Digital Atlas of the Zebra Finch Brain (Taeniopygia Guttata): a Dimensional and High
Resolution Photo Atlas
Harvey J.
1
Karten ,
Agnieszka
1
Brzozowska-Prechtl ,
James
1
Prechtl ,
Haibin
2
Wang ,
and Partha P.
2
Mitra
1) Department of Neuroscience, University of California at San Diego, La Jolla, CA 92093, USA
2) Cold Spring Harbor Laboratory, Cold Spring Harbor, NY 11724, USA
Introduction
Neuroanatomical research is undergoing major change, driven by the
availability of automated scanning microscopes, ability to digitally store
and analyze tera/petabyte scale data sets, and to make these high
resolution images available through the internet. Here we present the
first high resolution Nissl stained digital images of the brain of the zebra
finch, which is the mainstay of songbird research.
The zebra finch has proven to be the most widely used model organism
for the study of the neurological and behavioral development of
birdsong. A unique strength of this research area is its integrative nature,
encompassing field studies and ethologically grounded behavioral
biology, as well as neurophysiological and molecular levels of analysis.
The availability of dimensionally accurate and detailed atlases and
photographs of the brain of male and female animals, as well as of the
brain during development, can be expected to play an important role in
this research program. Traditionally, atlases for the zebra finch brain
have only been available in printed format, with the limitation of low
image resolution of the cell stained sections.
One important feature is the fissure where the tectum and forebrain
come together. That fissure is not an unambiguous straight line, but on
average, it should be about 60° with respect to the plane of section as
shown in the left panel above.
As the first step in preparing the atlas, we photographed the block face
of serial images of frozen brain. Sections were cut at 30 micrometers,
and every other section was photographed with a digital camera
mounted above the microtome. These images provide a dimensionally
accurate set of registered images that were used to construct the initial
phase of the atlas.
Serial sections were collected, mounted on glass slides, stained with
either Nissl stain or myelin stains.
The staining pattern for Tyrosine Hydroxylase in the transverse plane is
shown below. TH is a rate limiting enzyme in the catecholamine synthesis
pathway and is therefore used as an indicator for catecholamines
(dopamine, norepinephrine, and epinephrine).
We have also obtained connectional data in a series showing
retinal projections in the zebra finch. This work is still preliminary
and not shown here. But a data set on chick using the same
technique provides some excellent examples of single axon tracing
of retinal efferent terminals [4], which is available at
www.brainmaps.org.
The cytoarchitectural high resolution photographs and atlas presented
here aim at facilitating electrode placement, connectional studies, and
cytoarchitectonic analysis. This initial atlas is not in stereotaxic
coordinate space. It is intended to complement the stereotaxic atlases of
Akutegawa and Konishi [1], and that of Nixdorf and Bischof [2].
The advantages of a digital atlas over a traditional paper-based atlas are
three-fold.
• The digital atlas can be viewed at multiple resolutions. At low
magnification, it provides an overview of brain sections and regions,
while at higher magnification, it shows exquisite details of the
cytoarchitectural structure.
• It allows digital “re-slicing” of the brain. The original photographs of
brain were taken in certain selected planes of section. However, the
brains are seldom sliced in exactly the same plane in real
experiments. Re-slicing provides a useful atlas in user-chosen
planes, which are otherwise unavailable in the paper-based version.
• It can be made available on the internet. High resolution histological
datasets can be independently evaluated in light of new experimental
anatomical, physiological and molecular studies.
Method
We have developed a method for reproducibly sectioning the brain
transversely in approximation of the “Frankfurt” plane of section.
Although the method is not perfect, it adds substantial value. It involves
orienting brain on the microtome according to a few anatomical features
on the lateral surface of the brain.
Discussion
The outer border of the brain, as well as large numbers of internal
structures, could be identified in photographs of the frozen block face.
The Nissl stained sections can thus be registered to the matching block
face photographs.
The lower right panel is enlarged below to reveal greater details of the
image with high resolution and contrast.
Finally, the complete series of the Nissl and Myelin stained tissue were
photographed with a scanning digital Aperio scanner at a resolution of
0.5 micrometers per pixel. The resulting RGB images were typically
about 1.4 Gigabytes per section, depending upon the area of individual
sections.
These images are available at www.zebrafinch.org/atlas, where they can
be viewed in multiple resolutions. The images were partitioned into
stacks of JPEG files. When being delivered over internet, they were
stitched together at client side via Brain Maps API [3] that allows
zooming to various resolution levels.
Acknowledgement
Results
This work is supported by the NIH (NS50436). PPM is also
supported by the W. M. Keck Foundation and the Crick-Clay
Professorship.
The brains are sectioned in a standardized “Frankfurt” plane that, for
anatomical and cytoarchitectural purpose, is preferable to that used in
the Stereotaxic plane of the Nixdorf and Bischof atlas. Using “video
template” imaging, we have been able to orient and section the brain in
a replicable plane in six cases. Some variance inevitably exists in light of
the variation in brain size even within a single cohort of male zebra finch.
The following figure shows a Nissl stained brain section at increasingly
higher resolutions. The cell bodies are clearly visible at the highest
resolution level (0.5 µm/pixel).
Although the results presented here are still very preliminary, the
high quality of the images clearly demonstrates the feasibility and
benefits of the encompassing cytoarchitectural study. The variance
across brains may be reduced by selecting all animals within a
limited range of body weight. Extending these methods to
experimental studies of molecular anatomy, connectional studies
and gene knockouts will provide an invaluable tool for data
assessment. On the computational side, we plan to segment and
register the brain sections automatically, as an intermediate step to
preparing outline drawings and annotations. The algorithms and
programs used here can be applied to other anatomy projects,
such as high-resolution digital mouse brain atlas that is currently in
progress in our lab and that of E.G. Jones at UC Davis
(www.brainmaps.org).
References
[1] Akutagawa E. and Konishi M., stereotaxic atalas of the brain of zebra finch, unpublished.
Next, we show a screenshot from a web browser with an image of a brain
section opened. Images at various resolution levels can be displayed
using the zoom-in and -out buttons at the upper left corner. The small
window at the lower right corner displays the location of the image within
the brain section.
[2] Nixdorf-Bergweiler B. E. and Bischof H. J., A Stereotaxic Atlas of the Brain Of the Zebra
Finch, Taeniopygia Guttata, http://www.ncbi.nlm.nih.gov.
[3] Brain Maps API, www.brainmaps.org.
[4] Karten H. J., single axon tracing of retinal efferent terminals in chick, www.brainmaps.org.