Microtubule-Associated Protein 2 Expression in Canine Glioma

Canine glioma is considered a potential model for human glioma, with recent studies of occurrence, therapy, and reclassification supporting the value of the canine model. The current diagnosis of canine glioma is based on morphologic criteria and immunohistochemistry (IHC), including oligodendrocyte transcription factor 2 (Olig2), glial fibrillary acidic protein (GFAP), and 2′, 3′ cyclic nucleotide phosphatase (CNPase). Microtubule-associated protein 2 (MAP2) is a proven marker of human glioma and is used to complement the diagnosis and its specific immunoreactivity pattern contributes to the differentiation of astrocytomas from other glial tumors. The objective of this study was to evaluate whether canine gliomas express MAP2 and to explore differences in the pattern of immunolabeling between different gliomas. Seventy-eight cases of canine glioma were evaluated for MAP2 expression by immunohistochemistry. A glial origin was supported by Olig2 IHC in all cases. MAP2 immunolabeling was evaluated on a semi-quantitative basis, including the percentage of immunolabeled neoplastic cells, as well as the signal intensity, distribution, and pattern of immunolabeling. MAP2 was expressed in all cases, with significant correlation between diagnosis and signal intensity (P = 0.04). MAP2 immunolabeling distribution was dominated by diffuse (34/78; 44%), followed by patchy (20/78; 26%), multifocal to coalescing (16/78; 21%), and scattered (8/78; 10%). All oligodendrogliomas (53/53; 100%) and undefined gliomas (12/12; 100%) revealed a combination of perinuclear and cytoplasmic immunolabeling, and all but 3 astrocytomas had a combination of perinuclear and cytoplasmic processes immunolabeling (10/13; 77%). Significant correlation between immunolabeling pattern and diagnosis was obtained (P = 0.001). The study demonstrates that MAP2 is expressed in canine gliomas and the pattern of expression can also be applied to help distinguish astrocytomas from oligodendrogliomas and undefined gliomas.


INTRODUCTION
Canine brain tumors are a potential naturally occurring model for their counterpart human disease and occur at a comparable incidence between the two species (1)(2)(3)(4)(5). Recent efforts to reclassify canine gliomas (6) have bridged the gap between canine and human glioma histopathology by employing a streamlined and more comprehensive tumor classification scheme. This reclassification has standardized the diagnoses and grading of canine glioma to oligodendroglioma, astrocytoma, or undefined representing either high-or low-grade neoplasms (6). Primary tumors of the central nervous system (CNS) in dogs are dominated by meningioma (∼50%) (7), followed by gliomas (∼40%) (8,9), and choroid plexus tumors (∼7%) (10,11). Over 50% of gliomas occur in brachycephalic breeds, such as boxers, Boston terriers, and bulldogs. Even among the brachycephalic breeds, boxers, bulldogs, and Boston terriers have a reported higher incidence for astrocytomas in particular (12). The overall incidence of primary CNS tumors in dogs ranges from 3 to 5% of all neoplasms and is similar to their human counterparts (1,13). The prognosis of dogs with primary CNS tumors is guarded to poor, and numerous cases are euthanized after the initial diagnosis. Therapeutic management of these cases, when elected, follows similar protocols as in humans, including surgical resection, radiation, and chemotherapy. Histologic characterization of the neoplasm is integral to deciding on the therapeutic course (14), but therapy is often attempted without fulfilling this step (15).
A variety of morphological and immunohistochemical (IHC) similarities between canine and human gliomas have been reported, particularly in astrocytomas (16)(17)(18)(19). Olig2, a transcription factor, consistently labels canine glioma (20) with additional differentiation aided by expression of CNPase, for oligodendroglioma, and GFAP, for astrocytoma (6,21). Furthermore, expression of growth factors and receptors, such as VEGF, EGFR-1, and PDGFR alpha, which correlate with higher grades for astrocytomas in humans, are also documented in dogs (18). In human neuropathology, molecular techniques are the norm for determining the diagnosis and prognosis; however, such techniques are not yet widely used in the veterinary field. Recent studies do indicate a wide variety of cell signaling abnormalities in canine glioma, many similar to those described in human glioma (22).
The cellular origin of gliomas is still a heavily debated topic (23,24). Current theories are focused on neural stem cells (23); however, there is strong evidence that mature astrocytes can dedifferentiate and regain immature glial properties and be a source of neoplastic transformation (25,26). While the cell of origin remains unclear in the dog, canine oligodendrogliomas do express a number of immunohistochemical markers of oligodendrocyte precursor cells including SOX10 and NG2, supporting a role for altered progenitor cells in the pathogenesis (27).
Microtubule-associated proteins (MAPs) are a group of cytoskeletal proteins of the MAP2/Tau family (28) that regulate homeostasis of microtubules. MAP2 binds to and increases the rigidity of the microtubules. Previous research has indicated that MAP2 plays an important role in forming dendritic protrusions in astrocytes and oligodendrocytes (28,29). As cytoskeletal microtubules are essential for cell migration, interrupting microtubular dynamics would offer the potential of altering the infiltrative nature of glioma (30). MAP2 is found in two main forms, low molecular weight (MAP2c and MAP2d) and high molecular weight (MAP2a and MAP2b) protein (28). In the adult brain, MAP2a, 2b, and 2c are confined to neurons whereas MAP2d is associated with oligodendroglial precursors (31). The purpose of this study was to evaluate the expression of MAP2 in canine gliomas and determine if expression differences exist amongst oligodendroglioma, astrocytoma, and undefined gliomas.

METHODS Cases
A total of 78 canine gliomas were included in this study. Cases were retrieved from the pathology archives of Cornell University College of Veterinary Medicine, New York State Animal Health Diagnostic Center; University of Georgia College of Veterinary Medicine, Athens Diagnostic Laboratory; Colorado State University College of Veterinary Medicine; and Texas A&M College of Veterinary Medicine. All cases were fixed in 10% neutral buffered formalin, embedded in paraffin, sectioned at 5 µm, and stained with hematoxylin and eosin. All cases were classified and graded according to Koehler et al. (6).

Immunohistochemistry Protocols and Evaluation
Using an online sequence alignment tool (https://blast.ncbi. nlm.nih.gov/Blast.cgi), the MAP2 sequences for the dog were compared to sequences for rat and human and sequence homology indicated that an antibody for MAP2 that was derived in a rat would be predicted to label canine tissue. IHC for MAP2 (Sigma-Aldrich, St. Louis, MO; catalog number M4403) and Olig2 (Abcam Cambridge, MA, catalog number EPR2673) were performed using the Leica BOND Max IHC Staining System (Leica Systems, Buffalo Grove, IL). Pretreatment with heat-induced antigen retrieval was performed for 30 min using Tris/EDTA pH 9 (Bond Epitope Retrieval Solution 2, Leica, catalog #AR9640) for MAP2 and Olig2. Endogenous peroxidase activity was blocked with a 3% peroxide solution for 5 min (Leica, catalog #DS9800). The antibodies were diluted at 1:1,000 for 60 min (MAP2) and 1:200 for 60 min (Olig2). Biotin-free PowerVision poly-polymeric horseradish peroxidase anti-mouse (MAP2) (Leica, catalog #PV6114) or anti-rabbit (Olig2) (Leica, catalog #PV6119) IgG reagent was then applied to the slides for 30 min, followed by incubation with Bond Polymer Refine Detection for 10 min (Leica, catalog #DS9800). Tissues were developed with 3,3diaminobenzidine (DAB) (Leica, catalog #DS9800) for 10 min. The slides were counterstained with hematoxylin (Leica, catalog #DS9800) for 5 min, dehydrated, cleared, and mounted. A tissue microarray composed of a number of canine tissues including kidney, lung, liver, spleen, brain, heart, small intestine, testis, lymph node, and pancreas was used to determine the specificity of the antibody to canine nervous tissue. In the tissue microarray, labeling was consistently present in the brain sections. Positive controls consisted of normal canine brain where MAP2 immunolabeling is found in the  cytoplasm of neurons. Negative controls consisted of an isotypematched antibody. The degree of signal intensity was measured on a semiobjective scale of 0-3 according to the following criteria: 0 = no labeling; 1 = less than 25% of neoplastic cells with immunolabeling; 2 = 25-75% of neoplastic cells with immunolabeling; 3 = more than 75% of neoplastic cells with immunolabeling. Percentage of positive cells were determined based on the overall immunoreaction within the entirety of the neoplasm. The overall strength of the signal intensity was scored based on reasonable staining differences from 0 to 3, with 0 corresponding to no detectable signal, and 1, 2, and 3 corresponding to low, intermediate, and strongest signal intensity, respectively. The distribution of the immunolabeling within the neoplasm was evaluated as diffuse (throughout the neoplasm), multifocal to coalescing (large areas of immunolabeling within the mass that are confluent), patchy (small clusters of neoplastic cells with immunolabeling), and scattered (mostly individualized cells with immunolabeling). The cellular pattern of immunolabeling was also evaluated as nuclear (Nc), perinuclear (PNc), cytoplasmic (Ct), cytoplasmic processes (CtP), or any combinations of these. The IHC immunolabeling was scored independently by EAD and by a board certified veterinary pathologist (ADM). Intensity scores were then determined for each case through a combined review of EAD and ADM's original scoring. Olig2 expression was evaluated for percentage of neoplastic cells with immunolabeling based on the same scoring system employed for MAP2. All data were statistically analyzed using JMP TM software (www.jmp.com).
MAP2 cellular pattern of immunolabeling: All oligodendrogliomas (53/53; 100%) and undefined gliomas    (Figures 1C-F, 3C-F). Three astrocytomas (3/13; 23%) had similar PNc and Ct staining; however, the remaining astrocytoma cases had a combination of PNc and CtP staining (10/13; 77%; Figures 2B,D-F). A significant correlation between staining pattern and diagnosis was obtained (P = 0.001). Specifically astrocytomas were more likely to stain with a combination of PNc and CtP, and oligodendrogliomas and undefined gliomas were more likely to stain with a PNc and Ct pattern (Supplemental Figures 1, 2).

DISCUSSION
The classification of the gliomas included in this study was based on Koehler et al. (6). Based on these new criteria, glioma diagnoses are divided into oligodendroglioma, astrocytoma, and undefined, where the latter contains a roughly equal distribution of the two former glioma subtypes. Criteria considered for oligodendroglioma are: round nuclei, coarse chromatin, nuclear rowing, artifactual loss of cytoplasm, branching capillaries, and pseudo-rosettes. For astrocytoma, the criteria are: angular nuclei, open chromatin, pleomorphism, and a lower degree of cellular density than oligodendroglioma. The distribution of glioma subtypes in this study is similar to previous findings (6), with most cases being oligodendroglioma. Determining the exact prevalence and anatomic distribution of these tumors is challenging in veterinary medicine, due to data heterogeneity and inconsistent diagnostic criteria. While sample size is a limiting factor for strong statistical correlations, boxers were overrepresented in this study, similar to other previous reports (2,10,11). A susceptibility locus on canine chromosome 26 has been associated with a predilection for glioma in brachycephalic breeds (2,22), but the routine molecular analyses necessary to confirm this predilection are nascent in veterinary medicine. The average age of the dogs in our study was 7.6 years, similar with previous reports, with 1 year old and 15 years old the minimum and maximum age, respectively. Numerous monoclonal and polyclonal antibodies have been applied in the diagnosis of canine CNS tumors (8,33), which allows for a more accurate correlation with their human counterparts. MAP2 is a proven marker in human gliomas, consistently used for diagnostic and grading purposes (31,(34)(35)(36), but its potential application in canine gliomas had not been explored to date. While sample size is a limiting factor, all cases included in this study had at least some degree of immunolabeling for MAP2. Furthermore, the majority of gliomas were scored as either a score 2 (38.5%) or a score 3 (38.5%), with significant correlation between diagnosis of glioma and immunolabel intensity (P = 0.04). MAP2 expression in human glioma reveals different immunolabeling patterns between oligodendroglioma (i.e., sparse processes, that in our study corresponds to lack of CtP immunolabeling and prominent PNc and Ct immunolabeling) and astrocytoma (i.e., densely ramified astrocytic elements, that correspond to predominately CtP staining in majority of the canine astrocytoma in this study) (37). All oligodendrogliomas (53/53; 100%) and undefined gliomas (12/12; 100%) in this study had a combination of PNc and Ct immunolabeling. Except for three astrocytomas (3/13; 23%) that had a combination of PNc and Ct immunolabeling, the majority of astrocytomas had a combination of PNc and CtP staining (10/13; 77%), with robust highlighting of cytoplasmic processes. Significant correlation between staining pattern and diagnosis was obtained (P = 0.001).
Based on these findings, MAP2 appears to be a suitable diagnostic immunohistochemical marker for canine glioma, with strong potential of aiding in the diagnosis of astrocytoma based on the pattern of immunolabeling. However, the number of cases exhibiting multifocal to coalescing (16/78; 20.5%) and patchy immunolabeling (20/78; 26%) illustrate the heterogeneity present in these tumors and raise a concern that the use of tissue microarray for immunohistochemical analysis in canine glioma may yield results that are not entirely representative of the totality of the neoplasm (33,38). The three cases of astrocytoma with PNc and Ct staining were all from one location (the cerebellar white matter) and had an unusual pattern composed of sheets of gemistocyte-like morphology. Whether this corresponds to a specific subtype of astrocytoma remains to be determined; however, tumors of this morphology have been described at this location in the dog (6). No correlation could be demonstrated between tumor grade and MAP2 immunolabeling intensity score, and this may be due to the small size and heterogeneity of the sample. The distribution of the data is summarized in Supplemental Figures 1, 2.
The overall statistical significance is difficult to interpret due to a limited and somewhat heterogeneous sample size; however, the IHC results presented herein suggest that MAP2 can be a valuable marker in the evaluation of canine glioma. Furthermore, the staining pattern obtained in this study is similar with findings in human gliomas (37), making this marker a tool for better understanding protein expression in canine glioma, but also for differentiating astrocytomas from oligodendrogliomas and undefined gliomas. Future directions of study should be focused on understanding the role that neuronal differentiation has in the pathogenesis of canine glioma.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

ETHICS STATEMENT
This study protocol did not require ethics approval according to local legislation and national guidelines.

AUTHOR CONTRIBUTIONS
This study was conceived by AM and ED. Case acquisition and interpretation was done by AM, ED, BP, CF, and DR. Experiments were performed by AM and ED. Data interpretation was done by AM and ED. Paper writing and editing was done by AM, ED, BP, CF, and DR.