The Structure of the Spinal Cord Ependymal Region in Adult Humans Is a Distinctive Trait among Mammals

In species that regenerate the injured spinal cord, the ependymal region is a source of new cells and a prominent coordinator of regeneration. In mammals, cells at the ependymal region proliferate in normal conditions and react after injury, but in humans, the central canal is lost in the majority of individuals from early childhood. It is replaced by a structure that does not proliferate after damage and is formed by large accumulations of ependymal cells, strong astrogliosis and perivascular pseudo-rosettes. We inform here of two additional mammals that lose the central canal during their lifetime: the Naked Mole-Rat (NMR, Heterocephalus glaber) and the mutant hyh (hydrocephalus with hop gait) mice. The morphological study of their spinal cords shows that the tissue substituting the central canal is not similar to that found in humans. In both NMR and hyh mice, the central canal is replaced by tissue reminiscent of normal lamina X and may include small groups of ependymal cells in the midline, partially resembling specific domains of the former canal. However, no features of the adult human ependymal remnant are found, suggesting that this structure is a specific human trait. In order to shed some more light on the mechanism of human central canal closure, we provide new data suggesting that canal patency is lost by delamination of the ependymal epithelium, in a process that includes apical polarity loss and the expression of signaling mediators involved in epithelial to mesenchymal transitions.

However, a different organization of the ependymal region is found in adult humans ( Figure 1D-L). During childhood, it is similar to other mammals in structure and expression of some ependymal markers [5,[35][36][37][38], but during late infancy to adulthood, the central canal is lost in the majority of individuals being substituted by a structure largely different from the rest of the species [20,[39][40][41]. After the central canal is lost, it is substituted by large accumulations of ependymal cells not enclosing a lumen ( Figure 1D-I), intense astrogliosis ( Figure 1J,K) and perivascular pseudorosettes ( Figure 1L) [5,20] and does not proliferate in response to spinal cord injury [42]. The expression of other markers related to neural stem cells can be found in [5][6][7]19,20,35,36].
The trigger that makes humans lose the central canal after childhood, the mechanism by which this process develops and the consequences in the physiology of the spinal cord and its response to injury are still unknown. The study of this phenomenon is rather complex due to the paucity of human spinal cord samples available and the lack of animal models. In the search for a natural model, we found two candidates: the Naked Mole-Rat (NMR, Heterocephalus glaber) and the hyh (hydrocephalus with hop gait) mutant mouse [43]. The ependymal region of the Naked Mole-Rats has not been described to date, but the observation of images from their spinal cord in other publications suggested us that they might present no patent canal [44,45]. The hyh mice carry a hypomorphic missense mutation in the NAPA gene encoding soluble N-ethylmaleimide-sensitive factor (NSF) attachment protein alpha (SNAP-alpha) [43,46,47]. These mice have been shown to lose central canal during late embryonic stages [48], although this observation has not been further extended ever since.
In the current work, we confirm (and in the case of NMR we report for the first time) that adult individuals of these animals lack a central canal. However, we show that in both cases the structure substituting the former canal is largely different from that found in adult humans, not including large accumulations of cells, neither gliosis nor perivascular pseudorosettes. In addition, we provide new data on the possible mechanism of central canal closure in humans by studying human spinal cord postmortem samples with a partially patent canal. In adult humans, the new structure in the ependymal region (EPR) substituting former canal also includes strong astrogliosis (GFAP immunoreactivity, red) and the presence of (K,L) perivascular pseudorosettes, i.e., cells expressing vimentin (green) radially oriented around a central vessel (v), separated from it by a hypocellular GFAP + region. Pseudocolors in L highlight the GFAP hypocellular region (red) that surrounds the vessel (v), cells radially oriented (green) and ependymal cell groups around the perivascular pseudorosette (EP). dWM, dorsal white matter; Vim, vimentin.

Naked Mole-Rats
Naked Mole-Rats (NMR, Heterocephalus glaber) were bred at Queen Mary University of London and housed in their natal colonies as previously described (e.g., [49]). In this study, we used one non-breeding female (2 years of age) and 5 non-breeding males (ages: 2, 9 and 10 years). Non-reproductive status was verified behaviorally, morphologically and/or by examination of the reproductive tract as previously described [49,50]. The research was conducted in accordance with the U.K. Home Office Animals (Scientific Procedures) Act 1986. Because tissue sample collection was post-euthanasia, additional local ethical approval for NMR work was not required for this study. For the immunohistochemical study, animals were terminally anesthetized with sodium pentobarbitone (Pentoject; Animalcare, Ltd., New York, United Kingdom; 50 mg/kg) and perfused transcardially with phosphate-buffered saline (PBS), followed by 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). The spinal cords were postfixed overnight in the same fixative, and maintained in PBS. Spinal cords were then extracted and cervical, thoracic or lumbar fragments of the spinal cord (5 mm in length) were embedded in low melting agarose with 10% sucrose in 0.1 M PB. Serial transverse vibrating microtome sections (40 µm) were obtained and stored in Olmos solution at −18 • C until use.

Hyh Mice
Mutant hyh mice (hydrocephalus with hop gait, B6C3Fe-a/a-hyh/J strain) [43] carry a point mutation in the Napa gene that encodes α-Snap [46,47], a protein involved in membrane fusion events. Mice were obtained from The Jackson Laboratory (Bar Harbor, ME, USA) and bred at the Animal Experimentation Service of the University of Malaga. The housing, handling, care and processing of the animals were conducted in accordance with the European and Spanish laws (DC 86/609/CEE and RD 1201/2005, 2010/63/ EU) and was approved by the Institutional Animal Care and Use Committee of the University of Malaga, Spain (CEUMA) and the Regional Government Council (Junta de Andalucía, Spain) (protocol # 4-2015-A). Mutant hyh mice were identified by clinical inspection and genotyping [51]. The animals used in this study (5 males) were anesthetized with intraperitoneally administered Dolethal (sodium pentobarbital; Vétoquinol, Lure, France; 0.2 mg/g bodyweight) and intracardially perfused with 4% paraformaldehyde at 20 days after birth (postnatal age 20, P20). Spinal cords were extracted and cervical, thoracic or lumbar fragments of the spinal cord (5 mm in length) were embedded in low melting agarose with 10% sucrose in 0.1 M PB. Serial transverse vibrating microtome sections (40 µm) were obtained and stored in Olmos solution at −18 • C until use.

Chimpanzee
Postmortem chimpanzee spinal cord samples were kindly provided by Dr. Eva Martinez Nevado (Zoo-Aquarium, Madrid, Spain). A 25 year old female chimpanzee (Pan troglodites) from the Zoo inhouse colony, who was being treated for heart disease (cardiomyopathy, left apex thrombus and tricuspid valve insufficiency), worsened after anesthesia during health checking and finally died after 48 h. During autopsy, a 10 mm spinal cord block was obtained from low cervical/high thoracic levels and immersed in formalin for 3 days. Spinal block was then rinsed in PBS and serial transverse vibrating microtome sections (40 µm) were obtained and stored at −18 • C in Olmos solution until use.

Macaque
Macaque tissue was gently provided by Drs. Javier Cudeiro, Casto Rivadulla (NEU-ROcom, School of Health Sciences University of A Coruña, A Coruña, Spain) and Juan Aguilar (Hospital Nacional de Paraplejicos). Spinal cord was extracted postmortem from a 7 year old male macaque monkey (Macaca mulatta) used in previous studies [52]. All procedures followed the rules of the Physiological Spanish Society, the International Coun-cil for Laboratory Animal Science, and the European Union (No. 2010/63/EU) and were approved by the ethics committee for animal research of the University Hospital of A Coruña. The animal was intracardially perfused with 4% paraformaldehyde and postfixed in the same fixative for 3 days. Spinal cord was extracted and divided in 5 mm blocks. Serial transverse vibrating microtome sections (40 µm) were obtained and stored at −18 • C in Olmos solution until use.

Human Tissue
Human tissue was obtained from two public tissue biobanks: the HUFA BioBank (Biobanco del Hospital Universitario Fundación Alcorcón, Alcorcón, Spain) and the Neurological Tissue Bank (Banco de Tejidos Neurológicos, IDIBELL-Hospital Universitario de Bellvitge, Hospitalet de Llobregat, Barcelona, Spain). Tissue was provided in 5-8 mm thick formalin fixed blocks or vibratome free floating sections. Tissue donation always included written informed consent from donors while alive or from their families after death. Data from donors and handling of samples obtained from all the Biobanks included in this study were processed after approval by the Clinical Research Ethics Committee (CEIC) in Toledo (Spain), in accordance with Spanish law and International Guidelines (LOPD15/1999; RD 1720/2007; Declaration of Helsinki, 2008). Samples were obtained from ten deceased individuals without clinical or histopathological involvement of the spinal cord (Table 1). Upon receipt, tissue blocks were embedded in low melting agarose with 10% sucrose in 0.1 M PBS and cut into serial transverse sections (40 µm) with a vibrating microtome (Leica VT 1000 M). Sections were then stored at −18 • C in Olmos solution until use. Free floating sections were mounted on gelatin covered slides and dried overnight. After two rinses with distilled water, slices were stained with 0.01% toluidine blue (Merck, Madrid, Spain) in 0.2 M Walpole buffer for 20 min, dehydrated in graded ethanol solutions, cleared with xylene and covered with DPX mounting medium (VWR, Barcelona, Spain).

Immunohistochemistry
Immunohistochemistry was performed on sections rinsed in rinse solution (RS) containing 0.1 M phosphate-buffered saline (pH 7.4), 0.3% Triton X-100 and 0.3% bovine serum albumin (BSA). An additional methanol pretreatment (50% methanol 1 min, followed by 100% methanol 9 min) was used for better detection of transcription factors and cytoskeletal proteins, when needed. Only for human samples, sections were subjected to antigen unmasking, consisting of a 30 min pretreatment with a 0.05% solution of citraconic anhydride (#27430, Sigma-Aldrich, St. Louis, MO, USA) at 96 • C, followed by temperature re-accommodation at room temperature for at least 30 additional minutes [53,54]. The sections were then incubated for 2 nights with the primary antibodies (Table 2) diluted in RS with 0.3% Triton X-100 and 5% normal goat serum, or 5% BSA in cases where the antiserum was generated in goat. After extensive rinsing, the primary antibodies were detected using fluorophoreconjugated antibodies (Table 3). Some antibodies (phospho Smad3, SNAI1, activated Notch-NICD) required an intermediate incubation with biotinylated secondary antibodies followed by incubation with Alexa conjugated streptavidin (Table 3). Nuclear counterstaining with bisbenzimide was performed (BBZ, Hoechst 33258 pentahydrate, Invitrogen, Waltham, MA, USA; 1:5000). Sections were mounted with Immumount (Thermo Fisher, Waltham, MA, USA) and analyzed with a LEICA SP5 confocal microscope at the Microscopy Facility in the National Hospital for Paraplegics (Toledo, SESCAM). Images were transferred to ImageJ (NIH, Bethesda, MD, USA) for cropping; they were adjusted to optimize contrast and brightness. Noise was reduced using a median filter with Fiji (http://pacific.mpi-cbg.de), a scientific image processing application based on ImageJ (http://rsb.info.nih.gov/ij). Control experiments were performed to rule out the interference of non-specific staining in parallel with the complete assays. The incubation of tissue without primary antibodies, with all other steps being identical to those described above, eliminated all staining, with the exception of some blood cells and lipofuscin autofluorescence present mostly in human tissue.

Transmission Electron Microscopy
A series of tissue slices were immersed in a fixative containing 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M PB for 5 days. Sections were then rinsed in 0.1 M PB, postfixed in 1% osmium tetroxide in 0.1 M PB with 5% glucose, dehydrated in acetone, and embedded in Araldite resin (Araldite 502 Kit, #13900, Electron Microscopy Sciences, Hatfield, PA, USA). Semithin sections (2 µm) were obtained with a Leica EM UC6 Ultramicrotome and stained with 1% toluidine blue in 1% borax. For transmission electron microscopy, semithin sections were re-embedded, cut in ultrathin (65 nm) sections and mounted on formvar-coated grids as described before [9]. Grids were stained with uranyl acetate (#22400, Electron Microscopy Sciences) and then with lead citrate (#17810 Electron Microscopy Sciences) according to the protocol of Venable and Coggeshall [55]. Analysis of the samples was performed with a Jeol 1200EXII electron microscope at the Electron Microscopy Facility in the Cajal Institute (CSIC, Madrid, Spain). Unlike humans, no astrogliosis was observed in the central gray matter nor in the whole lamina X. In the central gray matter of NMR, GFAP expression was equivalent to that found in the normal parenchyma ( Figure 3M-P). Among the small ependymal-like clusters of the midline, GFAP was generally absent or scarcely expressed ( Figure 3M), but in older individuals (9-10 years old) strong GFAP immunoreactivity in some midline cells with long GFAP + processes can be found ( Figure 3N-P). Microglial cells surrounding these GFAP + cells show non-activated morphologies (Iba1 + cells, Figure 3N,P).
Finally, no perivascular pseudo-rosettes were found in the NMR ependymal remnant, unlike in humans. In the NMR, the distribution and density of vasculature showed no apparent abnormalities when stained with laminin ( Figure 3F-H).

Hyh Mice
All the individuals studied showed absence of the central canal ( Figure 4). Instead of a well-defined ependymal region surrounding it, cells were evenly distributed, not forming clusters or large accumulations such as those observed in adult humans. The observation of cell nuclei in the central gray midline using toluidine blue stained semithin sections showed morphologies resembling microglia, other glia and neurons, as well as small capillaries enclosed by endothelial cells (Figure 4G-J). Dorsoventrally oriented cell alignments such as those observed in NMR were not clearly distinguished. On the other hand, small groups of ependymal cells enclosing small cavities were occasionally found ( Figure 4L,O), forming pseudocanal-like structures. Cells at these structures express Sox9 and a small part, also nestin ( Figure 4N,O).   As in NMR, astrocytosis was not found in the central gray matter of hyh mice ( Figure 4K). GFAP expression followed the general distribution observed in many species: high expression in the outer rim of the spinal cord, including glia limitans, and normal astrocytic expression in the parenchyma, with no reactive morphology, including those in the central gray ( Figure 4K,L).
Distribution of vessels and capillaries showed a normal appearance and no perivascular pseudorosettes were found unlike in humans ( Figure 4G-J).

The Closure of the Central Canal in the Adult Human Ependymal Region Is Preceded by a Delamination That Shows Features of Epithelial to Mesenchymal Transition (EMT)
After substantiating that the structure substituting central canal in humans is vastly different from that of other animals that spontaneously lose it, we aimed to collect evidence on the possible mechanisms underlying this specific process that results in unique human features. For this, we studied rare human spinal cord samples that are at different stages of the central canal closure instead of the normal samples lacking the central canal. We studied slices with canal patency (very rare event; Figure 5A,B) and slices with domains depicting abnormal cell accumulation or partial closure of the canal ( Figure 5C-F).
In samples with a partially closed canal, we observed a general pattern that involved accumulation of cells mainly in the ventral domains, and the presence of ectopic ependymocytes found at long distances from the lumen ( Figure 5G-K). In some domains, ependymal delamination is observed after losing apico-basal polarity and the loss of polarity markers such as Tight Junction Protein ZO-1 (TJP1) (Figure 5L-Q).
Ependymal delamination and loss of apico-basal polarity apparently involved the acquisition of a mesenchymal-like phenotype and included the nuclear expression of mediators related to epithelial to mesenchymal transitions such as the phosphorylated form of Mothers against decapentaplegic homolog 3 (SMAD3; Figure 6A-D), Zinc finger protein SNAI1 (SNAI1; Figure 6E-H), and the Notch Intracellular Domain (NICD; Figure 6I-L).

Discussion
The absence of the central canal in humans is a phenomenon already described in old manuals [56][57][58] but almost completely neglected afterwards, except for a few reports that deepen into that singularity [5,19,20,[39][40][41]59,60]. This disregard by the scientific community may explain the current absence of animal models to study the process of central canal loss and the ignorance of how this unique human ependymal region impacts physiology and responses to spinal cord damage.
An approach to tackle this deficit could be the search for natural models in which animals spontaneously lose the central canal. After a thorough literature search, we found the Naked Mole-Rat (NMR) as candidate, a long-living rodent increasingly studied for its important particularities in the fields of aging, pain, cancer or social behavior [61][62][63][64]. We describe here for the first time that NMR indeed lack the central canal as adults, but this is replaced by a new organization of lamina X largely different from that in humans. This may suggest that the cause and/or the process of central canal loss could be different between NMR and humans. In the literature, we also found reports from other mammals such as porpoises [65,66], whales [67] and dolphins [68] describing an absence of central canal. Unfortunately, we could not obtain tissue samples for a further study, but the appearance of the lamina X in these species, according to the published images, looks similar to what we found in NMR, also separating them from the human case.
A few other mammals in which the central canal is absent are transgenic or mutant strains of laboratory mice. One of them, the hyh mice studied here, present a mutant variant of alpha-Snap protein, and show spinal cord central canal loss during late embryonic stages [48]. These mice suffer a continuing denudation of the ependymal lining that finally affects the aqueduct, leading to massive hydrocephalus and death during the first month of age [69,70]. This is indeed an important difference with humans (and NMR), since the loss of central canal in humans is a general feature in almost every individual, and the incidence of postnatal hydrocephaly in the general population is extremely low. When studying the central gray matter of adult hyh mice, we found that the canal is not replaced by a structure similar to humans. The histological features are much closer to NMR, although cell accumulations in the midline are not as clearly aligned as in NMR and some pseudocanals can be found, in contrast to NMR.
Other mutant mice in which the central canal disassembles or is completely lost are afadin mutant mice [71], Sox9 overexpressing mice [72] or mice with deletion of the Rho family guanosine triphosphatase (GTPase) 3 (Rnd3) [73]. Interestingly, many of these transgenic mice have in common the malfunction of cell adhesion related molecules, but none of them give rise to a human-like structure.
A different approach that would help to model human central canal closure is the experimental intervention to induce canal disassembling. There is only one such attempt published to date [40], using reovirus type 1 infection in rats for inducing ependymal damage, proliferation and canal closure. This strategy followed the rationale of the known tropism of viruses for ependymal cells, and the strong link between virus infection, inflammation and ependymal disassembling [74][75][76]. However, in all those reports, including the model by Milhorat et al. [77], severe complications were observed, mostly hydrocephaly, that are not found after human central canal loss. Moreover, the histological findings in [77] included the presence of pseudocanals, mild cellular accumulation and moderate gliosis, in a much lesser extent than those observed in humans.
Milhorat et al. hypothesize, based on their model, that "stenosis of the central canal in man is a pathological lesion involving ependymal injury and scarring", but this may require some nuance. First, the sole lesion or mechanical damage to the ependymal cells is not enough to explain central canal de-structuring, since this does not happen after spinal cord injury in experimental animal models. In rodents, for example, when the ependymal region is damaged, it proliferates and contributes to the production of new ependymal cells and, in a lesser extent, to glial scar [12,14,78,79], but this does not involve a disassembling of the remaining canal neither rostral nor caudal to the injury site. In addition, the previous descriptions of human ependymal region [5,20,[39][40][41] and the evidence we present here unveil a complex process that suggest an active delamination and transformation of the ependymal lining (induced by still unknown triggers), that ends up in a structure that includes astroglial reactivity, but is not limited to this, nor forms a proper scar.
What we observe here, after studying spinal cord levels with intermediate features between the fully patent canal and the total absence of canal, suggests that the process may probably begin with a ventral delamination of ependymal cells that may further extend to other aspects of the ependymal layer accompanied by the expression of factors involved in epithelial to mesenchymal transition, such as SNAI1, TGFb mediators (phospho-Smad3) and Notch signaling (NICD). Due to the low incidence of spinal ependymal tumors in humans [80], the epithelial to mesenchymal transition-like process that may underlie human central canal transformation, might probably reflect a transition to a fibrosis-like state (with gliosis), rather than to cancer formation [81][82][83] or to stemness [17]. Interestingly, EMT has been recently proposed to underlie the process of regeneration after spinal cord injury in zebrafish [84]. On the other hand, since the proliferation in the human ependymal region is rather low [5,20,35], the large accumulation of cells observed in adult humans at this region may also reflect a modulation of cell death, for which some of these factors, such as SNAI1 have been shown to be inhibitors [85].
The observation of ventral accumulation of ectopic ependymal cells can be observed in previous reports [35], and the delamination of ventral aspects (floor plate) is a phenomenon described in the normal formation of the ependymal lining during mammal development [86]. The involvement of factors such as Arx or Foxa2, specifically present in the ventral regions of the ependymal layer in the cases where it maintains patency, might be a topic to be explored in the future [6]. It must be also considered that other multiple steps and factors have been described in the normal formation of the central canal in rodents, in mechanisms that enables central canal formation without loss of ventricular layer integrity, including dorsal attrition, dorsal delamination of progenitors and a crucial role for Protein crumbs homolog 2 (Crb2) [86][87][88]. Whether some of these processes and factors are affected in animals spontaneously losing the central canal, or if they may be related to different programs for central canal disappearance that may explain differences between humans and the rest of the mammals, should be approached in future studies.
In conclusion, data from past reports and from this one may warrant caution when considering direct translation of the properties of the spinal cord ependymal region between species. It was already known that important differences in the identity, function and responses of ependymal cells in the spinal cord exist between regenerating (zebrafish) and non-regenerating animals [18] and even among species that regenerate, such as lizards and salamanders [89]. We suggest here that the adult human spinal cord is unique in this specific trait from the infancy, and shows vast morphological differences even with the few mammals that spontaneously lose central canal patency during their lifetime (NMR, cetaceans, mutant mice). These morphological and structural dissimilarities are also accompanied by genomic [6,19,20,60] and functional [7,35,42,90,91] differences. In order to successfully tackle the many unsolved issues in this field, (what is the trigger of the central canal loss in humans, the consequences of having this structure in the physiology of the cord and its response to damage . . . ), it would be desirable to achieve experimental models that better mimic the human situation in the future.  Informed Consent Statement: All the human samples used were post-mortem samples obtained from two public tissue biobanks. Tissue donation always included written informed consent from donors while alive or from their families after death. Data from donors, completely anonymous and coded, and handling of samples obtained from all the Biobanks included in this study were processed after approval by the Clinical Research Ethics Committee (CEIC) in Toledo (Spain), in accordance with Spanish law and International Guidelines (LOPD15/1999; RD 1720/2007; Declaration of Helsinki, 2008).

Data Availability Statement:
No new data were created or analyzed in this study apart from those presented in the manuscript. Data sharing is not applicable to this article.

Acknowledgments:
We thank Eva Martinez Nevado (Zoo-Aquarium, Madrid, Spain) for kindly providing us with postmortem chimpanzee spinal cord samples and Javier Cudeiro, Casto Rivadulla (NEUROcom, School of Health Sciences University of A Coruna, A Coruna, Spain) and Juan Aguilar (Hospital Nacional de Paraplejicos) for their kind gift of macaque spinal cord samples. Sebastian Pons (Institut de Biologia Molecular de Barcelona, CSIC, Spain) generously provided us with TJP1 antibody.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.