reported a response bias during a visuospatial discrimination task with illumination of a greenlight sensitive excitatory opsin, C1V1 that had been introduced into the lateral intraparietal cortex
(LIP). This effect is similar to what had been previously reported with sub-threshold microstimulation (Hanks, Ditterich et al. 2006), but the source of the stimulation could be related to
both the optogenetic inactivation and heating. Even though Dai et al. (2014) only used a total light
power of 1 mW, the tip of their fiber was very small (10 tm). Delivering 1mW over a 10 pm
diameter surface yields light power densities in excess of 10 W/mm 2, which is two orders of
magnitude greater than the light power densities used in this study.
Illuminating an un-injected area of cortex does not fully control for heating-effects with excitatory
opsins for several reasons. First, it is possible that the effects of the excitatory opsin were enhanced
by increased temperature. This synergy would not be present at the control site. Second, the
control area of the cortex may not evoke behavior or network activation as readily as the injection
site does. In fact, researcher typically map brain regions prior to injection and chose the area in
which neurons are the most task responsive and the most easily modulated by external stimuli
or microstimulation. Finally, repeated penetrations into the testing location likely changed the
composition of the extracellular matrix in that location over time. The tissue in the uninjected area
is likely healthier and may have different thermal properties as a result.
Appendix 7.4: BOLD fMRI activation results from heating alone
Heating can drive neuronal firing and cause spurious fMRI activation. Recently, Christie et al.,
(2012) drove profound fMRI responses in the naive rat brain by delivering blue light and inducing
a subsequent increase in baseline temperature (Christie, Wells et al. 2012). Activations in the fMRI
BOLD signal reflect increased blood flow consistent with increased brain metabolism, but blood
flow can increase for reasons other than brain activity, such as heat dissipation. Rat studies report
126
brain temperature increases locally and transiently with increased neuronal activity, thus,
triggering increased blood flow which is detected in the fMRI BOLD signal (Kiyatkin, Brown et
al. 2002, Trubel, Sacolick et al. 2005, Vanhoutte, Verhoye et al. 2006, Kiyatkin 2007).
Gertis et al. (2012) injected a virus containing ChR2 into the ventral pre-motor area (F5) and FEF
of two macaques and reported changes in fMRI activation patterns with illumination, although
the results were not fully consistent among subjects (Gerits, Farivar et al. 2012). Revising the
methodology (light powers of 80 - 300 mW / mm2 and 8 ms pulses repeated at 40 Hz for 16 s) and
findings from Gertis et al. (2012) in light of the Christie et al. (2012) study published subsequently,
it is difficult to attribute the fMRI changes in Gertis et al. (2012) exclusively to optogenetic
activation, particularly in the absence of electrophysiological or histological data. As described
above, the control of illuminating an un-injected site is not a sufficient control with activating
opsins. Overall, this demonstrated the need for additional thermal controls in newer primate
optogenetics studies.
Appendix 7.5: Results of temperature measurements in vivo
A primate-specific illuminator as described in Chapter 3 was inserted into the mouse brain
longitudinally such that all of the illuminator remained in the cortex. Four needle
thermocouples (Omega) were placed 1 mm away from the illuminator and the largest
temperature increase from all probes with illumination was used for the heating measurements
shown below in Figure 48 and Figure 49. Minimal temperature increase was observed with a 1 s
pulse of either 50 mW /mm2 or 100 mW/mm2 ; however if these pulses were repeated for long
durations additive heating results, as shown in 100 mW/mm 2 . The relaxation time for tissue was
determined to be approximately 20x the pulse duration so a pulse spacing of 20x pulse duration
on average was used between laser pulses in this study to prevent additive heating.
127
1
4-,
100 mW/mm2
i I
11 .1.1-
0.5F
.A.1 . I I.. I
I,
I.1
I
.
Q)
I
E
I
IF
I
II
1,111 -1
0
I F.
-1
II I
III
I
''III I I
I
50 mW/mm 2
4-J
Laser on
-1
0
i
I
I
1
2
3
I
I
4
5
6
Time after laser onset (s)
Figure 48: In vivo temperature measurements with the primate specific illuminator for a single laser
pulse
128
4
100 mW/mm
2
3CD
2-
0
1s
a)
0
laser pulses
repeated 25x
10
20
30
40
50
60
Time after pulse train onset (s)
Figure 49: Additive heating with frequently spaced laser pulses
129
70
Works Cited
Adamantidis, A. R., F. Zhang, A. M. Aravanis, K. Deisseroth and L. de Lecea (2007). "Neural substrates
of awakening probed with optogenetic control of hypocretin neurons." Nature 450(7168): 420-
424.
Airan, R. D., K. R. Thompson, L. E. Fenno, H. Bernstein and K. Deisseroth (2009). "Temporally precise
in vivo control of intracellular signalling." Nature 458(7241): 1025-1029.
Alilain, W. J., X. Li, K. P. Horn, R. Dhingra, T. E. Dick, S. Herlitze and J. Silver (2008). "Light-induced
rescue of breathing after spinal cord injury." J Neurosci 28(46): 11862-11870.
Amiez, C. and M. Petrides (2009). "Anatomical organization of the eye fields in the human and nonhuman primate frontal cortex." Progress in Neurobiology 89(2): 220-23 0.
Amir, Y., M. Harel and R. Malach (1993). "Cortical hierarchy reflected in the organization of intrinsic
connections in macaque monkey visual cortex." The Journal of Comparative Neurology 334(1):
19-46.
Andersen, R. A., C. Asanuma, G. Essick and R. M. Siegel (1990). "Corticocortical connections of
anatomically and physiologically defined subdivisions within the inferior parietal lobule." J
Comp Neurol 296(1): 65-113.
Anderson, J. C., H. Kennedy and K. A. Martin (2011). "Pathways of attention: synaptic relationships of
frontal eye field to V4, lateral intraparietal cortex, and area 46 in macaque monkey." J Neurosci
31(30): 10872-10881.
Angell-Petersen, E., H. Hirschberg and S. J. Madsen (2007). "Determination of fluence rate and
temperature distributions in the rat brain; implications for photodynamic therapy." J Biomed Opt
12(1): 014003.
Anikeeva, P., A. S. Andalman, I. Witten, M. Warden, I. Goshen, L. Grosenick, L. A. Gunaydin, L. M.
Frank and K. Deisseroth (2012). "Optetrode: a multichannel readout for optogenetic control in
freely moving mice." Nat Neurosci 15(1): 163-170.
Aravanis, A. M., L. P. Wang, F. Zhang, L. A. Meltzer, M. Z. Mogri, M. B. Schneider and K. Deisseroth
(2007). "An optical neural interface: in vivo control of rodent motor cortex with integrated
fiberoptic and optogenetic technology." J Neural Eng 4(3): S143-156.
Arrenberg, A. B., F. Del Bene and H. Baier (2009). "Optical control of zebrafish behavior with
halorhodopsin." Proc Natl Acad Sci U S A 106(42): 17968-17973.
Arrenberg, A. B., D. Y. R. Stainier, H. Baier and J. Huisken (2010). "Optogenetic Control of Cardiac
Function." Science 330(6006): 971-974.
Asaad, W. F. and E. N. Eskandar (2008). "Achieving behavioral control with millisecond resolution in a
high-level programming environment." J Neurosci Methods 173(2): 235-240.
Asaad, W. F. and E. N. Eskandar (2008). "A flexible software tool for temporally-precise behavioral
control in Matlab." J Neurosci Methods 174(2): 245-258.
Asaad, W. F., N. Santhanam, S. McClellan and D. J. Freedman (2013). "High-performance execution of
psychophysical tasks with complex visual stimuli in MATLAB." J Neurophysiol 109(1): 249260.
Ayling, 0. G., T. C. Harrison, J. D. Boyd, A. Goroshkov and T. H. Murphy (2009). "Automated lightbased mapping of motor cortex by photoactivation of channelrhodopsin-2 transgenic mice." Nat
Methods 6(3): 219-224.
Bahring, R., R. K. Meier and N. Dieringer (1994). "Unilateral ablation of the frontal eye field of the rat
affects the beating field of ocular nystagmus." Exp Brain Res 98(3): 391-400.
Bajada, S., F. L. Mastaglia, J. L. Black and D. W. Collins (1980). "Effects of induced hyperthermia on
visual evoked potentials and saccade parameters in normal subjects and multiple sclerosis
patients." J Neurol Neurosurg Psychiatry 43(9): 849-852.
130
Bays, R., G. Wagnieres, D. Robert, J. F. Theumann, A. Vitkin, J. F. Savary, P. Monnier and H. van den
Bergh (1997). "Three-dimensional optical phantom and its application in photodynamic therapy."
Lasers Surg Med 21(3): 227-234.
Bays, R., G. A. Wagnieres, D. Robert, J. C. Mizeret, D. Braichotte, H. van den Bergh, J.-F. Savary and P.
Monnier (1995). "Clinical measurements of tissue optical properties in the esophagus." 39-45.
Bernstein, J. G., P. A. Garrity and E. S. Boyden (2012). "Optogenetics and thermogenetics: technologies
for controlling the activity of targeted cells within intact neural circuits." Curr Opin Neurobiol
22(1): 61-71.
Bernstein, J. G., X. Han, M. A. Henninger, E. Y. Ko, X. Qian, G. T. Franzesi, J. P. McConnell, P. Stern,
R. Desimone and E. S. Boyden (2008). "Prosthetic systems for therapeutic optical activation and
silencing of genetically-targeted neurons." Proc Soc Photo Opt Instrum Eng 6854: 68540H.
Bi, A., J. Cui, Y.-P. Ma, E. Olshevskaya, M. Pu, A. M. Dizhoor and Z.-H. Pan (2006). "Ectopic
Expression of a Microbial-Type Rhodopsin Restores Visual Responses in Mice with
Photoreceptor Degeneration." Neuron 50(1): 23-33.
Binding, J., J. Ben Arous, J. F. Leger, S. Gigan, C. Boccara and L. Bourdieu (2011). "Brain refractive
index measured in vivo with high-NA defocus-corrected full-field OCT and consequences for
two-photon microscopy." Opt Express 19(6): 4833-4847.
Bizzi, E. (1968). "Discharge of frontal eye field neurons during saccadic and following eye movements in
unanesthetized monkeys." Exp Brain Res 6(1): 69-80.
Boussaoud, D., R. Desimone and L. G. Ungerleider (1991). "Visual topography of area TEO in the
macaque." J Comp Neurol 306(4): 554-575.
Boyden, E. S. (2011). "A history of optogenetics: the development of tools for controlling brain circuits
with light." F 1000 Biol Rep 3: 11.
Boyden, E. S. (2011). "A history of optogenetics: the development of tools for controlling brain circuits
with light." F 1000 Biology Reports 3(11).
Boyden, E. S., F. Zhang, E. Bamberg, G. Nagel and K. Deisseroth (2005). "Millisecond-timescale,
genetically targeted optical control of neural activity." Nat Neurosci 8(9): 1263-1268.
Brengelmann, G. L. (1993). "Specialized brain cooling in humans?" FASEB J 7(12): 1148-1152;
discussion 1152-1143.
Brown, M. T., K. R. Tan, E. C. O'Connor, I. Nikonenko, D. Muller and C. Luscher (2012). "Ventral
tegmental area GABA projections pause accumbal cholinergic interneurons to enhance
associative learning." Nature 492(7429): 452-456.
Brown, P. L., D. Bae and E. A. Kiyatkin (2007). "Relationships between locomotor activation and
alterations in brain temperature during selective blockade and stimulation of dopamine
transmission." Neuroscience 145(1): 335-343.
Bruce, C. J. and M. E. Goldberg (1985). "Primate frontal eye fields. I. Single neurons discharging before
saccades." J Neurophysiol 53(3): 603-63 5.
Bruce, C. J., M. E. Goldberg, M. C. Bushnell and G. B. Stanton (1985). "Primate frontal eye fields. II.
Physiological and anatomical correlates of electrically evoked eye movements." J Neurophysiol
54(3): 714-734.
Burgoon, P. W. and J. A. Boulant (2001). "Temperature-sensitive properties of rat suprachiasmatic
nucleus neurons." Am J Physiol Regul Integr Comp Physiol 281(3): R706-715.
Buschman, T. J. and E. K. Miller (2007). "Top-down versus bottom-up control of attention in the
prefrontal and posterior parietal cortices." Science 315(5820): 1860-1862.
Busto, R., W. D. Dietrich, M. Y. Globus, I. Valdes, P. Scheinberg and M. D. Ginsberg (1987). "Small
differences in intraischemic brain temperature critically determine the extent of ischemic
neuronal injury." J Cereb Blood Flow Metab 7(6): 729-738.
Cabanac, M. (1993). "Selective brain cooling in humans: "fancy" or fact?" FASEB J 7(12): 1143-1146;
discussion 1146-1147.
131
Cardin, J. A., M. Carlen, K. Meletis, U. Knoblich, F. Zhang, K. Deisseroth, L. H. Tsai and C. I. Moore
(2009). "Driving fast-spiking cells induces gamma rhythm and controls sensory responses."
Nature 459(7247): 663-667.
Cardin, J. A., M. Carlen, K. Meletis, U. Knoblich, F. Zhang, K. Deisseroth, L. H. Tsai and C. I. Moore
(2010). "Targeted optogenetic stimulation and recording of neurons in vivo using cell-typespecific expression of Channelrhodopsin-2." Nat Protoc 5(2): 247-254.
Carter, M. E. and L. de Lecea (2011). "Optogenetic investigation of neural circuits in vivo." Trends in
Molecular Medicine 17(4): 197-206.
Carter, M. E., 0. Yizhar, S. Chikahisa, H. Nguyen, A. Adamantidis, S. Nishino, K. Deisseroth and L. de
Lecea (2010). "Tuning arousal with optogenetic modulation of locus coeruleus neurons." Nat
Neurosci 13(12): 1526-1533.
Cavanaugh, J., W. M. Joiner and R. H. Wurtz (2012). "Suppressive Surrounds of Receptive Fields In
Monkey Frontal Eye Field." The Journal of Neuroscience 32(35): 12284-12293.
Cavanaugh, J., I. E. Monosov, K. McAlonan, R. Berman, M. K. Smith, V. Cao, K. H. Wang, E. S.
Boyden and R. H. Wurtz (2012). "Optogenetic inactivation modifies monkey visuomotor
behavior." Neuron 76(5): 901-907.
Chafee, M. V. and P. S. Goldman-Rakic (2000). "Inactivation of Parietal and Prefrontal Cortex Reveals
Interdependence of Neural Activity During Memory-Guided Saccades." Journal of
Neurophysiology 83(3): 1550-1566.
Chafee, M. V. and P. S. Goldman-Rakic (2000). "Inactivation of parietal and prefrontal cortex reveals
interdependence of neural activity during memory-guided saccades." J Neurophysiol 83(3): 1550-
1566.
Chan, A. W. S. (2009). "Transgenic Primate Research Paves the Path to a Better Animal Model: Are We
a Step Closer to Curing Inherited Human Genetic Disorders?" Journal of Molecular Cell Biology
1(1): 13-14.
Chan, A. W. S. and S.-H. Yang (2009). "Generation of transgenic monkeys with human inherited genetic
disease." Methods 49(1): 78-84.
Cheng, J., J. Zhang, C. Lu and L. Wang (2012). "Using optogenetics to translate the "inflammatory
dialogue" between heart and brain in the context of stress." Neuroscience Bulletin 28(4): 435-
448.
Chow, B. Y. and E. S. Boyden (2013). "Optogenetics and translational medicine." Sci Transl Med 5(177):
177ps175.
Chow, B. Y., X. Han and E. S. Boyden (2012). "Genetically encoded molecular tools for light-driven
silencing of targeted neurons." Prog Brain Res 196: 49-61.
Chow, B. Y., X. Han, A. S. Dobry, X. Qian, A. S. Chuong, M. Li, M. A. Henninger, G. M. Belfort, Y.
Lin, P. E. Monahan and E. S. Boyden (2010). "High-performance genetically targetable optical
neural silencing by light-driven proton pumps." Nature 463(7277): 98-102.
Christie, I. N., J. A. Wells, P. Southern, N. Marina, S. Kasparov, A. V. Gourine and M. F. Lythgoe
(2012). "fMRI response to blue light delivery in the naive brain: Implications for combined
optogenetic fMRI studies." Neurolmage 66C: 634-641.
Chuong, A. S., Miri, M. L., Busskamp, V., Matthews, G.A.C., Acker, L.C., Soresnsen, A.T., Young, A.,
Klapoetke, N. C., Henninger, M.A., Kodandaramaiah, S.B., Ogawa, M., Ramanlal, S. B., Forest,
C.R., Chow, B.Y., Han, X., Lin, Y., Tye, K.M., Roska, R., Cardin, J.A., Boyden, E.S. (2014).
"Non-invasive optogenetic neural silencing." submitted.
Ciocchi, S., C. Herry, F. Grenier, S. B. Wolff, J. J. Letzkus, I. Vlachos, I. Ehrlich, R. Sprengel, K.
Deisseroth, M. B. Stadler, C. Muller and A. Luthi (2010). "Encoding of conditioned fear in
central amygdala inhibitory circuits." Nature 468(7321): 277-282.
Claridge-Chang, A., R. D. Roorda, E. Vrontou, L. Sjulson, H. Li, J. Hirsh and G. Miesenb6ck (2009).
"Writing Memories with Light-Addressable Reinforcement Circuitry." Cell 139(2): 405-415.
Covington, H. E., M. K. Lobo, I. Maze, V. Vialou, J. M. Hyman, S. Zaman, Q. LaPlant, E. Mouzon, S.
Ghose, C. A. Tamminga, R. L. Neve, K. Deisseroth and E. J. Nestler (2010). "Antidepressant
132
Effect of Optogenetic Stimulation of the Medial Prefrontal Cortex." The Journal of Neuroscience
30(48): 16082-16090.
Cronemeyer, D. C. (1966). "Optical Absorption Characteristics of Pink Ruby." J. Opt. Soc. Am. 56(12):
1703-1705.
Dai, J., Daniel I. Brooks and David L. Sheinberg (2014). "Optogenetic and Electrical Microstimulation
Systematically Bias Visuospatial Choice in Primates." Current biology: CB 24(1): 63-69.
Dias, E. C., M. Kiesau and M. A. Segraves (1995). "Acute activation and inactivation of macaque frontal
eye field with GABA-related drugs." J Neurophysiol 74(6): 2744-2748.
Dias, E. C. and M. A. Segraves (1996). "The primate frontal eye field and the generation of saccadic eye
movements: comparison of lesion and acute inactivation/activation studies." Rev Bras Biol 56 Su
1 Pt 2: 239-255.
Dias, E. C. and M. A. Segraves (1999). "Muscimol-Induced Inactivation of Monkey Frontal Eye Field:
Effects on Visually and Memory-Guided Saccades." Journal of Neurophysiology 81(5): 2191-
2214.
Diester, I., M. T. Kaufman, M. Mogri, R. Pashaie, W. Goo, 0. Yizhar, C. Ramakrishnan, K. Deisseroth
and K. V. Shenoy (2011). "An optogenetic toolbox designed for primates." Nat Neurosci 14(3):
387-397.
Ding, L. and J. I. Gold (2012). "Neural Correlates of Perceptual Decision Making before, during, and
after Decision Commitment in Monkey Frontal Eye Field." Cerebral Cortex 22(5): 1052-1067.
Dixson, A. F. and C. M. Nevison (1997). "The socioendocrinology of adolescent development in male
rhesus monkeys (Macaca mulatta)." Horm Behav 31(2): 126-135.
Dow, B. M., A. Z. Snyder, R. G. Vautin and R. Bauer (1981). "Magnification factor and receptive field
size in foveal striate cortex of the monkey." Exp Brain Res 44(2): 213-228.
Eggert, H. R. and V. Blazek (1987). "Optical properties of human brain tissue, meninges, and brain
tumors in the spectral range of 200 to 900 nm." Neurosurgery 21(4): 459-464.
F., C. W. (1995). Summary of optical properties. Optical-Thermal Response of Laser-Irradiated Tissue.
W. A. J. a. v. G. M. J. New York, Platinum Press.
Felleman, D. J. and D. C. Van Essen (1991). "Distributed hierarchical processing in the primate cerebral
cortex." Cereb Cortex 1(1): 1-47.
Flock, S. T., B. C. Wilson and M. S. Patterson (1987). "Total attenuation coefficients and scattering phase
functions of tissues and phantom materials at 633 nm." Medical Physics 14(5): 835-841.
Freire, M. A., E. Morya, J. Faber, J. R. Santos, J. S. Guimaraes, N. A. Lemos, K. Sameshima, A. Pereira,
S. Ribeiro and M. A. Nicolelis (2011). "Comprehensive analysis of tissue preservation and
recording quality from chronic multielectrode implants." PLoS One 6(11): e27554.
Gattass, R., C. G. Gross and J. H. Sandell (1981). "Visual topography of V2 in the macaque." J Comp
Neurol 201(4): 519-539.
Gerits, A., R. Farivar, Bruce R. Rosen, Lawrence L. Wald, Edward S. Boyden and W. Vanduffel (2012).
"Optogenetically Induced Behavioral and Functional Network Changes in Primates." Current
Biology 22(18): 1722-1726.
Gottschalk, W. (1992). Ein Messverfahren zur Bestimmung der optischen Parameter biologischer Gevebe
in vitro. Ph.D., University of Karlsruhe.
Gradinaru, V., M. Mogri, K. R. Thompson, J. M. Henderson and K. Deisseroth (2009). "Optical
deconstruction of parkinsonian neural circuitry." Science 324(5925): 354-359.
Gradinaru, V., K. R. Thompson and K. Deisseroth (2008). "eNpHR: a Natronomonas halorhodopsin
enhanced for optogenetic applications." Brain Cell Biol 36(1-4): 129-139.
Gradinaru, V., K. R. Thompson, F. Zhang, M. Mogri, K. Kay, M. B. Schneider and K. Deisseroth (2007).
"Targeting and Readout Strategies for Fast Optical Neural Control In Vitro and In Vivo." The
Journal of Neuroscience 27(52): 14231-14238.
133
Gradinaru, V., F. Zhang, C. Ramakrishnan, J. Mattis, R. Prakash, I. Diester, I. Goshen, K. R. Thompson
and K. Deisseroth (2010). "Molecular and Cellular Approaches for Diversifying and Extending
Optogenetics." Cell 141(1): 154-165.
Gregoriou, G. G., S. J. Gotts, H. Zhou and R. Desimone (2009). "High-frequency, long-range coupling
between prefrontal and visual cortex during attention." Science 324(593 1): 1207-12 10.
Gregory Keating, E. and S. G. Gooley (1988). "Saccadic disorders caused by cooling the superior
colliculus or the frontal eye field, or from combined lesions of both structures." Brain Research
438(1-2): 247-255.
Guatteo, E., K. K. Chung, T. K. Bowala, G. Bernardi, N. B. Mercuri and J. Lipski (2005). "Temperature
sensitivity of dopaminergic neurons of the substantia nigra pars compacta: involvement of
transient receptor potential channels." J Neurophysiol 94(5): 3069-3080.
Gunaydin, L. A., 0. Yizhar, A. Berndt, V. S. Sohal, K. Deisseroth and P. Hegemann (2010). "Ultrafast
optogenetic control." Nat Neurosci 13(3): 387-392.
Gutierrez, D. V., M. D. Mark, 0. Masseck, T. Maejima, D. Kuckelsberg, R. A. Hyde, M. Krause, W.
Kruse and S. Herlitze (2011). "Optogenetic Control of Motor Coordination by Gi/o Proteincoupled Vertebrate Rhodopsin in Cerebellar Purkinje Cells." Journal of Biological Chemistry
286(29): 25848-25858.
Han, X. (2012). Chapter 11 - Optogenetics in the nonhuman primate. Progress in Brain Research. K.
Thomas and S. B. Edward, Elsevier. Volume 196: 215-233.
Han, X. and E. S. Boyden (2007). "Multiple-Color Optical Activation, Silencing, and Desynchronization
of Neural Activity, with Single-Spike Temporal Resolution." PLoS ONE 2(3): e299.
Han, X., B. Y. Chow, H. Zhou, N. C. Klapoetke, A. Chuong, R. Rajimehr, A. Yang, M. V. Baratta, J.
Winkle, R. Desimone and E. S. Boyden (2011). "A high-light sensitivity optical neural silencer:
development and application to optogenetic control of non-human primate cortex." Front Syst
Neurosci 5: 18.
Han, X., B. Y. Chow, H. Zhou, N. C. Klapoetke, A. Chuong, R. Rajimehr, A. Yang, M. V. Baratta, J.
Winkle, R. Desimone and E. S. Boyden (2011). "A high-light sensitivity optical neural silencer:
development, and application to optogenetic control of nonhuman primate cortex." Frontiers in
Systems Neuroscience 5.
Han, X., X. Qian, J. G. Bernstein, H. H. Zhou, G. T. Franzesi, P. Stern, R. T. Bronson, A. M. Graybiel, R.
Desimone and E. S. Boyden (2009). "Millisecond-timescale optical control of neural dynamics in
the nonhuman primate brain." Neuron 62(2): 191-198.
Hanes, D. P., W. F. Patterson, 2nd and J. D. Schall (1998). "Role of frontal eye fields in countermanding
saccades: visual, movement, and fixation activity." J Neurophysiol 79(2): 817-834.
Hanks, T. D., J. Ditterich and M. N. Shadlen (2006). "Microstimulation of macaque area LIP affects
decision-making in a motion discrimination task." Nat Neurosci 9(5): 682-689.
Hardingham, N. R. and A. U. Larkman (1998). "The reliability of excitatory synaptic transmission in
slices of rat visual cortex in vitro is temperature dependent." The Journal of Physiology 507(1):
249-256.
Haubensak, W., P. S. Kunwar, H. Cai, S. Ciocchi, N. R. Wall, R. Ponnusamy, J. Biag, H. W. Dong, K.
Deisseroth, E. M. Callaway, M. S. Fanselow, A. Luthi and D. J. Anderson (2010). "Genetic
dissection of an amygdala microcircuit that gates conditioned fear." Nature 468(7321): 270-276.
Herculano-Houzel, S. (2009). "The human brain in numbers: a linearly scaled-up primate brain." Front
Hum Neurosci 3: 31.
Horel, J. A., M. L. Voytko and K. G. Salsbury (1984). "Visual learning suppressed by cooling the
temporal pole." Behav Neurosci 98(2): 310-324.
Hubel, D. H. and T. N. Wiesel (1974). "Sequence regularity and geometry of orientation columns in the
monkey striate cortex." J Comp Neurol 158(3): 267-293.
Huber, D., L. Petreanu, N. Ghitani, S. Ranade, T. Hromadka, Z. Mainen and K. Svoboda (2008). "Sparse
optical microstimulation in barrel cortex drives learned behaviour in freely moving mice." Nature
451(7174): 61-64.
134
Jazayeri, M., Z. Lindbloom-Brown and G. D. Horwitz (2012). "Saccadic eye movements evoked by
optogenetic activation of primate V1." Nat Neurosci 15(10): 1368-1370.
Jing, W., W. Fabien, A. B. David, Z. Jiayi, 0. Ilker, D. B. Rebecca, V. N. Arto, W. Rick van, D. Ilka and
D. Karl (2012). "Integrated device for combined optical neuromodulation and electrical recording
for chronic in vivo applications." Journal of Neural Engineering 9(1): 016001.
Johansen, J. P., H. Hamanaka, M. H. Monfils, R. Behnia, K. Deisseroth, H. T. Blair and J. E. LeDoux
(2010). "Optical activation of lateral amygdala pyramidal cells instructs associative fear
learning." Proceedings of the National Academy of Sciences 107(28): 12692-12697.
Kahn, I., M. Desai, U. Knoblich, J. Bernstein, M. Henninger, A. M. Graybiel, E. S. Boyden, R. L.
Buckner and C. I. Moore (2011). "Characterization of the functional MRI response temporal
linearity via optical control of neocortical pyramidal neurons." J Neurosci 31(42): 15086-15091.
Keating, E. G. and S. G. Gooley (1988). "Saccadic disorders caused by cooling the superior colliculus or
the frontal eye field, or from combined lesions of both structures." Brain Res 438(1-2): 247-255.
Keller, E. L., K.-M. Lee, S.-W. Park and J. A. Hill (2008). "Effect of Inactivation of the Cortical Frontal
Eye Field on Saccades Generated in a Choice Response Paradigm." Journal of Neurophysiology
100(5): 2726-2737.
Kirchner, H., E. J. Barbeau, S. J. Thorpe, J. Regis and C. Liegeois-Chauvel (2009). "Ultra-rapid sensory
responses in the human frontal eye field region." J Neurosci 29(23): 7599-7606.
Kiyatkin, E. A. (2004). "Brain hyperthermia during physiological and pathological conditions: causes,
mechanisms, and functional implications." Curr Neurovasc Res 1(1): 77-90.
Kiyatkin, E. A. (2005). "Brain hyperthermia as physiological and pathological phenomena." Brain Res
Brain Res Rev 50(1): 27-56.
Kiyatkin, E. A. (2007). "Physiological and pathological brain hyperthermia." Prog Brain Res 162: 219243.
Kiyatkin, E. A. and P. L. Brown (2004). "Modulation of physiological brain hyperthermia by
environmental temperature and impaired blood outflow in rats." Physiol Behav 83(3): 467-474.
Kiyatkin, E. A., P. L. Brown and R. A. Wise (2002). "Brain temperature fluctuation: a reflection of
functional neural activation." European Journal of Neuroscience 16(1): 164-168.
Klapoetke, N. C., Y. Murata, S. S. Kim, S. R. Pulver, A. Birdsey-Benson, Y. K. Cho, T. K. Morimoto, A.
S. Chuong, E. J. Carpenter, Z. Tian, J. Wang, Y. Xie, Z. Yan, Y. Zhang, B. Y. Chow, B. Surek,
M. Melkonian, V. Jayaraman, M. Constantine-Paton, G. K. Wong and E. S. Boyden (2014).
"Independent optical excitation of distinct neural populations." Nat Methods 11(3): 338-346.
Koeppen, B. M. and B. A. Stanton (2009). Berne & Levy Physiology, Updated Edition, Elsevier Health
Sciences.
Kravitz, A. V., B. S. Freeze, P. R. L. Parker, K. Kay, M. T. Thwin, K. Deisseroth and A. C. Kreitzer
(2010). "Regulation of parkinsonian motor behaviours by optogenetic control of basal ganglia
circuitry." Nature 466(7306): 622-626.
Kubelka, P. (1948). "New Contributions to the Optics of Intensely Light-Scattering Materials. Part I."
Journal of the Optical Society of America 38(5): 448-448.
Lee, J. H., R. Durand, V. Gradinaru, F. Zhang, I. Goshen, D. S. Kim, L. E. Fenno, C. Ramakrishnan and
K. Deisseroth (2010). "Global and local fMRI signals driven by neurons defined optogenetically
by type and wiring." Nature 465(7299): 788-792.
Little, C. C. and J. Pearsons (1940). "The Results of a "Functional Test" in a Strain of Mice (C57 Black)
with a Low Breast Tumor Incidence." The American Journal of Cancer 38(2): 224-233.
Llewellyn, M. E., K. R. Thompson, K. Deisseroth and S. L. Delp (2010). "Orderly recruitment of motor
units under optical control in vivo." Nat Med 16(10): 1161-1165.
Lue, N., J. Bewersdorf, M. D. Lessard, K. Badizadegan, R. R. Dasari, M. S. Feld and G. Popescu (2007).
"Tissue refractometry using Hilbert phase microscopy." Opt Lett 32(24): 3522-3524.
Madisen, L., T. Mao, H. Koch, J. M. Zhuo, A. Berenyi, S. Fujisawa, Y. W. Hsu, A. J. Garcia, 3rd, X. Gu,
S. Zanella, J. Kidney, H. Gu, Y. Mao, B. M. Hooks, E. S. Boyden, G. Buzsaki, J. M. Ramirez, A.
R. Jones, K. Svoboda, X. Han, E. E. Turner and H. Zeng (2012). "A toolbox of Cre-dependent
135
optogenetic transgenic mice for light-induced activation and silencing." Nat Neurosci 15(5): 793-
802.
Masamizu, Y., T. Okada, H. Ishibashi, S. Takeda, S. Yuasa and K. Nakahara (2010). "Efficient gene
transfer into neurons in monkey brain by adeno-associated virus 8." Neuroreport 21(6): 447-451.
Maunsell, J. H. R. and S. Treue (2006). "Feature-based attention in visual cortex." Trends in
Neurosciences 29(6): 317-322.
McPeek, R. M., J. H. Han and E. L. Keller (2003). "Competition Between Saccade Goals in the Superior
Colliculus Produces Saccade Curvature." Journal of Neurophysiology 89(5): 2577-2590.
McPeek, R. M. and N. Takahashi (2006). "Deficits in covert attention after temporary inactivation of
macaque frontal eye field." Journal of Vision 6(6): 823.
Mesradi, M., A. Genoux, V. Cuplov, D. Abi Haidar, S. Jan, I. Buvat and F. Pain (2013). "Experimental
and analytical comparative study of optical coefficient of fresh and frozen rat tissues." Journal of
Biomedical Optics 18(11): 117010-117010.
Mobley, J. a. V.-D., T. (2003). Optical Properties of Tissue. Biomedical Photonics Handbook. T. VoDinh. Boca Raton, Florida, CRC Press. 1.
Mohler, C. W., M. E. Goldberg and R. H. Wurtz (1973). "Visual receptive fields of frontal eye field
neurons." Brain Res 61: 385-389.
Moore, T. and K. M. Armstrong (2003). "Selective gating of visual signals by microstimulation of frontal
cortex." Nature 421(6921): 370-373.
Moore, T. and M. Fallah (2004). "Microstimulation of the frontal eye field and its effects on covert spatial
attention." J Neurophysiol 91(1): 152-162.
Moore, T. and M. Fallah (2004). "Microstimulation of the Frontal Eye Field and Its Effects on Covert
Spatial Attention." Journal of Neurophysiology 91(1): 152-162.
Moran, M. M., H. Xu and D. E. Clapham (2004). "TRP ion channels in the nervous system." Current
Opinion in Neurobiology 14(3): 362-369.
Murphey, D. K. and J. H. R. Maunsell (2008). "Electrical microstimulation thresholds for behavioral
detection and saccades in monkey frontal eye fields." Proceedings of the National Academy of
Sciences 105(20): 7315-7320.
Niu, Y., Y. Yu, A. Bernat, S. Yang, X. He, X. Guo, D. Chen, Y. Chen, S. Ji, W. Si, Y. Lv, T. Tan, Q.
Wei, H. Wang, L. Shi, J. Guan, X. Zhu, M. Afanassieff, P. Savatier, K. Zhang, Q. Zhou and W. Ji
(2010). "Transgenic rhesus monkeys produced by gene transfer into early-cleavage-stage
embryos using a simian immunodeficiency virus-based vector." Proc Natl Acad Sci U S A
107(41): 17663-17667.
Ohayon, S., P. Grimaldi, N. Schweers and D. Y. Tsao (2013). "Saccade modulation by optical and
electrical stimulation in the macaque frontal eye field." J Neurosci 33(42): 16684-16697.
Osawa, S.-i., M. Iwasaki, R. Hosaka, Y. Matsuzaka, H. Tomita, T. Ishizuka, E. Sugano, E. Okumura, H.
Yawo, N. Nakasato, T. Tominaga and H. Mushiake (2013). "Optogenetically Induced Seizure and
the Longitudinal Hippocampal Network Dynamics." PLoS ONE 8(4): e60928.
Ozden, I., J. Wang, Y. Lu, T. May, J. Lee, W. Goo, D. J. O'Shea, P. Kalanithi, I. Diester, M. Diagne, K.
Deisseroth, K. V. Shenoy and A. V. Nurmikko (2013). "A coaxial optrode as multifunction writeread probe for optogenetic studies in non-human primates." Journal of Neuroscience Methods
219(1): 142-154.
Parthasarathy, H., J. Schall and A. Graybiel (1992). "Distributed but convergent ordering of corticostriatal
projections: analysis of the frontal eye field and the supplementary eye field in the macaque
monkey." The Journal of Neuroscience 12(11): 4468-4488.
Quintana, J., J. M. Fuster and J. Yajeya (1989). "Effects of cooling parietal cortex on prefrontal units in
delay tasks." Brain Research 503(1): 100-110.
Raimondo, J. V., L. Kay, T. J. Ellender and C. J. Akerman (2012). "Optogenetic silencing strategies differ
in their effects on inhibitory synaptic transmission." Nat Neurosci 15(8): 1102-1104.
Rengachary, S. S. a. E., R.G (2005) "Principles of Neurosurgery."
136
Roberts, T. F., S. M. H. Gobes, M. Murugan, B. P. Olveczky and R. Mooney (2012). "Motor circuits are
required to encode a sensory model for imitative learning." Nat Neurosci 15(10): 1454-1459.
Robles, F. E., S. Chowdhury and A. Wax (2010). "Assessing hemoglobin concentration using
spectroscopic optical coherence tomography for feasibility of tissue diagnostics." Biomed Opt
Express 1(1): 310-317.
Ruiz, 0., B. R. Lustig, J. J. Nassi, A. Cetin, J. H. Reynolds, T. D. Albright, E. M. Callaway, G. R. Stoner
and A. W. Roe (2013). "Optogenetics through windows on the brain in the nonhuman primate."
Journal of Neurophysiology 110(6): 1455-1467.
Schafer, R. J. and T. Moore (2011). "Selective attention from voluntary control of neurons in prefrontal
cortex." Science 332(6037): 1568-1571.
Schall, J. D., A. Morel, D. J. King and J. Bullier (1995). "Topography of visual cortex connections with
frontal eye field in macaque: convergence and segregation of processing streams." J Neurosci
15(6): 4464-4487.
Schiller, P. H. and I. Chou (2000). "The effects of anterior arcuate and dorsomedial frontal cortex lesions
on visually guided eye movements: 2. Paired and multiple targets." Vision Res 40(10-12): 16271638.
Schiller, P. H. and I. H. Chou (1998). "The effects of frontal eye field and dorsomedial frontal cortex
lesions on visually guided eye movements." Nat Neurosci 1(3): 248-253.
Schiller, P. H. and E. J. Tehovnik (2003). "Cortical inhibitory circuits in eye-movement generation." Eur J
Neurosci 18(11): 3127-3133.
Schone, C. and D. Burdakov (2012). "Glutamate and GABA as rapid effectors of hypothalamic
"peptidergic" neurons." Front Behav Neurosci 6: 81.
Sharp, A. A., A. M. Ortega, D. Restrepo, D. Curran-Everett and K. Gall (2009). "<emphasis
emphasistype="italic">In Vivo</emphasis> Penetration Mechanics and Mechanical Properties of
Mouse Brain Tissue at Micrometer Scales." Biomedical Engineering, IEEE Transactions on
56(1): 45-53.
Skinner, J. G. (1964). "Pumping Energy Distribution in Ruby Rods." Appl. Opt. 3(8): 963-965.
Smith, K. S., A. Virkud, K. Deisseroth and A. M. Graybiel (2012). "Reversible online control of habitual
behavior by optogenetic perturbation of medial prefrontal cortex." Proc Natl Acad Sci U S A
109(46): 18932-18937.
Sohal, V. S., F. Zhang, 0. Yizhar and K. Deisseroth (2009). "Parvalbumin neurons and gamma rhythms
enhance cortical circuit performance." Nature 459(7247): 698-702.
Sommer, M. A. and E. J. Tehovnik (1997). "Reversible inactivation of macaque frontal eye field."
Experimental Brain Research 116(2): 229-249.
Sommer, M. A. and R. H. Wurtz (1998). "Frontal eye field neurons orthodromically activated from the
superior colliculus." Journal of Neurophysiology 80: 3331-3335.
Sommer, M. A. and R. H. Wurtz (2001). "Frontal eye field sends delay activity related to movement,
memory, and vision to the superior colliculus." Journal of Neurophysiology 85(4): 1673-1685.
Sommer, M. A. and R. H. Wurtz (2002). "A pathway in primate brain for internal monitoring of
movements." Science 296(5572): 1480-1482.
Stanton, G. B., C. J. Bruce and M. E. Goldberg (1995). "Topography of projections to posterior cortical
areas from the macaque frontal eye fields." The Journal of Comparative Neurology 353(2): 291305.
Stanton, G. B., S. Y. Deng, E. M. Goldberg and N. T. McMullen (1989). "Cytoarchitectural characteristic
of the frontal eye fields in macaque monkeys." The Journal of Comparative Neurology 282(3):
415-427.
Star, W. M. (2011). Chapter 6: Diffusion Theory of Light Transport. Optical-Thermal Response of LaserIrradiated Tissue. A. J. Welch and M. J. C. van Gemert, Springer.
Star, W. M., J. P. Marijnissen and M. J. van Gemert (1987). "Light dosimetry: status and prospects." J
Photochem Photobiol B 1(2): 149-167.
137
Sterenborg, H. J., A. E. Saarnak, R. Frank and M. Motamedi (1996). "Evaluation of spectral correction
techniques for fluorescence measurements on pigmented lesions in vivo." J Photochem Photobiol
B 35(3): 159-165.
Stirman, J. N., M. M. Crane, S. J. Husson, A. Gottschalk and H. Lu (2012). "A multispectral optical
illumination system with precise spatiotemporal control for the manipulation of optogenetic
reagents." Nat. Protocols 7(2): 207-220.
Story, G. M., A. M. Peier, A. J. Reeve, S. R. Eid, J. Mosbacher, T. R. Hricik, T. J. Earley, A. C.
Hergarden, D. A. Andersson, S. W. Hwang, P. McIntyre, T. Jegla, S. Bevan and A. Patapoutian
(2003). "ANKTM1, a TRP-like channel expressed in nociceptive neurons, is activated by cold
temperatures." Cell 112(6): 819-829.
Tamura, K., Y. Ohashi, T. Tsubota, D. Takeuchi, T. Hirabayashi, M. Yaguchi, M. Matsuyama, T. Sekine
and Y. Miyashita (2012). "A glass-coated tungsten microelectrode enclosing optical fibers for
optogenetic exploration in primate deep brain structures." Journal of Neuroscience Methods
211(1): 49-57.
Tehovnik, E. J. and M. A. Sommer (1997). "Effective spread and timecourse of neural inactivation caused
by lidocaine injection in monkey cerebral cortex." J Neurosci Methods 74(1): 17-26.
Tehovnik, E. J., M. A. Sommer, I. H. Chou, W. M. Slocum and P. H. Schiller (2000). "Eye fields in the
frontal lobes of primates." Brain Research Reviews 32(2-3): 413-448.
Thomas, N. A. and L. J. Elias (2011). "Upper and lower visual field differences in perceptual
asymmetries." Brain Research 1387(0): 108-115.
Thompson, K. G., K. L. Biscoe and T. R. Sato (2005). "Neuronal Basis of Covert Spatial Attention in the
Frontal Eye Field." The Journal of Neuroscience 25(41): 9479-9487.
Tomita, H., E. Sugano, Y. Fukazawa, H. Isago, Y. Sugiyama, T. Hiroi, T. Ishizuka, H. Mushiake, M.
Kato, M. Hirabayashi, R. Shigemoto, H. Yawo and M. Tamai (2009). "Visual properties of
transgenic rats harboring the channelrhodopsin-2 gene regulated by the thy-1.2 promoter." PLoS
One 4(11): e7679.
Tomita, H., E. Sugano, H. Isago, T. Hiroi, Z. Wang, E. Ohta and M. Tamai (2010). "Channelrhodopsin-2
gene transduced into retinal ganglion cells restores functional vision in genetically blind rats."
Experimental Eye Research 90(3): 429-436.
Tonnesen, J., A. T. Sorensen, K. Deisseroth, C. Lundberg and M. Kokaia (2009). "Optogenetic control of
epileptiform activity." Proc Natl Acad Sci U S A 106(29): 12162-12167.
Tonnesen, J., A. T. Sorensen, K. Deisseroth, C. Lundberg and M. Kokaia (2009). "Optogenetic control of
epileptiform activity." Proceedings of the National Academy of Sciences 106(29): 12162-12167.
Tootell, R. B., E. Switkes, M. S. Silverman and S. L. Hamilton (1988). "Functional anatomy of macaque
striate cortex. II. Retinotopic organization." J Neurosci 8(5): 1531-1568.
Trubel, H. K. F., L. I. Sacolick and F. Hyder (2005). "Regional temperature changes in the brain during
somatosensory stimulation." J Cereb Blood Flow Metab 26(1): 68-78.
Tsai, H. C., F. Zhang, A. Adamantidis, G. D. Stuber, A. Bonci, L. de Lecea and K. Deisseroth (2009).
"Phasic firing in dopaminergic neurons is sufficient for behavioral conditioning." Science
324(5930): 1080-1084.
Tsunematsu, T., T. S. Kilduff, E. S. Boyden, S. Takahashi, M. Tominaga and A. Yamanaka (2011).
"Acute optogenetic silencing of orexin/hypocretin neurons induces slow-wave sleep in mice." J
Neurosci 31(29): 10529-10539.
Van Essen, D. C., W. T. Newsome and J. H. Maunsell (1984). "The visual field representation in striate
cortex of the macaque monkey: asymmetries, anisotropies, and individual variability." Vision Res
24(5): 429-448.
Vanhoutte, G., M. Verhoye and A. Van der Linden (2006). "Changing body temperature affects the T2*
signal in the rat brain and reveals hypothalamic activity." Magn Reson Med 55(5): 1006-1012.
Volgushev, M., I. Kudryashov, M. Chistiakova, M. Mukovski, J. Niesmann and U. T. Eysel (2004).
"Probability of Transmitter Release at Neocortical Synapses at Different Temperatures." Journal
of Neurophysiology 92(1): 212-220.
138
Williams, H. P. (2001). "Sir Harold Ridley's vision." British Journal of Ophthalmology 85(9): 1022-1023.
Wilson, B., M. Patterson and D. Burns (1986). "Effect of photosensitizer concentration in tissue on the
penetration depth of photoactivating light." Lasers in Medical Science 1(4): 235-244.
Wise, S. P., D. Boussaoud, P. B. Johnson and R. Caminiti (1997). "PREMOTOR AND PARIETAL
CORTEX: Corticocortical Connectivity and Combinatorial Computations 1." Annual Review of
Neuroscience 20(1): 25-42.
Witten, I. B., S.-C. Lin, M. Brodsky, R. Prakash, I. Diester, P. Anikeeva, V. Gradinaru, C. Ramakrishnan
and K. Deisseroth (2010). "Cholinergic Interneurons Control Local Circuit Activity and Cocaine
Conditioning." Science 330(6011): 1677-1681.
Xu, H., I. S. Ramsey, S. A. Kotecha, M. M. Moran, J. A. Chong, D. Lawson, P. Ge, J. Lilly, I. SilosSantiago, Y. Xie, P. S. DiStefano, R. Curtis and D. E. Clapham (2002). "TRPV3 is a calciumpermeable temperature-sensitive cation channel." Nature 418(6894): 181-186.
Yaroslavsky, A. N., P. C. Schulze, I. V. Yaroslavsky, R. Schober, F. Ulrich and H. J. Schwarzmaier
(2002). "Optical properties of selected native and coagulated human brain tissues in vitro in the
visible and near infrared spectral range." Phys Med Biol 47(12): 2059-2073.
Yizhar, 0., L. E. Fenno, T. J. Davidson, M. Mogri and K. Deisseroth (2011). "Optogenetics in neural
systems." Neuron 71(1): 9-34.
Zhang, F., M. Prigge, F. Beyriere, S. P. Tsunoda, J. Mattis, 0. Yizhar, P. Hegemann and K. Deisseroth
(2008). "Red-shifted optogenetic excitation: a tool for fast neural control derived from Volvox
carteri." Nat Neurosci 11(6): 631-633.
Zhang, Y.-P., N. Holbro and T. G. Oertner (2008). "Optical induction of plasticity at single synapses
reveals input-specific accumulation of aCaMKII." Proceedings of the National Academy of
Sciences 105(33): 12039-12044.
Zhang, Y., E. Ivanova, A. Bi and Z.-H. Pan (2009). "Ectopic Expression of Multiple Microbial
Rhodopsins Restores ON and OFF Light Responses in Retinas with Photoreceptor Degeneration."
The Journal of Neuroscience 29(29): 9186-9196.
Zhao, S., C. Cunha, F. Zhang, Q. Liu, B. Gloss, K. Deisseroth, G. Augustine and G. Feng (2008).
"Improved expression of halorhodopsin for light-induced silencing of neuronal activity." Brain
Cell Biology 36(1): 141-154.
Zhou, H. and R. Desimone (2011). "Feature-Based Attention in the Frontal Eye Field and Area V4 during
Visual Search." Neuron 70(6): 1205-1217.
Zorzos, A. N., E. S. Boyden and C. G. Fonstad (2010). "Multiwaveguide implantable probe for light
delivery to sets of distributed brain targets." Opt Lett 35(24): 4133-4135.
139