Phenotypical peculiarities and species‐specific differences of canine and murine satellite glial cells of spinal ganglia

Abstract Satellite glial cells (SGCs) are located in the spinal ganglia (SG) of the peripheral nervous system and tightly envelop each neuron. They preserve tissue homeostasis, protect neurons and react in response to injury. This study comparatively characterizes the phenotype of murine (mSGCs) and canine SGCs (cSGCs). Immunohistochemistry and immunofluorescence as well as 2D and 3D imaging techniques were performed to describe a SGC‐specific marker panel, identify potential functional subsets and other phenotypical, species‐specific peculiarities. Glutamine synthetase (GS) and the potassium channel Kir 4.1 are SGC‐specific markers in murine and canine SG. Furthermore, a subset of mSGCs showed CD45 immunoreactivity and the majority of mSGCs were immunopositive for neural/glial antigen 2 (NG2), indicating an immune and a progenitor cell character. The majority of cSGCs were immunopositive for glial fibrillary acidic protein (GFAP), 2',3'‐cyclic‐nucleotide 3'‐phosphodiesterase (CNPase) and Sox2. Therefore, cSGCs resemble central nervous system glial cells and progenitor cells. SGCs lacked expression of macrophage markers CD107b, Iba1 and CD204. Double labelling with GS/Kir 4.1 highlights the unique anatomy of SGC‐neuron units and emphasizes the indispensability of further staining and imaging techniques for closer insights into the specific distribution of markers and potential colocalizations.

ganglia are mainly composed of neuronal somata, Schwann cells and SGCs. A unique anatomical feature of SGCs represents the tight glial sheath they form around neurons, a characteristic not seen in any other glial cell type of the CNS or PNS. 3,4 Typically, several SGCs enclose one neuronal soma as well as the initial portion of the axon. 3,4 The close morphological contact between SGCs and neurons already alludes to an intimate functional interdependence. [4][5][6] The enveloping SGCs thus seem to possess a comparable role to astrocytes in the CNS in preventing direct contact between blood vessels and neurons. 4,[6][7][8] A thin layer of connective tissue separates each neuron-SGC unit. Sensory neurons do not form synapses to each other. However, intercellular communication is thought to be achieved by exchanging signals through and with SGCs. SGCs and neurons are believed to communicate via transmission of chemical substances such as ATP and Ca 2+ as well as receptor-dependent activation of intracellular pathways. 7,[9][10][11] Beyond that, SGCs have proven to interconnect with each other through gap junctions. 12,13 SGCs display important modulatory and protective functions for controlling and maintaining the microenvironment of neurons, comparable to central glial cells. Generally, the response of SGCs to injury, for example to the peripheral nerve is multifaceted. For instance, murine and rat SGCs begin to proliferate, become hypertrophic and upregulate the expression of glial fibrillary acidic protein (GFAP) in pathologic conditions. 14-17 A recent study investigating transcriptional changes in SGCs following peripheral nerve injury demonstrates that SGCs are also engaged in 'injury-induced immune-related processes in the DRG'. 18 Furthermore, human trigeminal SGCs express a variety of Toll-like receptors (TLR) and produce cytokines after stimulation with eligible TLR ligands. Thus, SGCs might also play an important role in triggering and managing the inflammatory response to pathogens. 19 Interestingly, in vitro studies of SGCs indicate that they represent multipotent glial cells or may even display developmentally arrested Schwann cells. [20][21][22] Moreover, SGCs seem to be susceptive to being differentiated towards specific phenotypes resembling that of oligodendrocytes, oligodendrocyte precursor cells and astrocytes, in vitro. 20,22,23 These features could make SGCs promising candidates for further research in regeneration and reparation after CNS injury.
Overall, SGCs appear to be a plastic cell population with multiple functional roles. Although the interest in this cell population is growing, few studies have specifically dealt with canine SGCs (cSGCs), and current knowledge of cSGCs is scarce in comparison to murine (mSGCs) and rat SGCs. [23][24][25] The dog is of particular interest as it represents a suitable translational large animal model for certain canine and human CNS diseases, including spinal cord injury. Dogs show comparable pathogenic mechanisms, lesion distribution and morphology as well as clinical manifestations. [26][27][28] A better understanding of cSGCs and their potential regenerative properties will be beneficial for future applications in regenerative medicine. The goal of this study is to provide a detailed phenotypical analysis of cSGCs in direct comparison to mSGCs. Furthermore, the study aimed for an in-depth characterization of the expression and anatomical localization of selected markers using different staining and visualization methods. Finally, it was aspired to investigate potentially different functional subsets of SGCs and at the same time for detecting possible interspecies differences.

| Animals and tissue sampling
Cervical, thoracic and lumbar SG of six female and two male, adult C57BL/6 wildtype mice were harvested for this study. Likewise, cervical, thoracic and lumbar SG from two female and two male, Alternatively, fresh-frozen tissue was collected using optimal cutting temperature (OCT) compound (Tissue-Tek ® OCT ™ Compound, Sakura, Alphen aan den Rijn, Netherlands) and snap-freezing in liquid nitrogen. The fresh-frozen, OCT-embedded (FFOE) SG were cut into approximately 5 µm sections on a cryostat (Leica, CM1950), mounted on SuperFrost-Plus® slides, fixed in acetone (Roth C. GmbH & Co. KG) for 10 minutes and stored at −80℃ until use for IF staining.

| Immunohistochemistry
For the phenotypical characterization of cSGCs and mSGCs, formalinfixed, paraffin-embedded (FFPE) sections were immunostained performing the avidin-biotin-peroxidase complex (ABC) method. The used primary antibodies including dilutions and antigen retrieval technique are listed in Table 1. Sections were deparaffinized in Roticlear® (Roth C. GmbH & Co. KG) and rehydrated with graded series of alcohols. Endogenous peroxidase was blocked using 0.5% H 2 O 2 in 85% ethanol for 30 minutes at room temperature (RT). Antigen retrieval was achieved by boiling in citrate buffer (pH = 6.0) for 20 minutes in a microwave (800 W). Non-specific binding sites were blocked with inactivated serum from the respective host species of the secondary antibody, followed by overnight incubation of the primary antibodies

| Immunofluorescence
Departing from the protocol for FFPE material indicated above, pri-  After incubation with the appropriate secondary antibodies on the next day, the directly labelled Kir 4.1 antibody was incubated for another 24 hours.

| Laser scanning confocal microscopy and 3D reconstruction
In order to further illustrate and confirm the localization of selected For the 3D images and movies, a series of optical sections (z-stacks) were collected and analysed with LAS X 3D version 3.1.0 software from Leica. Z-stack pictures were used and the background set to black by standard software settings.

| Picture analysis
Images of immunohistochemical and-fluorescence staining were captured with a BZ-9000E microscope (Keyence Deutschland GmbH). Selected sections were also analysed in 3D-reconstructed images of confocal laser microscopy.

| Canine SG
For antibodies creating a distinct signal in IHC of FFPE material, three SG of each of the four dogs were analysed. For each antibody, a maximum of ten randomly selected high power fields (40X) were evaluated. Immunopositive and immunonegative SGCs within the pictures as well as the number of associated neurons were counted manually using Fiji Is Just ImageJ software. 29 For CD204 and Iba1, IF double labelling with GS of one representative SG of each dog was performed in order to rule out falsepositive results. Prior to quantitative analysis, exemplary sections were investigated using laser scanning confocal microscopy. A maximum of ten pictures per SG was examined, and the percentage of immunopositive SGCs was calculated. Furthermore, prior to quantification, 3D-reconstructed images of these staining were analysed to substantiate results from 2D images.

| Colocalization analysis
To further substantiate a potential overlap of selected markers, that could represent potential functional subsets (NG2 with Kir 4.1 and CD45 with Kir 4.1), the EzColocalization plugin for ImageJ (version 1.53c; http://imagej.nih.gov.ij/) was exemplary applied to selected zstack images captured with a confocal microscope. 30 Furthermore, z-stack images of a marker anticipated not to co-localize with SGC markers (CD204 and GS) were included as an internal control. All single images from z-stacks were investigated using Spearman's rank correlation neural progenitor markers (NG2, nestin, Sox2) in order to investigate a potential multipotent character of SGCs. 20,22 The results of the quantitative analysis of all applied markers (Table 1) are depicted in Table 2.  Figure 1A,B is provided in Video S1; a movie of 3D confocal reconstructions of Figure 2A, In addition, an antibody specifically targeting an extracellularly located epitope of Kir 4.1 (APC-165, Alomone laboratories Ltd.) was also found in mSGCs and cSGCs (data not shown). This is particularly interesting for prospective in vitro studies, because extracellularly located epitopes can be targeted in order to separate the desired cell population from others.
majority of cSGCs lacked Iba1 expression. In fact, it seems that resident macrophages next to SGCs are the main source of the obtained signals ( Figure 4; a movie of 3D confocal reconstructions of Figure 4 B is provided in Video S5).
In the canine SG, a limited number of cells expressed CD204.
Again, IF double labelling with GS as well as confocal microscopy of a representative SG revealed that most likely resident macrophages in close vicinity to SGCs were immunopositive for CD204. No SGCs showed co-labelling of GS and CD204 (Figure 4; a movie of 3D confocal reconstructions of Figure 4 A is provided in Video S6).
73.89% of mSGCs were immunopositive for CD45 in IF staining of FFOE material ( Figure 5; a movie of 3D confocal reconstructions of Figure 5 is provided in Video S7). Using splenic tissue as positive control, anti-CD45 antibody staining pattern observed in canine tissue did not reflect the organ-typical structure and was therefore considered to have low sensitivity in canine tissue. 55.6% of the analysed mSGCs expressed NG2. The 3D-reconstructed image ( Figure 6; a movie of 3D confocal reconstructions of Figure 6 is provided in Video S8) confirms the co-labelling of mSGCs with the SGC-specific marker Kir 4.1 and NG2. IF staining of FFOE material created a more distinct staining pattern compared to IHC of FFPE material. Hence, FFOE tissue processing proved to be best suitable for this antibody. The anti-NG2 antibody of this study did not work appropriately on canine tissue and was therefore excluded for this species. In the present study, 0% of mSGCs and 96.63% of cSGCs stained positive for Sox2 (Figure 7). IHC and IF produced a clear nuclear signal in Sox2 positive cSGCs (Figure 7; Figure S1). In IHC, 0% of the investigated mSGCs and cSGCs showed an immunoreaction for nestin (Figure 7).

| Colocalization analysis
Exemplary evaluation of confocal images of murine SG stained with was observed. Especially, the M2 value suggests that a colocalization for these markers is highly unlikely ( Figure S6).

| D ISCUSS I ON
Most of the recent studies focus on the phenotype of SGCs in an activated state, for example in the context of pain and nociception in response to injury. The aim of this study was to comparatively characterize mSGCs and cSGCs. Furthermore, the goal was to identify different subsets among SGCs that could be indicative of distinct functional aspects of SGCs.

| GS and Kir represent SGC-specific markers in mice and dogs
Several studies have characterized GS as an SGC-specific marker in murine and rat sensory ganglia that identifies SGCs in situ and in vitro. 6,9,[35][36][37][38][39][40][41] GS catalyses the conversion of the excitatory neurotransmitter glutamate to glutamine 42 and is expressed by SGCs together with glutamate receptors and glutamate transporters. 43 In the CNS, astrocytes represent the main glial cell population expressing GS. 44 Interestingly, this study demonstrates that the been mentioned before. 45 Based on the results obtained from this study, GS can be used as a SGC-specific marker not only in murine but also canine SG.
The inwardly rectifying potassium channel Kir 4.1 is responsible for K + buffering, which regulates excitability of neurons, too. 46 In the CNS, this subunit of potassium channels is again mainly expressed by astrocytes. 46 Previous studies investigated the expression of Kir 4.1 by SGCs and its functional significance. 34,47,48 In response to injury to the SG or the peripheral nerve, Kir 4.1 expression was significantly reduced leading to an increased excitability of neurons. 47,49,50 It is assumed that Kir 4.1 is the main channel responsible for potassium influx and hence regulation of extracellular potassium concentrations in the SG. 34 In this study, the majority of mSGCs expressed Kir 4.1.
Interestingly, cSGCs consistently expressed Kir 4.1, too, which has not been described yet. This indicates a similar role of cSGCs in regulation of potassium concentration and therefore excitability of sensory neurons. Altogether, Kir 4.1 is considered a highly suitable SGC-specific marker in the SG of both species. Moreover, GS and Kir 4.1 are also described to be specifically expressed by human SGCs of SG. 51

| Glial cell characteristics of SGCs
GFAP is involved in the structure and function of the cytoskeleton and therefore also in cell motility and migration. In astrocytes, an increased expression of GFAP indicates an activated state and plays an important role in the formation of the thickened and elongated processes. 52 Similarly, GFAP expression is upregulated in activated, injured murine and rat SGCs of sensory ganglia, while it is often below detectable level in a non-activated state. 16,35,[53][54][55] In contrast to mSGCs, the majority of cSGCs within this study expressed GFAP, which is in accordance with previously published data. 23 It can be hypothesized that adult cSGCs possess a phenotype, which more closely resembles that of CNS glial cells, es-

| MSGCs and cSGCs do not exhibit macrophagerelated markers (Iba1, CD204, CD107b), but mSGCs display a subset of cells positive for common leukocyte antigen (CD45)
It has been proposed that SGCs influence the immune system or even display an immune cell character themselves. A recent study investigated the transcriptome of mSGCs. 18 Genes linked to the immune system were enriched. Therefore, it was hypothesized that

| Expression of neural progenitor markers could indicate a potential regenerative capacity of SGCs
SGCs are derived from neural crest stem cells. 62,63 There is evidence that SGCs might retain stem cell characteristics in adult animals and are capable of dedifferentiation under certain conditions. The transcription factor Sox2 governs neural differentiation and sustains the self-renewal of neural progenitor stem cells. 64 High Sox2 reactivity was found in the SGCs of adult rat SG 65 and of young adult C57BL/6 mice. 36 Furthermore, an increase in the expression of nestin was described under chronic pain conditions in mSGCs. 36 In human adult trigeminal ganglia, nestin expression by SGCs has been described, too. 66 Nestin is an intermediate filament, part of the cytoskeleton, and can be identified in a variety of cell types and stages including neural stem and progenitor cells. 67 The results of the present studies did not detect immunoreactivity for Sox2 or nestin in adult mSGCs. However, many also remain in their immature state and represent a life-long pool of adult progenitor cells. 69 Interestingly, an NG2/GSpositive subpopulation of adult mSGCs was identified, which could represent a functional subset with potential regenerative capacities among this cell population, too. In previous studies, SGCs from embryonic and post-natal rat SG differentiated into cells resembling the phenotype of oligodendrocytic precursor cells positive for NG2 and platelet-derived growth factor receptorα (PDGFRα). 20 In another study, SGCs of lumbar rat SG also expressed NG2. 70 However, NG2-expression in adult mSGCs could represent a transient occurrence depending on the developmental state like it has been described in rats, too. 20 In summary, this study clearly demonstrates the influence of tis- However, SRCC and MCC values are well suited to obtain a first impression of quantity and quality of colocalization using selected markers, as shown for CD45 and NG2 in mSGCs.
There was a striking difference in the expression of GFAP, CNPase and Sox2 between mSGCs and cSGCs, which could hint towards a different function and developmental stage of SGCs with cSGCs exhibiting a more pronounced glial differentiation.
Whether the observed species-specific phenotypical peculiarities will change under pathological conditions and whether some of the features could be harnessed to use a potential regenerative capacity of SGCs needs to be further evaluated. A more in-depth understanding of cSGCs is of particular value, since dogs represent represent a suitable translational animal model for comparable human diseases, for example spinal cord injury. 26,27 In conclusion, SGCs represent a fascinating cell population that expresses a wide variety of interesting markers. These features make them attractive candidates for ensuing in vitro studies and research addressing regenerative processes post-injury to, for example, the CNS/