Ethics

The experimental design using human iPS cells was approved by the Institutional Stem Cell Research Oversight (ISCRO) committee in the Johns Hopkins University. Use of anonymous human samples for laboratory research was approved by JHU Institutional Review Board on Human subjects (ESC-09-05-20-01). This included the acquisition of CD34+ cells for iPS cell generation from adult patients [20,22]. Fetal brain samples that were used as a reference for qRT-PCR analysis was supplied by the University of Washington NIH-supported fetal repository. In vivo study was carried out in accordance with the guidelines of the Johns Hopkins University for the Care and Use of the Laboratory Animals. The protocol was approved by the Committee on the Ethics of Animal Experiments with the Protocol Number: RA12M82.

Cell Culture

Three different iPS cell lines were used in this study: BC1 [22], MR31 [21] and A1-4 [20]. The sources of the iPS cells and the reprogramming methods used are summarized in S1 Table. The BC1 iPS cell line was generated from adult CD34+ peripheral blood monocytes using an episomal plasmid (pEB-C5) expressing 5 reprogramming vectors (OCT4, SOX2, KLF4, MYC, LIN28). Notably, whole-genome sequencing of the BC1 iPS line generated using an episomal plasmid vector revealed that the vector sequence was undetectable in the genome [22,23]. A1-4 was generated by reprogramming MSCs with retroviruses expressing OCT4, SOX2, KLF4, MYC, and the high mobility group A1 chromatin remodeling protein, HMGA1 [20]. MR31 was generated from the IMR90 fetal lung fibroblast cell line reprogrammed with retroviruses expressing OCT4, SOX2, and KLF4 [23]. iPS cells were cultured and differentiated into OPs as described previously [19,22,24,25].

Undifferentiated iPS cells were propagated on a mouse embryonic fibroblast (MEF) feeder layer. All lines were expanded in iPS cell growth media containing DMEM/F12 supplemented with 20% knockout serum, 5% nonessential amino acids, 5% Glutamax, and 20 ng/ml Fibroblast Growth Factor 2 (FGF2) [22]. The iPS cells were differentiated to OPs using methods optimized by our laboratory for ES cells [24–26]. First, neural differentiation of undifferentiated iPS cells was initiated by embryoid body (EB) formation in neurobasal media (Gibco) supplemented with Noggin and FGF2 (200 ng/ml and 20 ng/ml, respectively; R&D Systems). EBs were maintained in ultra-low adhesion tissue culture plate for 15 days. EBs were plated onto matrigel-coated tissue culture plates in neurobasal media supplemented with FGF2 (20 ng/ml) for 5 days, leading to differentiation into neural progenitors (NPs). NPs were expanded in this media for 5–7 days and glial progenitors (GPs) were subsequently induced by adding EGF (10 ng/ml; R&D Systems). NPs and GPs are distinguished by their morphology, marker expression and their differentiation capacity. Our laboratory has shown that NPs are able to differentiate into neurons and glial cells, while GPs primarily develop into either astrocytes or oligodendrocytes that are regulated by specific cell culture requirements [26,27]. After 7 days in GP media, cells are split with 1 mg/ml collagenase and transferred into new plates. OPs were induced by adding neurobasal medium supplemented with EGF (20 ng/ml) and PDGF-AA (20 ng/ml; Peprotech) for 25 days. To determine differentiation efficiency among iPS lines, cells were harvested and RNA extracted for quantitative RT-PCR (qRT-PCR) analysis or cells placed in chamber slides for immunocytochemistry and protein expression analysis. A subset of OP cultures were also sorted by O4 expression using magnetically-activated cell sorting (MACs) and further expanded in culture for use in cell transplantation. O4 is a sulfated galactosylceramide (sulfolipid) synthesized primarily by oligodendrocytes and their progenitors OPs that has been shown to play a major role in myelin function and stability [28]. O4 expression is also expressed by NG+ OPs in the human fetal brain [29]. Using this method, we are able to maintain and expand cultures containing over 98% O4+ OPs that proliferative for up to six months in culture.

Immunocytochemistrical Characterization and Cell Counting. Cells were plated onto chamber slides. Before they became confluent, cells were washed twice in phosphate buffered saline (PBS) and the excess PBS removed. Cells were then fixed in 4% paraformaldehyde (PFA) in PBS for 10 minutes at room temperature (RT). The fixative solution was removed and cells were washed twice with PBS for 5 min each. Cells were permeabilized and blocked with PBS solution containing 0.2% (w/v) Triton X-100 (Sigma) and 1% (w/v) bovine serum albumin (BSA; Sigma) for 5 minutes at room temperature (for cell surface markers this step was omitted as previously described by [29]. Primary antibody (1:50 dilution, see S2 Table for list of the primary antibodies) was added after dilution in antibody dilution buffer (PBS with 0.1% (w/v) BSA) and incubated for 1 hr at RT. Samples were washed three times with PBS for 5 minutes each. Appropriate secondary antibodies were added (Alexa 488 or Alexa 564 conjugated antibodies, 1:200; Molecular Probes) in antibody dilution buffer and incubated in a humidified dark chamber for 1 hr at RT. Cells were then washed three times with PBS for 5 minutes each. Nuclei were stained using 4',6-diamidino-2-phenylindole (DAPI) solution (Sigma) for 10 min at RT in a humidified chamber in the dark. DAPI solution was aspirated and three 0.5 ml PBS rinses were performed and slides coverslipped. Differentiation of each cell line was performed in triplicate. Cells exposed to secondary antibody alone were used as negative controls. The number of positively stained cells were counted using Metamorph Image Analysis Software (Universal Imaging, Inc.) from 3 randomly, chosen 10X fields per 24-well cell culture dish and each stain was repeated in three wells (n = 9).

Quantitative Real Time-PCR Analysis. Total RNA was isolated from cells stored in RNAlater (Qiagen) solution using MiniRNeasy kits (Qiagen 74124) and complementary DNA (cDNA) generated with SuperScript III First-Strand Synthesis System (Invitrogen 18080–051) reverse transcription kits following the manufacturer instructions. Real-time quantitative polymerase chain reaction (qPCR) analyses were performed using ABi7900HT (ABiosystems) in 96 well plates (Abi N801-0560) where each real-time amplification had a template equivalent to 5 ng of total RNA. Each primer set was tested in at least triplicate across 3 biological replicates. Negative controls included a reverse-transcription-negative control per sample and a no template control. For these analyses, total RNA extracted from a 92-day old fetal human brain was used as a positive control. Gene expression and amplification curves were normalized to β-actin. Taqman Assay-on-Demand designed oligonucleotides were used for the detection of the following genes (S3 Table). Using the ∆Ct method, quantification within the log-linear phase of the amplification curve acquired for each probe/primer set was normalized to β-actin and plotted.

Contusive Spinal Cord Injury

All experimental procedures in this study adhered to the guidelines delineated in the Rodent Survival Surgery guide and approved by the Institutional Animal Care and Use Committees at the Johns Hopkins University. An ARRIVE Guidelines Checklist for reporting in vivo experiments is provided in Supporting Information (S1 Checklist). A total of 63 rats were used in the following study, twelve rats were injected with either BC1-, MR31-, or A1-4-derived OPs, PBS alone or heat-killed BC1-derived OPs and 3 rats with laminectomy alone (sham) were used for comparisons with electron microscopy. Ten-week-old adult female Lewis rats weighing ~200–220g were used in the study. One of the immediate side-effects of SCI is difficulty urinating and thus the bladder of these rats must be manually expressed for days immediately after injury and are thus susceptible to infection. Therefore, this study utilized female rats which are less suspectible to urinary tract infections under these conditions [30]. All rats were immunocompromised three days prior to cell transplantation followed by daily injections of cyclosporine A (10mg/kg/d) until the end of the study, as previously described [24]. Rats (n = 12 per treatment) were anesthetized with ketamine (7.5 mg/kg; Phoenix Pharmaceutical) and xylazine (60 mg/kg; Phoenix Pharmaceutical). Laminectomy was performed at thoracic vertebra T8 without opening the dura mater. Injury was produced using a Multicenter Animal Spinal Cord Injury Study (MASCIS) Impactor, formerly called the NYU-Impactor. The MASCIS impactor model is a well-established model for studying contusive SCI in rats, and histopathological changes closely mirror those observed in human SCI [31,32]. This device precisely measures the movement of a 10 gram rod dropped from 6.25, 12.5, 25.0, or 50.0 mm heights onto a rat T8-T10 spinal cord after laminectomy to induce Mild (6.25 mm), Moderate (12.5 mm), Severe (25.0 mm) and Very Severe injury (50 mm) respectively. In this report, the MASCIS impactor consisting of a 10.0 g weight rod dropped from a 12.5 mm height was utilized to produce a moderate contusion injury, as previously described [33–37].

Cell Transplantation. Rats were anesthetized 24 hours after injury as described above, and the injury site on the spinal cord was exposed and the dura excised using a 30-gauge needle. O4+ OPs cultured in OP-inducing media for 25 days were used as the majority of cells (> than 98%) expressed OP marker O4. Viability was also assessed by sampling a subset of cells with 0.4% trypan blue (Sigma) prior to transplantation and only populations with at least 90% viability were transplanted except for heat-killed controls which had their membranes disrupted by heating samples at 65°C for 90 minute followed by sonication.

Twelve rats were randomly selected and injected with either 500,000 live iPS-OPs or heat-killed iPS-OPs in 2 μL volumes with PBS at a rate of 1 μL/minute into the posterior funiculi of T8 at a depth of 1.2 mm. This region is known as the injury epicenter. Cells were injected using a 10 μL Hamilton Gastight syringe with an attached borosillate glass tip (50–100 μm tip diameter) secured to a manual micromanipulator (World Precision Instruments) adjusted with a 80° tilting base. Briefly, the tip is lowered to a depth of 1.2 mm below the surface of the cord and is held in place for 2 minutes before the cell injection and 2 minutes after the cell injection. Cells are delivered under the control of a micro-syringe pump controller (World Precision Instruments). After injection, the dura is closed with a 9–0 suture, the muscles re-apposed, and the skin closed with sutures. A suture is also placed in the muscle adjacent to the site of injection to locate the injection site at the time of sacrifice. Sites are monitored for bleeding for the next six hours and for infections over the next five to seven days. Rats were housed two per cage in a pathogen-free isolation unit and given subcutaneous injections of isotonic saline (20 ml/kg), which was administered every morning for 7 days after surgery. The rats’ bladders were expressed twice daily, and the rats were inspected for weight loss, discomfort, dehydration, and autophagia, with appropriate veterinary care as needed. Tylenol (100 mg/kg) and gentamicin (5 mg/kg) were administered every morning for 7 days to minimize pain and potential for infection, respectively. In addition to these study groups, three additional uninjured rats were employed and served as sham controls (undergoing laminectomies only) to compare the extent of myelination in the injured SCIs after transplant. These rats were treated under identical conditions following surgery.

Motor Behavior Assessment. The Basso, Beattie, and Bresnahan (BBB) Locomotor Rating Scale [38,39] was used to validate the motor behavior of each animal before and after injury. Two well-trained examiners, who were blinded to the experimental group of the animals, evaluated and scored the motor behavior of each animal for the left and right sides individually. The left and right sides were averaged. Scoring was performed on each animal every three days for the first four weeks, and thereafter performed once per week.

Histological Analyses. Postmortem characterization of transplants was observed up to two months after transplantation. Rats were deeply anesthetized using isoflurane. Transcardial perfusion was performed with DPBS (Gibco) and paraformaldehyde (PFA) solution (4%; 15713-S, Electron Microscopy Sciences). Spinal cords from twelve rats within each group were randomly prepared for either frozen sectioning, paraffin embedding or prepared for electron microscopy and histological assessments were performed by a laboratory technician who was blinded to the experimental groups. Thus, four rats of each treatment group were either prepared for cryosection, paraffin sectioning, or embedded for electron microscopy. The spinal cord was then carefully extracted from the vertebrae column and post-fixed in 4% PFA, followed by 30% sucrose solution for 24 hours, and last embedded in paraffin or cryopreserved in freezing compound (O.C.T. Tissue-Tek). Frozen tissue was cut in 5 μm sections and prepared for immunohistochemistry. For analysis of frozen sections, 30 histological sections, sliced 5μm thick were analyzed that encompassed the injury epicenter. Immunoflourescent staining with different antibodies was performed on adjacent slices for each spinal cord. Immunostaining was performed by incubation with 1:100 dilution of primary antibodies (S2 Table) overnight at 4°C in 15% FBS, followed by three rinses in DPBS, and incubated with 1:500 dilution of Alexa 488 and Alexa 564 secondary antibodies in 15% FBS for 1 hour at room temperature and coverslipped with DAPI-supplemented mounting media (Vectashield by Vector Labs). DAPI was used to stain nuclei and to determine the percent of immuno-positive cells. Negative controls included primary serum and secondary antibody alone. Cords embedded in paraffin were cut in 10 μm longitudinal sections and stained with either hematoxylin and eosin (H&E) or luxol fast blue (LFB). Sections encompassing the injury epicenter and surrounding areas of injury were selected by consecutively numbering each cut beginning from the dorsal surface. and then every other slice was stained with either LFB or H&E. The demarcation of glial scarring and cavitation in controls allowed for easy assessment to determine the boundaries around the epicenter of injury. This allowed the ability to analyze similar numbers of cuts across treatment groups. To analyze staining of images across groups were taken at the same exposure time below the level of saturation and analyzed within the same gray scale. The optical intensities was measured in identically-sized fields surrounding the injury epicenter and background subtracted from a region on the same slide surrounding the tissue using Metamorph Image Analysis Software.

Survival and Differentiation of Ops. To determine whether injected human iPS-OPs could be detected at the end of the study, frozen sections were analyzed using the human cell-specific marker, human nuclear antigen HNA (Millipore) to determine the number of human cells present. To determine the number of OPs that differentiated into mature oligodendrocytes, we counted the number of HNA/DAPI-positive cells that showed co-localized staining with myelin basic protein (MBP), a protein specifically expressed by mature oligodendrocytes and not by OPs. To study how OP transplantation may alter the immune response after SCI, sections were stained for CD68, (also known as ED-1) to count the number of immune cells (macrophages, microglia and T-cells) located at the injury epicenter [40]. Astrocyte response and glial scarring was also quantified by studying the number of GFAP-expressing cells at the epicenter of injury and by the fluorescence staining intensity around the cavity, as previously described [41].

Tissue Sparing Analysis. Images were analyzed using Metamorph Image Analysis Software, which allowed the tracing of the spinal cord perimeter and the central cavity for each image captured, giving the total area for each of these measurements. This analysis was performed on longitudinal spinal cord sections generated from 8 mm rostral to 8 mm caudal to the injury epicenter and encompassing the entire width of the spinal cord, as the cavity very rarely extended beyond this distance. Each cord was analyzed using 10 μm sections stained with hematoxylin and eosin and photographed at 16X magnification.

Electron Microscopy. For electron microscopy, spinal cords from 4 rats from each treatment group were processed into small 3 mm³ blocks that surrounded the injury epicenter and fixed for 1 hour in a mixture of glutaraldehyde (1.5%) and paraformaldehyde (3%), followed by washing three times in 0.1 M sodium cacodylate and 3 mM CaCl₂. Samples were then post-fixed in potassium ferrocyanide (0.8%) and osmium tetroxide (1%) for 1 hour followed by 3X washes in 0.1 M sodium cacodylate, and 3 mM CaCl₂. Following a brief dH₂O rinse, samples were embedded in Eponate 12 (Pella), and cured at 60°C for 2 days. Spinal cord sections (80 nm in thickness) corresponding to the site of the lesion were cut on a Riechert Ultracut E with a Diatome diamond knife, collected on formvar-coated 1 X 2 mm² copper grids, and stained with uranyl acetate followed by lead citrate. Sections were examined on a Hitachi 7600 transmission electron microscope (TEM) operating at 80 kV.

Myelination was examined on TEM sections at 10000X magnification. Normal myelinated, demyelinated, and remyelinated axons were counted using the line-sampling technique detailed by Blight [42]. The average number of normal myelinated, demyelinated, and remyelinated axons within five 25x25 μm² areas along the radial line yielded an estimate of the total number of axons within the injury epicenter, calculated as the number of axons per square millimeter. At least 10% of the area of injury is encompassed in these calculations. The number of remyelinated and demyelinated axons was used to determine the oligodendrocyte remyelination efficiency.

Statistical Analysis

All data were expressed as the mean ± SEM. Statistical significance was determined between treatment groups for cell counting and histological analyses by one-way ANOVA combined with Tukey’s posthoc test for comparison between groups. Similarly, significant differences in BBB scores among groups were also evaluated using a one-way ANOVA followed by a two-way Student’s t-test assuming equal variances.

Morphological Characterization

We derived OPs from three different human iPS cell lines using our published protocol to generate OPs from human ES cells [19,24]. We found that all three iPS cell lines exhibited morphologic changes consistent with a transition from the undifferentiated pluripotent stage into mature oligodendrocytes (Fig. 1). The transition from pluripotency into neuroectoderm was initiated by formation of embryoid bodies (EBs) in neurabasal media under low attachment conditions (Fig. 1B). EBs were cultured over a 15-day period and were then plated on matrigel in neurobasal media supplemented with FGF2 (20 ng/ml). Within 24 hours, cells with a neuroectodermal morphology were observed with projections extending outwards from the center of the EBs (Fig. 1C) indicative of neural progenitors (NPs) in culture. Adding EGF to the media led the cells to differentiate into glial progenitors (GP) with a bipolar morphology (Fig. 1D) [43]. At this stage GPs are bipotent and depending on the culturing media can be induced to become either OPs or astrocytes [26]. For transplantation studies, GPs were induced to differentiate into OPs by the addition of PDGF-AA and the removal of FGF2 from the media. The resulting cells had the characteristic rounded shape with multiple filopodial-like extensions (Fig. 1E). Upon removal of PDGF-AA from the culture media and with addition of 40 ng/ml 3,395-Triido-L-thyronine (T3), iPS-OPs demonstrate the ability to generate mature oligodendrocytes exhibiting branched morphology with high purity (> than 70%) in unsorted iPS-OP cultures (Fig. 1F).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Bright field microscopy images of the various stages of oligodendrocyte differentiation from induced pluripotent stem cells of the BC1 line.

(A) The undifferentiated cells grew in colonies on a mouse embryonic fibroblast feeder layer, (B) Embroid bodies (EBs) were formed by suspending the undifferentiating cells in culture on ultra-low adherence plates, (C) 15-day EBs were plated on matrigel, leading to the extension of cells of neuroectodermal lineage. Upon passaging, these neural progenitors grew in a monolayer and were subsequently differentiated to glial progenitors (D) stage. Addition of platelet derived growth factor-AA (PDGF-AA) to the media led to glial progenitors (GPs) transitioning into the (E) oligodendrocyte progenitor (OP) stage, displaying a rounded morphology with multiple filopodial extensions. (F) Terminal differentiation was induced by the removal of PDGF-AA, leading to the appearance of mature oligodendrocytes displaying characteristic branching. Scale bars- A-B: 300 μm, C-F: 80 μm

https://doi.org/10.1371/journal.pone.0116933.g001

Quantitative Gene Expression Analysis

Quantitative real-time PCR (qRT-PCR) analysis was carried out to compare gene expression profiles in the cells as they progressed from a pluripotent state into OPs. To determine if the differentiated cells decrease expression of genes associated with pluripotency, mRNA expression levels of NANOG, OCT4 and SOX2 were analyzed (Fig. 2A-C). We found that the expression levels of these genes were significantly decreased or undetectable once cells reached the NP stage. Interestingly, OCT4 expression showed a significant increase in the retrovirally-derived MR31 cell line at the embryoid body stage as compared to undifferentiated cells. These results suggest a persistence of OCT4 expression in this cell line that remains undifferentiated under EB conditions. Nonetheless, OCT4 transcript levels decreased after EBs were cultured as NPs. SOX2 expression also increased in EBs of MR31 and BC1. This is not surprising as SOX2 plays an essential role in early neuroectodermal differentiation [44] in addition to pluripotency. NANOG expression diminished during OP differentiation in all three lines. Most importantly, NP cultures from all three lines demonstrated significantly low or undetectable levels of NANOG, SOX2 and OCT4 mRNA transcripts, which is essential in preventing teratoma formation in vivo.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Plots showing mRNA expression of various developmental markers for all three iPS lines at each stage ofoligodendrocyte differentiation, as measured using qRT-PCR and normalized to fetal fibroblast controls.

Expression of pluripotency associated genes (A) NANOG, (B) Octamer-binding transcription factor 4 (OCT4) also known as POU5F1 (POU domain, class 5, transcription factor 1) and (C) SRY (sex determining region Y)-box 2 (SOX2) were undetected in oligodendrocyte progenitors (OPs) whilemarkers indicative of the oligodendrocyte lineage were induced including (D) NG2, (E) Myelin-associated glycoprotein (MAG), and (F) Myelin-associated oligodendrocytic basic protein (MOBP). Delta CTs were calculated using β-actin as a relative standard and lines normalized to fetal fibroblast controls. Mean + SEM were calculated from two independent differentiation experiments for each line and each gene performed in technical triplicate.Positive control included total RNA from human fetal brain. Abbreviations are Undiff: Undifferentiated iPS cells, EB: embryoid body, NP: neural progenitor, GP: glial progenitor, OP: oligodendrocyte progenitor.

https://doi.org/10.1371/journal.pone.0116933.g002

In addition to downregulating pluripotent genes, NPs also exhibited upregulated markers of early neural differentiation. For instance, NG2 is expressed in both GPs and OPs found in fetal and adult spinal cord [29,45]. While undetectable in undifferentiated iPS, low levels were initially detected in NPs but significant rose 5 days after adding GP induction media in all three iPS-derived NP lines (Fig. 2D). The increase in NG2 expression was more pronounced for the MR31 and A1-4 lines (both of which were derived from retroviral delivery of pluripotency factors), as compared to the BC1 line. However, the differences in NG2 expression in GPs were transient, as NG2 transcript levels decreased sharply in OPs where mRNA levels were similar in all three lines (Fig. 2D). To further confirm the identity of the OPs, we examined expression of genes encoding the myelin oligodendrocyte basic protein (MOBP) and myelin associated glycoprotein (MAG). Both of these proteins are expressed by myelinating oligodendrocytes [46,47]. As a positive control, total RNA was extracted from a 92-day old fetal human brain, at which time PDGFR-α +/O4+ OPs are found in the cortical subventricular zone [29]. We found that MAG expression was induced at the OP stage in all three lines (Fig. 2E). In fact, there was an almost linear increase in MAG expression as the cells progressed from the NP to OP stage in BC1. MAG expression was also relatively low in the MR31 and A1-4 lines at the GP stage, but rose sharply during differentiation to the OP stage. The MOBP transcript levels were also induced in all three OP lines (Fig. 2F), with negligible expression at the NP and GP stages and up-regulation at the OP stage. Expression of MAG and MOBP was higher in the control fetal brain tissue isolated at 13.3 weeks in gestation. This is consistent with other reports that have shown the presence of myelin marker expression as early as 5 weeks in gestation [29].

Immunocytochemistry and Protein Marker Expression

To corroborate the results for mRNA analyses, we also confirmed protein expression using immunofluorescence microscopy. Prior studies of oligodendrocyte development show that the first step in OP development is the differentiation of A2B5+ and NESTIN+ NPs into glial progenitors that express NG2 and PDGFR-α [24]. We also observed robust A2B5 and NESTIN protein expression in the NPs derived from EBs (Fig. 3A-B). As these cells differentiate into the GP lineage, they gain NG2 and PDGFR-α protein expression (Fig. 3C-D). Quantitative analysis was also performed on immunofluorescence images for three independent experiments for each line at each stage of differentiation. Quantitative analysis showed that >80% of the NPs were A2B5+ and NESTIN+ (Fig. 3E). In contrast, as expected, proteins associated with more differentiated OPs such as the RIP antigen, which specifically recognizes 2',3'-cyclic nucleotide 3'-phosphodiesterase in OPs and mature oligodendrocytes [48] as well as O1, MAG and MOBP, were low or undetected in NPs. In addition to NESTIN and A2B5 expression, greater than 80% of GPs in all three lines acquired NG2 and PDGFR-α expression (Fig. 3F). Moreover, some GPs also began expressing O4, RIP and O1 while MAG and MOBP expression was not detected.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Indirect immunofluorescence detection of neural markers help define the initial stages of oligodendrocyte differentiation from undifferentiated iPS cells, into neural (NP) and glial (GP) progenitors.

(A-D) Photos are of cells derived from BC1 with the differentiated stage (underlined) and marker analyzed as indicated similar staining patterns were seen in all three lines. Nuclei were stained using DAPI (blue). Most neural progenitors (NPs) were (A) NESTIN+ and (B) A2B5+ while the majority of glial progenitors (GPs) began expressing early oligodendrocyte progenitor (OP) markers (C) platelet derived growth factor receptor-α (PDGFR-α) and (D) NG2. Lower panel shows quantification of the indirect immunofluorescence analyses calculating the percent of labeled cells in all markers. (E) 90% of all cells from each line expressed early markers of the neural lineage including NESTIN and A2B5. (F) Likewise the number of glial progenitors expressing markers of the early OP lineage such as PDGFR-α and NG2 were also similar among lines. Differentiation of each cell line was performed in triplicate and the percent of positively stained cells was determined from 3 randomly, chosen 10X fields per 24-well cell culture dish from all three replicates. Values in the graph represent mean ± SEM. * and † denote statistical significance compared to BC1 and MR31, respectively by ANOVA and Tukey post hoc tests (N = 9, P < 0.001). Scale bars: 100 μm.

https://doi.org/10.1371/journal.pone.0116933.g003

Differentiation into OPs was confirmed by proteins expressed in mid- to late-OP differentiation including RIP, O4, and O1 with downregulation of A2B5 and NESTIN staining. O1 is a marker of late OP differentiation, while O4 is expressed earlier [19]. These proteins were expressed in the majority of OPs derived from all three iPS cell lines (Fig. 4A-E). In addition to markers of early OP differentiation, iPS-OPs also expressed detectable levels of proteins associated with myelin and are used to identify mature oligodendrocytes that are capable of generating myelin, such as the MAG and MOBP proteins (Fig. 4F and 4G, respectively). Quantitative analyses also showed that the number of cells expressing O1 was higher than the number of OPs expressing O4 from the BC1 line (episomal, non-integrating plasmid delivery or iPS factors), which could indicate that BC1 derived iPS-OPs are more efficiently differentiated. In contrast, the A1-4 and MR31 lines had higher levels of O4, a marker of less differentiated OPs compared to O1. These proteins were not detected in the NPs or GPs (Fig. 3E). Of note, the neuronal marker, Neuronal Class III βTubulin (TUJ-1) and the astrocyte marker, glial fibrillary acidic protein (GFAP), were expressed in less than 1% of the cell population in all three iPS cells (Fig. 4H and 4I, respectively).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Indirect immunofluorescence marker characterization of oligodendrocyte progenitors derived from human iPS cells.

Nuclei were stained with DAPI (blue). Results show reduced expression of early markers (A) A2B5 and (B) NG2 and increased protein expression of mid to late oligodendrocyte progenitor (OP) differentiation markers (C) O4 (D) RIP (E) O1 (F) myelin-associated glycoprotein (MAG) and (G) myelin-associated oligodendrocytic basic protein (MOBP). Myelin basic protein (MBP) expression was not detectable by immunostaining. 1% or less of OP cells expressed markers of other neural lineages as denoted by (H) the neuronal maker neuron-specific class III beta-tubulin (TUJ1) and (I) the astrocyte marker glial fibrillary acidic protein (GFAP). (J) Quantification of the indirect immunofluorescence analyses calculating the percent of cells expressing a particular marker is shown in the lower panel. A higher percent of cells derived from BC1 expressed later OP markers MAG, MOBP and O1 while fewer cells from the BC1 lines expressed early neural progenitor (NP) and glial progenitor (GP) markers and early OP marker O4 compared to MR31 and A1-4. Differentiation of each cell line was performed in triplicate and the percent of positively stained cells was determined from 3 randomly, chosen 10X fields per 24-well cell culture dish from all three replicates. Values in the graphs represent mean ± SEM. * and † denote statistical significance compared to BC1 and MR31, respectively by ANOVA and Tukey post hoc tests (N = 9, P < 0.001). Scale bars: 100 μm.

https://doi.org/10.1371/journal.pone.0116933.g004

The more differentiated state of the majority of BC1-derived OPs is further supported by reduced NP marker expression such as NESTIN and A2B5, which were expressed by nearly twice as many cells in OP induction media (~60%) in the A1-4 and MR31 lines compared to BC1 OP cultures (~30%). Furthermore, more cells from the BC1-derived OP cultures expressed myelin-associated proteins, MAG and MOBP, and at higher levels compared to MR31 iPS cells. For instance, MAG was expressed in more than 80% of the BC1 derived-OPs versus less than 20% in OPs derived from the MR31 line. Similar results were also shown for MOBP, which was expressed in more than 80% of the BC1 derived-OPs versus 30% of MR31-derived OPs. The number of the positively stained cells was divided by the total cell number to estimate the percent of cells that expressed a marker. Each experiment was replicated at least 3 times from 3 independent differentiation experiments. Despite expression of MAG and MOBP proteins in the iPS-OPs, MBP immunostaining was not detected in any of the iPS-OPs lines. This funding corroborates what we have previously shown for human ES-derived OPs, where MBP expression was found exclusively in nonproliferating, terminally-differentiated oligodendrocytes with the addition of T3 to the cell culture media [24].

OPs Survive and Differentiate after Transplantation

iPS-derived OPs were transplanted into rat spinal cords 24 hours after a moderate contusive SCI. The BC1 line was used for the in vivo assessments because this line expressed the greatest amount of late-OP markers compared to the other lines tested. Immunohistological analysis of spinal cords two months after cell transplantation showed human iPS-OP survival and integration into the spinal cord without tumor formation (Fig. 5). The site of cell injection can be easily identified as the regions of clustered, transplanted cells in areas with damage to the endogenous tissue. Human cells were confirmed by human anti-nuclei antibody (HNA), which is not detected in spinal cords injected with heat-killed iPS-OPs (Fig. 5A). In contrast, patches of cells that co-express HNA (green labeled) and MBP (red labeled) were detected at the injection site encompassing the injury epicenter in spinal cords transplanted with iPS-OPs (Fig. 5B). HNA staining indicates survival of the transplanted iPS-OPs while co-localization of HNA and MBP protein confirms OP differentiation of these cells into mature oligodendrocytes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Immunohistological characterization in rat spinal cords at the epicenter of injury (asterisks) following transplantation of either (A) heat-killed iPS-OPs or (B) live iPS-OPs, 24 hours after injury.

Human specific anti-nuclei antibody (HNA) staining showed that cells transplanted at the site of injury could survive, differentiate and express myelin basic protein (MBP, red) in iPS-OPs group, compared to the heat-killed iPS-OPs group. (C) Heat-killed iPS-OPs treated spinal cords demonstrated more cavitation and higher glial fibrillary acidic protein (GFAP) (green) expression than (D) spinal cords treated with iPS-OPs. Heat-killed iPS-OPs also expressed less MBP, a mature oligodendrocyte marker, than live iPS-OPs (red). (E) Double immunostaining with CD68 (expressed by macrophages, microglial, and T cells) and MBP shows more CD68+/MBP+ cells in control animals with morphology indicative of active macrophages or microglial engulfment of OPs. (F) Spinal cords injected with iPS-OPs had significantly fewer CD68+/MBP+ cells. Tissues shown here were isolated approximately 2 months after transplantation. DAPI was used to stain rat and human nuclei (blue). Scale bars for A, C, D: 100 μm and B, E, F: 30 μm. (G) Quantitative image analysis shows that more than 70% of HNA+ cells were differentiated MBP+ oligodendrocytes. Only 4.4 ± 1.5% of grafted cells express glial fibrillary acidic protein (GFAP), while no HNA+ cells were detected with co-localized staining for neuronal marker neuron-specific class III beta-tubulin (TUJ1) or NESTIN. Values represent mean ± SEM. * denote statistical significance of human MBP+ oligodendrocytes and MBP- OPs compared to either human-derived neural progenitors (Nestin+), neurons (TUJ1+) or astrocytes (GFAP+) and † denotes statistical significance of human HNA+MBP+ compared to HNA+MBP- OPs by Tukey post hoc tests. Tissue slices encompassing the injury epicenter were stained in intervals for each antibody combination from 4 rats in each treatment group, P < 0.001). (H) Optical density quantification of GFAP staining, which shows significantly higher intensity of GFAP staining in heat-killed iPS-OP compared with iPS-OPs of both BC1 and MR31 lines. (I) Quantification of the number of CD68+ cells shows significantly higher CD68+ cells in heat-killed iPS-OPs, indicative of increased inflammation and macrophage activation, compared with iPS-OPs of both BC1 and MR31 lines. Values in the graphs represent mean ± SEM. * and † denote statistical significance compared to heat-killed iPS-OPs and BC1-OPs, respectively by Tukey post hoc tests (10 identically numbered tissue slices encompassing the injury epicenter from 4 rats were analyzed for each treatment group, P < 0.001).

https://doi.org/10.1371/journal.pone.0116933.g005

Next, we examined the extent of glial scarring by studying the number GFAP+ astrocytes and the level of staining around the site of injury. Results show increased migration of GFAP+ cells to the site of injury and higher intensity of GFAP staining surrounding cavities in heat-killed iPS-OP controls (Fig. 5C) compared to iPS-OP treated spinal cords (Fig. 5D). Immune cells have also been demonstrated in the spinal cords of immunocompromised rats that engulf transplanted cells and show immunostaining for OP markers [49,50]. To investigate this possibility, we dual-stained sections with MBP and CD68, a marker of multiple types of immune cells activated in the damaged spinal cord including microglia, macrophages and T cells. Under normal conditions these cells express basal levels of CD68, which then become elevated when they are activated [51]. Two months after transplantation, results showed higher numbers of CD68+ cells in regions treated with heat-killed iPS-OP controls (Fig. 5E) compared to live iPS-OPs. The presence of CD68+ cells in the spinal cord of control rats after cyclosporine is consistent with others that show that the same concentration of cyclosporine reduces but not eliminate CD68+ related activation in this tissue [52]. In Fig. 5F, expression of MBP was seen in the region of transplantation while only a few CD68 expressing cells could be seen in the same region of the iPS-OP transplants. Thus, it seems that the transplanted human cells expressing MBP were not being attacked by the host immune cells. This result may be due in part to the treatment of cyclosporine used in these rats to inhibit graft rejection. It is interesting that the same mechanism involved in cyclosporine treatment that prevents the rejection of the human cell transplants does not prevent the engulfment of endogenous rat oligodendrocytes, which occurs in the control groups. We observed a much larger number of rounded, foamy-like microglial shaped cells that were CD68+ and MBP+ in the control groups injected with heat-killed cells (Fig. 5E). These results suggest that human iPS-OPs may facilitate specific anti-microglial immune responses in the rat spinal cord after injury.

Quantitation analysis was also performed on a series of dual staining experiments from cryosections of spinal cords. HNA+ staining was compared with a variety of markers to identify the differentiated state of the injected iPS-OP BC1 line (Fig. 5G). Staining was also performed with iPS-OPs derived from the other 2 lines with similar outcomes (data not shown). These results showed that HNA+ cells that co-labeled with MBP antibody comprised 71.2 ± 10.19% of the total number of injected HNA+ cells found in the dorsal column. The remaining HNA- MBP positive cells suggest that endogenous rat OPs also infiltrate the injured region. Some reports have suggested that pluripotent stem cell-derived OPs can differentiate into astrocytes [53]. Therefore, the expression of the astrocyte marker GFAP was also studied in these sections. Cells that co-labeled with human nuclei and an astrocyte marker GFAP comprised 4.4 ± 1.5% of the total number of human nuclei+ cells. The small percentage of GFAP staining in the HNA+ cells affirms the robustness of this protocol for differentiating stem cells into OPs. On the other hand, only 0.4 ± 0.1% of transplanted cells co-expressed HNA and the neuronal marker TUJ-1, and there was no co-expression of HNA and the early neural marker (NESTIN) detected in our transplanted cells. Therefore, the OPs derived in culture were not a more primitive progenitor, which could generate a large number of neurons or astrocytes after transplantation. Importantly, these results show that a sufficient number of human iPS-derived OPs survive at least two months after being transplanted 24 hours post-SCI. These unprecedented results show the valuable potential of OPs for early interventions to treat SCI, which may enhance future outcomes of SCI patients by reducing glial scarring and dampening long-acting or overactive immune responses that are known to hamper recovery.

OPs Reduce Cavitation after SCI

To study cavitation and morphological changes at the injury epicenter following cell transplantation, longitudinal sections of injured spinal cords were analyzed by H&E staining (Fig. 6A-B) and electron microscopy. Cavity formation within the gray and white matter surrounding the epicenter of injury are normally observed in untreated spinal cords following a moderate contusive injury. However, two months after implantation, the cavities manifest differently in spinal cords transplanted with control heat-killed iPS-OPs compared to those transplanted with the BC1 iPS-derived OPs. The extent of cavitation was 5-fold larger and spread further from the epicenter in controls (Fig. 6A) compared to iPS-treated spinal cords (Fig. 6B). Longitudinal sections of the cords were also stained with Luxol fast blue (LFB) to analyze the extent of myelination after treatment in the areas surrounding the cavitation (Fig. 6C-D). As expected in all spinal cords, the intensity of LFB staining is greater in the white matter compared to the grey matter due to the presence of myelinated tracts. Spinal cords injected with heat-killed controls (Fig. 6C) showed decreased staining compared to those injected with live BC1 iPS-derived OPs in the white matter tracts (Fig. 6D). This finding suggests remyelination is occurring as a result of the injected cells. In addition, a greater number of cells with dark blue cytoplasmic staining could be seen in the control group compared to the treated group (indicated by arrows, Fig. 6C), which is representative of increased demyelination and macrophage phagocytosis of disintegrated myelin in the control group [54]. This raises the possibility that iPS-OPs may provide immunoprotection to the tissue similar to that observed after MSC and neural stem cell transplantation [55]. Consistent with these findings, the heat-killed control group demonstrated a significantly higher number of cells with dark cytoplasmic LFB staining compared to OPs-treated cords as shown in Fig. 6C-D.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Hematoxylin and eosin and luxol fast blue stained paraffin sections of spinal cords from animals subjected to moderate contusive SCI and treated with (A) Heat-killed iPS-OPs or (B) Live, intact iPS-OPs derived from the BC1 line.

As expected for a 12.5 mm contusive injury, cavity formation was seen in both groups; however, the cavitation in spinal cords treated with iPS-OPs did not form well defined glial-like scar structures around the cavity periphery, compared to the control group. Thus, controls showed more pronounced cavitation, which extended further from the epicenter along the central canal. This indicates that the transplantation of live iPS-OPs into the newly formed cavity 24 hours after injury limited the normal expected progression of this level of contusive injury. The asterisks indicate the center of the injury site. Yellow boxes in A and B correspond to adjacent tissue sections stained in C and D, respectively. In these areas immediately surrounding the center of injury, staining with Luxol Fast Blue (LFB) to identify myelin showed reduced staining in the control group (C) versus the iPS-OP transplanted spinal cords (D). Thus, higher levels of LFB in the iPS-OP treated rats may indicate the presence of greater amounts of myelin. The representative spinal cords shown in (A-D) were harvested approximately 2 months after transplantation. Dotted lines help delineate regions of white matter (W) and grey matter (G). (E) Area measurements across groups revealed nearly a 5-fold significant reduction in cavity size in the iPS-OP treated groups compared to the heat-killed controls. Optical density of LBF, normalized to the optical density of the sham control group, was calculated for the grey matter (F) and white matter (G) of spinal sections to quantify the amount of myelin present in each area of the cord. Values in the graphs represent mean ± SEM. * denote statistical significance compared to heat-killed iPS-OPs by Tukey post hoc tests. For each stain, 5 identically numbered tissue slices encompassing the injury epicenter from 4 rats were analyzed for each treatment group, P < 0.001.

https://doi.org/10.1371/journal.pone.0116933.g006

OPs Remyelinate Axons after SCI

Increased levels of myelin were demonstrated in rats treated with iPS-derived OPs using transmission electron microscopy (TEM) (Fig. 7A-C). The spinal cord injected with control heat-killed iPS-OPs showed no evidence of myelination with multiple bare axons or severe disruption of myelin sheaths encompassing the injury site. On the other hand, spinal cords injected with BC1-derived iPS-OPs showed evidence of remyelination, where the myelin sheaths are much thinner with loosely arranged lamellae compared to normal myelination detected in an the uninjured spinal cord. Quantification of myelination indicated that the number of remyelinated axons was significantly higher in the iPS-OP transplanted group compared with heat-killed iPS-OP transplanted group (Fig. 7D). iPS-OP transplanted animals also demonstrated significantly less demyelinated axons compared with heat-killed iPS-OP transplanted animals. Notably, heat-killed iPS-OPs-transplanted animals contained an average of 60 oligodendrocyte remyelinated axons per mm² whereas iPS-OP transplanted animals contained an average of 2760 oligodendrocyte remyelinated axons per mm², indicating a 46-fold increase in remyelination in iPS-OP transplanted animals over the heat-killed controls. In addition, more normally myelinated axons were found at the injury epicenter in iPS-OP transplanted spinal cords whereas none were visible in controls.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Transmission electron microscopy of transverse sections of the spinal cord showed remyelination of spared axons by BC1-derived iPS-OP cells after contusive SCI in rats.

(A) Normal, endogenous myelination of an uninjured rat spinal cord (laminectomy sham control) as indicated by white arrows, showing thick and tightly-packed myelin sheaths. (B) Remyelination observed in rats transplanted with BC1-derived OPs, approximately two months after transplantation. These are sites where oligodendrocytes are detected, as shown in Fig. 4. The relatively thinner and loosely packed myelin sheaths may be compared with the normal condition in (A), suggesting that this is remyelination by human OPs. (C) Multiple nude axons are visible in rats injected with heat-killed iPS-OPs, with no observable remyelination, approximately two months after transplant. (D) Quantification shows that transplantation of BC1-derived OPs resulted in a significant increase in the number of remyelinated axons and reduced demyelinated of axons compared with the heat-killed iPS-OP group. Values in the graphs represent mean ± SEM. * denote statistical significance compared to heat-killed iPS-OPs by Tukey post hoc tests. Ten images encompassing the injury epicenter from 4 rats were analyzed for each treatment group (P < 0.001). Magnification (A-C): 10000X.

https://doi.org/10.1371/journal.pone.0116933.g007

Motor Behaviorial Analysis

BBB analyses demonstrated in interesting trend in BBB scores among groups (Fig. 8). For instance, motor recovery was initially greater after transplantation in rats either injected with live or heat-killed iPS-OPs compared to the control group. However, this effect diminished after the first week and throughout the first month of recovery during which time rats injected with PBS alone showed higher scores. This trend was followed by rats injected with iPS-OPs which showed a significant increase in BBB scores 5 weeks after injection. BBB scores from rats injected with A1-4 derived OPs showed results similar to MR31 (data not shown). Rats injected with heat-killed iPS-OPs were similar to rats injected with PBS at the end of the study.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Behavioral assessment 8 weeks following moderate contusive injury (12.5mm) and cell transplantation in rats.

The average of the Basso, Beattie, and Bresnahan (BBB) Locomotor Rating Scale (BBB SCORE) for left and right limb movement was evaluated for 55 days following spinal cord injury in rats injected with either PBS (blue), 500,000 heat-killed iPS-OP cells (green), live BC1-OP (black), or MR31-OP (red) cells 24 hrs after injury. Values in the graphs represent mean ± SEM of both left and right limbs as there were no significant differences in response between right and left limb scores. Statistical significance is denoted by a colored asterisk for each cell line with matched color compared to the PBS control group determined by one-way ANOVA followed by a two-way Student’s t-test assuming equal variances (P < 0.025).

https://doi.org/10.1371/journal.pone.0116933.g008