Oligodendrogliogenesis and axon remyelination after traumatic spinal cord injuries in animal studies: A systematic review

Extensive oligodendrocyte death after acute traumatic spinal cord injuries (TSCI) leads to axon demyelination and subsequently may leave axons vulnerable to degeneration. Despite the present evidence showing spontaneous remyelination after TSCI the cellular origin of new myelin and the time course of the axon ensheatment/remyelination remained controversial issue. In this systematic review the trend of oligodendrocyte death after injury as well as the extent and the cellular origin of oligodendrogliogenesis were comprehensively evaluated. The study design was based on Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA)-guided systematic review. PubMed and EMBASE were searched with no temporal or linguistic restrictions. Also, hand-search was performed in the bibliographies of relevant articles. Non-interventional animal studies discussing different types of myelinating cells including oligodendrocytes, Schwann cells and oligodendrocyte progenitor cells (OPCs) were evaluated. The extent of oligodendrocyte death, oligodendrocyte differentiation and remyelination were the pathophysiological outcome measures. We found 12,359 studies, 34 of which met the inclusion criteria. The cumulative evidence shows extensive oligodendrocytes cell death during the first week post-injury (pi). OPCs and peripheral invading Schwann cells are the dominant cells contributing in myelin formation. The maximum OPCs proliferation was observed at around 2 weeks pi and oligodendrogliogenesis continues at later stages until the number of oligodendrocytes return to normal tissue by one month pi. Taken together, the evidence in animals reveals the potential role for endogenous myelinating cells in the axon ensheathment/remyelination after TSCI and this can be the target of pharmacotherapy to induce oligodendrocyte differentiation and myelin formation post-injury.


INTRODUCTION
Traumatic spinal cord injury (TSCI) is a devastating event involving approximately between 3.6 and 195.4 cases per million annually in different countries (Jazayeri et al., 2015). The acute trauma initiate a cascade of progressive events which last by one year after injury and cause secondary injury (Rowland et al., 2008). Based on different degenerative events during the secondary injury various regenerative and repair strategies have been proposed and are under development to induce axon regeneration and remyelination which can be categorize in 3 different classes: cell-based strategies (Keirstead et al., 2005, Priest et al., 2015, drug therapies (Nagoshi et al., 2015) and tissue engineering approaches (Piantino et al., 2006, Fuhrmann et al., 2016. In some cases the outcomes of animal studies have provided the rationale for clinical trials (clinicaltrials.gov). Although many studies propose the advantageous of acute intervention in order to restrict the progress of secondary events (Fehlings et al., 2016), the endogenous regenerative responses at later times post-injury may also provide some extra therapeutic opportunities.
Oligodendrocytes are myelinating cells residing in central nervous system (CNS). During development oligodendrocytes are mostly derived from neuroepithelial zones, where neuroepithelial stem cells (i.e. stem cells of the nervous system) differentiate into oligodendrocyte precursor cells (OPCs). Subsequently, OPCs differentiate into premyelinating oligodendrocytes and finally into mature myelinating oligodendrocytes (van Tilborg et al., 2018). Myelination is a dynamic and plastic process with an excess of OPCs being generated and remained throughout adulthood which are recruited to drive remyelination in case of injury or neurodegenerative diseases (Gautier et al., 2015).
After SCI oligodendrocyte death leads to axon demyelination and leave axons vulnerable to degeneration (Hill et al., 2001, Houle and Jin, 2001, Hassannejad et al., 2018. Spontaneous remyelination after injury has been revealed in some studies (Bartholdi and Schwab, 1998, Horky et al., 2006, McDonough et al., 2013, Hesp et al., 2015, Assinck et al., 2017. Mature oligodendrocyte, OPCs as well as Schwann cells (i.e. peripheral resident myelinating cells) have been introduced as the cells contributing in remyelination after SCI. It has been shown that after injury OPCs and Schwann cells that are quiescent in normal conditions become activated, develop a larger cytoplasm and thinner cell processes. They proliferate and recruited to the injury site to restore the damage caused by dying oligodendrocytes through expression of neurotrophic factors to establish a growth promoting environment and enwrapping neuronal axons to remyelinate nude axons (for excellent review, see de Castro et al. (2013)).
The extent of oligodendrocyte death, endogenous remyelination and contribution of each of the aforementioned cells in myelin formation have been evaluated separately in different time points from immediate to chronic phases. However, the spatial and temporal pattern of myelin formation as well as prominent cell type contributing in axon ensheathment/remyelination in each time point or various regions of cord have not yet fully addressed. In this study the fate of mature oligodendrocytes and their proliferative potential after injury, cell sources to compensate the extensive loss of oligodendrocyte and also the extent of remyelination after injury have been systematically evaluated in order to reveal the appropriate time points to augment spontaneous remyelination. Based on the present evidence myelinating cells respond to TSCI in a way that after initial loss the number of oligodendrocytes approaches to the normal levels after about one month post-injury. The majority of oligodendrocytes are lost at the epicentre and adjacent regions, however, OPCs proliferation and subsequent differentiation to myelinating cells (both oligodendrocytes and Schwan cells) are the main route to compensate the extensive demyelination after injury. Invading peripheral Schwann cells also contribute in remyelination. Taken together, the evidence in animals suggests an important potential role for endogenous myelinating cells in the axon ensheathment/remyelination after TSCI.

Search and Selection of Studies
We searched Embased and MEDLINE via Ovid SP on 9 November 2013 and we updated this search on 24 October 2015 and 7 September 2018. Keywords were determined based on research team discussion, experts' opinion, literature reviews, controlled vocabulary (Medical Subject Headings = MeSH and ExcerptaMedica Tree = EMTREE) and review of the records retrieved in the initial search. The search strategy was then developed with assistance of a medical information specialist. The following keywords and MeSH terms were used: "Spinal cord injury", "trauma", "cell", and "pathophysiology" Search results were exported into EndNote X5. Also, a detailed review of the reference lists of the most relevant publications was performed.

Eligibility Criteria
All the articles investigating the pathophysiology of TSCI in animal after acute traumatic mechanical spinal cord injury including compression, contusion, hemisection and transection were included. Publications that did not specify cell type, and exact time points for the pathophysiological events were not included. In addition, in order to increase the homogeneity of included studies, we excluded studies reporting on humans (i.e. all clinical studies and case reports), non-mechanical injuries, including thermal and ischemic injuries and other interventions. Finally, non-original articles, including review articles and letters were excluded. No restrictions in regard to period of publication and language were applied.

Selection Process
Two authors independently assessed each article based on the title, abstract and finally the fulltext (ASM & MM, SA & ZH). In a case of disagreement, a third independent person (VRM) assessed the article and the final decision was made accordingly.

Outcome Measures
Different types of myelinating cells including oligodendrocytes, Schwann cells and OPCs, which are believed to be committed to oligodendrocyte lineage, were evaluated. Cell death, differentiation and proliferation rates for the aforementioned cell types were the pathophysiological outcome measures.

Data Extraction and Quality Assessment
Data were extracted to a data extraction form developed in our group by two independent members. On this form, general information about each article including the title, authors' name, publication date, information about method including study design (type/severity/level of injury), time of assessments, animal number and animal characteristics (strain, sex, age or weight), method(s) of evaluation and finally the research question and relevant results were recorded.
In order to control the quality of the included articles, two independent reviewers evaluated each article according to the criteria for TSCI (Hassannejad et al., 2016). Briefly, the quality of each article was assessed based on 15 items including: 1. species; 2. using appropriate tests; 3. severity of injury; 4. level of injury; 5. age/weight; 6. number of animals per group; 7. designation of strain; 8. definition of control; 9. description of statistical analysis; 10. bladder expression; 11. genetic background; 12. method of allocation to treatments; 13. description of the reasons to exclude animals from the experiment during the study (attrition); 14. blindness of assessors of the original investigators; 15. regulations and ethics
The collected data included the results of immunohistochemistry, light and electron microscopy and flow cytometry. The markers and methods used to identify the cell types (i.e., oligodendrocytes and OPCs) and to detect apoptosis are summarized in Table 3. Table 3 The extent of oligodendrocytes death at the lesion center and adjacent regions after TSCI The earliest evaluated time point was 15 minutes pi, when the number of oligodendrocytes (CC1 + ) significantly decreased at the epicenter as well as 2 mm caudal and rostral to the injury site (Grossman et al., 2001). Reduction in the number of oligodendrocytes persisted ±3 mm away from the lesion site, however, it was not statistically significant comparing to the uninjured tissue at this time point (Grossman et al., 2001). By one hour pi, no noticeable changes were observed 3 to 10 mm rostral and caudal to the lesion epicentre (Wang et al., 2009).
Four hours pi, a further 50% of oligodendrocytes were lost at the epicenter (Grossman et al., 2001) and according to Li et al. (1999) a small, statistically non-significant portion of this reduction was due to apoptosis. At 6 hours pi, TUNEL + cells were detected at the epicenter and 1-2 mm distal to the injury, most of them displaying oligodendrocyte characteristics (doublelabelled with CNPase) (Casha et al., 2001, Grossman et al., 2001. Eight and 24 hours pi, a persistent significant reduction in the number of oligodendrocytes was seen at distances up to 2 mm rostral and caudal to lesion epicenter (Bartholdi andSchwab, 1998, Grossman et al., 2001).
At 24 hours pi, the number of apoptotic oligodendrocytes (TUNEL/CNPase + ) at the white matter (WM) of epicenter approximately doubled compared to controls (Huang et al., 2014).
Also, in spared WM at the epicenter, the density of mature oligodendrocytes decreased significantly by 24 hours pi (Lytle and Wrathall, 2007). At this time point, 1.5 mm rostral and caudal to the injury site, as well as epicenter, CC1 + oligodendrocytes were significantly reduced (Lytle and Wrathall, 2007). Accordingly, 24 hours pi PLP (proteolipid protein) mRNA expression was mostly absent in the central necrotic area which indicated a significant loss of oligodendrocytes in this area (Frei et al., 2000). Li et al. (1999) reported, using TUNEL assay, a statistically non-significant presence of apoptotic cells rostral (T1-T7) and caudal (T10-L2) (where the epicenter was T8-9) 24 hour pi after moderate and severe injuries (Li et al., 1999).
At 2 days pi, the total number of mature oligodendrocytes (CC1 + ) decreased to less than half of the naïve animals at the epicenter and ventral-lateral funiculi up to 14 mm distal to the injury site (Rabchevsky et al., 2007).
At day 3, the number of apoptotic oligodendrocytes (TUNEL/CNPase + ) was higher than day 1 as well as control animals, at the epicenter (Huang et al., 2014) and within 4 mm distal to the epicenter (Yong et al., 1998). From the third day, TUNEL-MBP + (Myelin basic protein) cells were found sparsely scattered in the WM remote from the epicenter (Abe et al., 1999).
It is likely that from the third day onwards, the zone of apoptotic oligodendrocytes spread to sites remote from the epicenter. At 4 days pi, the number of apoptotic cells significantly decreased in caudal (T12, and L2 where the injury is applied at T8-9) and rostral segments (T1, T5, and T7) in moderate and severe compression (Li et al., 1999). The labelled cells were randomly distributed in the ventral, lateral and dorsal columns of the WM; most of them were located in the subpial region. However, very few labelled cells were found in grey matter (GM) (Li et al., 1999). The mentioned trend of oligodendrocytes apoptotic death is represented in

Figure 2
At 7 days pi, the number of CC1 + cells reduced significantly in all zones up to 14 mm distal to the injury compared to control animals (Rabchevsky et al., 2007). Also, in ventrolateral WM, the density of mature oligodendrocytes was less than 50% of that in normal uninjured mice (Lytle and Wrathall, 2007). In epicenter cross-sections, the proportion of oligodendrocytes to all survived cell types was reduced from 93% to %86 cells per section.
Indeed, most oligodendrocytes in the impact site died within the first week after injury (McTigue et al., 2001). In another study, cell counts in the dorsal fasciculus above the lesion, as well as lateral tracts above and below the lesion, showed that TUNEL + cells (presumably oligodendrocytes, because they were negative for GFAP, neurofilament, OX-6 and BS-1 markers) first appeared by 3 days pi, becoming maximal in number between days 5 and 7 (Abe et al., 1999). Also, two other studies confirmed that 7 days pi was the time point when the number of apoptotic oligodendrocytes reached the maximum (Casha et al., 2001, Huang et al., 2014.
At 8 days pi, apoptotic oligodendrocytes were most numerous in 7 and 13 mm rostral to the epicenter . Also, eight days after T9 level hemisection in C57BL/6J mice, a selective loss of oligodendrocytes were observed 3-7 mm rostral and caudal to epicenter.
Staining for CC1 and caspase-3 (a protein that is associated with apoptosis) confirmed activated caspase-3 was expressed in oligodendrocytes on injured side of this area but neither in the intact side of the cord nor in control sections. From double-labelling of CC1 and caspase-3, it was also concluded that approximately 10-15% of oligodendrocytes in this region expressed both epitopes. At this time point in the areas up to 5 mm from the epicenter, a significant loss of oligodendrocytes was observed on the injured side. Moreover, ultrastructural examination at 5-7 mm away from the epicenter showed prominent apoptotic features (Dong et al., 2003). In transection models from day 2 to 8, in situ hybridization revealed that a large area surrounding the epicenter was completely free from MBP mRNA, indicating complete loss of oligodendrocytes within this zone (Bartholdi and Schwab 1998).
By 9 dpi, a significant increase in the number of apoptotic oligodendrocyte was seen in rostral (T1 to T7) and caudal segments (T10, T12, and L2) (Li et al., 1999).
The number of apoptotic oligodendrocytes varied between moderate and severe injuries. After moderate injury the highest number of apoptotic cell death occurred in T7 and T10 segments while in severe injury the lesion size extended and the highest number of apoptosis observed in the T7 and T12 segments. In all examined segments there were more apoptotic oligodendrocytes in severely than in moderately injured rats. However, the difference was not statistically significant (Li et al., 1999).
By 10 days pi, oligodendrocytes were still present in areas with degenerating axons, Wallerian degeneration (WD) zone (i.e., 3 to 10 mm caudal and rostral to injury epicenter). However, the number of oligodendrocytes was significantly reduced to less than half of oligodendrocytes in sham operated animals (number of Olig2 + cells: ~ 60, 50, and 100 in dorsal ascending tracs, corticospinal tract and sham, respectively) (Wang et al., 2009).
In the second week after injury, the number of apoptotic oligodendrocytes declined significantly in lateral tracts above and below the injury site and the dorsal column (Abe et al., 1999), but they were still present 1-2 mm distal to the epicenter (Casha et al., 2001). The epicenter itself was already devoid of oligodendrocytes; only an amorphous mass of cell debris and MBP-positive macrophages were observed at the lesion epicenter, where MBP mRNA was almost absent. The very few MBP mRNA expressing cells detected in the dorsal funiculi that were probably due to invading Schwann cells (Bartholdi and Schwab 1998).
The oligodendrocyte processes formed a defined border separating the lesion area from the spared tissue (Bartholdi and Schwab 1998). In WM areas remote from the lesion, signs of WD became very prominent; at the center of the degeneration zone MBP mRNA + oligodendrocytes were lost; however, levels of MBP mRNAs strongly increased at the borders of the lesion (Bartholdi and Schwab, 1998).
Yong et al. also detected a second peak of oligodendrocyte apoptosis at 14 days pi (Yong et al., 1998). Since, at this time point the authors did not specify the areas where the TUNEL/RIP + cells were present, this second peak could be due to death in the WD zones rather than the epicenter where the first peak of apoptotic oligodendrocytes was observed at 5-7 days pi.
Apoptotic cell death of oligodendrocytes continued by the third weeks pi (Crowe et al., 1997. At this time point, oligodendrocytes apoptosis had decreased slightly in rostral blocks, in contrast to 7 mm caudal to the lesion epicenter where apoptosis increased . Between 8 and 30 days pi, a delayed apoptosis of oligodendrocytes continued specifically in distal zones both rostral and caudal (Dong et al., 2003).
By the fourth week after injury, in WM at the epicenter, the density of apoptotic oligodendrocytes reduced to the normal level (Lytle and Wrathall, 2007); also at 1.5 mm rostral and caudal to the injury epicenter in two injured mice the density of oligodendrocytes returned to the levels of uninjured controls (Lytle and Wrathall, 2007). Thus by the fourth week, the number of oligodendrocytes seemed to stabilize and remain constant thereafter (McTigue et al., 2001). The summary of OL's death is schematically presented in Fig. 2.

The response of oligodendrocyte progenitor cells to TSCI
At 24 hours pi, the proportion of BrdU/Olig2 + cells dropped dramatically from ~ 90% in control to almost 40% at the epicenter (McDonough et al., 2013), which shows that OPCs that were proliferating before injury either die or do not activate a proliferation program to a significant extent after TSCI (Horky et al., 2006). Meanwhile, a peak of NG2 + cell proliferation was observed at the 1.5 mm away from the epicenter (Lytle and Wrathall, 2007), and a small increase in number of NG2 + cells at the WM 5 and 10 mm caudal to the epicenter was detected (Wu et al., 2005). The NG2 + cells as well as newly formed oligodendrocytes were surrounding lesion borders (Hesp et al., 2015).
At day 2 pi, the number of NG2 + cells were increased by 50% compared to controls at the distal ventrolateral funiculi but did not change at the epicenter (Rabchevsky et al., 2007). One to three days pi, the number of proliferating OPCs (BrdU/NG2 + cells) in spared WM increased slightly but insignificantly compared to control animals (25 and ~10 cells/mm 2 , respectively), whereas in GM (borders and spared GM), the number was significantly higher (>40 cells/mm 2 ).
In areas 2-4 mm distal to the epicentre, as well as lesion borders, half of the dividing cells were NG2 + at day 3 and some of the NG2 + cells were co-labelled with GFAP (marker of astrocytes) (Zai andWrathall, 2005, Ceruti et al., 2009). There were only a few proliferating CC1 + cells in the lesion area and WM distal to the epicentre (Ceruti et al., 2009).
Later, during 4-7 days pi, OPCs continued to proliferate at the WM and GM lesion borders and some evidence of oligodendrogliogenesis was revealed along the lesion borders (Tripathi and McTigue, 2007). At day 7, the total number of NG2 + cells were higher at the spared WM at the epicentre compared to control animals (Lytle and Wrathall, 2007). There was small reduction in NG2 + cells compared to day 2 (Rabchevsky et al., 2007), yet remained 8 times higher than control animal, comprising over 80% of the proliferating cells at the lesion borders (Ceruti et al., 2009). At the ventrolateral funiculi, an increase in total number of mature oligodendrocytes was observed (Rabchevsky et al., 2007). In another study, the number of proliferating NG2 + cells (BrdU/NG2) were reported to increase up to three times within the epicentre (both at the spared and lesion areas) and up to 6 times rostral and caudal to the epicentre (< 2 mm) (McTigue et al., 2001).
At the first week pi, in 5 mm caudal from the epicentre, an increased number of NG2 + cells was observed in WM. Also, NG2 + cells in the WM from subpial region 5 mm distal to the epicentre were predominantly co-labelled with GFAP (Wu et al., 2005).
The OPCs that had proliferated during the first week, kept proliferating during the second week (7-14 dpi) at the GM lesion borders, whereas at the WM lesion borders the number remained similar to 7 days pi. The proliferation of OPCs at the borders led to higher numbers of newly formed oligodendrocytes in this areas during the first three days ( ~17 cells/mm 2 ) compared to control animals (<5 cells/mm 2 ) (Tripathi and McTigue, 2007). In-situ hybridization of MBP mRNA at the lesion borders showed an increase in MBP expression in WM and GM at 8 days pi. These MBP-expressing cells, displayed the initial signs of remyelination by forming loose network of cells at the lesion borders (Bartholdi and Schwab, 1998) By the second week pi a continual trend in the reduction of the extent of NG2 + and Olig2 + cells was seen reaching that of normal controls and remaining at this level afterwards (McTigue et al., 2001, McDonough et al., 2013. Also, in the regions that OPCs were highly proliferative during 1-3 days (i.e. spared WM), the number of oligodendrocytes was doubled compared to day 3, indication of the oligodendrogliogenesis process. An excess of oligodendrocytes was also observed in spared GM (>40 cells/mm 2 ) where the OPCs proliferated within the first three days after injury. As expected, the greatest increase in number of oligodendrocytes was detected in the area where OPCs had proliferated i.e. WM and GM lesion borders (>50 cells/mm 2 ). The highest proliferation rate occurred 1-2 mm distal from the epicentre (Tripathi and McTigue, 2007). The new oligodendrocytes formed a border between the lesion epicentre and the intact regions (Tripathi and McTigue, 2007). Similarly, it has been reported that the number of oligodendrocytes was significantly increased in areas distant from the injury epicentre (> 6 mm), mainly in WM close to the border of the pia-matter (Rabchevsky et al., 2007, Hesp et al., 2015. By this time, the density of proliferating NG2 + cells was still higher at the epicentre and up to 2 mm away from the lesion epicentre (~3.8 and ~4.5-fold, respectively) compared to the control animals (McTigue et al., 2001). However, at 4 weeks pi, the density of the proliferating NG2 + cells at the epicenter was in the normal level (Horky et al., 2006).
At 4 weeks pi, the total number of NG2 + cells was higher in rostral and caudal spared WM, and at the epicentre (~ 76, ~ 63, and ~ 64 cells/mm 2 ) compared to control animals ( ~ 40 cells/mm 2 ) (Lytle and Wrathall, 2007). This observations validated that NG2 + cells were continuously proliferating during the first month while preserving their progeny (Hesp et al., 2015).
McTigue et al. reported a decrease in the total number of NG2 + cells at the epicentre and rostral to the levels even less than uninjured animals, and a slight decrease compared to week 2 but still higher than caudal sections of normal tissue (McTigue et al., 2001). However, these cells were still proliferating at higher rates than in controls (~20-25 NG2/BrdU + cells/section as opposed to ~5 cells/section in naïve) (McTigue et al., 2001). Similarly, the proliferation rate has been reported to decline after 4 weeks pi, but remaining 3-9 fold higher than control animals (Hesp et al., 2015). The highest number of NG2 + cells was observed 5-10 mm rostral and caudal to the epicentre (Wu et al., 2005), whereas, at the WM of the epicentre, the number of oligodendrocytes returned to normal amount (Lytle and Wrathall, 2007).
At 6 weeks pi, although the number of NG2 + cells was still higher than controls (Rabchevsky et al., 2007), a very few proliferating cells were seen distal (2-4 mm) to the epicentre. However, one-fourth of the total oligodendrocytes were proliferative in this region, which is presumably the result of the NG2 + cells proliferation at the early stages (Zai and Wrathall 2005). The number of oligodendrocytes had returned to normal, validating the recovery of oligodendrocyte population by this time (Rabchevsky et al., 2007). These newly formed oligodendrocytes ran processes along the axons, and initiated remyelination from 6 weeks pi to as late as 9 weeks pi (Hesp et al., 2015). The fluctuation in proliferation rate of OPCs in epicenter and distal regions has been schematically demonstrated in Fig. 3.
The vast majority of new oligodendrocytes at 3-4 months pi, was derived from OPCs ( Barnabe-Heider et al., 2010, Assinck et al., 2017. McTigue et al. reported that the proliferative response of OPCs remains elevated compared to normal animals up to 4 weeks pi, and returns back to normal levels at 10 weeks pi (McTigue et al., 2001). In mice, on the other hand, the total number of NG2 + cells escalated 4, 5, 6, 7, and 9 weeks pi in both WM and GM compared to uninjured control animals (Hesp et al., 2015). This shows a long-term maintenance of the OPCs population in both rats and mice and it overall reveals that proliferation of OPCs during the first two weeks is the source of newly formed oligodendrocytes and remyelination at later times. The oligodendrogliogenesis and total number of oligodendrocytes over time after injury outlined in Fig. 4.

Figure 4
OPCs are the main source of new mature oligodendrocytes after TSCI and the reduction in the extent of OPCs at chronic phase after injury is a result of oligodendrocyte differentiation. This assumption has been confirmed using mapping the fate of OPCs after injury (Assinck et al., 2017). There was significantly more oligodendrocytes at 12 weeks pi compared with 3 weeks pi. Reciprocally, there was lower percentage of recombined OPCs (17%) at 12 weeks pi compared with 3 weeks pi (35%) (Assinck et al., 2017).
Reported data were based on immunohistochemistry, light and electron microscopy techniques. Also in one study the fate and origin of Schwann cells at the desired time was elucidated using the technology of CreER transgenic mice. In the earlier studies the presence of Schwann cells in the CNS parenchyma was identified using ultrastructural evaluations, Schwann cells produce thicker and more compact sheaths comparing to oligodendrocytes.
While in the recent articles the expression of the P 0 was chosen as the marker of the myelinating Schwann cells. A schematic comparison between the morphology and composition of myelin generated by Schwann cells in the CNS versus that derived from OPC differentiation is depicted in Fig. 5.

Figure 5
Myelin production by Schwann cells is a protracted process, which occurs in moderate to severe spinal cord injuries (data were not available for mild injuries). The migration of Schwann cells to the spinal cord parenchyma was reported to be limited due to the presence of astrocytes (Franklin and Blakemore, 1993). Eventually, in moderate to severe injuries, where the supportive cells, particularly astrocytes, suffer considerable damage, Schwann cells can migrate to the CNS tissue (Salgado-Ceballos et al., 1998). The extensive presence of dividing Schwann cells suggests that these cells are at least partly responsible for axon ensheathment/remyelination after injury (Hui et al., 2010).
The number of proliferative Schwann cells increases over time after injury. In one study evaluating the proliferative potential of P75 + Schwann cells in Rhesus monkeys, a 14-fold increase in the number of BrdU positive Schwann cells was detected at 25 weeks compared to 7 weeks pi, while there was no proliferative Schwann cells in non-injured control (Yang et al., 2006). At one week pi, Schwann cells or their precursors appeared caudal to the lesion (0.5-2 mm from the epicenter). The precursors were small, undifferentiated cells lacking a basement membrane and were in association with sprouts of invading axons near the roots (Gilson andStensaas, 1974, Blight, 1985). Also, bipolar spindle-shaped cells with ovoid nucleus were randomly dispersed, which is a typical morphology of migrating Schwann cells (Brook et al., 1998). In zebrafish, division of Schwann cells was observed at day 10 after injury which probably is the onset for ensheathment/remyelination (Hui et al., 2010).
At the second week pi, the presence of Schwann cells were also detected in the dorsal funiculi (Bartholdi and Schwab, 1998), they possessed basement membrane, elongated nucleus and were closely associated with individual axons (Gilson andStensaas, 1974, Matthews et al., 1979).
By the third week, groups of axons myelinated by Schwann cells were frequent around the cystic cavity and base of the dorsal columns (Griffiths and McCulloch, 1983). Up to one month pi, many unmyelinated fibers were enveloped by Schwann cells, either isolated or in small clusters (Salgado-Ceballos et al., 1998). In addition, many axons appeared to be remyelinated by Schwann cells (Hui et al., 2010, James et al., 2011. Six weeks pi, at the dorsal column, there were fascicles with thin perineural sheath containing up to 50 axons and restricted to immediate regions of the lesion site (Matthews et al., 1979).
By two months pi, many of myelinated axons had grown toward the scar matrix. Those in dorsal region were apparently myelinated by Schwann cells (Matthews et al., 1979, Salgado-Ceballos et al., 1998. At this time, approximately one third of the fibers were myelinated and the ratio of myelinated axons by Schwann cells to oligodendrocytes was about 8:1 (Salgado-Ceballos et al., 1998). It is worth noting that in this report the authors assume that thick myelin sheaths are associated with Schwann cells.
As the injury progressed into chronic stages (3 and 4 months), many axons had dense, healthy myelin sheaths and it was often associated with the presence of Schwann cells (Salgado-Ceballos et al., 1998, James et al., 2011. The Schwann cells contribution in remyelination of the axons in dorsal column was also confirmed by co-staining of axons (NF 200) and Schwann cell-associated myelin (P 0 ) (James et al., 2011). In this study P 0 reactivity was detected at 4 and 12 weeks but not one week pi. Remyelination by Schwann cells continued by 12 months pi (Salgado-Ceballos et al., 1998).
Although the Schwann cells contribution in remyelination after SCI is a well-shown evidence, the origin of the present Schwann cells in the CNS parenchyma is still a contentious issue. In a recent study at 12 weeks pi fate mapping experiments revealed that the most majority of Schwann cells (about 70-80%) which contribute in myelin production after TSCI were derived from Olig2 + and PDGFRα + CNS OPCs rather than P 0 expressing peripheral myelinating Schwann cells (Assinck et al., 2017). OPC-derived myelinating Schwann cells were mainly found in the dorsal column (Assinck et al., 2017).

DISCUSSION
Extensive oligodendrocytes cell death during the first two weeks after injury leave a broad number of denuded axons in the epicenter as well as distal regions. Oligodendrocyte cell death occurs through both mechanisms of apoptosis and necrosis while apoptosis was initially detected in epicenter and was extended rostro-caudally.
Along with the mature oligodendrocytes, many OPCs are lost at early stages, however, as early as one day pi they start proliferation at the lesion borders and spared tissues. Later, differentiation of these newly formed OPCs compensate the lack of oligodendrocytes.
Oligodendrogliogenesis continues at later stages until the number of oligodendrocytes return to normal (by one month pi). Despite differences between animals after TSCI, the glial response to injury and in particular myelinating cells shares similarities.
Recently numerous studies in the field of spinal cord regeneration has been focused on oligodendrocyte protection and enhancing remyelination after TSCI. Demyelination is a part of secondary injury cascade which initiate immediately after injury. The present preclinical evidence reveals the potential of remyelination in adult tissue. Remyelination of spared and growing axons occurs through a two-step process: 1. providing a sufficient population of myelinating cells and 2. induction of remyelination.
Therefore, a key determinant of the success of remyelination after injury is the proliferative ability of myelinating cells. In this study according to the included articles it can be concluded that oligodendrocytes within a demyelinated region of spinal cord are not induced to divide in the presence of demyelinated axons and the survived oligodendrocytes are post-mitotic and do not contribute to the remyelination in adult CNS. However, it should be noted that the oligodendrocyte proliferation after TSCI was not fully refute and there are some studies discussing the proliferative ability of oligodendrocytes within a demyelinated region of spinal cord (Vick et al., 1992). According to the fact that the Schwann cells and OPCs proliferation after injury is a prerequisite to remyelination, the finite proliferative potential of oligodendrocytes after injury may be the cause of low contribution of mature oligodendrocytes in remyelination.
OPCs constitute the dominating proliferating cell population in the intact adult spinal cord (Horky et al., 2006, Sellers et al., 2009, Barnabe-Heider et al., 2010. Indeed, according to accumulating evidence in this study spinal cord injury results in changing the proliferation rate of OPCs over time and within the first two weeks the proliferation rate of OPCs reaches to a maximum at both regions of epicenter and lesion border (Fig. 3). Reduction in the proliferation is followed by the oligodendrocyte differentiation, as demonstrated by detection of an elevated number of oligodendrocytes one week after injury which reaches to the normal level by about one month pi. Endogenous remyelination after TSCI was also confirmed by detection of an increase in the expression of MBP at first week pi compared to earlier days (Bartholdi and Schwab, 1998, Horky et al., 2006, McDonough et al., 2013, Hesp et al., 2015, Assinck et al., 2017. However, Assinck et al. through fate mapping of the OPCs showed that although there was a significant increase in the extent of new myelin from 3 to 12 weeks pi, the total myelin did not demonstrate a significant change and they attribute this to the myelin turn over and replacement between 3 and 12 weeks pi (Assinck et al., 2017).
Schwann cells that normally reside in peripheral nervous system, participate in remyelination of the denuded axons within the cord through a protracted process. Abundant preclinical (Martin et al., 1996, Oudega and Xu, 2006, Guest et al., 2013, Deng et al., 2015, Kanno et al., 2015 and clinical studies (Saberi et al., 2008, Anderson et al., 2017 have been conducted to evaluate the safety and efficacy of Schwann cell transplantation for repair of injured spinal cord. After peripheral nerve injury Schwann cells dedifferentiate into non-myelinating Schwann cells and proliferate extensively. These newly formed non-myelinating Schwann cells migrate within the lesion and involved in nerve regeneration through expression of neurotrophic factors and neural cell adhesion molecules, recruitment of macrophages for removal of myelin debris as well as myelination or ensheathment of growing axons after differentiation into myelin-forming Schwann cells. A moderate improvement of motor function after Schwann cell transplantation was also concluded in a systematic review discussing the efficacy of Schwann cell transplantation for motor function recovery after spinal cord injuries in animal models (Hosseini et al., 2016). There is some evidence showing that spinal cord injuries and disruption of the glia limitations induce Schwan cells invasion within the CNS parenchyma and remyelination (Gilson and Stensaas, 1974, Matthews et al., 1979, Griffiths and McCulloch, 1983, Blight, 1985, Franklin and Blakemore, 1993, Bartholdi and Schwab, 1998, Salgado-Ceballos et al., 1998, Hui et al., 2010, James et al., 2011. Besides of cell migration evidence, considerable proliferation of Schwann cells in the lesion confirm the participation of these cells in axon ensheathment/remyelination after TSCI. Schwann cells mostly contributed in ensheathment/remyelination of axons in the dorsal column. Although in the most studies this observation has been attributed to the route of peripheral Schwann cells invasion, one recent study through a comprehensive evaluation of the fate of OPCs showed that CNS OPCs are the main origin of remyelinating Schwann cells in the dorsal column (Assinck et al., 2017). Axon ensheathment/remyelination by Schwann cells continued to one year pi.
In conclusion, the cumulative evidence shows that in animal models, the number of oligodendrocytes approaches to the normal levels after about one month pi. The majority of oligodendrocytes are lost at the epicentre and adjacent regions, however, OPCs proliferation and subsequent differentiation to myelinating cells (both oligodendrocytes and Schwann cells) are the main route to compensate the extensive demyelination after injury. Invading peripheral Schwann cells also contribute in remyelination. Therefore, pharmacotherapy for enhancing OPCs proliferation and differentiation to mature myelinating cells (oligodendrocytes or Schwann cells) can be the therapeutic target (Imamura et al., 2015) and the evidence-based trends obtained for oligodendrocyte apoptosis, proliferation and differentiation can be used as a guide to design preclinical and in the next step clinical studies.

Limitations of the work:
Although the evidence shows that oligodendrogliogenesis and axon ensheathment/remyelination after TSCI is a continual process by the chronic phase, we do not yet know whether the newly myelinated axons are functional, since mature oligodendrocyte can remyelinate even dead axons or artificial fibers (Lee et al., 2012, Lee et al., 2013.
Therefore, the functional benefit of the new formed myelin remains to be clear.

CONFLICT OF INTREST STATEMENT
The authors have no conflicts of interest.

AUTHOR CONTRIBUTIONS
All authors passed four criteria for authorship contribution based on recommendations of the      Table 1 Characteristics of included studies Table 2 Quality assessment of included studies Table 3 Summary of the markers and methods used to identify the cell types (i.e., oligodendrocytes and oligodendrocyte progenitor cells) and to detect apoptosis ? ? 1. Species; 2. Using appropriate tests; 3. Severity of injury; 4. Level of injury; 5. Age/weight; 6. Number of injured animals per group; 7. Designation of strain; 8. Definition of control (normal, sham operated or laminectomy); 9. Description of statistical analysis; 10. Regulation and ethics; 11. Bladder expression; 12. Blindness of assessor; 13. Genetic background; 14. Method of allocation to treatments; 15. Description of the reasons to exclude animals from the experiment during the study (attrition), where + means clearly specified, ? has not specified clearly, and -specified as it has not been considered in the study. Table 3 List of the markers and methods used to identify the cell types (i.e., oligodendrocytes and OPCs 1 ) and to detect apoptosis Cell types/Assessment type Markers/ Method of Assessment Mature oligodendrocytes CC1(anti-APC 2 ) CNPase 3 PLP 4 MBP 5 RIP 6