MDSC isolation and in vitro characterization
To determine whether any of the standard markers for MDSC characterization are predictive of high in vivo muscle regeneration, we examined 25 populations of MDSCs in terms of five variables: in vivo muscle regeneration efficiency, expression of CD34, expression of Sca-1, expression of desmin, and cell sex. Analysis of these variables revealed a large degree of heterogeneity in the MDSC populations. The distribution of the populations' regeneration indexes (RIs) and CD34, Sca-1, and desmin expression are shown in . The RI is a measure of how efficiently stem cells participate in skeletal muscle regeneration (Jankowski et al., 2002
; Deasy et al., 2005
), whereas CD34 and Sca-1 are stem cell markers, and desmin is a myogenic marker.
Figure 1. Variability among muscle-derived cell populations. (A) We examined 25 populations of MDSCs for five variables: in vivo muscle regeneration efficiency (RI), CD34 expression, Sca-1 expression, desmin expression, and cell sex. Histograms show the distribution (more ...)
We performed a Pearson product moment correlation analysis to identify significant relationships between any two variables and, in particular, to determine whether any of the markers correlate with high in vivo muscle regeneration. shows the correlation scatter matrix comprising scatter plots for all possible combinations of variables. Only cell sex correlated with the RI (P = 0.070); M-MDSC populations had a significantly lower RI on average than F-MDSC populations (P = 0.035; t
test). We also detected a significant correlation between cell sex and Sca-1 expression (P = 0.014): the mean expression of Sca-1 for all M-MDSC populations was significantly higher (M-MDSCs, 73 ± 16%; F-MDSCs, 52 ± 22%; mean ± SD; P = 0.001; t
test). We next separated the correlation matrices on the basis of sex; shows the scatter plots for M-MDSCs (10 populations are shown in each plot), and shows the scatter plots for F-MDSCs (15 populations are shown in each plot). The scatter plot of the RI versus Sca-1 for M-MDSCs indicates that a higher level of Sca-1 expression by M-MDSCs positively correlated with a higher RI (P = 0.021; ). This demonstrates that the significantly higher levels of Sca-1 expression by M-MDSCs (as compared with F-MDSCs) are not directly related to their low RIs (as compared with the RIs of F-MDSCs). The scatter plots of the F-MDSC populations did not reveal such a relationship (). Significant correlations between variables other than RIs are also indicated in , and correlation coefficients are listed in Table S1 (available at http://www.jcb.org/cgi/content/full/jcb.200612094/DC1
). We focused our subsequent analysis on the sex-related differences exhibited by MDSCs because cell sex was the only variable that correlated with the RI.
We used identical methods to isolate M- and F-MDSCs, and both exhibited stem cell characteristics in vitro ( and Fig. S1, available at http://www.jcb.org/cgi/content/full/jcb.200612094/DC1
). Morphological comparison of the M- and F-MDSC populations revealed no significant differences in cell size or cell shape (Fig. S1, A and B). Analysis of the cells' multilineage differentiation potential showed that with appropriate stimulation in vitro, both male and female cells expressed markers of myogenic (), osteogenic (Fig. S1 C), and adipogenic lineages (Fig. S1 C), although they did so at different rates and to different extents. On average, there was no significant difference in the mean amount of fast myosin heavy chain (MyHC) in M- versus F-MDSCs ().
Figure 2. M- and F-MDSCs demonstrate similar in vitro stem cell characteristics. (A) M- and F-MDSCs underwent myogenic differentiation, as shown by MyHC expression (red; m, M-MDSC; f, F-MDSC). (B) We did not detect a significant difference in the extent or rate (more ...)
Our analysis of short-term kinetics showed that M-MDSC and F-MDSC populations had similar population doubling (PD) times (PDTs; 13 ± 1.4 h and 14.2 ± 3.1 h, respectively), cell cycle times (12 ± 1.3 h and 12 ± 1.3 h, respectively), and mitotic fractions (0.91 ± 0.10 and 0.85 ± 0.08, respectively; P = 0.28; t
test) over a 3-d period (mean ± SD; ). However, after moderate expansion over 14 d or extended expansion over 3 mo (>150 PDs), M-MDSC populations exhibited higher growth rates than F-MDSC populations (M-MDSCs: PDT = ~13 h and mitotic fraction = 0.87–0.97; F-MDSCs: PDT = ~15 h and mitotic fraction = 0.81–0.87; and not depicted). After extended expansion, there was no significant difference in telomere length (Fig. S1 D) or telomerase activity (Fig. S1 E) as measured by flow cytometry. In addition, we observed a normal modal chromosome number for M- and F-MDSCs that were expanded to 90 PDs (unpublished data); this is consistent with our previous study of F-MDSCs (Deasy et al., 2005
In vivo skeletal muscle regeneration
After transplanting M- and F-MDSCs into the skeletal muscle of dystrophic mdx mice (i.e., mice that lack dystrophin at the sarcolemma of muscle fibers), we investigated the cells' abilities to regenerate dystrophin-expressing myofibers. Sex-matched experiments involving several cell populations obtained from various isolations revealed that the implantation of F-MDSC populations led to significantly better skeletal muscle regeneration in vivo, as determined by calculating each population's RI (the ratio of dystrophin+ fibers per 105 donor cells; ). Female cell populations regenerated significantly more dystrophin-positive myofibers (n = 15 F-MDSC populations; two to six muscles per population; RI = 230 ± 52; mean ± SEM) as compared with male populations (n = 10 M-MDSC populations; two to six muscles per population; RI = 95 ± 20; P = 0.035). Representative images of dystrophin-expressing myofibers within the mdx muscle tissue are shown in . Although the RIs of both M- and F-MDSC populations varied, no M-MDSCs exhibited high regeneration efficiency (i.e., RI > 203). In vitro immunostaining of myotubes after MDSC differentiation demonstrated that both M- and F-MDSCs are able to express dystrophin in myotubes containing a similar number of nuclei ().
Figure 3. F-MDSCs induce more efficient skeletal muscle regeneration than M-MDSCs. (A) Quantitation of engraftment in terms of the RI (RI = number of dystrophin-positive fibers per 105 donor cells). The overall mean RI is significantly higher for F-MDSCs (more ...)
We performed both sex-matched and sex-mismatched transplantations to determine whether the sex of the host tissue might play a role in the cells' differing regeneration abilities. We used two F- and two M-MDSC populations that had similar expression profiles for CD34, Sca-1, and desmin and exhibited similar short-term proliferation characteristics. Regardless of host sex, M-MDSC transplantation resulted in a low RI (160 ± 75 in female hosts and 105 ± 25 in male hosts; ). Two-way analysis of variance (ANOVA) confirmed these findings and revealed both a significant difference as a result of host sex (P = 0.048) and a significant effect as a result of cell sex (P < 0.001). These results are consistent with those observed in previous studies using sorted populations of M-MDSCs (Jankowski et al., 2002
; Jankowski and Huard, 2004a
) or using male hosts (Mueller et al., 2002
). M-MDSCs had a significantly lower RI than F-MDSCs (P < 0.05), and there was a significantly lower RI with the male host as compared with the female host (P < 0.01).
The difference caused by host sex suggests that the female microenvironment might be a more receptive environment for skeletal muscle regeneration by MDSCs and might play a role in RI variability. Hormonal differences between male and female hosts may influence the regeneration process; previous studies have described such differences in the responses of progenitor and stem cells to hormones (Jilka et al., 1992
; Steinlein et al., 1995
). We first tested the possibility that the low RI of M-MDSCs implanted in female mdx
hosts (as compared with female to female transplantations) might be caused by the immune rejection of the M-MDSCs as a result of H-Y antigenicity. We examined the immune response at the site of cell delivery by measuring the total cross-sectional area of CD4 expression after sex-matched and sex-crossed transplantations. We observed a larger area containing CD4-positive cells after sex-crossed transplantations ().
We next investigated the effect of host sex in the absence of an immune response by performing sex-crossed and sex-matched transplantations in immune-compromised animals. M-MDSC to female mdx/severe combined immunodeficiency (SCID) mouse transplantations again resulted in a low level of skeletal muscle regeneration (RI = 216 ± 39), which is similar to what we observed in the female mdx host. We observed no significant difference as a result of host sex (P = 0.235), yet the significant difference as a result of cell sex remained (P = 0.018; two-way ANOVA; ).
To determine whether estrogen stimulation improves the in vivo regeneration efficiency of MDSCs, we stimulated M- and F-MDSCs with two doses of 17-β-estradiol (10 or 100 nM; physiologic range of 10–100 nM; Maeda et al., 2000
). We confirmed the expression of estrogen receptors in M- and F-MDSCs by immunocytochemistry (Fig. S2 A, available at http://www.jcb.org/cgi/content/full/jcb.200612094/DC1
) and by microarray analysis (not depicted). After 2 wk in vitro with estradiol, M-MDSCs displayed lower proliferation rates than unstimulated M-MDSCs (although the decrease was not significant). Consistent with other results, comparison of the proliferation rates of the unstimulated cells showed that F- MDSCs proliferated more slowly than unstimulated M-MDSCs at all time points after 7 d (9 d, P = 0.070; 12 and 14 d, P < 0.05; t
test; Fig. S2 B). We compared the in vivo regeneration efficiency of stimulated and unstimulated M- and F-MDSCs. 2 wk after transplantation, F-MDSCs cultured in control conditions of 0 nM estradiol had a higher RI than similarly cultured M-MDSCs (557 ± 229 vs. 115 ± 36; P = 0.06). Stimulation with estradiol had no significant effect on the RIs of M-MDSCs; however a trend toward a decreased RI was detected in F-MDSCs stimulated with 10 or 100 nM estradiol (P = 0.09 and P = 0.07; n
= 3; t
test; Fig. S2 C).
Exploring alternate M-MDSC donor sources
Based on cell adhesion characteristics, the preplate technique or variations of this technique have been used to isolate cells from adult skeletal muscle tissue (Rando and Blau, 1994
; Qu et al., 1998
). These cells have demonstrated multilineage differentiation potential ( Lee et al., 2000
; Winitsky et al., 2005
; Sarig et al., 2006
; Nomura et al., 2007
). To determine whether M-MDSCs are a rare subpopulation with particular adhesion or isolation characteristics, we attempted to isolate potent M- MDSCs from three other methods: (1) other preplates or other subfractions of myogenic progenitors, (2) a defined FACS subfraction expressing the stem cell marker CD34, and (3) clonal populations.
The preplate technique used to isolate MDSCs from a skeletal muscle biopsy involves serial preplating of low-adherence cells found within the primary medium supernatant to separate these cells from cells adhered to the collagenated surface of the flask (). Usually, myogenic precursors with characteristics of myoblasts or satellite cells are found within the early preplates (pp3–6), whereas cells isolated from the later preplates (≥pp6) have stem cell characteristics. We compared the in vivo regeneration efficiency (RI) of M-MDSCs with that of male myoblasts obtained from pp3–6 but identified no significant differences ().
Figure 4. Examination of alternate M-MDSC sources. (A) The preplate technique to isolate subfractions of muscle-derived cells is based on their adhesion characteristics from muscle biopsy. Male myoblast subfractions also demonstrated low regeneration efficiency. (more ...)
To determine the effect of the expression of CD34 on regeneration efficiency, we used FACS to purify M- and F-MDSCs and obtain CD34+ fractions. The expression of Sca-1 and CD34 by unsorted M- and F-MDSCs was comparable (), and, in both populations, <15% of the cells were desmin positive. Our comparison of the RI of CD34-positive F-MDSCs (RI = 417 ± 56) with that of similarly sorted M-MDSCs (RI = 266 ± 61) revealed that the latter RI was significantly lower (P = 0.053; ).
To test whether there are fewer cells within the M-MDSC populations that are capable of contributing to muscle regeneration, we performed in vitro and in vivo clonal analysis of myogenic behavior. We used FACS to single-cell sort and obtain clones to determine the percentage of clones with myogenic potential in the M- and F-MDSC populations. Because some populations already had a high level of expression of the myogenic marker desmin, we analyzed the clones from these populations separately. At 2 wk after single-cell cloning, there was no significant difference in the percentage of clones among male or female cells derived from MDSC populations (with high or low desmin expression), which yielded myogenic colonies as defined by myotube formation (P = 0.776; one-way ANOVA; ). When we compared all three factors (cell sex, desmin expression, and week of analysis), we detected a significant effect (P = 0.066, P = 0.003, and P = 0.756, respectively) as a result of the desmin level and week of myogenic analysis but no effect as a result of the cell sex (three-way ANOVA). We transplanted eight representative clonal populations of MDSCs (four male and four female clones) into mdx animals to determine whether a clonal population could be identified within the M-MDSC populations that demonstrated a high RI. We confirmed that the overall trend we observe in (using 25 populations) is consistent with results obtained from clonal populations (). Although we did observe an F-MDSC clone that exhibited a lower RI as compared with its parent population (F3 vs. C parent population), we did not observe any male clone that exhibited a higher RI as compared with the parent population.
Microarray analysis and cell stress response
We next performed microarray gene ontology analysis of M- and F-MDSCs to explore global gene expression differences between the populations. We identified the expression of more general cell stress–related genes, including oxidative stress and antiapoptotic genes, in F-MDSCs than in M-MDSCs () using both gene ontology based on the NetAffx annotation database and GeneSifter. Of 45 general cellular stress–related genes, 29% of the genes were significantly increased in females, 7% were significantly higher in males (P < 0.05), and 64% showed no significant difference. There were also sexual dimorphisms in genes that were involved in responses to hypoxia and oxidative stress between M- and F-MDSCs. Of 99 apoptosis-related genes, 23% of the genes were significantly increased in females, 4% were higher in males, and 72% showed no significant difference. We confirmed the sex-related differences in the expression levels of several of these genes using RT-PCR (). We also compared microarray data of F-MDSCs that yielded a low level of regeneration with F-MDSCs that yielded a high level of regeneration. We found two stress-related genes (TRP53inp1 and SOD1) that showed significant trends in a direction opposite to what we expected. For example, SOD1 was significantly higher in F-MDSCs as compared with M-MDSCs; however, when we explored the data of F-MDSCs with low RIs versus those with high RIs, we observed that SOD1 was higher in F-MDSCs that had only a low level of regeneration (P < 0.05). Similarly, TRP53Inp1 was significantly higher in the low-regenerating M-MDSCs, but, when we examined the low- versus high-regenerating F-MDSCs, we observed significantly higher levels of TRP53Inp1 in the female populations with the best in vivo regeneration (). Although these trends are interesting, this analysis did not provide any clear indicators for genes of importance.
Figure 5. Differences in hypoxic and oxidative stress genes. (A) Microarray gene ontology analysis of M- and F-MDSCs. An increase in the expression of cell stress–related genes, including oxidative stress and antiapoptotic genes, was identified for F-MDSCs (more ...)
In particular, we found that the antiapoptotic factor Bcl2 was twofold higher in F-MDSCs than in M-MDSCs by microarray analysis (, arrow) and Western blotting (not depicted). To test whether the overexpression of Bcl2 in M-MDSCs could provide a gain of function in terms of in vivo skeletal muscle regeneration, we transfected the M-MDSCs with a Bcl2 plasmid. Western blot analysis and quantification showed higher levels of Bcl2 in M-MDSCs that were transfected to overexpress Bcl2 as compared with control M- or F-MDSCs (unpublished data). However, we did not observe a change in the RI of Bcl2-engineered M-MDSCs as compared with M-MDSC controls ().
Examination of cell fate at early time points
We transplanted M- and F-MDSCs, which were transduced with a lacZ
-encoding retroviral vector, into the gastrocnemius muscles of sex-matched mice, and we harvested at several time points between 16 h and 14 d. We quantified the number of lacZ
-expressing nuclei that were detected at the transplantation site. At 16, 24, 48, and 72 h, there were significantly more lacZ
nuclei detected in muscles of M-MDSC transplantations as compared with F-MDSC transplantations (P < 0.05; ). In separate experiments, we digested the skeletal muscle 24 and 48 h after the transplantation of M- and F-MDSCs, and we quantified the amount of lacZ
gene by quantitative RT-PCR. Similar to the histological analysis, we found more lacZ
in muscles transplanted with M-MDSCs as compared with F-MDSC transplantations, although this was not statistically significant (P > 0.05; ). In support of the finding of more M-MDSCs after transplantation, we detected higher levels of reduced glutathione, an antioxidant peptide whose levels are reported to be maximal in mitotic cells (Li and Oberley, 1998
). We used flow cytometry and monochlorobimane labeling to detect glutathione levels and found significantly more glutathione in M-MDSCs as compared with F-MDSCs (P < 0.01; ).
Figure 6. Examination of the transplantation site at early time points. (A) M- and F-MDSCs were transduced with the retroviral lacZ gene and transplanted into the gastrocnemius muscles of sex-matched mice, and the muscles were harvested at several time points between (more ...)
We also examined the amount of donor cell fusion or differentiation at the engraftment site by quantifying the percentage of lacZ nuclei that were located within muscle fibers. We examined the transplantation sites at 1, 2, and 5 d after cell injection. At 5 d after transplantation, we observed a trend toward significantly more donor cell fusion or differentiation with M-MDSCs as compared with F-MDSC transplantation to mdx muscle (P = 0.068; ).
Response to cell stress
To evaluate the effect of low oxygen conditions and oxidative stress on M- and F-MDSCs, we examined MDSC viability, phenotype, and myogenic differentiation under these conditions. We hypothesize that after MDSC transplantation, the transplanted cells will experience conditions of low physiologic oxygen environment and oxidative stress. Previous studies have demonstrated that inflammation occurs at the local microenvironment after cell transplantation (Guerette et al., 1997a
; Qu et al., 1998
; Bottino et al., 2002
; Suzuki et al., 2004
) and that oxidative stress occurs after an inflammation response and in the presence of low oxygen (Gute et al., 1998
; Dhalla et al., 1999
After exposure to physiologic oxygen for 24 h (2.5% O2), we observed a trend toward reduced proliferation in both M-MDSCs and F-MDSCs as compared with their controls (P = 0.18 and P = 0.079, respectively; ). There was no significant difference in the amount of cell viability or cell death in M-MDSCs as compared with F-MDSCs under conditions of low oxygen (P = 0.652; ). After exposure to oxidative stress (100 μM H2O2), we observed a significant decrease in cell numbers of both M-MDSCs and F-MDSCs (P = 0.039 and P = 0.001, respectively; ). However, when we compared the response of M- with F-MDSCs in terms of cell proliferation or total cell numbers after stress, we observed no difference between the populations; there was no difference in the percent change in cell viability for M- vs. F-MDSCs after oxidative stress (41 vs. 55% decrease; P = 0.325; ).
Figure 7. Behavioral response to low oxygen and oxidative stress. (A) After exposure to physiologic oxygen for 24 h (2.5% O2), we observed a trend toward reduced cell numbers in both M-MDSCs and F-MDSCs (P = 0.18 and P = 0.079, respectively). There (more ...)
We examined the expression of CD34, Sca-1, and desmin in the populations after exposure to cell stress. We did not observe a change in CD34 or Sca-1 expression in M- or F-MDSCs that were exposed to 2.5% O2 as compared with cells cultured in atmospheric O2. We did observe a trend in increased desmin expression in M-MDSCs that had initially low levels of desmin (P = 0.065; ). There was no similar change in desmin expression in a comparable population of F-MDSCs. After treatment with hydrogen peroxide, we observed trends toward an increase in CD34 and Sca-1 in all populations; however, after oxidative stress, there was no significant difference in the phenotype of M-MDSCs as compared with F-MDSCs ().
Next, we tested the ability of the cells to undergo myogenic differentiation after exposure to low oxygen or hydrogen peroxide. We plated M- and F-MDSCs at high density. After 4 d in culture, we transferred cells to either low oxygen or applied 100 μM H202 growth media. Control MDSCs were maintained at atmospheric oxygen with no H202. After 24 and 48 h, we performed immunochemistry for fast MyHC expression.
Under control conditions, there was no significant difference in the percentage of nuclei that colocalized with MyHC in M- or F-MDSC populations at 24 or 48 h (27–45% of the nuclei colocalized with MyHC). We observed a trend toward increased myogenic differentiation in M-MDSCs as compared with F-MDSCs after 48 h in low oxygen (). The sex-related difference in the differentiation response was also observed after exposure to hydrogen peroxide for 24 (P = 0.060) and 48 h (). In comparison with F-MDSCs, we observed an increase in myogenic differentiation with M-MDSCs as compared with their untreated controls ().