Validating indicators of CNS disorders in a swine model of neurological disease

Genetically modified swine disease models are becoming increasingly important for studying molecular, physiological and pathological characteristics of human disorders. Given the limited history of these model systems, there remains a great need for proven molecular reagents in swine tissue. Here, to provide a resource for neurological models of disease, we validated antibodies by immunohistochemistry for use in examining central nervous system (CNS) markers in a recently developed miniswine model of neurofibromatosis type 1 (NF1). NF1 is an autosomal dominant tumor predisposition disorder stemming from mutations in NF1, a gene that encodes the Ras-GTPase activating protein neurofibromin. Patients classically present with benign neurofibromas throughout their bodies and can also present with neurological associated symptoms such as chronic pain, cognitive impairment, and behavioral abnormalities. As validated antibodies for immunohistochemistry applications are particularly difficult to find for swine models of neurological disease, we present immunostaining validation of antibodies implicated in glial inflammation (CD68), oligodendrocyte development (NG2, O4 and Olig2), and neuron differentiation and neurotransmission (doublecortin, GAD67, and tyrosine hydroxylase) by examining cellular localization and brain region specificity. Additionally, we confirm the utility of anti-GFAP, anti-Iba1, and anti-MBP antibodies, previously validated in swine, by testing their immunoreactivity across multiple brain regions in mutant NF1 samples. These immunostaining protocols for CNS markers provide a useful resource to the scientific community, furthering the utility of genetically modified miniswine for translational and clinical applications.


Introduction 60
Animal models are essential tools for studying the underlying mechanisms of disease as 61 well as providing a platform for preclinical research and drug discovery. Historically, rodents 62 have been one of the primary model systems for studying disease and driving drug discovery, 63 largely due to the widespread availability of well-described and validated reagents for use in 64 these model organisms. However, there are increasing instances where rodent models either 65 fail to recapitulate aspects of human disease, or that treatments that are efficacious in a rodent 66 model fail to translate to viable human therapies. This has led to development of large animal 67 models of disease, such as genetically modified swine, which are more similar to humans 68 anatomically, genetically, physiologically, and metabolically. [1][2][3][4] This increased similarity is 69 especially important when studying neurological disorders, where the anatomical and 70 physiological differences found in the rodent systems, cannot recapitulate human disease. 71 Successful genetically modified miniswine models have been established to study a number of 72 human diseases including atherosclerosis, cancer, ataxia telangiectasia, cystic fibrosis, and 73 neurofibromatosis type 1. 1,[4][5][6][7] However, while these new larger animal systems can better 74 recapitulate many of the hallmarks of human disease, there are limited tools and reagents that 75 have been well described and validated in these models. Herein, we test a number of 76 antibodies relevant to the study of the brain and neurological disorders in porcine brain tissue. 77 We focus on reagents specific to neurons and glia (astrocytes, microglia, and 78 oligodendrocytes) as these cells function together to support and protect neurons. When this 79 function is disrupted, glia are implicated as causal agents in neurological disease, including 80 Alzheimer's, Huntington's, and Parkinson's disease. 8,9 For example, dysregulation of 81 oligodendrocytes, the myelinating cells of the CNS, can lead to loss of myelination. In other 82 cases, a decrease in the density of oligodendrocytes in the prefrontal cortex may lead to 83 schizophrenia, bipolar disorder, and major depressive disorder. 10 Astrocytes, the most 84 abundant cell-type in the CNS, serve as a neuronal support cell to promote survival, 85 synaptogenesis and synapse pruning. In response to injury, astrocytes proliferate and/or are 86 "activated" [indicated pathologically by an upregulation of glial fibrillary acidic protein (GFAP)] 87 during various neurodegenerative diseases such as ALS and Parkinson's disease. 8 Microglia, 88 the primary immune cells in the CNS, can sense changes in their environment and either 89 promote healthy neurons or provide protection to neurons that have been injured or diseased. 11 90 Microglia produce proinflammatory agents that recruit inflammatory cells that are toxic to 91 neurons, contributing to neurodegenerative diseases like multiple sclerosis. 92 Neurons are highly involved in signal transmission within the CNS. 12 They release 93 chemical neurotransmitters that affect signaling between neurons and play a role in various 94 physiological functions of the CNS. A loss of neurons in specific regions of the CNS may cause 95 certain affects, for example, loss of dopaminergic neurons in the substantia nigra in patients 96 with Parkinson's disease, causes reduced balance and motor coordination. 13 97 Several studies of swine models of CNS injuries/disorders have tested of the specificity 98 of antibodies within the CNS. 14-18 However, two of these studies have focused solely on the 99 spinal cord and not the brain itself. 15,16 Of the remaining studies, only one of these describes 100 the impact of CNS injuries/disorders by addressing axonal injury and astrocytic/microglial 101 reactivity. 14 The other brain immunohistological study was similar to ours as it validated 102 antibodies, however this validation was compared to a general histological stain (Giemsa). 17 103 Moreover, we recently published a comprehensive study on antibody immunoreactivity in swine 104 tissues, specifically wild type swine, 19 but more studies are needed to validate markers that 105 have a role in CNS disorders, specifically in genetically modified swine that recapitulate 106 characteristics of a human disease. Anti-GFAP and anti-Iba1 antibodies have been used in a 107 number of swine studies, however, none have explored expression/activation within specific 108 brain regions (that may be impacted due to disease). 109 As some antibodies are predominantly reactive in a disease state, here we use a 110 recently developed miniswine model of NF1 to validate a number of CNS cell-specific 111 antibodies. 1 As patients with NF1 experience a host of CNS-specific impairments, this porcine 112 model is ideal for validation of neurologically relevant antibodies. We validate and explore the 113 expression of antibodies implicated in glial inflammation, oligodendrocyte differentiation, 114 neuronal signaling, and nociceptive function. Taken together, we provide a powerful set of tools 115 to researchers modeling neurological dysfunction in porcine models of disease. 116 117

Materials and Methods 118
Animal Tissue 119 All miniswine were maintained at Exemplar Genetics under an approved IACUC 120 protocol. All mice were maintained in an AAALAC accredited facility in strict accordance with 121 National Institutes of Health guidelines, and studies were approved by the Sanford Institutional 122 Animal Care and Use Committee (USDA License 46-R-0009). 123

Tissue Microarray 124
Regions from formalin fixed cortex (CTX), cerebellum (CB), hippocampus (HPC), 125 thalamus (THAL), corpus callosum (CC), and cerebral aqueduct (CGG) of a 15-month old, male 126 NF1 miniswine 1 were isolated and placed in tissue cassettes. These regions were selected due 127 to their relevance to neurologic disease in relation to macrocephaly (CC), 20 white matter 128 abnormalities (CTX, CB, and CC) 21 brain lesions (CB and THAL), 22 abnormal physiology 129 (HPC), 23 and aqueductal stenosis (CGG). 24 The tissue cassettes were processed in a Lecia 130 ASP300 Tissue Processor (Lecia Biosystems Inc, Buffalo Grove, IL) and embedded in paraffin. 131 Sections from each paraffin block were cut with a Leica RM2125 (Lecia Biosystems Inc, Buffalo 132 Grove, IL). Subsections of interest were marked on each slide and a circular biopsy was taken 133 from the paraffin block that matched the marked region. The paraffin biopsies were placed into 134 a tissue microarray mold and re-embedded in paraffin to create a paraffin microarray block. 135 Sections of the paraffin microarray block were cut and floated onto slides. 136 137

Validation of antibodies 138
Details regarding each of the antibodies used in this study are listed in Table 1. When 139 possible, we selected antibodies predicted to work in swine or constructed with a porcine 140 immunogen, however, very few of these antibodies exist. Therefore, we primarily selected 141 focused on antibodies known to react in multiple mammalian species (such as mouse, rat and 142 human), as the degree of homology between swine and the aforementioned mammalian 143 proteins is fairly high (89-100% similarity), 25 especially for evolutionarily conserved genes. 144 More consistent immunopositive results in swine were obtained by selecting antibodies in this 145 manner instead of antibodies that only react in human or mouse tissue. 146 The antibodies that we validated were known to react in postnatal mouse tissues based 147 on information provided from the manufacturer. According to the gene expression database at 148 the mouse genome informatics website (http://www.informatics.jax.org), 26 the markers that we 149 validated were known to have expression in the forebrain of the mouse (NG2), cerebral cortex 150 and hippocampus (doublecortin, GAD67), and corpus callosum and hippocampus (myelin PLP). 151 Other studies have found CD68 (macrosialin) expression in the corpus callosum and striatum of 152 C57BL/6 mice, 27 Olig-2 expression in corpus callosum and ventral forebrain regions, 28 O4 153 protein expression in cerebral cortex and above the cingulum, 29 and tyrosine hydroxylase 154 protein expression in the forebrain-cerebral cortex and hippocampus of mice. 30 As a positive 155 control, a coronal section of mouse brain was immunolabeled alongside the miniswine tissue, to 156 verify the proper reactivity, localization, and expression of the antibody in question. 157 Immunogen peptides sequences were obtained from manufacturer's documentation and 158 compared to swine (Sus scrofa) protein sequences from the Refseq database using the Basic 159 Local Alignment Search Tool (BLAST) from NIH ( Table 2). 160

Immunohistochemistry 161
Paraffin tissue arrays on slides were deparaffinized in xylene, rehydrated in ethanol, and 162 rinsed in double distilled water. Antigen retrieval was performed at 90C for 20 minutes using 163 sodium citrate buffer, pH 6. Then, slides were rinsed in 1xTBST, endogenous peroxidases were 164 blocked in Bloxall™ (Vector Laboratories, Burlingame, CA) for 10 minutes, and rinsed again in 165 1xTBST. For antibodies raised in rabbit, blocking serum from an ImmPRESS™ HRP Anti-166 Rabbit IgG (Peroxidase) Polymer Detection Kit (Vector Laboratories) was incubated on slides 167 for 20 minutes at room temperature (RT). Slides were drained and the primary antibody was 168 incubated on the slides at 4C overnight. Negative controls without primary antibody were run in 169 parallel with normal host IgG. Slides were rinsed in 1xTBST, and ImmPRESS™ (Peroxidase) 170 Polymer was incubated on slides for 30 minutes at RT. Slides were rinsed in 1xTBST and 3,3'-171 diaminobenzidine (DAB) from Vector laboratories was added to the slides for 2 to 10 minutes 172 until the DAB activation occured. Slides were then washed with DI water, stained with Mayer's 173 hematoxylin, washed with running tap water, dipped in 0.25% Lithium carbonate, rinsed in DI 174 water, dehydrated with ethanol, cleared with Xylene and mounted with DPX mounting media. 175

For antibodies raised in mouse, an ImmPRESS™ HRP Anti-Mouse IgG (Peroxidase) Polymer 176
Detection Kit (Vector Laboratories) was used, followed by the described DAB staining. For 177 antibodies raised in rat, 1% goat serum with Triton was used as a blocking solution for at least 1 178 hour at RT and a Goat Anti-Rat IgG H&L (HRP) (Abcam ab97057) was used as a secondary 179 and incubated for 1.5 hours at RT before continuing with the described DAB staining. Tissue 180 sections were viewed with an Aperio Versa slide scanner (Lecia Biosystems Inc, Buffalo Grove, 181 IL) and images were extracted with Leica's ImageScope software. 182

Dorsal root ganglia (DRG) Immunostaining 183
Miniswine DRGs were fixed in 4% PFA for 24 hours, cryoprotected in 30% sucrose (m/v) 184 in PBS for 48 hours and frozen at -70°C in 2-methylbutane chilled with dry ice. Samples were 185 cut into 20µm-thick sections. Sections were subsequently blocked in 3% BSA, 0.1% Triton X-186 100 in PBS for 1 hour at room temperature then incubated with primary antibodies (anti-TRPV1, 187 Neuromics GP14100; anti-CGRP, Abcam ab16001) diluted in blocking solution overnight at 188 +4°C. After 3 washes in PBS, slides were incubated with secondary antibody diluted at 1/1000 189 in blocking solution for 2 hours at room temperature, washed and counterstained with DAPI. 190 For the negative control, primary antibodies were omitted. Images were acquired on an Axio 191 Imager 2 (Zeiss), using a 10X objective controlled by the Zen software (Zeiss). to that seen previously in human and swine cerebellum. 19 Immunopositive cells were present in 202 the NF1 miniswine cortex, cerebellum and thalamus, mirroring what has been observed in 203 identical regions of the human brain. 32 204 Cytoplasmic ionized calcium binding adaptor molecule 1 (Iba1) immunostaining was 205 observed in both non-reactive, ramified cells with numerous branching processes and more 206 reactive amoeboid-like cells (arrows) in the cortex, cerebellum and thalamus ( Fig. 1D-F). 207 Filamentous immunopositive structures likely represent cross-sectional microglial processes. 208 The cytoplasmic localization is similar to published results in human and swine cerebrum (also 209 referred to as AIf1). 19 We expected to document Iba1 staining in various regions of the mutant 210 miniswine brain, as injury and inflammatory factors activate Iba1+ microglia in affected areas of 211 the brain, such as the cerebellum of sheep exposed to LPS, 33 and thalamus of wild type mice 212 exposed to traumatic brain injury. 34 213 Large areas within the cortex, cerebellum and hippocampus were immunopositive for 214 anti-myelin basic protein (MBP) antibodies, which label mature oligodendrocytes in white matter 215 tracts ( Fig. 1G-I, G'-I', arrows). A higher magnified image of the tracts documents the 216 filamentous morphology of the cytoplasmic and membranous localization of MBP to the myelin 217 sheath (Fig. 1G'-I'). We see a similar localization pattern to the results shown in cerebral white 218 matter of humans and swine. 19 The results are consistent with the expectation of MBP 219 immunostaining in any white matter tracts of the brain, which include regions of the cortex, 220 cerebellum, and hippocampus. 221 Detection of microglia and oligodendrocyte cell lineage in the NF1 miniswine brain 222 Reactive microglia, indicated by cytoplasmic CD68 + immunostaining, were localized to 223 the cytoplasm (arrows) in the cortex, cerebellum, and hippocampus ( Fig. 2A-C). We observed a 224 similar cytoplasm localization in mouse cerebral cortex (Fig. 2D). CD68 immunoreactivity has 225 been documented in microglia within the cerebral cortex of humans, 32 and in the multiple brain 226 regions of aging wild type mice and mice exposed to an neurological insult such as LPS. 35 227 Tracing cell type lineage can be incredibly informative in determining mechanisms of 228 disease and appropriate intervention points. Oligodendrocytes have a well-studied lineage and 229 are of particularly important for the pathology of NF1. For example, oligodendrocyte precursor 230 cells (OPCs) have been identified as the cell of origin for gliomas in NF1 conditional knockout 231 mice 36 , with NF1 mutant mice expressing more OPCs in the brain. As expected, we observed 232 membrane and cytoplasmic neural/glial antigen 2+ (NG2) immunostaining in the cortex, 233 cerebellum and thalamus, indicating the presence of oligodendrocyte progenitors (Fig. 2E-G,  234 arrows). We see a similar membrane and cytoplasmic localization in mouse cerebral cortex 235 (Fig. 2H). NG2+ immunostaining has been documented in the cerebral cortex and cerebellum 236 in human tissues, 32 and throughout the rodent brain, including the thalamus of wild-type rats. 37 237 Oligodendrocyte marker 4+ (O4) immunostaining to the membrane indicates the presence of 238 pre-oligodendrocytes in the cortex, cerebellum, and cerebral aqueduct ( Fig. 2I-K, arrows). 239 Compared to mouse cerebral cortex, we see a similar membrane localization pattern (Fig. 2L). 240 O4+ immunostaining has been found in the corpus callosum in young rat pups, 38 in the rat 241 cerebellum, 39 and in the midbrain (substantia nigra pars compacta) of control and neurotoxin 242 exposed C57BL/6 mice. 40 243 Detection of matured oligodendrocytes in the NF1 miniswine brain 244 The nucleus and cytoplasm of multiple oligodendrocytes (in various stages of 245 differentiation) were immunopositive for anti-Olig2 antibodies in the cortex, cerebellum, and 246 cerebral aqueduct (Fig. 3G-I, arrows). Comparatively, we see evidence of nuclear 247 immunostaining in mouse cerebral cortex (Fig. 3D). Though Olig2 was shown to be expressed 248 in all oligodendrocyte lineages 41 , it is known to traditionally label immature oligodendrocytes, 249 thus is most abundant in the developing brain. In the adult human brain, Olig2 + cells can be 250 found in the cerebral cortex, in molecular and granular layer cells of the cerebellum, 32 and in 251 similar regions in mice. 28,42 As a marker of mature oligodendrocytes, we also tested a myelin 252 proteolipid protein (Myelin PLP) antibody with inconsistent results. There was faint membrane 253 expression in the white matter tracts of the cortex (Fig. 3E), cerebellum (Fig. 3F) and corpus 254 callosum (Fig. 3G) to the neuropil, though there appears to be non-specific staining in other 255 cells within the CNS. We see similar staining to the neuropil and non-specific immunostaining in 256 mouse cerebral cortex (Fig. 3H). 257 Immunolabeling of various neuronal subtypes and neurotransmitters in the NF1 miniswine brain 258 Spatial learning and memory deficits are a prominent feature of neurological disease, 259 specifically affecting dopamine, GABA, and glutamate signaling in hippocampal neurons. 23,43 260 Dysregulated GABA signaling in the CNS, which causes an increase of GABA-mediated 261 inhibition, has been implicated as a cause of learning defects in mice models of Huntington's 262 disease and NF1. 23,44 Loss of dopamine signaling in hippocampal neurons has been proposed 263 to be the reason for the spatial learning and memory defects in mutant NF1 mutant mice and a 264 treatment of dopamine to these mice rescued the long-term potentiation response to normal 265

levels. 43 266
In our study, we report cytoplasmic doublecortin (DCX) + immunostaining in the cortex, 267 cerebellum and hippocampus of the adult miniswine brain, indicating the presence of immature 268 neurons ( Fig. 4A-C). Cytoplasmic immunostaining of DCX was also found in mouse cerebral 269 cortex (Fig. 4D). This was as expected, as DCX+ immunostaining has been documented in the 270 cerebral cortex and hippocampus in mice, 26  Chronic pain is usually a component of many neurological diseases that affects 289 approximately 20-40% of patients. 49 We investigated the utility of pain perception-nociceptive 290 markers in dorsal root ganglion isolated from our mutant miniswine. Sliced DRGs from 291 miniswine were immunostained with antibodies raised against calcitonin gene related peptide 292 (CGRP), a pro-nociceptive neurotransmitter, and transient receptor potential cation channel 293 subfamily V member 1 (TRPV1), a capsaicin receptor. Unlike immunostaining patterns 294 observed in rodents and Rhesus monkeys, in which subpopulations of neurons were stained by 295 each antibody; 50,51 all DRG neurons were stained by antibodies against CGRP and TRPV1 with 296 different fluorescent intensities in miniswine (Fig. 5A-B). For one of the antibodies, differences 297 in staining pattern were observed between batches. Control omitting primary antibodies 298 revealed no non-specific staining due to the secondary antibodies ( Fig. 5C-D). However, 299 comparison of the immunogen sequences with swine (Sus scrofa) protein databases in Table 2, 300 revealed that several off-target proteins could also bound by these antibodies. Therefore, better 301 controls are necessary to validate these antibodies for use in swine. oligodendrocytes, Olig2+ all oligodendrocytes). These markers are widely studied as a hallmark 318 of neurodegeneration as they commonly are found in early stages of neurological disease as an 319 indicator that the microenvironment of the brain has been altered, and neuron function is likely 320 to be disrupted. For example, upregulated Olig2 is an indicator that OPCs are activated in 321 response to autoimmune regulated demyelination in multiple sclerosis. 52 These markers 322 continue to be pathological markers in later stages of neurological diseases such as 323 Alzheimer's. In postmortem Alzheimer's patients, dense populations of CD68+ microglia exist in 324 the hippocampus, 53 and reduced NG2 immunoreactivity was found in brain tissue. 54 325 Glial markers are very effective tools to document the progression of a disease. Studying their 326 mechanisms of activation and differentiation during the pathogenesis of neurodegeneration will 327 help to develop therapeutics that treat and prevent these diseases. 328 Neuronal inflammation, which can be caused by astrogliosis and activated microglia, can 329 lead to disrupted neuronal transmission. Disrupted neuronal signaling and loss of neurons can 330 lead to learning deficits, cognitive dysfunction, and loss of motor control. The neuronal markers 331 that we studied, which are involved in GABA (GAD67+) and dopamine (tyrosine hydroxylase+) 332 neurotransmission have benefits to neurological disease research that studies activity and 333 function of neurons in conjunction with cognitive and motor impairment. For example, altered 334 GABA level and synthesis is implicated in Huntington's disease (HD), and transgenic mice that 335 express exon 1 of the HD gene had reduced levels of GAD67 in brain tissues, 55 and these mice 336 have difficulty performing spatial cognition tasks. 44 337 To study these dysfunctions in swine models, that provide more translatable human 338 therapies, reagents specific to swine need to be validated. Ours is the first study validating 339 antibodies specific for CD68, NG2, O4, Olig-2, GAD67, tyrosine hydroxylase, and myelin PLP in 340 the CNS of mutant miniswine; and the first study validating GFAP, Iba1, MBP, and doublecortin 341 in the cerebral cortex, cerebellum, thalamus, and hippocampus of miniswine. GFAP was 342 validated in 100 day old swine in the neocortex. 17 Our previous publication only validated 343 publication, we chose an anti-MBP antibody that was specific to a different amino acid chain 345 (aa82-87 vs aa182-197), a different dilution (1:100 vs 1:600), and a different secondary (HRP 346 conjugated secondary vs HRP Labeled Polymer). Our antigen retrieval process was longer for 347 anti-Iba1 and anti-GFAP (20 minutes vs 5 minutes), the temperature of our antigen retrieval 348 process was less (approximately 95C vs 110-125C), and we incubated our primary antibody 349 overnight compared to 1 hour. We also chose a less concentrated dilution for anti-GFAP 350