JOIHWULOF

NE E
Journal of Neuroscience Methods 61(1995) 5-14

Embedding, sectioning, immunocytochemical and stereological methods

that optimise research on the lesioned adult rat spinal cord
D.A. Stuart *, D.E. Oorschot
Department of Anatomy and Structural Biology and the Neuroscience Centre, University of Otago Medical School, P.O. Box 913,
Dunedin, New Zealand

Received 13 September 1994;accepted 19 January 1995

Abstract

Numerous methods have been developed to embed, section and immunocytochemically label nervous tissue and the method
chosen depends upon numerous factors. However, many of these methods have technical drawbacks that make them difficult to
use in studies using injured/lesioned tissue. We present here, methods for embedding, sectioning and immunocytochemically
labelling lesioned adult spinal cord tissue at the light microscope level. We have developed a novel, gelatine-embedding
technique for vibratome sectioning which overcomes many of the difficulties encountered with lesioned tissue. Individualised
immunocytochemical protocols have also been developed for the antibodies GFAP (to label astrocytes), MBP (to label myelin)
and CNP-ase (to label oligodendrocytes). Sequential pre-treatment with proteinase-K, methanol and sodium borohydride
achieved optimal GFAP localisation. MBP and CNP-ase were optimally localised after sequential pre-treatment with proteinase-K
(at different concentrations) and sodium borohydride. Methanol pre-treatment resulted in a loss of immunoreactivity for these
latter two antibodies. Each protocol achieved full immunocytochemical penetration throughout 40 pm vibratome sections. These
techniques enable the new unbiased stereological tools (that is, the Cavalieri and optical disector principles) to be utilised.

Keywords: Gelatine; Vibratome sectioning; Immunocytochemistry; Optical disector; Cavalieri method

1. Introduction lesioned spinal cord using specific immunocytochemi-

The injured and/or lesioned spinal cord is being
studied more extensively now as a result of promising
in vitro (Carbonetto et al., 1987; Caroni and Schwab,
1988; Bastmeyer et al., 1991, 1993) and in vivo
(Richardson et al., 1980; David and Aguayo, 1981;
Fernandez et al., 1990; Schnell and Schwab, 1993)
work. This work has demonstrated that the central
nervous system has some capacity to regenerate follow-
ing injury. The glial cell and myelin contribution to this
regenerative response is likely to be crucial (see, for
example, Oorschot and Jones, 1990; Schnell and
Schwab, 1993). As part of our research in this field, we
have sought to examine the effect of growth factors on
the glial cell response in the adult rat thoracic spinal
cord after a lesion. Detection of the glial cell response
requires localisation of glial sub-types in control and
* Corresponding author. Tel.: (64) 3-479-7382;Fax: (64) 3-479-7254.
cal markers. Quantitation of this response is best un-
dertaken by estimating the total number of glial cells
rather than glial cell densities (as in, for example,
Vijayan and Cotman, 1987) since densities may lead to
misleading biological conclusions (see Oorschot, 1994).
The total number of glial cells can be unbiasedly and
efficiently determined by using the Cavalieri method in
combination with the optical disector method (see
Gundersen, 1986; Janson and Meller, 1993; Oorschot,
1994). The Cavalieri method unbiasedly estimates the
total volume of the structure of interest (in this case,
the thoracic spinal cord), while the optical disector
method unbiasedly estimates the number of glial cells
in a sub-volume of the thoracic spinal cord (i.e., the
glial cell density, or the N,). The product of the total
volume and the N, yields an unbiased estimate of the
total number of glial cells.
The use of lesioned spinal cord tissue, and of the
optical disector method to determine total glial cell
numbers, imposes certain methodological require-

0165-0270/95/$0950 0 1995 Elsevier Science B.V. All rights reserved

SSDI 0165-0270(9.5)00017-8
(1 il. Stuart, D.E. Oorschor
ments. These are as follows: (1) determination of total
numbers requires complete longitudinal serial section-
ing of the spinal cord; (2) sectioning of the lesioned
spinal cord requires minimal damage and the mainte-
nance of tissue integrity; (3) application of the optical
disector requires thicker (40-50 pm> sections, as well
as immunocytochemical labelling of the glial cells
throughout the depth or thickness of these sections;
and (4) immunocytochemical processing of lesioned
tissue requires the inactivation of intrinsic im-
munoglobulins in order to minimise non-specific bind-
ing.
A small number of other researchers have embed-
ded, sectioned and immunocytochemically labelled in-
jured spinal cord tissue (i.e., Cammer et al., 1985a,b;
Mikoshiba et al., 1985; Trapp et al., 1988 and Kidd et
al., 1990). However, none of these published methods
fulfilled all of the above requirements. In this paper,
we present the methods that we have developed which
fulfii the above requirements, namely: (1) complete
serial sectioning of the lesioned adult rat spinal cord at
40 and 50 pm with minimal damage to the tissue; (2)
immunocytochemical labelling of the glial components
(namely astrocytes, oligodendrocytes and myelin using
the commercially available antibodies, glial fibrillary
acidic protein (GFAP), 2’,3’-cyclic nucleotide 3’-phos-
phodiesterase (CNP-ase) and myelin-basic-protein
(MBP) respectively) throughout the depth of these
sections; and (3) inactivation of host imrnunoglobulins
to minimise non-specific binding during immunocyto-
chemical processing. These methods should be invalu-
able for future research on the injured spinal cord
whenever optimal immunocytochemical labelling of
glial cell types needs to be combined with efficient
stereological counting methods like the optical disec-
tor.
2. Materials and Methods
2.1. Animals
Adult male albino Wistar rats, 3 months old, weigh-
ing 270-390 g were used in this study. All surgery was
performed using sterile instruments and aseptic tech-
niques. The rats were anaesthetised by an intraperi-
toneal injection of Hypnorm/ Hypnovel (3.3 ml/kg)
after a previous injection of atropine (0.05 mg/kg).
After opening the dorsal skin and musculature in the
midline, a laminectomy was performed at the level of
thoracic vertebra 6 (T6). The dura was then opened
and the dorsal right quarter of the spinal cord was
severed to the central canal. During this procedure, an
ALZBT mini-osmotic pump was impIanted into a sub-
cutaneous pocket in the dorsal abdomen. The pump
contained either a growth factor or artificial cere-
/Jowrrul of Newosciencr Methods 61 (1995) .5-I4
brospinal fluid (ACSF). A cannula from the pump to
the lesion site continuously infused the growth factor
or ACSF to the lesion site for a specified number of
days. Following surgery, the animals had full control of
both bladder and bowel function, and after 24 h walked
around with only a slightly unstable gait. All proce-
dures were ethically approved by the University of
Otago’s Committee on Ethics in the Care and Use of’
Laboratory Animals.
2.2. Perfusion and fucation
After a specified numbers of days, each animal was
anaesthetised by an intraperitoneal injection of Hyp-
norm/ Hypnovel(3.0 ml/ kg) and perfused via the heart
with saline followed by 4% paraformaldehyde and 0.1%
glutaraldehyde in 0.1 M phosphate buffer (pH 7.4).
Two to 3 h later the spinal cord from T5 to T7 was
removed, weighed and placed in post-fixative, 4%
paraformaldehyde in 0.1 M phosphate buffer,
overnight. The spinal cord was then stored in phos-
phate buffered saline (PBS) until further processing.
The primary dependence on fixation by paraformalde-
hyde in this schedule ensures a high degree of uniform
tissue penetration (and hence fixation) as well as the
preservation of tissue immunoreactivity (Berod et al.,
1981). The extremely good tissue preservation by glu-
taraldehyde (due to its ability to rapidly form cross-Iinks
with tissue proteins) is also harnessed, but only to a
minimal degree (i.e., O.l%, since higher concentrations,
and thus more rapid, fixation may limit glutaraldehyde
penetration into the deeper tissue (resulting in non-
uniform fixation) and may limit accessibility of the
immunochemical reagents to the antigens of primary
interest). The resounding effectiveness of low glu-
taraldehyde plus paraformaldehyde in preserving tissue
structure and antigenicity can be gleaned from its
widespread use in immunocytochemical studies on the
brain and spinal cord (for e.g., Kimura et al., 1981;
Slovitzer and Nilaver, 1987; Trapp et al., 1988; Kidd et
al., 1990, Martinez et al., 1991). Thus, variations in
fixative concentrations were not carried out in this
study.
2.3. Embedding of spinal cord for sectioning
The use of a low glutaraldehyde concentration re-
sulted in fixed tissue that was still comparatively soft.
In order to reproducibly cut sections of equivalent
thickness, the spinal cord needed to be either sur-
rounded in a harder medium or hardened itself to
stabilise the tissue during the sectioning process. In
addition, 40- or 50-pm sections needed to be sectioned
to enable efficient and effective use of the new unbi-
ased stereological method, the optical disector. The
embedding procedures thus trialled included, embed-
 D. Stuart, D.E. Oorschot
ding in agar or gelatine, followed by sectioning on a
vibratome, and freezing the tissue followed by cryosec-
tioning on a sledge microtome.
To embed the tissue in agar, Davis bacteriological
agar (Davis-Germantown) was dissolved in PBS (pH
7.4) by heating until bubbling, stirring at all times. The
solution was then cooled to around 37°C. Next, a small
amount was poured into the bottom of a mould, cooled
at 4°C for 1 min, the tissue was blotted dry, placed
onto the agar base and agar poured around the tissue
to cover it completely. Once completely set, the hard-
ened block was then removed from the mould, trimmed
and stored in PBS ready for sectioning.
To embed the tissue in gelatine a 5-16% solution
was made by heating double-distilled water to 60°C
then adding the gelatine, stirring continuously until
dissolved. The gelatine was then cooled to around
37°C the tissue was placed into a sealable glass con-
tainer filled with the gelatine (around 20 times the
volume of the tissue) and kept at 37°C overnight.
Graded concentrations of gelatine solutions were also
experimented with (e.g., 5%, then lo%), leaving each
at 37°C overnight. The next day a fresh lo-16% gela-
tine solution was made. This final gelatine solution was
cooled to below 37°C poured into a mould so as to
completely cover the spinal cord and cooled at 4°C
until it began to set. The tissue was lifted using tweez-
ers and positioned so that it was ‘floating’ and ran-
domly orientated within the setting gelatine. The spinal
cord embedded in gelatine was then kept at 4°C for
2-3 days to set completely. The gelatine block was
then removed from the mould and a smaller block cut
around the tissue leaving about l-2 cm of gelatine
which was placed into 4% neutral buffered formalin
(NBF) at 4°C for 2 72 h, until hardened. Once hard-
ened the blocks were stored in PBS at 4°C until re-
quired.
For tissue to be frozen-sectioned, it was first cry-
oprotected to prevent the formation of ice crystals
during the freezing process by soaking it in a series of
graded sucrose/NBF solutions (0.5 M, 0.7 M and 0.8
M sucrose>. Initially, the tissue was frozen by first
blotting it dry and simply surrounding it in CO, (dry
ice). An improvement of this method was to surround
the tissue in OCT compound (i.e., Tissue Tek, Miles)
within a small foil mould and freeze this in a slurry of
acetone and dry ice. A further improvement was to
slowly freeze the tissue (surrounded by Tissue Tek
within a foil mould) in liquid nitrogen-cooled iso-pen-
tane. This final method allowed multiple pieces of
tissue to be frozen at once and stored at -80°C until
required.
2.4. Sectioning
Prior to vibratome serial sectioning of the entire
spinal cord embedded in agar or gelatine, the tissue
/Journal of Neuroscience Methods 61 (1995) 5-14 I
blocks were thoroughly rinsed in PBS. The block was
trimmed to fit onto the chuck, ensuring the top and
bottom were as flat and as parallel as possible. By
trimming gently sloping edges, especially the front and
back, movement of the block on the chuck while sec-
tioning was minimised. The base of the tissue block
was blotted dry, supa glue was spread over the chuck
surface and the block gently pressed onto the chuck.
Sufficient time was allowed for this to adhere securely.
Wilkinson Sword blades were used, cleaning the oil
from the blade with acetone first. Both agar and gela-
tine-embedded tissue were sectioned longitudinally at
40 or 50 pm using a LANCER series 1000 Vibratome
at a speed of 1.5-2 and amplitude of 8-9. The water
bath was filled with PBS, cold (4°C) for agar and either
cold or room temperature for gelatine. The serial sec-
tions were collected from the bath using a brush. The
1st to 4th sections were consecutively placed in pottles
labelled l-4 containing PBS. This was repeated for the
5th-&h, 9th-12th sections, and so on, such that Pottle
1 contained sections 1, 5, 9, 13, and so on. In this way,
each pottle contained a systematic series of every 4th
section. The sections were stored in these pottles until
required.
For frozen serial sectioning, a sledge microtome was
used. The frozen spinal cord tissue (surrounded by
Tissue Tek) was adhered to the corner of a cooled
wooden block by droplets of Tissue Tek. The tissue
was surrounded at all times by dry ice to keep it frozen.
Serial sections were cut at 40 pm by smoothly dragging
the knife across the frozen tissue. The sections were
then lifted from the knife with a wet brush, systemati-
cally collected and stored in 0.1 M phosphate buffer.
2.5. Immunocytochemistry
Commercially available antibodies were utilised for
labelling cells: GFAP for astrocytes (Sigma), CNP-ase
for oligodendrocytes (Sigma) and MBP for myelin
(Boehringer Mannheim Biochemica). Immunostaining
for all antibodies was performed using the biotin-
streptavidin-peroxidase method (based on Hsu et al.,
1981), and all immunocytochemistry was carried out on
free-floating sections at room temperature unless oth-
erwise stated. One pottle of sections was used per
antibody (see Section 2.4). Thus, the sections processed
for immunocytochemistry represented every 4th section
through the entire spinal cord. Following immunostain-
ing, all sections were mounted onto gelatine subbed
slides and allowed to dry at room temperature.
For the optical disector to be used efficiently the
antibody must penetrate the full depth of the section.
To achieve this a specific immunocytochemical proto-
col was developed for each antibody. This involved
trialling different reagents at variable concentrations
and incubation times in order to: (1) satisfactorily
remove the gelatine from the sections without destroy-
ii D. Stuart, D.E. Oorschot
ing either their antigenicity or the sections themselves,
(2) improve the depth of penetration of antibodies, (3)
improve the intensity of the immunocytochemical reac-
tion and (4) decrease or eliminate non-specific and
background staining.
Formaldehyde fixatives primarily fix tissue by form-
ing cross-linking ‘methylene bridges’ both within and
between adjacent protein molecules within the tissue
(Farmilo and Stead, 1989). These cross-links can cover-
up or mask antigenic sites and also render tissue pro-
teins more hydrophobic. Hydrophobicity is one of the
natural forces that confers stability on the tertiary
structure of peptides but may also take place between
different protein molecules. The immunoglobulins are
particularly hydrophobic (Boenisch, 1989). Non-specific
and background staining, commonly a problem in im-
munocytochemistry, is a result of hydrophobic and
electrostatic forces between protein molecules. Reduc-
ing the incidence of these forces will thus enable the
demonstration of antigenic sites, improve tissue pene-
tration of antibodies and reduce non-specific binding.
Proteinase K (PK, Sigma), a proteolytic enzyme, was
used at variable concentrations as a pre-treatment in
all the immunocytochemical protocols to remove the
gelatine from the sections. PK was diluted in pre-
warmed PK buffer (i.e., 0.606 g Tris[hydroxymethyl]-
aminomethane in 100 ml of double-distilled water,
followed by pH adjustment to 8.5 using 2 N HCl,
addition of 0.186 g EDTA and then readjustment of
the pH to 7.4). The sections were then incubated in PK
at 37°C for specific intervals. Proteolytic digestion also
compensates for the impermeable nature of formalde-
hyde fixatives by ‘etching’ the tissue, breaking some of
the cross-links, and thus allowing hidden structures to
be revealed (Farmilo and Stead, 1989). As a result, this
will also allow for further penetration of the antibodies
and help decrease non-specific binding.
Since the immunoreactive sites were localised by a
peroxidase-labelled antibody as the final step in the
procedure, endogenous peroxidase was quenched by
two different methods depending upon the antibody.
Either 3% H,O, alone or 3% H,O, in acid methanol
was trialled.
Sodium borohydride (NaBH,, BDH) is an anti-fixa-
tive and also breaks cross-links created by the aldehyde
fixatives. It was trialed here to further facilitate tissue
penetration by the antibodies.
A 3% normal sheep serum block was utilised to
occupy non-specific binding sites. The bonds at these
sites are weaker than antigen-antibody reactions, thus
allowing the specific sites to react with the highly
competitive primary antibody (Polak and Van Noor-
den, 1987). Sheep serum was used so that when the
secondary antibody (sheep anti-mouse, Amersham) was
applied, it would not bind to any sheep serum that may
have remained on the section.
/Journal of Neuroscience Methods 61 (1995) 5-14
The diluent for the antibodies and the washes was
always PBS, with the addition of Triton X-100 (a
non-ionic detergent, Sigma) and bovine serum albumin
(BSA) or non-fat milk powder. The presence of saline
breaks weak electrostatic forces between non-specifi-
cally bound proteins and the addition of detergent
reduces hydrophobic interactions between tissue and
reagent proteins. The addition of BSA to the antibody
incubation steps (PBS(A)), and milk powder to the
wash steps (PBS(B)) serve as an additional source of
protein that competitively binds to non-specific binding
sites, thereby preventing non-specific binding of the
antibodies.
All preliminary studies were carried out on unle-
sioned control tissue and non-specific binding had been
eliminated under these circumstances. When testing
the protocols upon lesioned tissue however, positive-
staining of negative controls was encountered. This
occurred for all antibodies and was probably due to the
immune response naturally occurring following injury
to the spinal cord. The secondary antibody was proba-
bly binding to immune-induced F, receptors. To over-
come this, the addition of 2% non-immune rat sera
(Sigma) to the secondary antibody diluent was trialled.
To ensure binding of the tertiary streptavidin anti-
body (conjugated to horseradish peroxidase, Amer-
sham), the wash preceding this incubation was PBS
only (i.e., it contained no non-fat milk powder). This
eliminates the presence of naturally occurring strepta-
vidin and biotin within the milk powder, which itself
would bind to the secondary antibody thereby blocking
the sites.
The peroxidase of the bound streptayidin antibody
was localised by incubating the sections in 50 mM
Tris-HCl buffer (pH 7.6) containing 0.05% 3,3’di-
aminobenzidine tetrachloride (DAB) and 0.02% H,O,
for 10 min to yield a brown reaction product.
2.6. Negative controls for immunocytochemistry
To verify that the primary antibody was binding
specifically, the primary antibody was replaced with
heat-inactivated normal serum from the species in
which the primary serum was derived (mouse serum for
all the antibodies). A dilution of 1:1200 of mouse
serum yielded a final protein concentration that was
either equivalent to, or more than, the final protein
concentration for each of the diluted antibodies.
2.7. Counterstaining
Sections immunocytochemically labelled for GFAP
and CNP-ase were mounted onto gelatine subbed
slides, air dried for 30 min and counterstained for glial
cell nuclei. As the immunocytochemical reaction prod-
uct is brown, various blue/purple stains were experi-
 D. Stuart, D. E. Oorschot
mented with. The counterstaining protocol needed to
be easily reproducible and easily adjusted. Toluene
Blue, Haematoxylin (Gills numbers 1, 2 and 3), Cresyl
Fast Violet and Cresyl Violet Acetate were all trialled
at varying concentration and incubation times with
Cresyl Fast Violet at 0.1% being optimal. The counter-
staining worked best when the sections, after being air
dried, were taken through a graded series of ethanol
first to maximally rehydrate them before staining rather
than placing them straight into distilled water. This
procedure also eliminated any substances left over
from the immunocytochemistry that would otherwise
compromise the uptake of the stain.
2.8. Penetration of antibodies through sections
For the optical disector to be used efficiently the
antibody must penetrate the full depth of the section.
To establish the depth of penetration, the sections
were viewed using a X 100 oil immersion objective (NA
1.4) on an Olympus BH-2 microscope. The image was
projected onto an adjacent television monitor via a
video camera connected to the microscope with the
final magnification being X4500. By focusing through
the section, the depth of penetration (or the presence
of complete penetration) and the intensity of the im-
munocytochemical labelling could be easily discerned.
The depth of penetration through the section in the z
plane was measured by an electronic microcator that
was connected to the microscope stage.
3. Results
3.1. Embedding and sectioning
Agar is often utilised as the embedding medium for
vibratome sectioning (Borgens et al., 1986). Using agar
as the embedding medium for this spinal cord tissue
posed the major problem of the tissue pulling out of
the agar once just over half of the tissue had been
sectioned. This resulted in thick/ thin sections being
cut due to the tissue moving within the agar block.
None of the following attempts worked well enough to
consider using; a solution of 6% agar made up in
phosphate buffer, cooling the agar as much as possible
before embedding to reduce the amount of shrinkage
that would occur as it set; ‘dipping’ the tissue into agar
to coat it before embedding; using 7% agar, and/or
dipping the tissue into lower percentages of agar first;
and embedding the tissue at different orientations. It
was found that the dural sheath was the cause of the
movement of the tissue. The spinal cord is secured to
the dura at two lateral points via the denticulate liga-
ments. When these are removed during the normal
process of sectioning, the spinal cord is no longer
/Journal of Neuroscience Method.7 61 (I 995) 5-14 9
secured within the dural sheath or the agar block,
resulting in sections of variable thickness (or even
making it impossible to complete the sectioning due to
it pulling from the agar). Thus an attempt was made to
manually remove the dura from the fixed spinal cord
before embedding (it was not possible to remove the
dura immediately after perfusion). This proved suc-
cessful if the dura was completely removed but was
extremely difficult to do without damaging the tissue
because it was still quite soft. Consequently, agar em-
bedding was abandoned in search of an embedding
medium or method that was more rigid, and that
reduced or eliminated any movement of the tissue.
Frozen sectioning using the sledge microtome was
experimented with and found to be most successful on
unlesioned tissue. It fulfilled the above requirements in
that the tissue was quite rigid and secure during the
sectioning process. All of the tissue was able to be
consistently sectioned at 40 pm and the immunocyto-
chemical staining was very good. A major drawback
was discovered when lesioned spinal cord tissue was
frozen sectioned. The sections would tear apart at the
lesion site as they were sectioned due to the instability
of the lesion site itself. This was difficult to overcome
so another embedding and sectioning method was
sought.
Gelatine embedding also fulfilled the requirements
to eliminate movement of the tissue while sectioning.
Gelatine embedding is an infiltration process rather
like wax embedding, binding the tissue and embedding
medium together, in this case holding the lesion site
and the dural sheath rigid. Initial trials involved using
16% and 12% gelatine without further hardening but
this was found to be too soft to section. After this,
16%, 12%, 10% and 8% gelatine solutions hardened in
4% NBF after being trimmed into blocks was trialled.
Only the 8% block was too unstable, while all the
others sectioned very well with no movement of the
tissue within the block and excellent sectioning through
the lesion site. After sectioning, the tissue and the
gelatine cannot be separated either by teasing with a
brush or by heating the sections in PBS in an attempt
to redissolve the gelatine. Therefore the lowest concen-
tration of 10% gelatine was chosen over the 12% and
16% solutions to minimise the severity of post-treat-
ments that were required to remove the gelatine from
the sections after sectioning. For these optimal 10%
blocks, it was not necessary to infiltrate the tissue with
5% gelatine as part of the embedding procedure.
3.2. Immunocytochemistry
Free-floating 40 pm frozen sections were initially
used to test immunocytochemical labelling of cells.
GFAP was experimented with first and found to work
well provided all of the solutions were freshly made.
 I0 D. Stuurt, D.E. Oorschor /Jortrtrd of Neuroscience Methods 61 (1995) 5-l-/

However, frozen sectioning was not continued as some 3.3. Counter-staining

distortion of the tissue due to the freezing process was
evident and there was an inability to section the le- Cresyl Fast Violet (CFV) was the counterstain used
sioned spinal cord satisfactorily. Immunocytochemical for this study. Individual nuclei were easily distinguish-
labelling was then experimented with using free-float- able and the staining protocol was easily adjusted for
ing gelatine embedded, vibratome sections (40 pm) any changing circumstances. A stock solution of 1%
and after much trial and .error, each antibody had its CFV was made up and diluted to 0.1% as required.
own immunocytochemical protocol, as listed in Table The mounted sections were air dried at room tempera-
1. From this table it can be seen that the optimal PK ture for 30 min before counterstaining. The protocol
concentration for the first treatment step varied for was as follows:
each antibody. For the next step of quenching endoge-
nous peroxidases, the methanol/H,O, combination Absolute ethanol 1 min
was more efficient than H,O, alone at quenching 95% Ethanol 1 min
peroxidases, but resulted in a loss of immunoreactivity 70% Ethanol 1 min
for CNP and MBP sections and was thus omitted from 50% Ethanol 1 min
these protocols. For the GFAP protocol, the success of Distilled water 1 min
this incubation step meant no H,O, treatment was 0.1% CFV 20 min
needed after exposure to NaBH,. For all three proto- Distilled water 1 min
cols, exposure to NaBH, improved antibody penetra- 70% Ethanol 1 min
tion and immunolocalisation of glial cell sub-types (see 95% Ethanol 1 min
Figs. 3 and 2). After incubation with the primary anti- 95% Ethanol + 0.3% glacial acetic acid 5-30 min
body, the inclusion of 2% rat sera with the secondary 95% Ethanol 1 min
antibody was highly effective in reducing immune-in- Absolute ethanol 1 min
duced binding in lesioned spinal cord tissue (see Fig. Xylene 2 X 2 min
1). Slightly longer incubation times for CNP-ase im- Mount in DPX
proved the penetration and intensity of the labelling
(Fig. 1). Since MBP primarily localised myelin and not 3.4. Antibody penetration, tissuethickness and applicabil-
oligodendrocytes, a counterstain for glial nuclei was ity of the disector and Cavalien. methods
not undertaken (Fig. 1). For all 3 protocols, their
respective negative controls exhibited no immunoreac- For all 3 antigens, immunoreactivity was evident
tivity (Fig. 1). throughout the sections and was highly specific (see

Table 1
Specific immunocytochemical processing schedule for each glial cell antibody
Activity CNP-ase GFAP MBP
Protease K: 30 min@37”C 6.3 Ups 1.68 Ups 8.4 ups
PBS(B) wash: 3 X 10 min YES YES YES
Methanol + 0.1% concHCl + 3% H,Oz: 1 hour NO YES NO
1% NaBH, in Phosphate Buffer: 30 min YES YES YES
PBS(B) wash: 3 x 10 min YES YES YES
3% H,O, in ddH,O: 30 min YES NO YES
PBS(B) wash: 3 x 10 min YES NO YES
Normal Sheep Serum in PBS(A) (1: 30) 1 hour 30 min 30 min
Primary Antibody in PBS(A) 1:500 I:400 1:200
and negative control in PBS(A) (1: 1200)
17-20 hours overnight at 4°C
PBS(B) wash: 3 x 20 min YES YES YES
Sheep-anti-mouse (1 : 200) + 2% Rat Sera in PBS(A) 1 hour 40 min 40 min
PBS only wash: 3 x 10 min YES YES YES
Conjugated Streptavidin-HRP in PBS(A) (1 :300) 40 min 30 min 30 min
PBS only wash: 3 x 5 min YES YES YES
DAB incubation: 10 min YES YES YES
PBS only wash: 3 x 5 min YES YES YES
Mount sections onto gelatine subbed slides YES YES YES
and air dry for 30 min
Key: Protease K: Ups = Number of units/mg solid of activity in 300 ~1 of PK buffer per section = units per section
PBS(A) = PBS + 1% Bovine Serum Albumin + 0.5% Triton X-100
PBS(B) = PBS + I % Non-fat milk powder + 0.5% Triton X-100
 D. Siuart, DE. Oorschot /Journal of Neuroscience Methods 61 (199s) 5-14 11

Fig. 1). At the end of immunocytochemical processing,
the thickness of the vibratome sections had diminished
consistently to around 25 pm (i.e., a 40% reduction of
the original thickness of 40 pm). This has also been
noted by others for frozen sections (Janson and Moller,
1993). This consistent reduction does not compromise
the use of the optical disector. Since the sections<were
a systematic fraction of a complete series of sections
(i.e., every 4th section, with the pottle containing these
sections being randomly selected), the total volume of
the spinal cord and the number of glial cells within a
subvolume of these sections (i.e., the NV) could thus be
determined using the Cavalieri and optical disector
methods respectively. For the MBP sections, penetra-
tion throughout the sections meant that the volume
fraction (i.e., V,) could be systematically sampled in

Fig. 1. Immunocytochemical localisation of GFAP, GNP-ase and MBP in vibratome sections of the adult rat spinal cord. A: GFAP
counterstained negative control, x 110. B: GFAP reactivity showing astrocytes and their nuclei (see arrow), X 110. C: GNP-ase counterstained
negative control, x110. D: GNP-ase reactivity showing oligodendrccytes, their processes and nuclei (see arrows), X110. E: MBP negative
control, x55. F: MBP reactivity showing the myelin (at arrows) in the grey matter (GM) and the white matter (WM), X55.
I?. D. Stuart. D.E. Oorschot /Journal of Neuroscience Methoti.~ 61 (1995) S-14

each section. This meant that unbiased stereological 4. Discussion

methods, yielding reliable results of the total numbers
or total volumes, could now be applied to the lesioned The methods in this paper enable optimal immuno-
spinal cord. cytochemical localisation and stereological assessment
of the total number of astrocytes and oligodendrocytes
in the lesioned rat spinal cord.
Crucial to these methods is the ability to serially
section the lesioned spinal cord with minimal damage
to the tissue. This was achieved by embedding the
tissue in gelatine, and involved a modification of the
method developed by Gurusinghe and Ehrlich (1986)
for normal avian brains. The distinct advantage of
gelatine embedding over agar is that the spinal cord is
infiltrated rather than encased, thus providing com-
plete and overall stability. In contrast, stability is much
reduced after agar embedding due to the spinal cord
being anchored to the dural sheath solely by the den-
ticulate ligaments. During sectioning, the cord is ini-
tially stable, but once these anchoring sites are re-
moved, instability ensues with consequent section un-
evenness. Other advantages with gelatine are (1) the
gelatine fills in the lesion site keeping it in a fixed
position (which was not the case during frozen section-
ing), (2) the gelatine can be resoftened into liquid if
another attempt at embedding is required, and (3) this
method requires no freezing and consistent and repeat-
able sectioning using the vibratome is possible. These
advantages have also been highlighted for lesioned
hypothalamus by Griffioen et al. (1992).
From Table 1, it is noteworthy that each glial cell
antibody (i.e., GFAP, CNP-ase and MBP) required a
specific immunocytochemical protocol in order to be
effectively localised and to be seen throughout the
depth of a vibratome section.
As a first step in these protocols, gelatine embed-
ding provided the challenge of how to efficiently re-
move the gelatine from the sections without damaging
the tissue or destroying the antigenic sites. This was
accomplished by using a pre-treatment with PK. The
results are reproducible provided the temperature and
time of incubation, and the volume of the incubation
solution per section, is strictly adhered to. It was neces-
sary to use different units per section of PK for each
antibody (see Table 1). This indicates that each glial
cell antigen has a different sensitivity to PK. Others
have also noted that the optimal concentratrion of
proteolytic enzyme depends very heavily upon the sen-
sitivity of the antigenic site (Finley and Petrusz, 1982).
Therefore many different concentrations of PK at vari-
Fig. 2. Immunocytochemical localisation of GFAP in 40 pm vi- able incubation times were trialled in order to optimise
bratome sections illustrating the penetration of the antibody the outcome. When too little a concentration of PK
throughout the middle 10 pm of this 2.5 pm section. These 3 photos was used, the innermost structures were not stained
are identical except for the depth. The arrowed astrocyte is in focus and when too much was used, the enzyme degraded
in photo (A) and out of focus in photos (B) and (C). The double-
arrowed astrocyte is out of focus in photo (A), clearly visible in photo
the tissue to shreds.
(B) and out of focus again in photo (0, where another astrocyte fat For GFAP, 1.68 units per section gave excellent
arrowhead) is clearly visible. All X 110. immunoreactivity, while increasing the units has a di-
 D. Stuart, D.E. Oorschot
rect relationship with loss of labelling. CNP-ase and
MBP both required higher concentrations of PK to
visualise the antigenic sites and to improve the inten-
sity of the labelling. These results are all due to effec-
tive degradation by PK of fixative-induced protein
cross-linkages.
The addition of methanol to the CNP-ase and MBP
protocols (see Table 1) resulted in a loss of immunore-
activity in the middle of the sections, while labelling
was throughout the sections when methanol was omit-
ted. The methanol could be either removing or chang-
ing the antigenic sites or preventing the penetration of
the antibodies. This phenomenon occurred with the
two antibodies that stain for similar structures so it is
possible that the methanol had a direct effect upon the
antigenic sites. When the methanol step was omitted,
the 3% H,O, block was placed after the sodium boro-
hydride step.
All of the antibodies were diluted in PBS in order to
reduce non-specific binding due to electrostatic forces.
A non-ionic detergent, Triton X-100 was also added to
reduce hydrophobic interactions while bovine serum
albumin was added as a competitive protein also to
reduce non-specific binding. The washes were PBS
with Triton X-100 and milk powder as the source of
protein rather than bovine serum albumin (due to the
expense of bovine serum albumin). The addition of 2%
rat sera to the secondary antibody diluent was highly
effective in reducing what was probably immune-in-
duced F, receptor binding in lesioned spinal cord tis-
sue (see Fig. 1).
For the penetration of immunochemical reagents it
was found there was not a direct relationship between
incubation time and depth of penetration. Increasing
the incubation time of an antibody did not necessarily
increase the penetration of that antibody. Other fac-
tors such as antibody size, hydrophobic and electro-
static forces may effect how far an antibody can pene-
trate into the tissue.
The consistent penetration of 3 glial cell antibodies
through thick 40 pm sections is extremely useful. This
is because it allows the use of the efficient optical
disector method for determining the glial cell density
as a first step in total glial cell number determination.
Prior to publication of the optical disector method by
Gundersen (1986), the physical disector method was
published by Sterio (1984, alias Gundersen). In con-
trast to the optical disector method, where cells are
counted as one focuses through one thick 2.5-50 pm
section, the physical disector method is dependent on
counting cells in two (usually adjacent) thin 4-5 pm
sections (see Sterio, 1984; Oorschot, 1994; Raadsheer
et al., 1994). Its great strength is that the use of two
sections ensures the counts are independent of cell size
and shape (unlike previous methods, see Oorschot et
al., 1991; Oorschot, 19941, but a limitation is that it can
/Journal of Neuroscience Methods 61 (1995) 5-14 13
be time-consuming with respect to finding or following
the same neurons in consecutive sections (e.g., Moller
et al., 1990; Raadsheer et al., 1994). Thus the optical
disector is a very attractive variant of the original
physical disector principle, provided that the immuno-
cytochemical reagents can penetrate throughout the
depth of the section.
When penetration of immunocytochemical reagents
is not possible in thick sections, then the physical
disector is the method of choice (see Raadsheer et al.,
1994). This may often be the case for the neurotrans-
mitters of neurons since they are enclosed not only
within the cell membrane, but also within vesicular
membranes inside the ceil membrane. This is in con-
trast to glial cell antigens which are either expressed by
the cell membrane or are enclosed only by the cell
membrane. Thus, barriers to the penetration of im-
munoreagents to the antigen source are likely to be
less for glial cells in comparison to neuronal transmit-
ters.
The methods described in this paper now permit the
optical disector method to be used for astrocytes, oligo-
dendrocytes and myelin within the thoracic spinal cord
of the adult rat. With a slight modification to the PK
protocol, this method is also very applicable to the
detection of astrocytes within the cerebrum of normal
and injured young adult rats (Oorschot and Galvin,
1995). It therefore appears to be useful for glial cell
detection throughout the central nervous system.
There are, of course, 3 glial cell types (astrocytes,
oligodendrocytes and microgiia). The individual im-
munocytochemical protocol for microglia using OX-42
(Serotec) as the specific antibody is currently under-
way.
Acknowledgements
The generous support of the New Zealand Neuro-
logical Foundation (to D.A.S., 1991-1993 Miller Post-
graduate Scholar) and of the Dean of the University of
Otago’s Health Research Council Fund (to D.E.O.) is
gratefully acknowledged. We would also like to thank
Dr. Ian McLennan and Dr. Glenn Buchan for helpful
discussions.
References
Bastmeyer, M., Beckmann, M., Schwab, M.E. and Stuermer, C.A.O.
(1991) Growth of regenerating goldfish axons is inhibited by rat
oligodendrocytes and CNS myelin but not by goldfish optic nerve
tract oligodendrocytelike cells and fish CNS myelin. J. Neurosci.,
1 I : 626-640.
Bastmeyer, M., Bahr, M. and Stuermer, C.A.O. (1993) Fish optic
nerve oligodendrocytes support axonal regeneration of fish and
mammalian retinal ganglion cells. Glia, 8: l-11.
i‘l D. Stuorr, D.E. Oorschot /Journul
Berod, A., Hartman, B.K. and Pujol, J.F. (1981) Importance of
fixation in immunohistochemistry: Use of formaldehyde solutions
at variable pH for the localization of tyrosine hydroxylase. J.
Histochem. Cytochem., 29: 844-850.
Borgens, R.B., Blight, A.R. and Murphy, D.J. (1986) Axonal regener-
ation in spinal cord injury: a perspective and new technique. J.
Comp. Neurol., 250: 157-167.
Boenisch, T. (1989) Background. In: S.J. Naish (Ed.), Immunochemi-
cal Staining Methods, DAKO, California, pp. 24-29. I
Cammer, W., Sacchi, R. and Kahn, S. (1985a) Immunocytochemical
localization of 5’-nucleotidase in oligodendroglia and myelinated
fibres in the central nervous system of adult and young rats. Dev.
Brain Res., 20: 89-96.
Cammer, W., Sacchi, R. and Sapirstein, V. (1985b) Immunocyto-
chemical localization of carbonic anhydrase in the spinal cords of
normal and mutant (shiverer) adult mice with comparisons among
fixation methods. J. Histochem. Cytochem., 33: 45-54.
Carbonetto, S., Evans, D. and Cochard, P. (1987) Nerve fiber growth
in culture on tissue substrata from central peripheral nervous
systems. J. Neurosci., 7: 610-620.
Caroni, P. and Schwab, M.E. (1988) Antibody against myelin-associ-
ated inhibitor of neurite growth neutralizes nonpermissive sub-
strate properties of CNS white matter. Neuron, 1: 85-96.
David, S. and Aguayo, AI. (1981) Axonal elongation into peripheral
nervous system ‘bridges’ after central nervous system injury in
adult rats. Science, 214: 931-933.
Farmilo, A.J. and Stead, R.H. (1989) Fixation in immunocytochem-
istry. In: S.J. Naish (Ed.), Immunochemical Staining Methods,
DAKO, California, pp. 24-29.
Fernandez, E.. Pallini, R. and Mercanti, D. (1990) Effects of topi-
cally administered nerve growth factor on axonal regeneration in
peripheral nerve autografts implanted in the spinal cord of rats,
Neurosurgery, 26: 37-42.
Finley, J.C.W. and Petrusz, P. (1982) The use of proteolytic enzymes
for improved localization of tissue antigens with immunocyto-
chemistry. In: G.R. Bullock and P. Petrusz (Eds.), Techniques in
Immunocytochemistry, Vol. 1, Academic Press, London, pp. 239-
249.
Griffioen, H.A., Van der Beek, E. and Boer, G.J. (1992) Gelatin
embedding to preserve lesion-damaged hypothalami and intrac-
erebroventricular grafts for vibratome slicing and immunocyto-
chemistry. J. Neurosci. Methods, 43: 43-47.
Gundersen, H.J.G. (1986) Stereology of arbitrary particles. A review
of unbiased number and size estimators and the presentation of
some new ones, in memory of William R. Thompson. J. Microsc.,
143: 3-45.
Gurusinghe, C.J. and Ehrlich, D. (1986) Gelatin embedding of cen-
tral nervous system tissue improves the quality of vibratome
sections. Stain Technol., 61: 324-326.
Hsu, S.M., Raine, L. and Ganger, H. (1981) The use of the avidin-
biotin-peroxidase complex (ABC) in immunoperoxidase tech-
niques: A comparison between ABC and unlabeled antibody
(PAP) procedures. J. Histochem. Cytochem., 29: 577-580.
Janson, A.M. and Moller, A. (1993) Chronic nicotine treatment
counteracts nigral cell loss induced by a partial mesodiencephalic
hemitransection: an analysis of the total number and mean vol-
ume of neurons and glia in substantia nigra of the male rat.
Neuroscience, 57: 931-941.
Kidd, G.J., Hauer, P.E. and Trapp, B.D. (1990) Axons modulate
myelin protein messenger RNA levels during central nervous
system myelination in vivo. J. Neurosci. Res., 26: 409-418.
of Neuroscience Methods 61 (199.5) j-14
Kimura, H., McGeer, P.L., Peng, J.H. and McGeer, E.G. (1981) The
central cholinergic system studied by choline acetyltransferase
immuno-histochemistiy in the cat. J. Comp. Neural., 200: 151-
201.
Martinez, A., Lopez, .I., Barrenechea, M.A. and Sesma, P. (1991)
Immunocytochemical and ultrastructural characterization of en-
docrine cells in chicken proventriculus. Cell. Tiss. Res., 263:
541-548.
Mikoshiba, K., Kohsaka, S., Hayakawa, T., Takamatsu, K. and
Tsukada, Y. (1985) Immunohistochemical localization of 2’,3’-
cyclic nucleotide 3’-phosphodiesterase and myelin basic protein in
the central nervous system of the jirnpy and the normal mouse. J.
Neurochem., 44: 686-691.
Moller, A., Strange, P. and Gundersen, H.J.G. (1990) Efficient
estimation of cell volume and number using the nucleator and the
disector. J. Microsc., 159: 61-71.
Oorschot, D.E. (1994) Are you using neuronal densities, synaptic
densities or neurochemical densities as your definitive data?
There is a better way to go. Prog. Neurobiol., 44: 233-247.
Oorschot, D.E. and Galvin, K.A. (199.5) The effect of hypoxic-
ischemic brain injury on the total number of astrocytes in the rat
basal ganglia: is there hyperplasia? Proc. Anst. Neurosci. Sot., 6:
234.
Oorschot, D.E. and Jones, D.G. (1990) Axonal regeneration in the
mammalian central nervous system. A critique of hypotheses. In:
F. Beck, W. Hild, W. Kriz, R. Ortmann, J.E. Pauly and T.H.
Scheibler (Eds.), Advances in Amlrtmy, Embryology and Cell
Biology, Vol. 119, Springer, Berlin, I31 pp.
Oorschot, D.E., Peterson, D.A. and Jones, D.G. (1991) Neurite
growth from, and neuronal survival within, cultured explants of
the nervous system: a critical review tjf morphometric and stereo-
logical methods and suggestions for rhe future. Prog. Neurobiol.,
37: 526-546.
Raadsheer, F.C., Oorschot, D.E., Verwer, R.W.H., Tilders, F.J.H.
and Swaab, D.F. (1994) Age-related increase in the total number
of corticotropin-releasing hormone neurons in the human par-
aventricular nucleus in controls and Alzheimer’s disease: compar-
ison of the disector with an unfolding method. J. Comp. Neurol.,
339: 447-457.
Richardson, P.M., McGuinness, V.M. and Aguayo, A.J. (1980) Axons
from CNS neurons regenerate into PNS grafts. Nature, 284:
264-265.
Schnell, L. and Schwab, M.E. (1993) Sprouting and regeneration of
lesioned corticospinal tract fibres in the adult rat spinal cord.
Eur. J. Neurosci., 5: 1156-1171.
Slovitzer, R.S. and Niiaver, G. (1987) Immunocytochemical localiza-
tion of GABA-, cholecystokinin-, vasoactive intestinal peptide-,
and somatostatin-like immunoreactivity in the area dentata and
hippocampus of the rat. J. Comp. Neural., 256: 42-60.
Sterio, D.C. (1984) The unbiased estimation of number and sizes of
arbitrary particles using the disector J. Microsc., 134: 127-136.
Trapp, B.D., Bernier, L., Andrew, S.I<. and Colman, D.R. (1988)
Cellular and subcellular distribution of 2’,3’-cyclic nucleotide
3’-phosphodiesterase and its mRNA in the rat central nervous
system, J. Neurochem., 51: 859-868.
Vijayan, V.K. and Cotman, C.W. (1987) Hydrocortisone administra-
tion alters glial reaction to entorhinal lesion in the rat dentate
gyrus. Exp. Neurol., 96: 307-320.