Abstract
Muscarinic receptor antagonists act as potent inducers of oligodendrocyte differentiation and accelerate remyelination. However, the use of muscarinic antagonists in the clinic is limited by poor understanding of the operant receptor subtype, and questions regarding possible species differences between rodents and humans. Based on high selective expression in human oligodendrocyte progenitor cells (OPCs), we hypothesized that M3R is the functionally relevant receptor. Lentiviral M3R knockdown in human primary CD140a/PDGFαR+ OPCs resulted in enhanced differentiation in vitro and substantially reduced the calcium response following muscarinic agonist treatment. Importantly, following transplantation in hypomyelinating shiverer/rag2 mice, M3R knockdown improved remyelination by human OPCs. Furthermore, conditional M3R ablation in adult NG2-expressing OPCs increased oligodendrocyte differentiation and led to improved spontaneous remyelination in mice. Together, we demonstrate that M3R receptor mediates muscarinic signaling in human OPCs that act to delay differentiation and remyelination, suggesting that M3 receptors are viable targets for human demyelinating disease.
SIGNIFICANCE STATEMENT The identification of drug targets aimed at improving remyelination in patients with demyelination disease is a key step in development of effective regenerative therapies to treat diseases, such as multiple sclerosis. Muscarinic receptor antagonists have been identified as effective potentiators of remyelination, but the receptor subtypes that mediate these receptors are unclear. In this study, we show that genetic M3R ablation in both mouse and human cells results in improved remyelination and is mediated by acceleration of oligodendrocyte commitment from oligodendrocyte progenitor cells. Therefore, M3R represents an attractive target for induced remyelination in human disease.
Introduction
Regenerative therapies that promote remyelination in multiple sclerosis (MS) have the potential to restore lost neurological function and prevent chronic demyelination-associated disease progression. To this end, both enhancing endogenous repair and stem/progenitor cell transplantation have been proposed (Franklin and Goldman, 2015). As quiescent oligodendrocyte progenitor cells (OPCs) are observed in chronic demyelinated lesions in MS (Wolswijk, 1998; Kuhlmann et al., 2008), induction of differentiation is considered a reasonable pharmacologic approach (Huang et al., 2011a). Similarly, while transplanted human OPCs (hOPCs) have the capacity to myelinate large areas of the hypomyelinated rodent brain, they do so very slowly over several months (Windrem et al., 2008; Wang et al., 2013). Their slow rate of repair likely contributes to the fact that only ∼25% of transplant recipients benefit from such treatment. Thus, accelerated differentiation would likely improve the outcome of cell therapy. Indeed, as remyelination acts to protect the demyelinated axons (Irvine and Blakemore, 2008), approaches that accelerate remyelination will yield additional functional benefits by acting to prevent the progressive neurodegeneration that is associated with chronic demyelination (Trapp and Nave, 2008).
Nonselective muscarinic receptor (MR) antagonists have been shown to improve remyelination in rodents (Deshmukh et al., 2013; Mei et al., 2014; Najm et al., 2015). Clemastine, a nonselective MR and H1 histamine receptor antagonist, substantially induced OPC differentiation and remyelination following demyelination in adult spinal cord (Mei et al., 2014). However, clemastine also has potent activity on several other GPCRs, including σ1/2 (Gregori-Puigjané et al., 2012) and P2X7 (Nörenberg et al., 2011). Likewise, benztropine, a nonselective dopaminergic and MR antagonist, improved remyelination in the cuprizone model of demyelination (Deshmukh et al., 2013). Therefore, for success of future therapy, it is critical to determine which MR subtype or subtypes signal to block oligodendrocyte differentiation and inhibit repair.
Recently, we found that solifenacin, an M1/3R selective antagonist, induced oligodendrocyte differentiation by transplanted hOPCs in a mouse model of hypomyelination (Abiraman et al., 2015). While rat OPCs express all five MRs (Ragheb et al., 2001; De Angelis et al., 2012), in this study, we found that hOPCs only expressed M1R and M3R. Previous data have demonstrated high M3R expression in OPCs (Zhang et al., 2014; Abiraman et al., 2015) with little or no expression among other glial cell types (Zhang et al., 2014). Thus, we hypothesized that M3R signaling in OPCs acts to delay differentiation and, if specifically targeted in OPCs, could induce remyelination. We used genetic approaches in human and mouse OPCs to test this hypothesis. In primary human CD140a/PDGFαR+ OPCs, lentiviral (LV) knockdown (KD) of M3R induced differentiation to postmitotic oligodendrocytes in vitro and resulted in precocious myelin formation following transplantation into hypomyelinating shiverer/rag2 mice. Furthermore, we found that conditional deletion of M3R in adult NG2-expressing OPCs induced oligodendrocyte differentiation following focal demyelination in mice, likewise leading to improved remyelination.
Materials and Methods
Human CD140a/PDGFαR cell preparation.
Fetal brain tissue samples, between 17 and 22 weeks of gestational age, were obtained from Advanced Bioscience Resources with consent from patients under protocols approved by the University at Buffalo Research Subjects Institutional Review Board. Forebrain samples were minced and dissociated using papain and DNase as previously described (Conway et al., 2012). Magnetic sorting of CD140a/PDGFαR+ cells was performed as described previously (Pol et al., 2013).
RNA sequencing.
RNA was extracted from 4 × 105 human CD140a+ OPCs following isolation using EZNA Total RNA Kit I (Omega Bio-Tek) according to the manufacturer's protocols. RNA sequencing was performed at the UB sequencing core on an Illumina HiSeq2500 using 100 cycle paired-end sequencing. Sequences were aligned to the UCSC Hg19 mouse genome using tophat (version 2.1.1), and counts per gene determined using htseq (version 0.6.1). R/Bioconductor was used for subsequent analysis. Following loading of read counts using DESeq2, fragments per kilobase per million mapped fragments were calculated. hOPCs isolated from three separate fetal samples between 20 and 21 weeks gestation age were used.
Single-cell RNA-seq was performed on individual CD140a+O4+ OPCs isolated by FACS (Abiraman et al., 2015) using the Fluidigm C1 system per the manufacturer's instructions. To achieve detection of low expressed transcripts, each cell-specific library was sequenced to achieve a depth of 106 transcripts/cell. Single-cell capture was ensured by visual inspection, before lysis and PCR amplification using SMARTER Ultra low RNA kit (Clontech). Quality and quantity of single-cell cDNA were determined by Bioanalyzer HS DNA chip (Agilent Technologies) and PicoGreen assay (Invitrogen). Sequencing libraries were prepared using Nextera XT DNA kit (Illumina), barcoded and sequenced on a HiSeq2500 Sequencer using 100 cycle paired-end sequencing. Bioinformatics was performed as described above.
shRNAi LV preparation.
TRC shRNAi plasmids targeting CHRM3/M3R mRNA (TRCN0000011259 and TRCN0000011260) or GFP (TRCN0000072185) were obtained in a pLKO.1 LV vector (Open Biosystems, RHS4533). Lentiviruses were prepared as previously described (Sim et al., 2006). Briefly, following triple transfection of HEK 293T cells with pLKO.1 and packaging plasmids pLP/VSVG (Invitrogen) and psPAX2 (AddGene), viral supernatant was collected at 48 and 72 h, and concentrated by ultracentrifugation. Viral titer was determined by counting colony-forming units following viral infection and selection of HEK 293T cells with 1 μg/ml puromycin (AMRESCO). KD efficiency was determined in M3R-expressing CHO cells by Western blot, using rabbit IgG anti-M3R (1:2000, Millipore catalog #AB9018-50UL, RRID:AB_2080197), and in hOPCs by qRT-PCR, as detailed below.
Infection and puromycin selection.
Following MACS isolation, cells were plated at 5 × 104 cells/ml in 24-well plates coated with poly-L-ornithine and laminin, in serum-free media (SFM) (Sim et al., 2011) supplemented with 20 ng/ml PDGF-AA (PeproTech) and 5 ng/ml NT-3 (PeproTech). After 24 h, cells were supplemented with 5 ng/ml FGF-2 and infected with M3R KD or control (TRC GFP shRNAi) lentivirus. The GFP shRNAi was selected as an appropriate control as it represents a validated nontargeting control siRNA that is matched for GC content with M3R shRNAi (33%–43%). At 24 h after infection, media was replaced with SFM supplemented with 20 ng/ml PDGF-AA and 5 ng/ml NT-3, and cells were allowed to expand for 48 h. Following expansion, cell cultures were additionally supplemented with 400 ng/ml puromycin and allowed to reach ∼80% confluence before being passaged for differentiation analysis, calcium imaging, or transplantation into mice.
Immunocytochemistry.
Puromycin-selected OPCs infected with either M3R KD or control lentivirus were seeded at 3 × 104 cells/ml in 24-well plates in SFM supplemented with 20 ng/ml PDGF-AA and 5 ng/ml NT-3. Immunocytochemical analysis was performed as described previously (Conway et al., 2012). Live staining using mouse anti-O4 antibody (1:25) was followed, after fixation, by OLIG2 (1:1000, Millipore catalog #AB9610, RRID:AB_570666). Alexa-488-, Alexa-594-, and Alexa-647-conjugated secondary antibodies (Invitrogen) were used at 1:500 dilutions. Differentiation was assessed as the percentage of O4+OLIG2+ cell of all OLIG2+ cells in culture. Morphological analysis of O4-expressing oligodendrocytes was performed based on criteria described by Huang et al. (2011b). Statistical significance was determined using one-way repeated-measures ANOVA with Tukey's multiple-comparison test (GraphPad Prism 6.0e).
qRT-PCR analyses.
For qPCR, RNA was obtained following puromycin selection for in vitro studies or immediately before transplantation into mice. cDNA was synthesized using SuperScript III reverse transcriptase (Invitrogen). Human-specific primers for SYBR Green-based PCR were described previously (Abiraman et al., 2015). Samples were run in triplicate and gene expression calculated by ΔΔCt analysis using 18S as a reference. Statistical significance was tested on log2-transformed data using ANOVA.
Cloning of LV-EF1a:GCaMP6s.
We PCR/TOPO cloned the coding region of GCaMP6s from pLP-CMV-GCaMP6s-CAAX (AddGene, #52228) (Tsai et al., 2014) into pCR2.1 TOPO4 plasmid (Invitrogen). The GCaMP6s fragment was then subcloned using unique 5′ SpeI and 3′ PspXI restriction sites into LV pTRIP-EF1a (derived from pTRIP-EF1a; gift of Abdel Benraiss, University of Rochester) (Sevin et al., 2006). Lentiviruses were prepared as described above. Titration of virus was performed on matched mCherry-expressing virus using flow cytometry for mCherry fluorescence and directly compared with GCaMP6s virus using qRT-PCR for WPRE sequence (Geraerts et al., 2006). GCaMP6s expression was confirmed using CHO cells stably expressing the human M3R (Dörje et al., 1991) by fluorescence imaging following addition of carbachol or CaCl2. For all imaging experiments, hOPCs were infected at one multiplicity of infection for 24 h followed by complete media replacement. One hour before imaging, media was replaced with phenol red-free media.
Calcium imaging using LV-EF1a:GCaMP6s.
All calcium imaging experiments were performed at 10× using an Olympus IX51 with Prior XYZ stage equipped with a Hamamatsu ORCA-ER camera using a 1× TV lens. All phase images and fluorescent time-lapse acquisitions were performed at room temperature and captured using μManager (Edelstein et al., 2010). Oxotremorine-M (Tocris Bioscience) was thawed immediately before each experiment. Two fields per preparation were imaged for each condition, at 2 s intervals for 10–12 min. Drug addition occurred 1 min after the start of imaging. Post hoc immunofluorescence confirmed that 88 ± 1.3% of cells imaged were Olig2+O4−.
Phase images were used to generate ROIs corresponding to the soma of every cell by thresholding and supervised analysis. Rolling ball subtraction was performed on fluorescence image frames and mean pixel intensity calculated for each cell using ImageJ. Baseline cellular GCaMP6s fluorescence was determined immediately before drug addition on a per-cell basis. Analysis of calcium wave characteristics, such as amplitude, peak number, frequency, and subsequent statistics, was performed in R (complete analysis code is available on request). Briefly, calcium response curves were loess fitted (zoo) (Zeileis and Grothendieck, 2005) and the local minima and maxima calculated. Local maxima was considered a peak if its amplitude increased >35% from its local minima. Response duration was measured from the onset of the first peak to the end of the last peak. Area under the curve was calculated for the duration of the response. For each parameter, a linear model was used for two-way ANOVA with Tukey HSD analysis, fitted using virus (control or M3R KD), oxotremorine dose, and the interaction of these variables as predictors, as well as the source human sample to consider individual tissue sample variability.
Transplantation into shiverer/rag2 mice.
All animal experiments were performed according to protocols approved by the University at Buffalo Institutional Animal Care and Use Committee. Three fetal brain samples between 17 and 20 weeks gestational age were used for transplantation experiments. Cell preparation and transplantation were performed as described previously (Abiraman et al., 2015). Briefly, hOPCs infected with M3R KD or control lentivirus were seeded at ∼5 × 104 cells/ml in SFM supplemented with 20 ng/ml PDGF-AA and 5 ng/ml FGF-2, allowed to expand for 48 h, then cultured in SFM supplemented with 20 ng/ml PDGF-AA and 5 ng/ml NT-3 until confluent. Postnatal day 2–3 shiverer/rag2 mice were anesthetized using hypothermia and 5 × 104 cells were injected per site, bilaterally at a depth of 1.1 mm into the rostral corpus callosum (∼2.5 mm posterior to bregma in day 2–3 pups).
Analysis of human myelination in vivo.
Animals were killed at 12 weeks after transplant and perfused with 0.9% saline followed by 4% PFA. Cryopreserved coronal sections of mouse forebrain (16 μm) were cut and sampled every 160 μm. Immunohistochemistry was performed, as described previously (Sim et al., 2011). Primary antibodies used were as follows: mouse IgG1 anti-human nuclear antigen (hNA) (1:400, Millipore catalog #MAB1281, RRID:AB_94090), rat IgG2a anti-MBP (1:200, Abcam catalog #ab7349, RRID:AB_305869), mouse IgG2b anti-CC1 (1:50, Millipore catalog #OP80, RRID:AB_2057371), mouse IgG1 anti-human specific GFAP (1:800, Covance Research Products catalog #SMI-21R-500, RRID:AB_509979), and mouse IgG1 anti-neurofilament (1:800, 1:1 mix of Covance Research Products catalog #SMI-311, RRID:AB_2315332 and catalog #SMI-312, RRID:AB_2314902). Goat secondary antibodies were used as described above. Images of entire coronal sections were captured under matching conditions using the 10× objective of an Olympus IX51 with Prior XYZ motorized stage and stitched using Fiji (Schindelin et al., 2012). For each animal, the best cross-sectional region of myelination was determined by blinded assessment of rostral brain coronal sections every 160 μm. Quantification of myelination and differentiation was performed within these areas in each animal. To determine myelination, the portions of MBP+-immunoreactive area (μm2) and the number of hNA+ cells were quantified across the corpus callosum. To assess human CC1+ oligodendrocyte differentiation, CC1+/hNA+ cells were quantified across the corpus callosum. MBP ensheathment of host axons was assessed in the corpus callosum as previously described (Wang et al., 2014). Briefly, a 2 μm stack of 20 optical sections every 0.1 μm was taken, and the proportion of ensheathed axons that crossed 3 perpendicular sampling lines placed randomly over each image was counted (Zeiss LSM 510 Meta confocal).
M3R conditional knock-out animals and lysolecithin-induced demyelination.
Young adult (8- to 11-week-old) female Balb/C mice were purchased from Envigo. NG2creER:Rosa26YFP animals were a gift from Akiko Nishiyama (University of Connecticut, Storrs, CT) (Zhu et al., 2011). The generation of floxed M3R mice (M3Rfl/fl) has been described previously (Gautam et al., 2006). Animals were bred to generate NG2CreER offspring homozygous for the M3R floxed allele (NG2creER:Rosa26YFP:M3Rfl/fl). Conditional gene knock-out by cre-mediated recombination in NG2-positive OPCs was achieved by daily intraperitoneal administration of tamoxifen (200 mg/kg, Sigma-Aldrich) for 5 d, the last of which occurred 7 d before surgery.
Focal demyelination of the young adult (8- to 11-week-old) mouse spinal cord was induced as previously described (Zhao et al., 2006). Briefly, animals were anesthetized under isoflurane (5% for induction, 1.5% for maintenance), and 0.5 μl 1% lysolecithin (Lα-lysophosphatidylcholine, Sigma-Aldrich) was directly injected into the dorsal and ventral funiculi of the spinal cord between two thoracolumbar vertebrae. Preoperative and postoperative analgesia was provided by subcutaneous injection of buprenorphine (0.1 mg/kg).
Spinal cord processing and analysis by immunohistochemistry and in situ hybridization.
Animals were killed at 14 d postlesion (dpl) by transcardial perfusion of saline followed by 4% PFA under deep anesthesia. The spinal column was extracted, postfixed for 30 min in 4% PFA, the spinal cord isolated, cryoprotected in a sucrose gradient (6% sucrose overnight, followed by 15% sucrose overnight), and frozen in OCT medium (Tissue-Tek). Serial 16-μm-thick transverse sections were cut using a Leica cryostat and stored at −80°C.
To identify lesions, one slide from each series was dehydrated in an ethanol gradient, rehydrated, and stained for 15 min in a solution containing 0.2% (w/v) Eriochrome cyanine (Sigma-Aldrich), 5.6% (w/v) FeCl3·6H2O (Fisher Scientific), and 0.5% (v/v) H2SO4. Slides were differentiated in 5.6% (w/v) FeCl3·6H2O until proper depth of color was achieved, dehydrated by sequential ethanol gradient followed by xylene, and mounted in OpticMount I (Mercedes Medical). Slides immediately adjacent to the lesion centers were used for all immunohistochemical and in situ hybridization procedures. Additional primary antibodies were as follows: mouse anti-Gfap (1:300, Millipore catalog #MAB360, RRID:AB_11212597), rabbit anti-Iba1 (1:300, Wako catalog #019-19741, RRID:AB_839504), and rabbit anti-Ng2 (1:200, Millipore catalog #AB5320, RRID:AB_91789). In situ hybridization, using digoxigenin-labeled Plp1 antisense RNA probe (Fancy et al., 2004), was performed as described by Sim et al. (2002). The density of positive labeled cells was calculated at 20× magnification following imaging and montaging of entire spinal cord sections. White matter lesions with cross-sectional area <10,000 μm2 were excluded from the analysis, as were lesions that extended into the gray matter. Analysis was performed by an investigator blinded to the identity of the sections.
Electron microscopy.
Mice were perfused with 4% glutaraldehyde and spinal cords removed (n = 5 per group); 1-mm-thick blocks surrounding the spinal cord lesion were processed through osmium tetroxide, dehydrated through ascending ethanol washes, and embedded in TAAB resin (TAAB Laboratories). The 1 μm sections were cut, stained with 1% toluidine blue (Sigma-Aldrich), and examined by light microscopy. Selected blocks were trimmed, ultrathin sections cut, and examined by electron microscopy (Hitachi, H600). Images were acquired at 2900× magnification. Analyses of remyelinated axons and g-ratios were performed blinded. For remyelination counts, a minimum of 300 axons were counted for each animal from at least six different fields, with 3 or 4 animals per treatment group. Analysis of g-ratios represents 200 axons from each animal.
Experimental design and statistical analysis.
The following experimental design was followed for in vitro experiments assessing the effect of M3R KD on hOPC differentiation. Three experimental groups were tested (control, M3R A, and M3R B) at a single time point with immunocytochemical and RNA-based endpoints. As cultures were tested from matched individual human tissue samples, comparison between individual groups was performed by repeated measures one-way ANOVA followed by Dunnett's post hoc tests. Calcium imaging experiments consisted of two independent experimental variables: oxotremorine dose and virus (M3R KD vs control virus). As such, two-way ANOVA was used with Tukey HSD post tests. In animal experiments, two experimental groups were tested (control or M3R deletion, either shRNAi or conditional knock-out). Student's t tests were performed to compare distinct experimental endpoints. Group sizes can be found in the experimental corresponding Results section. Statistical analyses were performed in GraphPad Prism (version 6.0e), or for calcium imaging experiments using R (3.4.3).
Results
M3R expression in hOPCs delays onset of oligodendrocyte differentiation in vitro
We recently found that the M3R was the most differentially expressed G-protein coupled receptor in human CD140a+O4+ OPCs (Abiraman et al., 2015). To better determine the expression of MR subtypes, we performed RNA-seq analysis of CD140a+-sorted OPCs. Among the five MR subtypes, only M1R and M3R were expressed at detectable levels by human CD140a+ OPCs (3.1 ± 0.5 and 3.0 ± 0.4; mean ± SEM; fragments per kilobase per million mapped fragments, respectively, RNA-seq, n = 3 individual patient samples). To define whether M3R was expressed by all OPCs or in a heterogeneous pattern, we performed single-cell RNA-seq (Fluidigm C1). Intriguingly, among single CD140a+O4+ FACS-sorted OPCs, we detected M3R in 13 of 48 cells but could not detect M1R with this method. Based on these expression data, we focused on M3R function and hypothesized that M3R signaling acts to delay differentiation of human primary OPCs.
We used LV-mediated shRNAi KD to genetically ablate M3R expression and signaling in hOPCs. Following a screen of several shRNAi vectors in M3R-expressing CHO cells, we found that two distinct M3R shRNAi vectors, labeled A and B, significantly reduced M3R mRNA (CHRM3 qPCR, post hoc p = 0.002 and p = 0.01, respectively, n = 3) and protein (Fig. 1). Similarly, these two shRNAi vectors induced a significant KD of M3R in CD140a/PDGFαR+ hOPCs (36.0 ± 12.2% and 56.3 ± 8.9% KD for A and B, respectively, n = 4–6) relative to nontargeted control siRNA. Importantly, M1R expression was not affected following M3R KD (M1R expression ratio vs control infected cells was 0.83 ± 0.69 and 1.07 ± 0.19 following infection with shRNAi A and B, respectively; n = 3, p > 0.7, Dunnett's post hoc).
M3R KD induces human CD140a/PDGFαR+ OPC differentiation in vitro. A, Western blot confirmation of M3R KD by LV shRNAi in CHO-M3R cells. B–D, Fetal PDGFαR/CD140a+ hOPCs were infected with M3R (A, B) or control shRNAi LV, selected with puromycin and switched to mitogen-free media to encourage oligodendrocyte differentiation. Oligodendrocyte differentiation, determined by the proportion and morphological complexity of O4+ oligodendrocytes (green) and OLIG2+ cells (red), was increased in M3R KD hOPCs (B–D, quantified in E, mean ± SEM). MBP qRT-PCR, normalized to control OPCs in PDGF-AA, showed increase myelin protein gene expression following M3R KD (F, mean ± SEM). *p < 0.05 versus Ctrl (two-way ANOVA followed by Bonferroni post hoc test). ***p < 0.001 versus Ctrl (two-way ANOVA followed by Bonferroni post hoc test). There were 3 fetal samples (18- to 20-week gestational age). Scale bar, 20 μm.
Next, to determine whether M3R KD would influence hOPC differentiation in vitro, we infected freshly isolated CD140a+ fetal hOPCs with M3R KD or control virus, performed puromycin selection, and switched the resulting cultures to mitogen-free media to promote oligodendrocyte differentiation. After 4 d, the proportion of O4+ oligodendrocytes differentiating from M3R KD-infected hOPCs (33.9 ± 8.0% and 30.9 ± 10.9% for A and B, respectively) was significantly increased compared with nontargeted control cells (10.9 ± 5.0%, Tukey's post hoc adjusted p = 0.01 and 0.03, n = 5; Fig. 1B–E). In addition, we observed substantially more O4+ cells with a complex branched morphology (Fig. 1E), consistent with more advanced differentiation in M3R KD-transduced cells. In hOPCs cultured in the presence of PDGF-AA, M3R KD was not sufficient to induce oligodendrocyte differentiation. MBP mRNA was not significantly upregulated in either control or M3R KD groups unless cultured in differentiating conditions (Fig. 1F). However, whereas control cells exhibited a threefold increase in MBP in the absence of PDGF-AA, M3R KD cells upregulated MBP by >25-fold, indicating that M3R signaling acts to slow synthesis of myelin gene protein mRNA as well as morphological maturation.
M3R regulates the calcium response following muscarinic agonist treatment in hOPCs
We next sought to determine the contribution of M3R on downstream signaling in OPCs following MR activation. Muscarinic agonist treatment of hOPCs results in robust intracellular calcium release (Cohen and Almazan, 1994; Haak et al., 2001). To determine the contribution of M3R signaling in muscarinic-evoked Ca2+ responses, we tested whether M3R KD in hOPCs altered intracellular Ca2+ release using LV GCaMP6s (Chen et al., 2013). Upon muscarinic agonist treatment (oxotremorine-M; Oxo-M), GCaMP6s fluorescence rapidly increased in hOPCs (Fig. 2), indicative of an increase in intracellular Ca2+ concentration [Ca2+]i. The vast majority of responding cells (>85%) exhibited a distinctly oscillatory pattern of [Ca2+]i response lasting up to 5 min in duration (Fig. 2A,B). The proportion of responding hOPCs increased in a dose-dependent manner in both control and M3R KD cells (Fig. 2C). While we noted a downward shift in the dose–response curve for M3R KD cells, M3R KD did not affect the EC50 of Oxo-M (log(EC50) = −5.74 ± 0.20 vs −5.93 ± 0.18 for control and M3R KD, respectively; two fetal preparations, n > 30 cells per condition/dose). This is consistent with shRNAi-mediated reduction in M3R receptor expression leading to a reduction in signaling efficacy while not altering the ligand/receptor affinity.
M3R modulates muscarinic-evoked Ca2+ response in hOPCs. Fetal CD140a+ hOPCs were cultured and infected with intracellular Ca2+ reporter GCaMP6s and M3R or control shRNAi LV. Time-lapse microscopy of Ca2+ response following Oxo-M treatment was recorded and analyzed. A, Pseudo-color representation of Ca2+ response from control (top) and M3R KD (bottom) hOPCs. B, Normalized intensity plots for control (top) and M3R (bottom) KD hOPCs. C, Dose–response curves of percentage [Ca2+]i responding cells following Oxo-M addition. Per-cell quantification of average Ca2+ peak amplitude (D), total number of Ca2+ spikes (E), and area under the curve (F) in Ca2+-responsive hOPCs. M3R KD (red) resulted in a highly significant decrease in Ca2+ response at higher Oxo-M doses. *p < 0.05 (two-way ANOVA with Tukey's HSD post test). **p < 0.01 (two-way ANOVA with Tukey's HSD post test). ***p < 0.001 (two-way ANOVA with Tukey's HSD post test). There were two fetal preparations (n > 30 cells per condition/dose, mean ± SEM). Scale bar, 25 μm.
Interestingly, among activated oscillating hOPCs, M3R KD led to a striking attenuation of Ca2+ spike amplitude following muscarinic stimulation (Fig. 2D). This resulted in a complete loss of dose dependency for average peak amplitude and highly significant reduction on spike amplitude in M3R KD hOPCs (Tukey's HSD; 2.5 μm, p = 0.00013, 7.9 μm, p = 1.5 × 10−11, 25 μm, p = 3.1 × 10−9). The minimal response observed in M3R KD cells could correspond to activity of residual M3R expression or to activity of other receptors, such as M1R. While response duration and frequency were only mildly reduced, M3R KD induced a highly significant decrease in the total number of Ca2+ spikes elicited by Oxo-M at higher doses (Fig. 2E). The combination of these alterations in [Ca2+]i response profile translated into a significant difference in the total cellular calcium response, as assessed by the area under the curve (Fig. 2F). Together, these data indicate that M3R is critical in mediating the muscarinic agonist-induced Ca2+ response in hOPCs.
M3R KD promotes oligodendrocyte differentiation following transplantation into hypomyelinating shiverer mice
Next, we investigated whether M3R signaling is active in hOPCs in vivo and acts to impair remyelination by transplanted human cells. hOPCs infected with control or M3R KD LV were transplanted into neonatal hypomyelinated shiverer/rag2 mice (105 cells per animal) as described previously (Abiraman et al., 2015). Twelve weeks after implantation, transplanted human engrafted hOPCs were widely distributed within the recipient corpus callosum and had begun to differentiate into MBP-expressing oligodendrocytes (Fig. 3). Similar to observations with pharmacological MR antagonists (Abiraman et al., 2015), M3R KD did not significantly influence the density of human hNA+ cells (958 ± 121 vs 764 ± 66 hNA+ cells/mm2 for M3R KD and control, respectively, n = 3).
Long-term LV KD of M3R promotes remyelination by transplanted hOPCs. Shiverer/rag2 pups were transplanted with hOPCs following LV infection and selection. At 12 weeks after implantation, human cells were stained for hNA (red) and MBP (green) (A). Immunostaining for differentiated oligodendrocytes (CC1, green) showed that M3R KD increased the resultant density of human hNA+CC1+ oligodendrocytes (arrows) in the corpus callosum following engraftment (B, quantified in C). Analysis of MBP synthesis revealed that, compared with controls, M3R KD cells produced substantially more MBP-stained fibers per transplanted cell in the corpus callosum (D, quantified in E). Within myelinated regions, M3R KD led to an increased number of MBP (green) ensheathed axons (neurofilament, red) compared with controls (F, quantified in G). Dotted lines indicate outline of the corpus callosum. *p < 0.05 (t test). n = 3 animals per group, mean ± SEM shown. Scale bars: A, 500 μm; B, 20 μm; D, 10 μm; F, 20 μm.
Importantly, M3R KD significantly increased the production of human oligodendrocytes (hNA+CC1+: 309 ± 71 vs 133 ± 33 CC1+hNA+ cells/mm2 for M3R KD and control, respectively, p = 0.034, n = 3; Fig. 3B,C). The extent of human oligodendrocyte maturation and myelination was measured by MBP immunostaining, which is completely absent in shiverer mice. The increased density of human oligodendrocytes resulted in substantially more MBP-positive fibers in animals receiving M3R KD-infected cells (Fig. 3A,D,E). Within the corpus callosum, the area of MBP+ immunostaining per human cell was increased by >5-fold in M3R KD recipients (p = 0.033, n = 3; Fig. 3E). Moreover, when quantified by confocal microscopy, M3R KD in transplanted hOPCs resulted in a 30% increase in the fraction of ensheathed axons within engrafted regions of the corpus callosum (54.2 ± 3.8% vs 41.7 ± 1.2% for M3R KD and control, respectively, t test, p = 0.034, n = 3; Fig. 3F,G). Together, these data demonstrate that M3R signaling delays hOPC differentiation and myelination in vivo.
M1/3R-selective antagonist accelerates oligodendrocyte differentiation during remyelination
While the shiverer model is ideal for assessing effects on human oligodendrocyte differentiation and transplant-mediated myelination, the shiverer hypomyelinating environment is not a model of demyelination. We therefore tested whether M3R regulates oligodendrocyte differentiation following lysolecithin-induced demyelination in mouse spinal cord. We initially tested whether a selective M1/M3R antagonist (solifenacin) would enhance oligodendrocyte differentiation in a similar manner to clemastine (Mei et al., 2014), which can potently antagonize several GPCRs (Nörenberg et al., 2011; Gregori-Puigjané et al., 2012). Demyelination was induced in adult mouse spinal cord by injection of lysolecithin into the dorsal funiculus and ventrolateral white matter. Animals received solifenacin (10 mg/kg daily, s.c.) or saline daily until death at 14 dpl, during the onset of remyelination in this model (Fancy et al., 2009, 2011). Solifenacin treatment increased the density of Plp1-positive oligodendrocytes at 14 dpl compared with saline-treated controls (Fig. 4; solifenacin: 1211 ± 68 vs saline: 836 ± 45 cells/mm2, t test p = 0.02, n = 3). Similarly, solifenacin treatment increased the density of CC1+ oligodendrocytes by immunofluorescence (Fig. 4B,G; solifenacin: 1224 ± 46 vs saline: 1016 ± 56 cells/mm2, t test p = 0.04, n = 3). Interestingly, we found no difference in Olig2+ oligodendrocyte lineage cell density (Fig. 4C,H; saline: 1643 ± 207.9 cells/mm2; solifenacin: 1627 ± 267.9 cells/mm2, p = 0.97, n = 3), suggesting that solifenacin similarly enhanced differentiation. Solifenacin treatment did not influence lesion size (data not shown), microglial response (Iba1, Fig. 4D), or astrogliosis (GFAP, Fig. 4E) compared with saline (n = 3). These results suggest that M1/3R selective antagonists have therapeutic potential and are consistent with a role for either M1R or M3R in OPC and oligodendrocyte differentiation.
Systemic solifenacin treatment enhances OPC differentiation following focal demyelination of mouse spinal cord. Young adult (8- to 10-week-old) mouse spinal cords were lesioned via direct injection of lysolecithin into dorsal white matter. Animals received daily subcutaneous administration of either isotonic saline (top row) or 10 mg/kg solifenacin (bottom row) from day of lesion until death at 14 dpl. In situ hybridization for Plp1 (A, quantified in F) and CC1 immunostaining (C, quantified in H) revealed increased density of oligodendrocytes in solifenacin animals relative to controls (quantified in F and H, respectively) with no observable effect on overall density of oligodendrocyte lineage cells (Olig2 in B, quantified in G), microglial response (Iba1, D), or astrogliosis (Gfap, E). *p < 0.05 (Student's t test). n = 3 per group, mean ± SEM shown. Scale bars: A, B, 40 μm; C–E, 100 μm.
Conditional knock-out of M3R in adult OPCs accelerates oligodendrocyte differentiation and remyelination following demyelination
Based on the role of M3R in human progenitors and solifenacin treatment during remyelination, we hypothesized that conditional M3R knock-out in adult OPCs would be sufficient to increase oligodendrocyte differentiation and remyelination and that this would not be compensated by M1R signaling. To test this hypothesis, we investigated the effect of conditional M3R deletion from adult NG2-expressing OPCs (Fig. 5). We generated homozygous NG2CreERT2:M3Rfl/fl:Rosa26-Yfp (M3R-inducible conditional knock-out [icKO]) mice to ask whether M3R deletion in NG2+PDGFαR+ OPCs could accelerate oligodendrocyte differentiation in a solifenacin-like manner. Tamoxifen was administered for 5 d to 8- to 9-week-old mice, and lysolecithin demyelinating lesions were induced 7 d after the last injection to ensure clearance of tamoxifen (n = 4 animals per group). To assess the role of M3R signaling on OPC differentiation, mice were killed at 14 dpl. Consistent with previous reports (Zhu et al., 2011), we observed YFP expression in 54.6 ± 6.1% of Olig2+ oligodendrocyte lineage cells, suggesting a relatively high rate of cre-mediated recombination (n = 4).
Conditional M3R knock-out in OPCs enhances their differentiation following focal demyelination of mouse spinal cords. Genetic ablation of M3R in NG2+ OPCs was induced in NG2CreERT2:Rosa-YFP:M3Rfl/fl young adult (8- to 10-week-old) mice via daily intraperitoneal injection of 200 mg/kg tamoxifen. Mice received either tamoxifen or canola oil (vehicle) for 5 d, and their spinal cords were lesioned via direct injection of lysolecithin 1 week after the final day of intraperitoneal injections. Animals were killed at 14 dpl, and spinal cord tissue was collected for histological analysis. In situ hybridization for Plp1 revealed increased oligodendrocyte density in M3R icKO animals relative to WT controls (A, quantified in B). We observed a significantly greater proportion of oligodendrocyte lineage cells that were oligodendrocytes in M3R icKO animals relative to controls (immunostaining for CC1 in E, quantified in F). M3R deletion did not influence the overall density of oligodendrocyte lineage cells (Olig2 in C, quantified in D). OPC-specific conditional M3R knock-out had no observable effect on microglial response (Iba1, G) or astrogliosis (Gfap, H). *p < 0.05 (Student's t test). n = 4 or 5 per group, mean ± SEM shown. Scale bar, 200 μm.
Next, we assessed the overall effect of M3R icKO in NG2-expressing OPCs on oligodendrocyte differentiation. M3R icKO substantially increased the density of Plp1+ oligodendrocytes in 14 dpl lesions relative to controls (Fig. 5A,B; wild-type [WT]: 599.8 ± 59.9 cells/mm2, n = 5; icKO: 791 ± 49 cells/mm2, n = 4, p = 0.04). This was not accompanied by a reduction in Olig2+ cell density, suggesting that the effect of M3R ablation was due to induced differentiation (Fig. 5C,D; WT: 1072 ± 155 cells/mm2; icKO: 1117 ± 56 cells/mm2, n = 4, p = 0.8). Consistent with this interpretation, we observed that a significantly greater proportion of Olig2+ cells were CC1+ oligodendrocytes in icKO animals relative to controls (Fig. 5E,F; WT: 54.9 ± 5.7% CC1+/Olig2+; icKO 70.4 ± 2.0% CC1+/Olig2+, p = 0.04, n = 4). Importantly, we found that the majority of CC1+ oligodendrocytes also expressed YFP+ (54.9 ± 0.5%, n = 3), indicating that they had undergone cre-mediated recombination. In addition, we observed a cell-autonomous mechanism of M3R on OPC differentiation. A greater proportion of recombined (YFP+) oligodendrocyte lineage cells differentiated to CC1+ oligodendrocytes compared with nonrecombined OPCs (CC1+YFP+/YFP+: 73.7 ± 3.3% CC1+; CC1+YFP−/YFP−: 57.1 ± 3.2% CC1+, p = 0.02, n = 3). We found no difference in Iba1 immunolabeling of microglia (Fig. 5G) or in GFAP immunostaining for astrocytes (Fig. 5H) between icKO and WT groups. To assess the effect of M3R icKO on OPC recruitment to demyelinated lesions, we euthanized mice at 5 dpl and determined oligodendrocyte lineage density by Olig2 immunostaining. We found no significant difference in Olig2+ cell density between WT and icKO animals (WT: 522 ± 84 cells/mm2; icKO: 604 ± 59 cells/mm2, n = 3, p = 0.47). These data indicate that ablation of M3R signaling in OPCs enhances their differentiation with no effect on recruitment, potentially by inducing more rapid cell cycle exit and terminal differentiation at the cost of proliferation.
To examine whether enhanced differentiation of M3R-deficient OPCs correlated with accelerated remyelination, we repeated our transgenic animal lesion paradigm and processed spinal cords for myelin ultrastructure analysis by electron microscopy. At 14 dpl, during the onset of remyelination, we found that M3R icKO significantly increased the proportion of remyelinated axons relative to WT controls (Fig. 6) (WT: 52.9 ± 1.2% remyelinated, n = 3; icKO: 62.1 ± 1.7% remyelinated, n = 4, p = 0.009). Consistent with an acceleration of OPC differentiation, and remyelination following M3R icKO, myelin g-ratio analysis revealed significantly decreased g-ratio, which represents the relative thickness of myelin in remyelinated fibers compared with axonal diameter (Fig. 6C). Together, these data demonstrate that M3R signaling in OPCs delays efficient remyelination in adult spinal cord.
Conditional ablation of M3R in OPCs accelerates remyelination. OPC-specific M3R knock-out was induced in adult NG2CreERT2:Rosa-YFP:M3Rfl/fl mice by daily intraperitoneal administration of 200 mg/kg tamoxifen or vehicle for 5 d. Animals were lesioned 1 week after the last day of injection. Animals were killed at 14 dpl and spinal cord tissue processed into resin for electron microscopy. Analysis revealed that M3R icKO in OPCs significantly increased the proportion of oligodendrocyte (OL) remyelinated axons in lesions compared with WT animals (A, quantified in B, mean ± SEM). C, Quantification of myelin sheath thickness in WT (green) and M3R iCKO (red) at 14 dpl by g-ratio analysis. SC, Schwann cell-remyelinated fiber. **p < 0.01 (Student's t test). n = 3 or 4 animals, ≥300 axons per animal. Scale bar, 2 μm.
Discussion
In this paper, we demonstrate that M3R signaling acts to delay OPC differentiation and ultimately contributes to delayed remyelination. We found that M3R signaling is functionally important and conserved in both mouse and human progenitors. The net effect of inhibition of muscarinic signaling is the acceleration of OPC and oligodendrocyte differentiation, leading to improved remyelination and myelin repair. In hOPCs transduced with KD virus or murine OPCs with conditional knock-out, M3R ablation alone was sufficient to induce precocious oligodendrocyte differentiation. Furthermore, fate-mapping of M3R-null OPCs confirmed that this was mediated by a cell-autonomous mechanism.
While the functional importance of specific MR subtypes in OPCs is not known, recent data indicate that M1R plays a key role in the maturation of immature oligodendrocytes (Mei et al., 2016). While we did not assess the role of M1R directly in this study, Mei et al. (2016) reported negative results using oligodendrocyte lineage cells isolated from germline M3R knock-out mice. The apparent difference between these studies might be due to compensatory changes that occur following germline deletion in neural stem cells versus specific ablation in OPCs or due to differences in the developmental stages of cells examined in vitro. Mei et al. (2016) isolated mouse progenitors/immature oligodendrocytes using O4 from postnatal day 7 mouse brain, which represents a later stage of oligodendrocyte development than the human fetal CD140a/PDGFαR+ used in the current study. Likewise, conditional deletion of M1R was achieved using the CNP-cre mouse (Lappe-Siefke et al., 2003), which drives recombination primarily in immature oligodendrocytes after the formation of OPCs (Ye et al., 2009). Thus, it appears that the role of M3R is restricted to earlier PDGFαR/NG2 stages of the oligodendrocyte lineage, whereas M1R regulates differentiation in committed progenitors and immature oligodendrocytes. Direct comparison of conditional KO mice using the same cre driver lines and the use of CRISPR-mediated deletion of M1R and M3R in human IPSCs will provide a more definitive understanding of the contribution of these inhibitory receptors.
The pattern of MR subtype mRNA expression is different between human and mouse OPCs, and the reported effects of MR agonists on proliferation appear to be different between species (Cohen et al., 1996; De Angelis et al., 2012; Abiraman et al., 2015). Our data indicate that M3R is critical for calcium-related signaling induced by muscarinic agonists in OPCs, suggesting that, at least in human primary OPCs, M3R is the operant receptor. Quantitative differences in expression signature and specifically in receptor subtype expression likely underlie some of the differences observed between human and rodent (Sim et al., 2009; Dietz et al., 2016). Such differences will be of upmost importance when designing therapeutic approaches that target signaling pathways to be translated from preclinical to human patients.
Interestingly, MR expression is highly variable within the FACS-isolated CD140a+/O4+ hOPC population with at least one-fourth of hOPCs expressing high levels of M3R/CHRM3 mRNA. Along with similar heterogeneity of other neurotransmitter receptor expression, it appears that a mosaic of OPCs in fetal forebrain exist with differing responsiveness to environmental and neuronal stimuli. Whether the variability in M3R expression represents stochastic expression within a single population of hOPCs or is representative of a specific stage of oligodendrocyte lineage differentiation remains to be determined. Intriguingly, a direct comparison of M3R-positive and M3R-negative hOPCs identified Gαq-associated transcripts as the most significantly regulated gene set among pathways in the mSigDB (q = 1.33 × 10−7; ST_GAQ_PATHWAY) (Subramanian et al., 2005). This may suggest that M3R-expressing hOPCs are primed to respond via M3R/Gαq-coupled signaling.
The muscarinic antagonist clemastine is under active clinical investigation as a remyelinating agent in MS (www.clinicaltrials.gov, NCT02040298 and NCT02521311). Clemastine has shown great promise in mouse models of demyelination and MS (Mei et al., 2014, 2016; Liu et al., 2016). However, there are some concerns that centrally acting nonspecific MR antagonists induce cognitive impairment due to inhibition of M1R signaling (Anagnostaras et al., 2003). Furthermore, the use of antimuscarinic agents is associated with mild cognitive impairment and dementia in the aging population (Cai et al., 2013). M1R is the most abundant subtype in forebrain and hippocampus (Levey et al., 1995). M1R expression is decreased in schizophrenia (Dean et al., 2002), and MR antagonists have been shown to worsen Alzheimer's disease (Perry et al., 2003). Together, these findings suggest that M1R-acting drugs may not be optimal targets for MS therapy. On the other hand, M3R antagonists are typically well tolerated clinically with a manageable peripheral side effect profile (Chapple, 2006). It is not clear whether M3R-selective antagonists would avoid the cognitive adverse side effects. As such, the development of new selective and potent M3R antagonists might yield an effective approach to improve both spontaneous and transplant-mediated repair while minimizing CNS anticholinergic side effects.
The cellular source of ACh that activates M1/3R following demyelination and impairs OPC differentiation and remyelination is unknown. ACh fibers are widely distributed in the forebrain and spinal cord, and both axons and dendrites have the capability to release ACh (Garzón and Pickel, 2000). Moreover, these fibers can signal extrasynaptically via volume-based neurotransmission (Yamasaki et al., 2010). The differential pattern of acetylcholinesterase and ChAT activity in human brain development (Court et al., 1993) suggests that neuron-derived ACh may act directly on OPC-expressed MRs. OPC differentiation is modulated by M1/3R antagonists in corpus callosum (Deshmukh et al., 2013; Abiraman et al., 2015), midbrain (Abiraman et al., 2015), and spinal cord (Deshmukh et al., 2013; Mei et al., 2014). This is consistent with a widespread role for ACh/MR signaling in OPCs throughout the CNS. However, we cannot rule out other cellular sources of Ach, such as lymphocytes (Kawashima et al., 1989; Rinner and Schauenstein, 1993).
Muscarinic agonist treatment increases InsP3 and [Ca2+]i in OPCs (Cohen and Almazan, 1994). In this study, we developed a new viral GCaMP6s reporter under the control of EF1α to visualize Ca2+ signaling in individual hOPCs. CD140a-sorted hOPCs did not exhibit spontaneous calcium waves in vitro. Although we did not attempt to measure rare and small-magnitude calcium events that can occur in OPCs (Haak et al., 2001), we did not observe spontaneous Ca2+ waves that have been described in migrating murine OPCs grown in serum-containing media (Paez et al., 2009). The lack of spontaneous Ca2+ waves allowed us to determine whether increased [Ca2+]i is regulated by M1R and/or M3R, both Gq-coupled receptors. M3R KD largely inhibited muscarinic agonist increased [Ca2+]i, suggesting that M3R is the principal mediator of this response in CD140a+ hOPCs. The possibility that MR-coupled Ca2+ signaling directly contributes to the inhibition of OPC differentiation represents an intriguing contrast to the role of Ca2+ entry via voltage-operated calcium channels (Cheli et al., 2015) or calcium-permeable glutamate channels (Wake et al., 2011; Lundgaard et al., 2013) that are known to induce differentiation. Thus, neuronal activity, which drives OPC proliferation and differentiation in the developing brain (Barres and Raff, 1993; Gibson et al., 2014), may act in a neurotransmitter-specific manner to regulate diverse Ca2+-dependent signals to it turn regulate OPC homeostasis. The spatiotemporal regulation of Ca2+ entry provides the basis for activation of distinct signaling stimuli (Di Capite et al., 2009), with store-operated calcium entry being coupled with Gq-coupled GPCRs, such as M3R. It therefore noteworthy that store-operated calcium entry via TRPC1 blocks OPC differentiation (Paez et al., 2011).
The mechanisms that match OPC number to white matter volume are not well understood. In primate development, before OPC specification, the subventricular zone splits into two distinct regions separated by cholinergic fibers emanating from the ganglionic eminence (Smart et al., 2002). Moreover, a distinct population of cholinergic interneurons has been identified in the rodent SVZ (Paez-Gonzalez et al., 2014). As such, cholinergic signaling is well placed to prevent premature OPC differentiation and allow for sufficient proliferation of the progenitors needed for a larger volume of white matter. Cholinergic signals may act to prevent premature demyelination, which would limit the number of progenitors and delay colonization of larger tissue/lesion volume necessary for human development and remyelination in disease.
Finally, by genetically deleting M3Rs in human cells, we could engineer cells before transplantation to eliminate this otherwise inhibitory environmental cue. Lentiviral-mediated mCherry expression in hOPCs continues for at least 1 year after transplantation, suggesting that ex vivo manipulation can modify hOPC gene expression for extended periods of time. While we did not directly assess M3R KD after transplantation, leaving open the possibility that shRNAi expression may decline, KD of inhibitory M3R signaling substantially augmented human oligodendrocyte formation over an 8 week period. This type of ex vivo genetic modification may represent an important means by which transplanted cells could overcome the inhibitory environment present in most MS lesions. As such, we have established a proof-of-principle that rendering OPCs unresponsive to potentially negative environmental stimuli before implantation represents a viable approach to increase the capacity for remyelination by transplanted hOPCs. M1R and M3R are representative of an emerging set of pharmacological targets that relay inhibitory signals for differentiation in demyelination and may be targeted to improve remyelination MS lesions.
Footnotes
↵* R.R.W and J.J.P are co-first authors.
This work was supported by National Institute of Neurological Disorders and Stroke Grant R01NS104021, National Center for Advancing Translational Sciences UL1TR001412, National Multiple Sclerosis Society RG 5505-A-2 and RG 5110A1/1, the Kalec Multiple Sclerosis Foundation, the Change MS Foundation, the Skarlow Memorial Trust, and the Empire State Stem Cell Fund through New York State Department of Health Contract C028108. J.J.P. received additional support from National Institute of General Medical Sciences R25GM09545902 and National Center for Advancing Translational Sciences UL1TR001412-S1. The single-cell RNA-seq was performed by the Genomics Shared Resources at Roswell Park Cancer Institute, which is supported by National Institutes of Health/National Cancer Institute Grants P30 CA016056 (RPCI Cancer Center Support Grant). We thank the Confocal Microscope and Flow Cytometry Facility in the School of Medicine and Biomedical Sciences, University at Buffalo for assistance; and Ted Szczesny and Ed Hurley for assistance with electron microscopy.
The authors declare no competing financial interests.
- Correspondence should be addressed to Dr. Fraser J. Sim, Department of Pharmacology and Toxicology, School of Medicine and Biomedical Sciences, University at Buffalo, 3435 Main Street, Buffalo, NY 14214. fjsim{at}buffalo.edu
