Nestin‐expressing cell types in the temporal lobe and hippocampus: Morphology, differentiation, and proliferative capacity

Abstract Nestin is expressed in immature neuroepithelial and progenitor cell types and transiently upregulated in proliferative neuroglial cells responding to acute brain injury, including following seizures. In 36 temporal lobe (TLobe) specimens from patients with TLobe epilepsy (age range 8–60 years) we studied the number, distribution and morphology of nestin‐expressing cells (NEC) in the pes, hippocampus body, parahippocampal gyrus, amygdala, temporal cortex and pole compared with post mortem control tissues from 26 cases (age range 12 gestational weeks to 76 years). The proliferative fraction of NEC was evaluated in selected regions, including recognized niches, using MCM2. Their differentiation was explored with neuronal (DCX, mushashi, βIII tubulin, NeuN) and glial (GFAP, GFAPdelta, glutamine synthetase, aquaporin4, EAAT1) markers, both in sections or following culture. Findings were correlated with clinical parameters. A stereotypical pattern in the distribution and morphologies of NEC was observed, reminiscent of patterns in the developing brain, with increased densities in epilepsy than adult controls (p < .001). Findings included MCM2‐positive radial glial‐like cells in the periventricular white matter and rows of NEC in the hippocampal fimbria and sulcus. Nestin cells represented 29% of the hippocampal proliferative fraction in epilepsy cases; 20% co‐expressed βIII tubulin in culture compared with 28% with GFAP. Significant correlations were noted between age at surgery, memory deficits and nestin populations. TLobe NEC with ongoing proliferative capacity likely represent vestiges of developmental migratory streams and resident reactive cell populations of potential relevance to hippocampal epileptogenesis, TLobe pathology, and co‐morbidities, including memory decline.

migratory streams and resident reactive cell populations of potential relevance to hippocampal epileptogenesis, TLobe pathology, and co-morbidities, including memory decline.
The aim of this study was to explore the distribution and morphology of nestin-expressing cell (NEC) populations in HS compared with control groups, in recognized niches as well as other temporal lobe (TLobe) regions. In addition, we aimed to explore their proliferative capacity, evidence to support neuronal versus glial maturation and their relation to the underlying histopathology and clinical features, in particular, memory dysfunction.

| Case selection and regions studied
The study has ethical approval. Brain samples from 62 cases were with HS were also included in the study. PM brain samples from five developmental fetal controls, and 15 normal adult controls were used for comparison (Table 1; see Supporting Information Table 1 for full details of all cases). HS was classified according to International League Against Epilepsy (ILAE) guidelines (Blumcke et al., 2013). Standard

| Immunohistochemistry
Immunochemistry for nestin was carried out on 5 mm-thick formalinfixed, paraffin-embedded sections from all regions in each of the cases.

| Quantitative Analysis and Regions of Interest
The distribution and morphology of NEC was evaluated in regions of interest (ROIs) including known neurogenic niches:

| Cell Culture
To investigate the lineage of NEC in vitro, cells were cultured from 0.5 to 1.2 g of fresh tissue taken from the hippocampus sample (pes and HB including all subfields, WM and dentate gyrus), and the gray and WM of the TPole from six cases with mesial TLE and Type 1 HS (cases EAC1-6). Up to 0.2 g of tissue from the amygdala was available for culture from two cases (EAC4-5). Dissociated cells were cultured in MACS Neuro Medium (Miltenyi Biotec Ltd., Surrey, UK) supplemented with 1% MACS NeuroBrew-21, 1% penicillin and streptomycin, 10 ng/ ml EGF, 10 ng/ml bFGF, and 10% fetal bovine serum (detailed in Supporting Information Methods). After 4 weeks in culture, cells were immunolabeled for nestin and GFAP or bIII-tubulin and immunolabeled cells were visualized using confocal scanning laser microscopy (LSM710; Zeiss, Germany). An average area of 45 6 4 mm 2 was imaged from each culture. Images were imported into the image analysis software Definiens Tissue Studio 3.6 and Definiens Developer x64 (Definiens AG; Munich, Germany) for automated quantification as previously described in (Liu et al., 2014b). Final results were expressed as percentages of NEC that co-expressed GFAP or bIII-tubulin, percentages of hoechst-positive cells that expressed nestin and GFAP, bIII-tubulin or both per culture. The average area of NEC from each region of each case was also recorded and the number of primary branches from each NEC, were manually counted.

| Clinical Data
Age at onset and duration of epilepsy, seizure types and laterality of HS were obtained from the clinical records (Table 1 and Supporting   Information Table S1). Results from standardized pre-operative psychometric memory and naming tests and measures of memory decline at one year post-operatively were also reviewed (Supporting Information   Table S1 and Supporting Information Methods).
Statistical analysis was carried out between pathology groups, regions and zones and with clinical and psychometric data using non-parametric tests with SPSS (IBM, California, USA; version 22); uncorrected p values of < .05 were regarded as significant. For cell culture data, non-parametric (Kruskal-Wallis and Spearman correlation) were used to determine whether the areas and the percentages of immunolabeled or co-localized cells differed significantly between regions or correlated with age at surgery.

| Nestin expression: Developmental control
In fetal brains of 12-14 gestational weeks, NEC and immunolabeled radial processes from these cells, were numerous in the SVZ of the lateral ventricle (Supporting Information Figure S1a Figure S1a, inset). Of note, the subpial surface of the  Figure S1a).

| Nestin expression: Operated epilepsy cases
T lobe: Similar patterns of NEC regional distribution were noted across

| Hippocampal body
Small multipolar NEC showed a stereotypical distribution pattern (

| PES hippocampus
PES samples were available in 17 patients; In 5/11 Type 1 HS cases (based on HB), neuronal loss in the PES was restricted to CA4 region

| PM cases (epilepsy and controls)
In the six epilepsy PM cases with HS, there was an impression of fewer NEC overall compared with surgical HS, despite strong nestin capillary endothelial labeling (Figure 1d). In bilateral HS cases, similar patterns were seen in both hippocampi. NEC in the SVZ of the HB were present in 4/6 cases, including cases in their 70s (EPM1 and 4); in one case labeling of NEC in the FZ/SPL of the hippocampus was striking (case EPM4). NECs were less frequently present in CA1 or the SGZ of the hippocampus than surgical cases. Thread-like nestin 1 fibers were present in 2/6 cases in the PHG WM, but chains of NEC not apparent.
Labeling of NEC in the cortex of the PHG was present in 1/6 cases.

| Differentiation and maturation of NEC in fixed tissue sections
Neuronal and glial markers were used to further characterize regional NEC in selected cases (Supporting Information Table S1). endothelium and expression in NEC, including bipolar cells, was not observed (Figure 4t,u). PDGFRb highlighted pericytes and scattered small NG2-like glial cells in the PVWM, SVZ, and SGZ as previously reported (Garbelli et al., 2015;Shepherd et al., 2013); there was occasional coexpression of nestin in these cell types (Figure 4v). and oligodendroglia (olig2 1 ;11.4%) in similar regions (Figure 5 b-d,f).

| Clinico-pathological correlations
Semi-quantitation of NEC in the PES hippocampus showed an inverse correlation with age of onset (p 5 .021) and a linear correlation with duration of epilepsy (p 5 .039); no significant differences were noted between pediatric and adult HS Type 1 cases. Quantitative analysis of hippocampal regions showed a positive correlation between age at surgery and mean Nestin 1 /lm 2 (p 5 .03) and a trend for a negative correlation with the proliferative fraction (MCM2 1 ) of NEC (p 5 .05). In culture, significantly more cells cultured from brain tissue of younger than older patients (r s 5 20.468, p 5 .040; Figure 6a). Furthermore, the average size of NEC negatively correlated with age (r s 5 20.889, p 5 .000; Figure 6d), with larger NEC grown from younger than older patients. There was no correlation between NEC and seizure types. (f). NeuN: Expression was mainly restricted to mature, large neurons but occasional expression in small NEC was observed (arrowheads). GFAP: (g) Expression was more extensive in all compartments than nestin and in this field of the SPL, NEC are mainly GFAP negative. (h) In the SGZ multipolar NEC co-expressing GFAP were seen (arrowhead). GFAP-delta: (i) There was more extensive coexpression with nestin compared with GFAP, particularly in CA1 astrocytes (arrowhead) but (j) NEC-negative cells were also observed. (k) Nestin-positive bipolar cells and threads in the PVWM were GFAP (shown) and delta negative. Aq4: (l-n). Aq4 showed extensive labeling of astrocytic processes through all regions (shown here for the SPL, SGZ, and PVWM) but in all regions NEC lacking Aq4 expression were noted (arrowheads). GS: (o) In the SPL (shown here) and cortex numerous astroglial cell bodies were positive for GS forming an extensive plexus. A striking proportion of NEC were GS negative and also observed in the (p) SGZ (arrowhead) as well as (q)

| NEC in developmental and mature niches
Nestin is a marker of progenitor cells of both neuronal and glial lineage in the developing brain and has been widely used to distinguish undifferentiated from differentiated cell types (Kawaguchi et al., 2001;von Bohlen und Halbach, 2011). Previous studies have described bipolar cells with parallel processes extending from the SVZ inward into the developing hippocampus, declining in number between 14 and 22 weeks post-conception (Yang et al., 2014 (Brus, Keller, & Levy, 2013) and long nestin 1 processes (Kamphuis, 2012). In the adult human brain, the existence of a RMS has been long debated and is considered likely to have diminished proliferative capacity (Brus et al., 2013;Curtis et al., 2007;Nogueira et al., 2014;Sanai et al., 2004;van Strien, van den Berge, & Hol, 2011). Vestigial structures around an extension of the lateral ventricle with migrating cells in chains, bipolar cells and thread-like fibers have been reported (Curtis et al., 2007). In rodents, the hippocampal fissure, equivalent to the sulcus in humans, has also been identified as a niche and further potential source of progenitor cells in adult animals . A further migratory stream has been described in mice extending from the hippocampal fimbra to dentate, composed of radial-glial like cells, which persists into adulthood (Belmadani et al., 2015). Recent studies of human hippocampal development also identify the fimbria as a further source of progenitor cells for the dentate gyrus (Cipriani et al., 2015). These observations highlight the hippocampal formation as unique in both development and maturation with the close apposition of several neurogenic regions. Furthermore, these regions parallel the distribution of persistent NEC populations we observed in the adult human epileptic hippocampus.

| Persistent radial glial in adult human brain
The persistence of NEC in the normal adult rat brain has been previously described (Hendrickson, Rao, Demerdash, & Kalil, 2011). Small multipolar NEC distributed throughout the forebrain, distinct from other astroglial, microglia, and oligodendroglial cell types (Hendrickson et al., 2011) compare with the small multipolar NEC we currently observe in the TLobe cortex, WM, and mesial structures. In rodent as well as human PM samples from the basal ganglia, small numbers of larger, NeuN co-expressing NEC have been identified (Hendrickson et al., 2011); we observed similar cells but mainly localized in the amydgala. In adult PM tissues, NEC in the hippocampal SVZ, SPL, and fimbria, extending to hypothalamic structures, have previously been described, interpreted as persistent neurogenic streams (Nogueira et al., 2014). In TLE tissue, radial glial-like cells, bipolar and astroglial like NEC have been reported in patients less than 14 years (Kruglyakova et al., 2005) and NEC in the SGZ SGZ in TLE patients <2 years (Blumcke et al., 2001).
In this study we demonstrated persistence of NEC in a stereotypical pattern. Striking observations included chains or rows of NEC in hippocampal, PHG WM, elongated nestin 1 fibers and bipolar radialglial like NEC, often in proximity to the recess of the lateral ventricle; these features are suggestive of a residual/vestigial TLobe RMS. We also identified continuous streams of NEC extending from the fimbrial or supbial surface of the HB or pes to the dentate gyrus as well as surrounding the para-laminar nucleus of the amygdala. Supportive data from our and previous in vitro studies (Arsenijevic et al., 2001;Roy et al., 2000) also demonstrate that NEC may be cultured from a number of brain regions including amygdala, hippocampus, temporal cortices, and the subcortical WM of mesial TLE/HS patients, even at the age of 52 years. Cultured cells showed a range of morphologies comparable to histology, including bipolar cells with long processes of >1 mm in culture, particularly from the TLobe and hippocampus. The elongated nestin 1 fibers and bipolar cells are reminiscent of developmental radial glial fibers (Tabata, 2015). We consider these elongated NEC as unlikely to represent endothelial cells or pericytes as they occasionally showed terminal branches, varicosities and were negative for relevant FIG URE 6 Immunophenotype and morphology of NEC in culture. (a) Scatterplot showing a slight decrease in the density of cells cultured as the age of patients at surgery increased (p 5 .040). Stacked bar chart (b) and an image (c) showing the varying proportions of uni-, bi-, tri-, and multi-polar NEC cultured from different brain regions (p 5 .000). NEC in C were derived from the hippocampus of 29-years old patient with mesial TLE/HS (EAC3). *The longest aspect or length of this bipolar NEC was over 1500 mm. (d) Scatterplot showing the area of NEC cultured from all regions of all cases. Larger NEC were more likely to be cultured from resected brain samples of patients who were younger at the time of surgery (p 5 .000). (e) Pie chart showing that approximately half of the cells in culture were immune-positive for nestin. The majority of NEC (green) were fate-determined as neuronal or glial cells, and co-expressed with bIII tubulin (f, red) or GFAP (g-h, red) respectively. (f) An image showing nestin and bIII tubulin immune-positive cells with different degrees of primary branching from the hippocampus of patient EAC3. *These BIII tubulin cells have a larger soma than the neighboring bi/tri-polar NEC, and expressed no or very little nestin protein. Images acquired from GFAP-expressing NEC cultured from the hippocampus (g) and WM (h) of a 52-years old patient with mesial TLE/HS (EAC6; GFAP, red; nestin, green). *NEC with no processes were more often observed in the amygdala and hippocampus, and they scarcely expressed GFAP. Both small and large multipolar cells expressing nestin and GFAP were commonly observed in the WM. Scale bar 5 200 mm vascular markers. Furthermore a regional predilection of these bipolar NEC to the temporal WM of the PHG and inferior TLobe WM was a consistent finding.

| NEC as injury-responsive progenitor cells
We also observed that bipolar NEC in the TLobe were licensed for replication. Proliferation of NEC types is well recognized at injury sites as part of the normal process of brain repair (Burda, Bernstein, & Sofroniew, 2016). An in vitro slice culture study in TLE patients following surgery showed an upregulation of NEC in relation to vessels (Verwer et al., 2015). In our previous study of the time course of NEC proliferation in relation to cortical injuries following depth electrode recordings, cells typically exhibited a radial-glial like bipolar morphology and a close relationship to new-formed vessels in acute lesions (Goc et al., 2014). Quantitative studies comparing nestin expression in surgically resected TLE to PM control tissue in culture showed significantly higher levels in epilepsy cases (Verwer et al., 2015). We also demonstrated in our series significantly more NEC in epilepsy than controls. Although we cannot exclude other factors, including differences in fixation times and potentially anti-epileptic drugs and other medications may influence staining and proliferation, our evidence supports that NEC are seizure-responsive populations.
There is compelling evidence that seizures alter rates of hippocampal neurogenesis [reviewed in (Jessberger & Parent, 2015;Zhong et al., 2016)]. MCM2 1 NEC were present in all zones in epilepsy cases and with significantly higher proliferative fractions of 6.6% compared with 3.3% in controls. This compares to our previous observations of proliferative fractions of 50% at acute injury sites to under 10% in chronic scars (Goc et al., 2014) suggesting a higher baseline activation of NEC in epilepsy as in previous studies (Verwer et al., 2015). The regions we noted with higher Nestin 1 /MCM2 1 cell densities were the hippocampal sulcus and PVWM. Crespel et al. (2005), also reported frequent PCNA 1 cells in the adult hippocampal sulcus in TLE, proposing this region as another neurogenic region. In another study MCM2 1 was not noted in hippocampal NEC in ten HS cases but NG21/Olig21 cells were the most proliferative type (Geha et al., 2010). We were unable to achieve reliable NG2 immuolabeling, but confirmed that NEC contributed to the hippocampal proliferative pool, comparable to Iba1 1 , GFAP 1 , and Olig2 1 glial cell types. Of note, bipolar NEC in the WM were MCM2 1 in over half the epilepsy cases compared with only 20% of controls. Neural stem cells in the SVZ are derived from embryonic radial glial cells (Kriegstein & Alvarez-Buylla, 2009); nestin 1 radial glial with ongoing stem-cell capacity have been previously reported in adult rat brain (Gubert, Zaverucha-do-Valle, Pimentel-Coelho, Mendez-Otero, & Santiago, 2009) and progenitor cells successfully isolated from the human adult WM (Lojewski et al., 2014). Our observations of a morphologically similar proliferating radial glial cell pool in the adult TLobe WM in epilepsy is therefore of potential relevance.

| Differentiation of NEC and relevance to comorbidity and pathology in TLE
Studies in chronic TLE suggest a diminished capacity for neurogenesis (Marucci et al., 2013;Paradisi et al., 2010) which may result from a preferential switch of precursor cells to glial over neuronal differentiation (Hattiangady & Shetty, 2010). We explored the differentiation of NEC in both tissue sections and cell culture. Expression of neuronal markers (bIII tubulin and mushahsi) was noted but there was more evidence for glial maturation. Bipolar NEC exceptionally did not coexpress GFAP in sections, as previously noted (Kruglyakova et al., 2005), although GFAP expression was observed in cell culture. In contrast to a previous study (Nogueira et al., 2014)  identifies subsets of specialized neurogenic astrocytes (Kamphuis et al., 2012;van den Berge et al., 2010). We also noted a paucity of labeling with GS, Aq4, and EAAT1, markers of mature, functioning astrocytes in many NEC. Furthermore, co-expression of pS6 in hippocampal NEC supports mTOR pathway activation in these cells (Liu et al., 2014a), also known to modulate glial cell function (Wang, Sha, Sun, Shen, & Xu, 2016). Hippocampal astrogliopathies are now considered central to pro-epileptogenic mechanism in experimental models (Bedner et al., 2015;Kielbinski, Gzielo-Jurek, & Soltys, 2016;Pekny et al., 2016). Our identification of NEC in the hippocampus, including CA1 in HS cases, as functionally immature, activated cells could be of pathophysiological relevance. Furthermore, glial proliferation following injury is also known to impair memory function (Sajja, Hlavac, & VandeVord, 2016); our identification of higher proliferative fractions of NEC in epilepsy patients with memory deficits, although only a small series, also warrants further investigation.
We noted the pattern of HS in the PES differed from the body, with preferential neuronal loss and gliosis in CA4 more than CA1 [Type 3 vs. Type 1 sclerosis in the ILAE classification system (Blumcke et al., 2013)]; this may relate to different distributions of injury responsive NEC. Furthermore, proliferative NEC with capacity and potential for glial/neuronal differentiation may contribute to the array of microscopic malformative and developmental tumors reported in TLobe in epilepsy (Supporting Information Figure S2).
In summary, NEC in the TLobe with ongoing proliferative capacity, likely represent the remnants of developmental migratory streams.
NEC populations represent reactive cell populations in the adult with capacity for glial and neuronal differentiation and may be relevant to