Perfusion with 10% neutral-buffered formalin is equivalent to 4% paraformaldehyde for histopathology and immunohistochemistry in a mouse model of experimental autoimmune encephalomyelitis

Abstract

Intravascular (IV) perfusion of tissue fixative is commonly used in the field of neuroscience as the central nervous system tissues are exquisitely sensitive to handling and fixation artifacts which can affect downstream microscopic analysis. Both 10% neutral-buffered formalin (NBF) and 4% paraformaldehyde (PFA) are used, although IV perfusion with PFA is most commonly referenced. The study objective was to compare the severity of handling and fixation artifacts, semiquantitative scores of inflammatory and neurodegenerative changes, and quantitative immunohistochemistry following terminal IV perfusion of mice with either 10% NBF or 4% PFA in a model of experimental autoimmune encephalitis (EAE). The study included 24 mice; 12 were control animals not immunized and an additional 12 were immunized with PLP_139–151 subcutaneously, harvested at day 20, and fixed in the same fashion. Equal numbers (4 per group) were perfused with 10% NBF or 4% PFA, and 4 were immersion-fixed in 10% NBF. NBF-perfused mice had less severe dark neuron artifact than PFA-perfused mice (P < .001). Immersion-fixed animals had significantly higher scores for oligodendrocyte halos, dark neuron artifact, and perivascular clefts than perfusion-fixed animals. Histopathology scores in EAE mice for inflammation, demyelination, and necrosis did not differ among fixation methods. Also, no significant differences in quantitative immunohistochemistry for CD3 and Iba-1 were observed in immunized animals regardless of the method of fixation. These findings indicate that IV perfusion of mice with 10% NBF and 4% PFA are similar and adequate fixation techniques in this model.

Central nervous system (CNS) tissues are very sensitive to handling, fixation, and processing artifacts which can affect downstream histopathologic analysis.^{1,2,7,9,12,16} These artifacts include the presence of contracted, intensely basophilic neurons (known as dark neuron artifact); artifactual neuronal and white matter vacuolation; clear halos surrounding oligodendrocytes; perivascular cleft formation; and retraction spaces. It is important to reduce artifacts in diagnostic and experimental studies because they may be misinterpreted as antemortem lesions¹¹ and they may obscure or complicate the diagnosis of other lesions (eg, if artifactual vacuolation is present, the pathologist may not appreciate subtle vacuolation occurring secondary to neurological disease).²¹ To minimize CNS artifacts in mouse models of human disease, intravascular (IV) perfusion with fixative is recommended.^7,9,22 Sources state that perfusion with either 4% paraformaldehyde (PFA) or 10% neutral-buffered formalin (NBF) can yield good results,^10,17 although perfusion with 4% PFA is most commonly referenced in the neuroscience literature. There may be concern among researchers that perfusion with NBF could yield inferior results compared to PFA, especially for immunohistochemistry given the presence of approximately 1% methanol in NBF.¹³ A previous study comparing immersion fixation in 10% NBF versus 4% PFA found differences in histologic appearance and intensity of immunohistochemical labeling in mouse tissues injected with human uterine cervical squamous cell carcinoma cells.¹⁴ To our knowledge, no study has directly compared IV perfusion with 4% PFA, IV perfusion with 10% NBF, and immersion fixation of the mouse brain in respect to semiquantitative histology scoring and quantitative immunohistochemistry.

PFA is supplied as a powder of polymerized formaldehyde and when prepared yields a solution of higher purity and lacks methanol, which is present in 10% NBF.⁵ Ten percent of the NBF is a solution containing 3.7% formaldehyde (1:10 dilution of a 37% formaldehyde stock solution) and is most commonly used for immersion fixation in both clinical and experimental pathology laboratories.¹⁴ Unlike 4% PFA, 10% NBF is stable at room temperature and is not light sensitive, making it more readily available and easier to use.⁸

Experimental autoimmune encephalomyelitis (EAE) is an inducible rodent disease model of multiple sclerosis. This disease model recapitulates many of the key histopathological lesions of multiple sclerosis including inflammation of the CNS, gliosis, demyelination, and axonal loss⁴ and was therefore chosen to help address the experimental question posed for this study.

The goal of this study was to determine, in a mouse model of EAE, whether fixative choice and perfusion method affect fixation and processing artifacts, histopathology severity scores, and immunohistochemistry results. Our hypothesis was that IV perfusion would yield fewer artifacts than immersion fixation, and that IV perfusion with 10% NBF and 4% PFA would give similar results.

Methods

Animals

Female 6- to 8-week-old SJL/J mice were purchased from The Jackson Laboratory (Bar Harbor, ME), maintained in barrier rooms at the University of Iowa under 12-hour light/dark cycle, and fed ad libitum. Animals were humanely cared for and studied as approved by the University of Iowa’s Institutional Animal Care and Use Committee. All methods were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory animals (NIH Publications No. 8023, revised 1978).

On day 1, mice were injected subcutaneously with saline (n = 12) or immunized with 50 μg of PLP_139–151 (HSLGKWLGHPDKF) peptide emulsified in 1:1 volume with complete Freund’s adjuvant supplemented with 4 mg/ml Mycobacterium tuberculosis (n = 12).^15,19 At day 20, all mice were humanely euthanized via carbon dioxide asphyxiation and CNS tissues were fixed in one of the 3 ways: postmortem IV perfusion with 10% NBF (n = 4 from each group); postmortem IV perfusion with 4% PFA (n = 4 from each group); or immersion fixation in 10% NBF (n = 4 from each group).

Histopathology

After euthanasia, perfusion was performed using a gravity-fed system through the left ventricle which perfused at a pressure of 100 cm of water at a rate of 10 ml/min with first chilled phosphate-buffered saline (PBS) and then 40 ml of chilled fixative.⁶ Approximately 30 minutes following perfusion, brains were removed, without opening the dura, and immersion-fixed in 10% NBF or 4% PFA for 48 to 72 hours at room temperature on an orbital shaker for NBF and in the refrigerator without shaking for PFA perfusion-fixed samples. PFA was freshly prepared in phosphate-buffered saline the day before perfusion and refrigerated prior to use. For immersion-fixed animals, the brain was fixed in situ in the intact skull for 24 to 48 hours and then removed and fixed an additional 48 to 72 hours in the same fixative used for perfusion at room temperature on an orbital shaker. Heads were skinned but overlying muscle was not removed and the skulls were left intact. For all mice, spinal cords were fixed in situ by immersion of the intact vertebral canal in the same fixative used for perfusion, with removal of the overlying skin but not muscle, and decalcified prior to processing at room temperature. Briefly, fixed tissues were washed with tap water and placed in 200 ml of 14% EDTA (479 mmol/L) pH 7.3 (Sigma, St. Louis, MO; catalog number E6758) for decalcification at room temperature with continuous shaking for 4 days, then washed thoroughly with tap water for 3 hours, and sections were placed into cassettes with the use of a microtome blade. Fixed tissues were routinely processed, embedded, sectioned at 5 μm, and stained with hematoxylin and eosin. For brain, coronal sections were obtained at 3 levels, including striatum; hippocampus and thalamus; and cerebellum and medulla, using a mouse brain mold for consistency. For spinal cord, a combination of longitudinal and coronal sections was obtained at 3 levels (cervical, thoracic, and lumbosacral). Slides were analyzed by 3 board-certified veterinary pathologists (J.M.S., K.N.G.C., and E.R.) blinded to experimental manipulation and fixation method, and the following parameters were scored: inflammation (Fig. 1), necrosis/demyelination, dark neuron artifact (Fig. 2), oligodendrocyte halos (Figs. 3, 4), perivascular clefts (Figs. 5, 6), and vacuolation (Supplemental Figs. 1–3). All parameters were scored on a 0–4 scale, with 0 indicating the absence of the lesion and 4 indicating severe changes (Table 1. Figures were prepared from scanned glass slides and image white balance, brightness, and contrast were adjusted using auto-corrections applied to the entire image (Adobe Photoshop Elements).

Figures 1–6.

Figure 1. Experimental autoimmune encephalomyelitis, brain, mouse. Inflammatory cells expand the meninges and perivascular space at the level of the hippocampus. Figures 2–6. Normal controls, brain, mouse. Figure 2. Dark neuron artifact (arrows) characterized by shrunken, intensely basophilic neuronal cell bodies. Figure 3. Perfusion with neutral-buffered formalin (NBF). Oligodendrocytes (arrow) do not have halo artifact. Figure 4. Immersion fixation in NBF. Oligodendrocytes (arrows) have halo artifact characterized by clear spaces around the cell. Figure 5. Perfusion with NBF. Vascular lumens are expanded. Figure 6. Immersion fixation in NBF. There are artifactual perivascular clefts (clear spaces around blood vessels) and oligodendrocyte halos.

Table 1.

Histologic scoring system for lesions and artifacts in the brain and spinal cord.

Inflammation
0	None
1	Mild meningitis, incomplete, with scattered parenchymal inflammation, <3 cell layers thick
2	Meningitis 3–6 cell layers thick, multifocal to coalescing, with occasional perivascular cuffs
3	Extensive meningitis, continuous, with frequent perivascular cuffs, 3–6 layers thick and parenchymal involvement (gray matter)
4	Severe, diffuse, thick cuffs and meningoencephalitis with regionally extensive myelomalacia
WM necrosis/demyelination
0	None
1	Scattered inflammation WM and axonal loss, <5%
2	Focal to multifocal mild inflammation of WM with necrosis, axonal loss, affecting 5%–20% WM
3	Multifocal moderate inflammation WM with necrosis and axonal loss/rarefaction WM, affecting 20%–40% WM
4	Extensive severe WM necrosis/axonal loss and inflammation, >40% WM
Perivascular clefts, oligodendrocyte clefts, dark neuron artifact, and vacuolation
0	None
1	Minimal—occasional (2–5 per section), scattered, mild when present
2	Mild—<10% of tissue, mild when present
3	Moderate—10%–25%, regionally extensive, moderate when present
4	Severe—>25%, severe

Abbreviation: WM, white matter.

Immunohistochemistry was performed on paraffin-embedded sections from brain and lumbar spine using antibodies for Iba-1 (1:1500 dilution; WAKO Chemicals, Richmond, VA; catalog number 019–19741) and CD3 (1:200 dilution; NeoMarkers Company, Fremont, CA; catalog number RM09107-S). Briefly, antigen retrieval was done using pH 6.0 Citrate buffer in a decloaking chamber (Biocare, Pacheco, CA; catalog number RD913) for 5 minutes at 125°C followed by blocking and incubation of the primary antibody for 60 minutes at room temperature. Rabbit Envision HRP (Dako Company, Santa Clara, CA; catalog number K400211–2) was used for detection for both protocols.

Quantitative Immunohistochemistry

Image analysis was performed using whole-slide digital images and automated image analysis. All slides were scanned in bright field with a 20× objective using a Nanozoomer Digital Pathology slide scanner (Hamamatsu, Bridgewater, NJ). Whole-slide digital images were imported into Visiopharm software (Hoersholm, Denmark) for analysis. The software converted the initial digital imaging into gray scale values using 2 features, RGB-R with a mean filter of 5 pixels by 5 pixels and an RGB-B feature. Visiopharm was then trained to identify positive labeling and the background tissue counterstain using a project-specific configuration based on threshold pixel values. Images were processed in batch mode using this configuration to generate the desired outputs (eg, area of Iba-1 and ratio of Iba-1 to total tissue area). Based on output for each brain analyzed, the mean value for each treatment group was calculated by averaging the results for the individual brains in each group.

For quantitative measurements of spinal cord, the Visiopharm Image Analysis module was used to define regions of interest by manually drawing around the spinal cord to exclude the vertebral body. As with the whole-slide analysis for brain, positively labeled versus unlabeled tissue was segmented using a project-specific configuration to generate the desired outputs.

Statistics

The median scores of the 3 veterinary pathologists for each parameter on each slide were calculated and used for statistical analysis. For a given fixation group, results from all levels of the CNS in all mice were combined on initial analysis, and then the analysis was repeated using scores from either brain or 3 levels of spinal cord. For inflammation and demyelination/necrosis analysis, only scores from the EAE mice were used; for all other parameters, scores from both control and EAE mice were used in the analyses.

For statistical analysis of quantitative immunohistochemistry results, values for CD3 and Iba-1 immunolabeling from 4 animals in each of 6 groups (immersion-fixed EAE, NBF perfusion-fixed EAE, PFA perfusion-fixed EAE, immersion-fixed control, NBF perfusion-fixed control, and PFA perfusion-fixed control) were used. Brain sections were analyzed separately from lumbar spinal cord, although for the NBF perfusion-fixed EAE group, only 2 animals had Iba-1 labeling on brain and 3 had Iba-1 labeling on lumbar spinal cord.

Data were analyzed in GraphPad Prism (version 7; San Diego, CA) by one-way analysis of variance using a nonparametric test (Kruskal-Wallis test) followed by direct comparisons between all groups (Dunn’s multiple comparison test). For quantitative immunohistochemistry analyses, the ratio of positive labeling to total tissue area was calculated for each animal and the mean value was calculated by treatment group and results were analyzed using the same statistical test described above. Statistical results with a P-value ≤.05 were considered statistically significant.

Results

First, we investigated the consistency of scores among pathologists. Pathologists had good score agreement, with >96% of all pathologist scores within 1 point of the median score, for all parameters except for vacuolation (94% agreement within 1 point). Specifically, scores for white matter necrosis/demyelination, perivascular clefts, inflammation, and dark neuron artifact had >98.5% agreement within 1 point of the median score.

Next, we looked at the effect of fixation method and type on scores for tissue artifacts and lesions. Parameters scored on a 0–4 scale included: inflammation (Fig. 1), necrosis/demyelination, dark neuron artifact (Fig. 2), oligodendrocyte halos (Figs. 3, 4), perivascular clefts (Fig. 6), and vacuolation (Supplemental Figs. 1–3). Of these, the method of fixation did not affect the scores for inflammation or necrosis/demyelination in EAE-induced animals, for either the brain or the spinal cord or both (Figs. 7, 8). For oligodendrocyte halos and perivascular clefts, there were significant differences (P < .05) between immersion fixation with 10% NBF and either method of perfusion fixation (10% NBF or 4% PFA) for both the brain and the spinal cord as well as for all CNS regions combined (Figs. 3, 4, 9, and 10).

Figures 7–10.

Histopathologic scores of central nervous system tissues. There was no significant effect of fixation type on histopathologic scores for inflammation (Figure 7) or necrosis/demyelination (Figure 8). Scores for both oligodendrocyte halos (Figure 9) and perivascular clefts (Figure 10) were significantly higher in immersion-fixed compared to perfusion-fixed tissues (*P < .05). For all graphs, each data point represents the median of individual pathologists scores (n = 3) for brain, cervical spinal cord, thoracic spinal cord, and lumbar spinal cord in each mouse (n = 4 mice per group). NBF, neutral-buffered formalin; PFA, paraformaldehyde.

Dark neuron artifact scores varied significantly between fixation types (P < .0001), with NBF perfusion-fixed sections having significantly less dark neuron artifact than either immersion-fixed sections or PFA perfusion-fixed sections (Supplemental Figs. 4–6). This was true for both the spinal cord (P < .0001) and the brain (P = .002) (Figs. 2 and 11–13). No significant difference was detected between immersion-fixed sections and PFA perfusion-fixed sections.

Figures 11–16.

Histopathologic scores for dark neuron artifact and vacuolation in all central nervous system (CNS) tissues (Figures 11, 14), brain alone (Figures 12, 15), and spinal cord sections alone (Figures 13, 16). NBF perfusion-fixed sections had significantly less dark neuron artifact than either immersion-fixed sections or PFA perfusion-fixed sections, for both the spinal cord (P < .0001) and the brain (P = .002). Vacuolation scores were also different between fixation types (P < .0001), and PFA perfusion-fixed samples had significantly more vacuolation than either immersion-fixed samples or NBF perfusion-fixed samples for both the combined CNS samples and the brain sections. Each data point represents the median of individual pathologists scores (n = 3) for brain in each mouse (n = 4 mice per group) or cervical spinal cord, thoracic spinal cord, and lumbar spinal cord in each mouse (n = 3 sections of spinal cord and n = 4 mice per group). *P < .05. NBF, neutral-buffered formalin; PFA, paraformaldehyde.

Vacuolation scores also varied significantly between fixation types (P < .0001), with PFA perfusion-fixed samples having significantly more vacuolation than either immersion-fixed samples or NBF perfusion-fixed samples, for both the combined CNS samples and the brain sections (Figs. 14–16; Supplemental Figs. 1–3). For spinal cord sections, degree of vacuolation in the NBF-perfused samples was significantly less than the degree of vacuolation in the PFA-perfused samples, although there was no significant difference between the immersion-fixed NBF samples and the perfusion-fixed PFA samples (Figs. 14–16).

Finally, we examined whether the percentage of positive labeling on immunohistochemistry was affected by fixation method. Quantitative immunohistochemistry showed no significant effect of fixation method on CD3 or Iba-1 immunolabeling in either brain or spinal cord of EAE-induced animals, although there were clear differences between EAE-induced animals and control animals (Figs. 17–24).

Figures 17–20.

Control (Figures 17, 18) or experimental autoimmune encephalomyelitis (Figures 19, 20), brain, mouse, with immersion fixation. Immunohistochemistry for CD3 at the level of mesencephalic aqueduct (Figures 17, 19) and Iba-1 at the level of the thalamus (Figures 18, 20). Immunopositive cells are brown with hematoxylin counterstain. m, meninges; h, hippocampus.

Figures 21–24.

Quantitative analysis of CD3 and Iba-1 immunoreactivity in central nervous system tissues. There was no significant effect of fixation method on the ratio of positive labeling to total tissue, for either CD3 in the brain (Figure 21) or spinal cord (Figure 22) or for Iba-1 in the brain (Figure 23) or spinal cord (Figure 24). Immunoreactivity did differ between EAE-induced mice and control mice. Each data point represents the ratio of positively labeled tissue to total tissue area in either the brain or the lumbar spinal cord in each mouse (n = 4 mice per group). Graphs show mean ± standard error of the mean. EAE, experimental autoimmune encephalitis; Immersion, immersion-fixed tissues; NBF, perfusion-fixed in 10% NBF; PFA, perfusion-fixed in 4% PFA.

Discussion

The goal of this study was to compare the severity of handling and fixation artifacts, semiquantitative scores for inflammation and necrosis, and quantitative immunohistochemistry following terminal IV perfusion of mice with either 10% NBF or 4% PFA in a model of EAE. No significant differences were identified between fixation groups for inflammation or demyelination. These findings indicate that histopathological changes for inflammation and demyelination, which are both cardinal findings in EAE, were not significantly affected by fixation method.

Oligodendrocyte halos and perivascular cleft formation, both of which have been described as fixation artifacts following immersion fixation,⁷ were significantly decreased in perfusion-fixed brains and spinal cords compared with immersion-fixed tissue, as expected.¹⁸ Dark neuron artifact (shrunken, intensely dark neuronal cell bodies) is also described as a common handling artifact and can be elicited by postmortem manipulation of CNS tissues.^7,18 In this study, perfusion fixation with NBF significantly decreased the number of dark neurons in both the brain and spinal cord sections in comparison with perfusion with PFA or immersion fixation. This was unexpected as dark neuron artifact is commonly attributed to handling CNS tissues prior to fixation, but it has also been documented that perfusion fixation does not eliminate dark neuron artifact, especially when there is not sufficient time (>4–24 hours) between perfusion and removal of the brain from the calvarium.^1,7,11 In this study, approximately 30 minutes was allotted between perfusion and removal of the brain, after which the brains were immersed in the same fixative used for perfusion, and this was consistent across both fixative types. The significance of our findings is unknown, but they may be affected by the short time between perfusion and removal of brains. However, these findings suggest that perfusion fixation with NBF is just as effective as with PFA for evaluation of neuronal morphology and necrosis in this model and study parameters.

Vacuolation scores had poorer agreement than the other parameters: approximately 6% of individual pathologist scores differed from the median score of all 3 pathologists by ≥2 points. Vacuolation differed significantly by fixation type and was generally more prominent in PFA perfusion-fixed samples, which also was an unexpected finding. The poor agreement in scoring may be related to more prominent vascular profiles in some perfused animals which may have been interpreted by some pathologists as mild vacuolation and to different interpretations of scoring criteria. Vacuolation of primarily the white matter has been documented to occur following holding tissues in 70% ethanol for longer than 48 hours in an automatic tissue processor.²⁰ In this study, all tissues were processed overnight and were not held in alcohol-based solutions, so this is likely not the cause of the differences noted between fixative types.

We did note a few shortcomings to the studies presented here, including that we only used a single mouse model of EAE, when there are multiple mouse strains and different ways of inducing EAE. There are also additional methods for immersion fixation that can minimize artifacts such as in situ fixation of the brain within the calvarium, followed by decalcification and sectioning⁷ which were not studied here. We also did not characterize all histopathologic parameters that could be affected by fixative type and/or method. For example, the morphology of microglia has been reported to be affected by fixation method.³

In general, fixation method did not appear to affect histopathologic analysis of severity scores of inflammation and demyelination/necrosis and immunolabeling following induction of EAE in this model. This finding indicates that regardless of fixation method, key pathologic findings remain intact. This study identifies that both PFA and NBF are reasonable and similar fixatives to use for IV perfusion of CNS tissues in mice.