PROTOCOL
Immunogold quantification of amino acids and
proteins in complex subcellular compartments
Linda H Bergersen1, Jon Storm-Mathisen1 & Vidar Gundersen1,2
1Department
of Anatomy and the CMBN, University of Oslo, POB 1105 Blindern, 0317 Oslo, Norway. 2Department of Neurology, Rikshospitalet, Oslo, Norway.
Correspondence should be addressed to V.G. (vidar.gundersen@medisin.uio.no).
© 2008 Nature Publishing Group http://www.nature.com/natureprotocols
Published online 10 January 2008; doi:10.1038/nprot.2007.525
An increasing number of imaging techniques are in use to study the localization of molecules involved in cell-to-cell signaling. Here
we describe the use of immunogold procedures to detect and quantify molecules on electron micrographs. To measure the areas
of the subcellular compartments under investigation, the protocol uses an overlay screen with an array of regularly spaced points.
On the basis of this, the densities of the gold-labeled molecules can be calculated. Despite the limited lateral resolution of the
immunogold method as used by many investigators (B30 nm), it is possible to measure the content of molecules associated with
tiny tissue compartments, e.g., synaptic vesicles and different types of membrane, such as plasma membranes and vesicle
membranes. The quantification protocol can be carried out without using computer programs. The entire protocol can be completed
in B15 d.
INTRODUCTION
In the brain, neurons contact each other through synapses. A
synapse consists of a presynaptic terminal and a postsynaptic
dendrite, separated by the synaptic cleft, which is only B20-nm
wide1. Tightly surrounding the synapses are astrocytes, which wrap
their tiny processes around the pre- and postsynaptic neuronal
elements2–4. Thus, the membranes of postsynaptic dendrites, presynaptic terminals and astrocytic processes are in close apposition,
they can be situated within a distance of B20 nm.
Localizing molecules such as transporters or receptors to these
membranes require microscopic techniques with a high resolution.
Confocal immunofluorescence microscopy is often used to detect
membrane proteins. However, even though modern optical imaging techniques can provide information about sub-resolution
events5, they cannot offer the resolution necessary to localize a
molecule reliably to defined membranes at the synapse. This can be
obtained by electron microscopic methods, of which immunogold
cytochemistry shows the highest lateral resolution, which theoretically is B23 nm6, depending on the type of secondary antibodies
used and on the size of the gold particle (see Box 1). As the distance
between apposing membranes at the synapse is often shorter than
the immunogold resolution, the distribution of the labeling for a
given protein across the membranes does not directly show to
which membrane the protein belongs.
Likewise, as the diameters of synaptic vesicles in the presynaptic
terminals are only B30 nm2,7,8, and the vesicles are often closely
MATERIALS
REAGENTS
. Experimental animals (Wistar rats) ! CAUTION Experiments involving live
rodents must conform to local and national regulations.
. Pentobarbital ! CAUTION Toxic.
. 25% (wt/vol) glutaraldehyde (GA; Electron Microscopic Sciences, cat. no.
16210). ! CAUTION Toxic (irritant, allergen, carcinogen).
. Paraformaldehyde (PFA; Electron Microscopic Sciences, cat. no. 19210).
! CAUTION Toxic.
. Dextran 70 (Mw 70000; Sigma-Aldrich, cat. no. 31390)
. Lowicryl HM20 (Polysciences)
. Sodium azide (Sigma-Aldrich, cat. no. 71290)
144 | VOL.3 NO.1 | 2008 | NATURE PROTOCOLS
packed, localizing transmitters or membrane proteins to individual
synaptic vesicles poses resolution problems that cannot be solved by
confocal microscopy. One must also bear in mind that transmitter
amino acids, such as glutamate, are present at quite high concentrations also in the cytoplasmic matrix of the terminals, making it
difficult to distinguish between the vesicular and cytosolic content
of glutamate. As discussed in this protocol, these problems can be
addressed with the immunogold method combined with various
ways of quantification.
Furthermore, whereas confocal immunofluorescence microscopy is dependent on the use of markers to identify subcellular
compartments of a cell, such compartments can be directly
visualized by the electron microscopic immunogold method. In
addition, immunogold electron microscopy gives the possibility to
quantify reliably and compare the contents of molecules between
different membrane domains or between different intracellular
compartments of the same and different cell types.
Here we present quantitative postembedding electron microscopic immunogold protocols that we have used to localize
transmitters, such as glutamate and GABA, and vesicular neurotransmitter transporters to synaptic and synaptic-like vesicles, as
well as transmitter receptors to distinct membrane domains. These
protocols can be used to locate and quantify other types of antigens,
such as constituents of membranes and cytosol, assuming that
appropriate antibodies are available.
. Glycerol (Sigma-Aldrich, cat. no. G5516)
. Na2HPO4 and NaH2PO4 (Sigma-Aldrich, cat. nos. 71496 and 71642)
. Triton X-100 (Sigma-Aldrich, cat. no. 234729)
. Tris–HCl/Tris–base (Sigma-Aldrich, cat. nos. T3253 and T1503)
. Sodium borohydride (Sigma-Aldrich, cat. no. 452882)
. Gly (Sigma-Aldrich, cat. no. 50050)
. 30% (wt/vol) H2O2 (Sigma-Aldrich, cat. no. 95302)
. NaOH (Sigma-Aldrich, cat. no. 71692)
. Absolute ethanol (Arcus Products)
. Human serum albumin (HSA; Sigma-Aldrich, cat. no. A1653)
. Polyethylene glycol (Sigma-Aldrich, cat. no. P2263)
PROTOCOL
BOX 1 | RESOLUTION OF THE IMMUNOGOLD METHOD
Secondary
antibody
coupled to a
collodial gold
particle
(usually 5 –15 nm)
© 2008 Nature Publishing Group http://www.nature.com/natureprotocols
In practice
about 30 nm
(10 nm gold
particle)
Primary antibody
Theoretically the distance between the center of a gold particle and the epitope is B23 nm for
gold particles with a diameter of 15 nm6. However, previous immunogold experiments have
shown that the lateral resolution is somewhat lower27–29. From these studies and depending on
the size of the gold particle the lateral resolution of the immunogold method can be
estimated to be B30 nm. In our experience, even if we use Fab fragments of the secondary
antibodies and 10 nm gold particles we get a lateral resolution of B30 nm. This means that
in many immunogold setups gold particles situated at some distance from the epitope (B30 nm)
still could represent a specific signal. It is essential to take into account this inherent feature
of the immunogold method when one wants to use the method for quantitative purposes. However,
the lateral resolution can be increased by using colloidal gold particles directly coupled to the
primary antibodies, making it possible to omit the secondary antibodies, and/or by using
smaller gold particles.
Antigen
. Lead citrate (Electron Microscopic Sciences, cat. no. 17800)
. Uranyl acetate (Sigma-Aldrich, cat. no. 73943)
. Parafilm (Pechiney Plastic Packing)
. Primary antibodies
. Secondary antibodies coupled to colloidal gold particles
EQUIPMENT
. Peristaltic pump (Watson-Marlow U/D 323; Watson-Marlow Bredel)
. Cryofixation unit (Reichert KF80, Reichert)
. Cryosubstitution unit (Reichert)
. Ultramicrotome (Reichert)
. Diamond knife for ultrathin sectioning (Diatome)
. Nickel mesh grids (300–600 mesh) (300 mesh square; Electron Microscopic
Sciences, cat. no. G300-Ni)
. Tweezers (Dumont tweezers; Electron Microscopic Sciences, cat. no. 72800-D)
. Electron microscope (Philips CM10 or Teknai 12) with a software (AnalySIS,
Olympus) for image acquisition
REAGENT SETUP
0.1 M sodium phosphate buffer (PB) Dissolve 11.5 g Na2HPO4 and 2.3 g
NaH2PO4 in 1,000 ml ultra filtered water (UFWater; MilliQ, 18MO). Adjust pH
to 7.4 by mixing the alkaline and acidic component.
Fixative Freshly prepare 500 ml of a 4% (wt/vol) PFA and 0.1% (vol/vol)
GA solution or a 1% (wt/vol) PFA and 2.5% (vol/vol) GA solution in PB.
Dissolve 20 g PFA in PB, add 2 ml 25% GA or dissolve 5 g PFA in PB,
add 50 ml 25% GA and bring the volumes to 500 ml. m CRITICAL This
solution must be made fresh. To dissolve the PFA efficiently, the solution
should be heated to B70 1C under constant stirring with a magnetic
stirrer in a fume hood. After the PFA solution is cooled, filter it to avoid
precipitates in the fixative.
2% (vol/vol) H2O2 Dilute 66 ml 30% H2O2 in 0.1 M PB and bring the volume
to 1,000 ml.
0.05 M Tris buffer Dissolve 0.97 g Trizma base and 6.61 g Trizma–HCl in
UFWater and bring the volume to 1,000 ml. Check that pH is B7.4.
0.3% (wt/vol) or 0.9% (wt/vol) NaCl Dissolve 3 g or 9 g NaCl in UFWater and
bring the volume to 1,000 ml.
Tris-buffered saline Triton (TBST) Mix 100 ml 0.05 M Tris and 900 ml 0.9%
or 0.3% NaCl and add 1 ml Triton X-100. Triton is a detergent that can
contribute to an enhancement of the immunogold signal. ! CAUTION Avoid skin
and eye contact.
Blocking solution: TBST with 2% (wt/vol) HSA To 1 ml solution, add 0.02 g
HSA.
Sodium borohydride/Gly solution Dissolve 10 mg sodium borohydride and
37.5 mg Gly in 10 ml TBST. m CRITICAL Use fresh. The solution makes bubbles
of nascent hydrogen gas, so make it in a 50-ml tube.
Contrast-enhancing agents Make a 1% (wt/vol) solution of uranyl
acetate (1 g in 100 ml) and a 0.3% (wt/vol) lead citrate (0.3 g in 100 ml)
solution in UFWater. Uranyl and lead are used as contrast-enhancing agents.
They are heavy metals that bind to proteins, nuclear components
and different types of membrane. m CRITICAL Protect from light.
Use aluminum foil to cover the top of the Petri dish. Do not breathe
on solution when working. CO2 causes precipitation of lead carbonate
on the grids. ! CAUTION Uranyl is radioactive and should be stored
covered by a lead sheathing. Lead and uranyl are toxic, avoid contact with
skin and eye.
Sodium ethanolate Make a saturated solution of 100 g NaOH in 700 ml
absolute ethanol. Make the solution 3 d in advance, leave it on a magnetic
stirrer for 1 d.
PROCEDURE
Day 1: Perfusion fixation
1| Anesthetize the experimental rats (pentobarbital, 0.3 ml per 100 g) according to animal welfare laws.
2| Perfuse the rats transcardially with the fixative through a peristaltic pump (50 ml min1 for adult rats, 5 ml min1 for P14-28
rats) for 10 min after a brief flush (for a few seconds) of 2% dextran 70 in PB.
m CRITICAL STEP Because GA is required for efficient fixation of free amino acids to tissue proteins, in the immunogold protocol for
detection of amino acids we use a mixture of 2.5% GA and 1% FA, a fixation that gives optimum retention of tissue amino acids9.
Such a high GA concentration will quench the immunogold signal for most membrane proteins. Thus, in the protocol for immunogold
detection of membrane proteins we use a mixture of 0.1% GA and 4% FA. It is useful to add a low concentration of GA to the fixative
also for protein immunocytochemistry, because this gives a better preservation of tissue morphology.
3| Place the fixed rats with the brains in situ in plastic bags and leave overnight at 4 1C.
Day 2: Dissection and cryoprotection
4| Dissect out the brain area of interest and make small specimens (o1 mm3).
NATURE PROTOCOLS | VOL.3 NO.1 | 2008 | 145
PROTOCOL
’ PAUSE POINT The fixed brain specimens can be stored for some weeks in the fixative diluted 1:10 in PB. If so, sodium azide
(0.1%, wt/vol) should be added to the storing solution. In amino acid immunogold cytochemistry, we often leave the fixed
tissue in the undiluted fixative until embedding.
© 2008 Nature Publishing Group http://www.nature.com/natureprotocols
5| Rinse the specimens in PB and cryoprotect in increasing concentrations of glycerol (10%, 20% and 30% (vol/vol) in PB;
0.5 h for each at room temperature (B20 1C) and overnight in 30% at 4 1C).
Days 3–7: Embedding procedure
6| Embed the tissue in Lowicryl HM20, which is a water-insoluble acryl-based resin that polymerizes at low temperatures
catalyzed by UV light. We use a freeze-substitution embedding protocol in which the tissue is first quickly frozen (to avoid
formation of ice crystals) in liquid propane (170 1C). Then the specimens are dehydrated and treated with uranyl acetate
(1.5% dissolved in anhydrous methanol) to increase the contrast of the tissue (45 1C, 30 h). The infiltration with the resin
(24 h) and the first phase of UV light ‘curing’ occurs at 45 1C (24 h), before continuing UV irradiation at 0 1C (35 h). Freezeembedding with Lowicryl HM20 preserves the antigens to a large extent and results in a superior immunogold sensitivity, but
the ultrastructure may be inferior compared with ‘conventional’ osmicated and epoxy embedded tissue.
Day 8: Ultrathin sectioning
7| Cut ultrathin sections on an ultramicrotome with a diamond knife at 70–100 nm. We choose the ultrathin sections that
have a bright gold color. Place the ultrathin sections on nickel mesh grids (routinely 300 mesh).
m CRITICAL STEP Often the ultrathin sections accidentally detach during the immunogold-labeling procedure. To reduce this risk
we coat the mesh grids with a glue (PAP pen; Daido Sangyo) before attaching the ultrathin sections to the grids.
8| Leave the ultrathin sections to dry on the grids at room temperature (B20 1C) for at least 12 h before using them in the
immunogold procedure.
’ PAUSE POINT The grids with the ultrathin sections can be put in a grid box (Leica Microsystem) and stored dry in darkness
for several months.
Days 9–10: Immunogold procedure
9| Prepare the primary-antibody solutions before starting the immunogold procedure. If we use amino acid antibodies that
turn out to show slight cross-reactivities against other amino acids in our extensive test systems, we add soluble aldehyde
complexes of these amino acids to the primary-antibody solution. Next, we make negative control solutions by adding the
complexes of aldehydes and the amino acid used for amino acid antibody production or the peptide used for generation of the
protein antibodies to the primary-antibody solutions (as described in Box 2). These additives are put in the primary-antibody
solutions at least 3 h before applying the antibodies to the tissue and test sections.
m CRITICAL STEP The entire procedure is normally performed at room temperature (B20 1C).
10| Prepare an ‘incubation chamber’ for the immunogold labelings. We use a large Petri dish with Parafilm placed in the
bottom. Droplets (30–50 ml per grid) of the different incubation solutions and the rinsing solution are placed on the Parafilm
with a pipette. The Parafilm lies on a moistened filter paper cut to cover the base of the dish, ensuring that the droplets are not
dried up during the incubations (Fig. 1).
11| Etch the resin with hydrogen peroxide (15 min) or sodium ethanolate.
! CAUTION Take care when moving the grids with the tweezers.
m CRITICAL STEP Expose the ultrathin sections to sodium ethanolate for only about a second, or else the sections will be dissolved.
12| Rinse in PB in 10-ml glass beakers (three times 20 s). Perform all 20-s rinsing by immersing the grids, held by the tweezers,
in 10–20 ml TBST/UFWater in three subsequent glass beakers.
13| Incubate the sections in borohydride/Gly to neutralize free aldehyde groups in the fixed tissue for 10 min. This may reduce
background labeling.
14| Rinse in TBST in 10-ml glass beakers (three times 20 s).
15| Incubate the sections in the blocking solution for 10 min. The soluble proteins in the blocking solution will compete with
proteins in the tissue for nonspecific binding of the antibodies. Consequently, the proteins will reduce the nonspecific
background labeling. An alternative or supplementary way of reducing the background is by increasing the ion concentration in
the blocking solution. We normally use 2% HSA and 0.3% NaCl in the blocking solution, especially if low labeling sensitivity is a
problem. If high background labeling is a problem we may increase these concentrations to 3% and/or 0.9%, respectively.
146 | VOL.3 NO.1 | 2008 | NATURE PROTOCOLS
PROTOCOL
© 2008 Nature Publishing Group http://www.nature.com/natureprotocols
BOX 2 | CONTROL OF PRIMARY-ANTIBODY SPECIFICITY
For amino acid immunogold cytochemistry the antibody specificity is tested by use of ultrathin test sections. These are made of conjugates of
different amino acids linked to brain macromolecules by glutaraldehyde–formaldehyde, which in turn are freeze-dried and embedded in epoxy
resin20. A simpler test system can be prepared by gelatin embedding amino acids or other compounds aldehyde fixed to proteins (e.g., bovine
serum albumin) before embedding in resin21. In each immunogold experiment such ultrathin test sections are processed along with the
ultrathin tissue sections, ascertaining the specificity of the labeling produced. For example, we can show that our glutamate and GABA
antibodies react only with the glutamate and GABA conjugates, respectively. As an extra negative control, we ensure that the amino acid
immunoreactivities of tissue and test sections are blocked by adding soluble complexes of glutaraldehyde and formaldehyde (weight proportion
2.5:1) and the amino acid (0.3 mM) against which the antibodies were raised to the primary-antibody solution before performing the
immunogold procedure.
Importantly, before using them in the immunogold procedure all our amino acid antibodies are run through an extensive specificity testing
against B40 low molecular weight molecules endogenous to the brain (for details, see, e.g., refs. 21, 22, 23). We may find that some amino acid
antibodies show a slight cross-reactivity against other amino acids. Such cross-reactivities can be abolished by adding soluble aldehyde
complexes of these ‘cross-reacting’ amino acids to the primary-antibody solution8,24,25.
For detection of proteins, we test the antibodies on Western blots of brain homogenates before applying the antibodies on the tissue
sections. Only antibodies that produce bands of appropriate molecular mass are used in the immunogold experiments. Like in the amino acid
protocol, if available we preabsorb the anti-protein antibodies with the peptide used for immunization as a specificity control. It should be
emphasized, however, that this is a poor specificity test: antibodies cross-reacting with a different protein will also be removed by the
preabsorption. In addition, we test the specificity of the secondary antibodies by omitting the primary antibodies from the immunogold
protocol. If available the specificity of the antibodies to a given protein can be tested in cultured cells genetically engineered to express the
protein26 (a prerequisite is that the cells normally do not express the protein) or in animals in which the protein is knocked out.
16| Transfer the sections to the droplets with the primary antibodies diluted in the blocking solution. We often incubate
overnight in room temperature (B20 1C). The antibodies must be diluted to an optimal working concentration, which is
determined for each type of antibody in separate experiments. Remember that separate ‘negative control solutions’ as described
in Box 2 should have been made in advance.
m CRITICAL STEP If background labeling is a problem, increase the dilution of the antibodies, reduce the incubation times or
incubate at 4 1C and ensure that the primary antibodies are specific (see Box 2).
17| Rinse in TBST (three times 20 s, one time 10 min, three times 20 s and one time 10 min). Perform the 10-min rinsing in
droplets of 50 ml TBST.
18| Incubate the sections in the blocking solution for 10 min.
19| Incubate the sections with the secondary antibodies diluted (as recommended by the manufacturer) in the blocking
solution for 2 h. We routinely use gold particles that are 10 nm in diameter, which we find give a better labeling intensity than
gold particles with larger diameters.
20| Rinse in UFWater as in Step 16.
21| Dry and check sections. Discard grids from which the sections have detached.
22| Incubate the sections with 1% uranyl acetate (15 min).
23| Rinse in UFWater (three times 20 s).
24| Dry sections.
25| Incubate the sections with 0.3% lead citrate (1 min).
26| Rinse in UFWater (three times 20 s).
27| Dry sections.
Figure 1 | Photograph of the immunogold incubation set up. The incubations
are performed in a large Petri dish. The grids with the ultrathin sections are
put in droplets on Parafilm, which is placed on top of a moisturized filter
paper (the short irregular black lines indicate the border between the Parafilm
and the filter paper). For rinsing, each section is put in a droplet of 50 ml
rinsing solution. For antibody incubations, the sections are arranged clockwise
(grid nos. 1–4) in a large drop of antibody solution (30 ml per section). Note
that the lid is covered with aluminium foil, which is used for light protection.
This is especially important in the contrasting step. The tweezers hold a grid
(Photograph by S.G. Owe).
PETRI
DISH
Lid with aluminium foil
Parafilm
Moistured
filter paper
Grids in droplets of rinsing solution
Tweezers with grid
Grids in large drop of antibody solution
NATURE PROTOCOLS | VOL.3 NO.1 | 2008 | 147
PROTOCOL
© 2008 Nature Publishing Group http://www.nature.com/natureprotocols
Days 11–12: Image acquisition
28| Take electron micrographs. We take electron micrographs with a Tecnai 12 or Philips CM10 electron microscope, usually at a
magnification of 460,000. The micrographs are stored as TIFF files and printed on paper on a high resolution printer. We
usually use the printed electron micrographs in the quantitative analysis.
Day 13– : Quantitative immunogold analysis
29| Perform quantitative analysis. Several problems can be addressed, examples of which are given below. One of the main
advantages with the immunogold method is that only epitopes exposed on the surface of the resin are labeled by the
antibodies, that is, when compared with preembedding electron microscopic methods, differences in hindrance of antibody
penetration into the various tissue compartments in the section are minimized. This allows a reliable quantification and
comparison of the density of gold particles between different tissue compartments. By counting the number of gold particles
representing a given antigen (transmitters, transporters or receptors) within the defined tissue compartments on an electron
micrograph and then measuring the areas of the compartments, it is possible to calculate the densities of an antigen in
different tissue compartments (option A). This is based on the ‘Delesse Principle’ (after the French geologist Delesse, 1847)
stating that the proportion a compound makes up of the volume of a structure is the same as the proportion it makes up of the
area of a section through the structure10. However, if we want to quantify the content of immunogold particles in tiny tissue
elements, such as membranes, we must take into account the lateral resolution of the immunogold method (B30 nm, see
Box 1). By doing so it is possible to address whether transporters and receptors are located to various types of membrane of
brain cells (e.g., plasma membranes and synaptic vesicle membranes), see option B. One important issue that can be addressed
with the immunogold method is whether a putative transmitter is located in synaptic vesicles. Again, because the lateral
resolution of the immunogold method (B30 nm) (Box 1) is similar to the diameter of synaptic vesicles (30–60 nm (refs. 2,7,8)),
the intervesicular distance, and the thickness of the ultrathin section, gold particles cannot be assigned to single synaptic
vesicles. Hence, a gold particle could be located outside the vesicle and still signal an epitope within the vesicle and vice versa.
Further adding to this problem is the fact that most amino acid transmitters are present in quite high extrasynaptic
Figure 2 | Electron micrographs showing the principles of the immunogold
a
b
quantification described in this protocol. (a) A micrograph of an axon and its
ensheathing myelin. The gold particles represent NMDA receptors. In (b) an
overlay with three types of regularly spaced points is superimposed on the
c
Myelin
micrograph. Within the unit (u) enclosed by stippled lines (area u2) there are
Axon
16 points not part of a line, 4 points at the end of a line and 1 encircled
u
Mito
point at the end of a line. The three types of points are suitable for measuring
areas of different sizes. The points are defined as ‘mathematical points’: the
Mito
u
intersections between the lines formed by the borders between black and
white, e.g., at the upper right areas of the crosses. Similarly, the location of
the gold particles is defined as the ‘mathematical point’ in its center. Thus,
d
e
f
g
there are no cases of ambiguous location of intersections of particles. When
Mito
estimating profile areas all points hitting the profile under investigation are
Ter
counted. For example, over part of the myelin, delineated in green, there are
19 intersections, whereas over the axon delineated in white there are 43
u
intersections. The area of the delineated myelin is 19/20 u2, whereas that of
Mito
the axon is 43/20 u2. In the myelin there are 11 gold particles giving a
uh
density of 11.6 particles/u2. In the axon there are no particles. The selected
*
myelin and axon areas were chosen as examples of how to use the
PSD
quantitative method. In our quantifications of cytoplasmic matrix, we include
*
Sp
the entire profiles under investigation, subtracting areas of mitochondria and
i
irrelevant areas with artifacts or poor morphology (e.g., the areas at the end
of the well defined myelin sheets in Figure 2a). (c) A higher magnification of
the area outlined by blue. The yellow line delineates the midline of the outer myelin membrane. As the resolution of the immunogold method is B30 nm
(see text and Box 1), gold particles situated within this distance can signal epitopes in the membrane. Thus, lines (red) are drawn 30 nm on each side of the
midline of the outer myelin membrane. The number of intersections (6) as well as the number of gold particles (3) are counted within this defined area, giving a
density of 10.0 gold particles/u2. (d) An electron micrograph showing a terminal (Ter) with synaptic vesicles (one is marked with an arrowhead) making a
synapse (asterisks) with an asymmetric specialization on a dendritic spine (Sp). The small gold particles in the terminal represent glutamate. In (e) a similar
overlay as in (b) is superimposed. (f) and (g) show a higher magnification of the area containing the vesicle marked with the yellow arrowhead. (f) shows a
micrograph without the overlay for better visualization of the vesicles. In (g) the membrane of the synaptic vesicles is delineated with a yellow circle. Taking
into account the lateral resolution of the immunogold method, a red circle is drawn 30 nm outside the yellow one. The numbers of gold particles (1) and of
intersections (3) are counted within the defined area, giving a density of 6.7 gold particles/u2. (h) and (i) are higher magnifications of the synaptic area
outlined in blue in (e). In (h) the density of gold particles associated with the membrane overlying the postsynaptic density (PSD) is estimated. The yellow line
covers the midline of the postsynaptic membrane. An area is made by drawing lines (red) 30 nm outside the yellow one. From the number of intersections
(5) and gold particles (2) within this area the density of gold particles in the postsynaptic membrane could be calculated (8 particles/u2). In (i) the
perpendicular distances from the center of the gold particles to the midline of the postsynaptic membrane are measured. These distances could be presented as
frequency distributions, allowing us to assess on which side of the membrane the epitope is localized. (The overlay is adapted from ref. 11.)
148 | VOL.3 NO.1 | 2008 | NATURE PROTOCOLS
PROTOCOL
BOX 3 | ASSESSMENT OF BACKGROUND LABELING
concentrations. To bypass these problems we use two different approaches (option C). The latter method of making an
association between a chemical component and a synaptic vesicle can also be used to study the association between two
proteins thought to be situated in the vesicle membrane (option D). Box 3 describes assessment of background labeling.
(A) Determining the density of substances in tissue compartments
(i) Measure the areas of the tissue compartments; this can be done in a number of different ways. We normally use a
stereological method in which we apply an overlay screen11 (Fig. 2). From the number of intersections (points) within a
given tissue profile, we can determine the area of this profile and calculate its density of immunogold particles (Fig. 2b).
This approach we have used to inter alia compare the density of glutamate NMDA receptors between axons and their
myelin sheets12 (Fig. 3a,b). Alternatively, after image acquisition the area determination of tissue profiles could be done
electronically using a computer program (ImageJ) (see ref. 13). The point counting method is particularly useful for small
and complicated components.
(B) Localizing transporter and receptor proteins to distinct membrane domains
(i) Draw a line 30 nm on each side of the membrane in question (exemplified in Fig. 2c,h) and determine the defined
areas using the stereological method with the overlay screen. In this way a stretch of membrane is ‘converted’ into an
area. All immunogold particles located within this area may belong to the membrane and are included in the
quantification. Similarly, particles due to ‘cross-firing’ from any significant immunoreactivity in the surroundings should be
subtracted. Exploiting this ‘trick’, we have quantified the densities of glutamate NMDA receptors in the outer and inner
membranes of the myelin sheets and in the membrane of the postsynaptic density of excitatory synapses12 (Fig. 3a,c), as
well as of the vesicular GABA transporters in synaptic-like microvesicles in cells of the endocrine pancreas14.
(ii) Because different types of membrane are very closely apposed at the synapse it is difficult to decide whether a gold
particle belongs to a certain membrane type. To answer this, determine the distribution of gold particles across a given
membrane (e.g., pre- or postsynaptic membranes, facing or overlying the postsynaptic density, respectively). First,
measure the perpendicular distance between the center line of the membrane and the center of the gold particles within a
distance of 50–100 nm in each direction from this line (Fig. 2i). Sort these distances into bins of various widths
(e.g., 10–20 nm). If the epitope is situated on the intracellular side of the membrane, this can reveal itself as a particle
distribution skewed to this side, thus indicating localization of the antigen on one rather than on the other of two closely
apposed membranes. Using this approach we have determined the distribution of the glutamate NMDA receptors across
extrasynaptic membranes of excitatory nerve terminals (Fig. 4)3, and over synaptic membranes of excitatory synapses
(Fig. 4b)15. The distributions of glutamate AMPA receptors16 and purinergic receptors17 have been similarly studied.
(C) Localizing transmitters to synaptic vesicles
(i) Count the number of gold particles over areas delimited by a circle drawn 30 nm outside the vesicle membrane
(exemplified in Fig. 2f,g), as well as over areas consisting of vesicle-free cytoplasmic matrix (gold particles situated closer
than 30 nm to other intracellular organelles or the plasma membrane are discarded). The two types of areas are
M
M
ye
Ax l
on
M
ito
om
M
ps im
d
ps m
d f
pf
f
Gold particles/µm2
Figure 3 | NMDA receptors are localized in myelin. (a) Electron micrograph
b 40
c 40
a
showing an axon (Axon) and its myelin (Myelin) sheath labeled for the NMDA
receptor subunit 1. (b) The quantitative method described in Figure 2b was
Myelin
used to show that the myelin contained a much higher density of gold
particles than did the axon, which showed a gold particle density similar to
20
20
Axon
the background levels found in mitochondrial outer membranes (Mito).
Mito
(Dark dots in the axoplasm represent cross-sectioned microtubules.) (c) The
Mito
method described in Figure 2c,h was used to show that the density of NMDA
0
0
receptor gold particles was similar in the myelin outer (Mom) and myelin inner
(Mim) membranes, and in the membrane of the postsynaptic density (psd mf)
of cerebellar mossy fiber synapses. These densities were much higher than the
density found in the postsynaptic membrane of parallel fiber synapses (psd pff ), supporting the specificity of the NMDA receptor labeling. (The micrograph and
histograms are from ref. 12.)
Gold particles/µm2
© 2008 Nature Publishing Group http://www.nature.com/natureprotocols
We determine the density of background labeling over empty resin in the ultrathin sections (usually over the lumen of blood vessels). This
estimates nonspecific affinity of the reactants for the embedding medium. Estimates of nonspecific background labeling of tissue can be
obtained by recording particle densities over tissue elements presumably lacking the antigen, e.g., mitochondrial outer membranes in the case
of receptors and transporters that are restricted to the plasma membrane. Knockout animals are useful if available. The background densities are
subtracted from the immunogold densities in the tissue profiles under investigation. In our experience, the immunogold background labeling
densities typically are in the range of 0.5–2 gold particles/mm2, depending on the antibodies. If the background labeling is higher than this,
actions to reduce it should be undertaken (see TROUBLESHOOTING). One must make every effort to ascertain that the antibodies are specific,
i.e., that they only react with the epitope that was used for producing the antibodies.
NATURE PROTOCOLS | VOL.3 NO.1 | 2008 | 149
a
Ter
Ast
Frequency of gold particles
0.2
Sp
b
0.1
Ter
Frequency of gold particles
PROTOCOL
Gran
0
0.15
0
–40 –20 0 20 40
Intra/Extra
Distance from terminal membrane
–80
0
–40
40
psd sc
Distance from postsynaptic membrane
Figure 4 | NMDA receptor distribution across membranes. (a) The electron micrograph shows immunogold labeling representing NMDA receptor subunit 2B in a terminal
(Ter) making an excitatory type of synapse with a dendritic spine (Sp) in the dentate gyrus. The inset is a larger magnification of the area outlined with dotted lines
showing intraterminal localization of gold particles signaling NMDA receptor 2B. The histogram shows the frequency distribution (see Fig. 2i) of the distances from the
gold particles to the midline of the terminal membrane shows that most of the gold particles are located at the intraterminal side of the membrane (a is from ref. 3).
(b) The electron micrograph shows that gold particles signaling NMDA receptors (small gold particles) (NR 1/2a/2b) are situated in synapses between a terminal (Ter)
positive for GABA (large gold particles) and a granule cell body (Gran) in the hippocampus. The histogram shows the frequency distribution of the intercenter distances
between the gold particles and the postsynaptic membrane. It indicates a postsynaptic localization of the NMDA receptors (b is from ref. 15).
determined using the stereological method with an overlay screen (Fig. 2d,e). Then, the gold particle densities over vesicles and vesicle-free cytosol are calculated. Thus, after subtracting the ‘contamination’ by the cytoplasmic matrix densities
from the vesicle densities, a ‘true’ ratio between the ‘corrected’ densities of vesicle gold particles and the cytoplasmic
matrix densities could be made. In this way, we have assessed the concentration of the inhibitory neurotransmitters
Gly and GABA in synaptic-like microvesicles in pancreatic endocrine cells (Fig. 5)14.
(ii) Use a computer program (e.g., MicroTrace) to digitize the localization of the centers of synaptic vesicles and gold
particles. Then calculate the distance from the center of each gold particle to the center of its nearest vesicle using
custom software. Sort the distances into bins (usually of 20 nm) and calculate the frequencies of intercenter distances for
each bin. These frequencies are compared with the intercenter distances between the vesicles and points randomly
scattered (by custom software) over the terminals. This type of analysis could also be made by manually measuring the
intercenter distances between gold particles and vesicles on the electron micrographs, and then using the overlay screen
to make random points over the terminal. With this approach we have studied the association of glutamate, GABA and
aspartate with synaptic vesicles in the hippocampus (Fig. 6)8,15,18 and in the olfactory bulb19.
(D) Co-localizing two vesicular membrane proteins
(i) Measure the distances between the centers of gold particles signaling two different synaptic vesicle membrane proteins.
This can be done manually or by using computer programs (see above). We have measured the distances from the centers
of gold particles signaling vesicular glutamate transporters to the centers of the closest particle signaling the vesicle
SNARE protein cellubrevin, as well as distances from the centers of vesicular glutamate transporter gold particles to random points (Fig. 7)2.
? TROUBLESHOOTING
TIMING
Steps 1–3: 1 d
Steps 4 and 5: 1 d
Step 6: 5 d
Steps 7 and 8: 1 d
Steps 9–27: 2 d
Step 28: 2 d
150 | VOL.3 NO.1 | 2008 | NATURE PROTOCOLS
b
B-cell
SG
1,000
500
0
Cytosol
a
SLMV
Figure 5 | GABA is localized in synaptic-like microvesicles in pancreatic B-cells.
(a) Immunogold labeling representing GABA is mainly located in B cells in the
islets of Langerhans. The gold particles are found with a high density in the
cytoplasm containing synaptic-like microvesicles, and with a low density in the
secretory granules (SGs). The inset shows the area outlined in dotted lines at
higher magnification. The profiles of two synaptic-like microvesicles, which are
delineated with white circles, are associated with gold particles representing GABA.
(b) Using the approach described in Step 29C(i) and Fig. 2g we found that the
density of GABA immunogold particles was several folds higher in the synaptic-like
microvesicles than in the intervening cytosol (a and b are from ref. 14).
Gold particles/um2
© 2008 Nature Publishing Group http://www.nature.com/natureprotocols
0.3
PROTOCOL
a
b
Glutamate
GABA
Glutamine
Frequency
mft
0.5
0
0 20 40 60
0 20 40 60
0 20 40 60
Step 29: The time spent on immunogold quantifications depends on the size of the material and on the questions under
investigation. If we measure the content of gold particles signaling one type molecule in three to four types of tissue element
from one brain region of one animal, we normally spend B2 d.
? TROUBLESHOOTING
No labeling
The primary antibodies could be diluted too much. We recommend titrating the antibody concentration in threefold increments
to optimize the immunoreactivity.
The concentration of soluble proteins or ions in the blocking solution could be too high, inhibiting the specific binding of
antibody. In our hands the effect of reducing the ion concentration is better than that of reducing the protein concentration.
The primary incubation times could be too short. As we routinely incubate overnight, we usually see little effect of further
increasing the incubation times.
The primary incubation temperature may be too low. The temperature can be increased to 37 1C.
Check that you have used the correct secondary antibodies. These must react with the immunoglobulins of the species in which
the primary antibodies were raised.
GA masks antigenic determinants. Omit GA from the fixative.
High background labeling
The primary antibodies could be too concentrated. Reduce the concentration of the antibodies in threefold steps.
The concentration of soluble proteins or ions in the blocking solution could be too low. Increase, especially, the ion
concentration.
The primary incubation times could be too long. The exposure times to the primary antibodies can be reduced down to 2 h.
The temperature during incubation with the primary antibodies could be too high. Background labeling can be decreased by
reducing the temperature to 4 1C.
Clustering of immunogold particles
Add polyethylene glycol (0.5 mg ml1) to the secondary antibody solution.
Spin the secondary antibody solution (1,000 r.p.m., 5 min) before applying it on the sections.
Suboptimal ultrastructure
If there are erythrocytes in the blood vessels this could be due to a failure of perfusion fixation. If so, new perfusion fixed
tissue must be prepared.
Frequency
a
b 0.4
Figure 7 | Co-localization of two synaptic vesicle proteins in synaptic-like
microvesicles in astrocytes. (a) Electron micrograph showing labeling for the
vesicular glutamate transporter 1 (VGLUT1) (small gold particles, some
Astro
indicated by short arrows) and cellubrevin (large gold particles, some
indicated with large arrows). The red lines give examples of the distances
0.2
between the centers of the VGLUT1 gold particles and the cellubrevin gold
particles. (b) The distances from each VGLUT1 gold particle to the
closest cellubrevin gold particle were measured using the approach
described in Step 29D. When these intercenter distances were sorted
0
into bins of 20 nm we could show that the first 20 nm bin contained
0
40 80 >120
the large majority of the distances. Next, the distribution of intercenter
distances between VGLUT1 and cellubrevin gold particles was significantly different from the distribution of distances between random points (black circles)
and cellubrevin gold particles (a and b are from ref. 2).
V
© 2008 Nature Publishing Group http://www.nature.com/natureprotocols
Figure 6 | Glutamate and GABA are concentrated in synaptic vesicles. (a) Electron
micrograph showing glutamate labeling of hippocampal mossy fiber terminals. The
inset is a higher magnification of the area outlined by the dotted rectangle showing
synaptic vesicles, the membranes of which are delineated by yellow circles, and gold
particles. The distances from the center of the gold particles to the center of the
vesicles are represented by red lines. (b) Using the method described in Step 29C(ii)
we could show that the intercenter distances between gold particles signaling
glutamate and synaptic vesicles were most frequently within 20 nm. Similar
frequency distributions were found for GABA, but not for the ‘metabolic’ amino acid
GIn. Both the glutamate and the GABA distributions were significantly different from
the distribution of random points (black circles), whereas this was not the case for
GIn (a and b are from ref. 18).
NATURE PROTOCOLS | VOL.3 NO.1 | 2008 | 151
© 2008 Nature Publishing Group http://www.nature.com/natureprotocols
PROTOCOL
ANTICIPATED RESULTS
With the use of these protocols it should be possible to get a quantitative measure of transporter and receptor proteins in
different ultrastructurally defined membrane domains and of transmitters in synaptic or synaptic-like vesicles.
In Figure 3a,b we have used the approach outlined in Step 29A to show that the density of glutamate NMDA receptor 1 is
much higher in the myelin sheets of the cerbellar white matter than in the axoplasm of axons that the myelin ensheaths and in
mitochondria.
In Figure 3a,c we have used the approach outlined in Step 29B(i) to show that the density of glutamate NMDA receptor 1 is
similarly high in the outer and inner membranes of the cerebellar white matter myelin as in the postsynaptic density of
excitatory synapses in the cerebellum.
In Figure 4a,b we have used the approach outlined in Step 29B(ii) to show that the glutamate NMDA receptor 2B is present
in the extrasynaptic part of membranes of excitatory nerve terminals (Fig. 3a) and that glutamate NMDA receptors are mainly
situated in the postsynaptic density in a subset of inhibitory synapses (Fig. 3b) in the hippocampus.
In Figure 5 we have used the approach outlined in Step 29C(i) to show that the concentration of the inhibitory neurotransmitter GABA is much higher in synaptic-like microvesicles than in the cytoplasmic matrix in endocrine pancreatic cells.
In Figure 6 we have used the approach outlined in Step 29C(ii) to show that the transmitters glutamate and GABA and
the transmitter candidate aspartate, but not the metabolic amino acid GIn, are associated with synaptic vesicles in
the hippocampus.
In Figure 7 we have used the approach outlined in Step 29D to show that the vesicular glutamate transporter VGLUT1 and
the vesicular SNARE protein cellubrevin are closely spaced around synaptic-like microvesicles in astrocytes in the hippocampus.
Published online at http://www.natureprotocols.com
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions
1. Peters, A., Palay, S.L. & Webster, H.D. The Fine Structure of the Nervous System:
Neurons and Their Supporting Cells. 3rd edn. (Oxford University Press, New YorkOxford, 1991).
2. Bezzi, P. et al. Astrocytes contain a vesicular compartment that is competent for
regulated exocytosis of glutamate. Nat. Neurosci. 7, 613–620 (2004).
3. Jourdain, P. et al. Glutamate exocytosis from astrocytes controls synaptic
strength. Nat. Neurosci. 10, 331–339 (2007).
4. Witcher, M.R., Kirov, S.A. & Harris, KM. Plasticity of perisynaptic astroglia during
synaptogenesis in the mature rat hippocampus. Glia 55, 13–23 (2007).
5. Jaiswal, J.K. & Simon, S.M. Imaging single events at the cell membrane.
Nat. Chem. Biol. 3, 92–98 (2007).
6. Ottersen, O.P. Quantitative electron microscopic immunocytochemistry of
neuroactive amino acids. Anat. Embryol. (Berl) 180, 1–15 (1989).
7. Harris, K.M. & Sultan, P. Variation in the number, location and size of synaptic
vesicles provides an anatomical basis for the nonuniform probability of
release at hippocampal CA1 synapses. Neuropharmacology 34, 1387–1395
(1995).
8. Gundersen, V. et al. Synaptic vesicular localization and exocytosis of L-aspartate
in excitatory nerve terminals: a quantitative immunogold analysis in rat
hippocampus. J. Neurosci. 18, 6059–6070 (1998).
9. Storm-Mathisen, J. & Ottersen, O.P. Immunocytochemistry of glutamate at the
synaptic level. J. Histochem. Cytochem. 38, 1733–1743 (1990).
10. Royet, J.P. Stereology: a method for analyzing images. Prog. Neurobiol. 37,
433–474 (1991).
11. Gundersen, H.J. et al. Some new, simple and efficient stereological methods
and their use in pathological research and diagnosis. APMIS 96, 379–394
(1988).
12. Káradóttir, R., Cavelier, P., Bergersen, L.H. & Attwell, D. NMDA receptors are
expressed in oligodendrocytes and activated in ischaemia. Nature 438,
1162–1166 (2005).
13. Talgøy Holten, A. & Gundersen, V. Glutamine as a precursor for transmitter
glutamate, aspartate and GABA in the cerebellum: a role for phosphate
activated glutaminase. J. Neurochem. Nov. 6 (Epub ahead of print)
doi: 10.1111/j.1471-4159.2007.05065.x (2007).
14. Gammelsaeter, R. et al. Glycine, GABA and their transporters in pancreatic islets
of Langerhans: evidence for a paracrine transmitter interplay. J. Cell Sci. 117,
3749–3758 (2004).
15. Gundersen, V., Holten, A.T. & Storm-Mathisen, J. GABAergic synapses in
hippocampus exocytose aspartate on to NMDA receptors: quantitative immunogold
evidence for co-transmission. Mol. Cell. Neurosci. 26, 156–165 (2004).
152 | VOL.3 NO.1 | 2008 | NATURE PROTOCOLS
16. Bergersen, L.H., Magistretti, P.J. & Pellerin, L. Selective postsynaptic colocalization of MCT2 with AMPA receptor GluR2/3 subunits at excitatory synapses
exhibiting AMPA receptor trafficking. Cereb. Cortex 15, 361–370 (2005).
17. Tonazzini, I., Trincavelli, M.L., Storm-Mathisen, J., Martini, C. & Bergersen, L.H.
Co-localization and functional cross-talk between A1 and P2Y1 purine receptors
in rat hippocampus. Eur. J. Neurosci. 26, 890–902 (2007).
18. Bergersen, L., Ruiz, A., Bjaalie, .JG., Kullmann, D.M. & Gundersen, V. GABA
and GABAA receptors at hippocampal mossy fibre synapses. Eur. J. Neurosci. 18,
931–941 (2003).
19. Didier, A. et al. A dendrodendritic reciprocal synapse provides a recurrent
excitatory connection in the olfactory bulb. Proc. Natl. Acad. Sci. USA. 98,
6441–6446 (2001).
20. Ottersen, O.P. Postembedding light- and electron microscopic immunocytochemistry of amino acids: description of a new model system allowing
identical conditions for specificity testing and tissue processing. Exp. Brain Res.
69, 167–174 (1987).
21. Gundersen, V., Danbolt, N.C. & Ottersen, O.P. & Storm-Mathisen J Demonstration
of glutamate/aspartate uptake activity in nerve endings by use of antibodies
recognizing exogenous D-aspartate. Neuroscience 57, 97–111 (1993).
22. Storm-Mathisen, J. et al. First visualization of glutamate and GABA in neurones by
immunocytochemistry. Nature 301, 517–520 (1983).
23. Ottersen, OP. & Storm-Mathisen, J. Glutamate- and GABA-containing neurons in
the mouse and rat brain, as demonstrated with a new immunocytochemical
technique. J. Comp. Neurol. 229, 374–392 (1984).
24. Ottersen, O.P., Storm-Mathisen, J., Madsen, S., Skumlien, S. & Strømhaug, J.
Evaluation of the immunocytochemical method for amino acids. Med. Biol. 64,
147–158 (1986).
25. Dale, N., Ottersen, O.P., Roberts, A. & Storm-Mathisen, J. Inhibitory neurones of a
motor pattern generator in Xenopus revealed by antibodies to glycine. Nature
324, 255–257 (1986).
26. Bergersen, L. et al. A novel postsynaptic density protein: the monocarboxylate
transporter MCT2 is co-localized with delta-glutamate receptors in postsynaptic
densities of parallel fiber-Purkinje cell synapses. Exp. Brain Res. 136, 523–534
(2001).
27. Chaudhry, F.A. et al. Glutamate transporters in glial plasma membranes: highly
differentiated localizations revealed by quantitative ultrastructural
immunocytochemistry. Neuron 15, 711–20 (1995).
28. Nagelhus, E.A. et al. Aquaporin-4 water channel protein in the rat retina and optic
nerve: polarized expression in Muller cells and fibrous astrocytes. J. Neurosci. 18,
2506–2519 (1998).
29. Matsubara, A., Laake, J.H., Davanger, S., Usami, S. & Ottersen, O.P. Organization
of AMPA receptor subunits at a glutamate synapse: a quantitative immunogold
analysis of hair cell synapses in the rat organ of Corti. J. Neurosci. 16, 4457–4467
(1996).