Loss of serum response factor induces microRNA-mediated apoptosis in intestinal smooth muscle cells

Serum response factor (SRF) is a transcription factor known to mediate phenotypic plasticity in smooth muscle cells (SMCs). Despite the critical role of this protein in mediating intestinal injury response, little is known about the mechanism through which SRF alters SMC behavior. Here, we provide compelling evidence for the involvement of SRF-dependent microRNAs (miRNAs) in the regulation of SMC apoptosis. We generated SMC-restricted Srf inducible knockout (KO) mice and observed both severe degeneration of SMCs and a significant decrease in the expression of apoptosis-associated miRNAs. The absence of these miRNAs was associated with overexpression of apoptotic proteins, and we observed a high level of SMC death and myopathy in the intestinal muscle layers. These data provide a compelling new model that implicates SMC degeneration via anti-apoptotic miRNA deficiency caused by lack of SRF in gastrointestinal motility disorders.

Serum response factor (SRF) is a muscle-specific transcriptional factor that drives smooth muscle cell (SMC)-specific gene expression, which is necessary for contractile and cytoskeletal functions. 1-3 SRF transcriptionally activates expression of SMC-specific genes by binding to the CArG box sequence [CC (A/T) 6 GG] in promoters and introns of most SMC-restricted genes. 1 Computational analysis of genome-wide CArG boxes (CArGome) in mice and humans has identified many SRF-dependent genes that encode for cytoskeletal/ contractile or adhesion proteins, suggesting that SRF is a master regulator of the actin cytoskeleton. 4 In addition, myocardin (MYOCD)-related transcription factors (MRTFs), which include MYOCD, MRTF-A/MKL1/MAL and MRTF-B/ MKL2, directly interact with SRF to activate a subset of SRF-modulated genes to promote myogenic differentiation and cytoskeletal structuring. 3 Although SRF is known primarily as the critical switch for induction of the muscle phenotype, it has recently been implicated in more diverse and multifunctional roles. For instance, there is now evidence that SRF is involved in carcinogenesis and tumor progression, induction of epithelial-to-mesenchymal transition with drug resistance in hepatocellular carcinoma and in lung fibrosis. 5,6 Furthermore, MYOCD, which is an integral part of the SRF/MRTF pathway, has also been recently implicated in apoptosis and autophagy of SMCs, 7 and SRF has been shown to attenuate Myc-induced apoptosis in mammary epithelial cells in culture. 8 However, as of yet, there has been no demonstration of apoptotic induction with knockout (KO) of Srf in SMCs.
MicroRNAs (miRNAs) are critical for SMC growth, differentiation and survival within the gastrointestinal (GI) track. 9 Furthermore, several hundred miRNAs that determine cellular fate and phenotype, including SMC-specific miR-143 and miR-145, are expressed in SMCs. 10 Depletion of Dicer, a RNase III that generates mature miRNAs, in mouse SMCs results in the degeneration of smooth muscle and severe myopathy within the GI tract, 9 and a similar phenotype results with Srf ablation in mouse SMCs. 11,12 The cellular mechanism of SMC loss in the Srf KO mice, however, remains poorly understood.
We report here a model that describes the functional role of SRF in suppression of apoptotic activity through SRF-induced miRNAs that target apoptotic proteins in SMCs. Our proposed model reveals how loss of SRF expression can lead to SMC death and intestinal myopathy in the development of GI motility disorders.

Results
Conditional deletion of Srf in adult SMCs results in the severe dilation of GI tract. Congenital Srf KO mice did not survive the prenatal stage. 13 Therefore, an inducible SMC-specific Srf KO mouse line was generated. Following  whereas the colon was packed with hard feces (Figures 1a  and b). Furthermore, the length of the GI tract from stomach to colon in the KO mice was significantly shorter than that of wild-type (WT) mice ( Figure 1c). Cross-sectional images of KO jejunum revealed hyperplastic and hypertrophic growth of the smooth muscle at PT15 (Figure 1d) in addition to severe circumferential dilation with thinning of the smooth muscle layers at PT21 (Figure 1e).
Cellular and phenotypic changes of KO SMCs. At PT21, SMCs in both circular and longitudinal muscle layers of Srf KO mice were significantly more disorganized with weaker expression of MYH11 compare with WT controls (Figure 2a). Notably, populations of SMCs in KO mice were abnormally enlarged and hypertrophic. In contrast, there was no significant qualitative or quantitative difference in the populations of myenteric (ICC-MY) and deep muscular plexus (ICC-DMP) interstitial cells of Cajal (ICCs) in WT and KO Srf mice as evidenced by anti-KIT staining ( Figure 2a). However, anti-PGP9.5 staining of neurons revealed abnormal myenteric (MY) ganglia in KO jejunum compared with WT ( Figure 2a). Specifically, Srf KO neuronal ganglia exhibited less cell bodies and more densely packed axons. Collectively, our data suggest that deficiency of Srf resulted in loss of contractile proteins, which may lead to morphological changes of SMCs and MY neurons. Ultrastructural examination of SMCs with electron microscopy revealed that Srf KO SMCs had normal well-developed contractile filaments at PT5, PT10 and PT15. However, at PT21, KO SMCs began to display irregular contours and processes compared with age-matched WT controls ( Figure 2b). Furthermore, the KO SMCs displayed well-developed rough endoplasmic reticulum and Golgi apparatus; widely distributed organelles and mitochondria; and less conspicuous contractile elements that were suggestive of an active synthesis state ( Figure 2b). The intercellular spaces of KO muscular tissue, which contained numerous apoptotic cells, were also markedly wider when compared with WT muscles (Figure 2b). There was, however, no evidence of necrosis in KO muscles such as cell swelling, plasma membrane breakdown or organelle leakage.
Ultrastructural analysis of Srf KO GI tissue also revealed degenerative changes within the MY ganglia (MG) with macrophage-like cells (MLCs; Figure 2b) located around damaged ganglia. In contrast to the degenerated SMCs and neurons in Srf KO intestinal tissue, the ICC-MY and ICC-DMP appeared relatively normal in Srf KO muscles at least up until PT21 (Figure 2b). Collectively, the ultrastructural data confirmed the apoptotic activity of SMCs and degeneration of neurons in the intestinal ganglia of Srf KO mice.
Time-lapsed examination of jejunal smooth muscle following tamoxifen-induced Srf KO revealed that SRF protein expression was significantly decreased in the longitudinal and circular muscle layers by PT10 and nearly undetectable by PT28 (Figure 3a). The number of SMCs expressing SRF in KO muscle layers also decreased significantly and gradually with the progressive time points compared with WT SMCs, whose SRF levels remained unchanged (Figure 3e). Noticeable degeneration of smooth muscle (as labeled with anti-SMA (ACTA2) and anti-MYH11 in Figure 3a-c), however, did not occur until PT21 and PT28.
Interestingly, terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) signal was detected in Srf KO SMCs at PT21, which indicated that the degenerative changes were at least partially mediated by apoptosis (Figures 3b and  d). Of note, there was little or no TUNEL signal in WT and KO SMCs at PT10, PT15 and PT28 suggesting that apoptotic degeneration was essentially complete by 3 weeks after tamoxifen injection ( Figure 3b). Furthermore, the average number of apoptotic cells detected in KO muscle layers at PT21 (Figure 3b Along with apoptosis induction, there was a slight increase in cellular proliferation in the Srf KO SMCs of both muscle layers as revealed by anti-Ki67 staining ( Figure 3c). More specifically, SMCs of Srf KO mice exhibited increased proliferative activity at PT21 but not at PT10 and PT15 when compared with WT ( Figure 3c). Overall, there were more proliferative cells in the muscle layers of KO than WT mice at the specified time points; particularly, there was a significant increase in cellular proliferation at KO PT21 ( Figure 3g). Notably, there was relatively high proliferative activity in the mucosal crypts, serosa and MY region at all-time points. Furthermore, there were significantly less cryptic epithelial cells in the KO mucosa suggesting that cryptic stem cells may need the supportive environment of SMCs for growth and survival ( Figure 3c). Altogether, our data indicated that in the absence of SRF, SMCs lost their contractile phenotype and underwent significant apoptosis at PT21 with a transient Genetic changes of SRF-induced miRNAs in KO SMCs.
To examine the effects of SRF KO on miRNA expression in SMCs, we sequenced the miRNAs expressed within the muscle layers of the more dilated upper and less dilated lower regions of the jejunum in WT and Srf KO mice at PT15 (Figure 1b). Of the approximately 50 million raw sequence reads that were acquired from these four samples, 44.8 million reads were mapped to the mouse genome, which led to the identification of 9-11 million annotated miRNAs from each sample ( Interestingly, there was a high correlation coefficient (0.989) between the normalized miRNA expression profiles of WT upper and lower jejunum indicating a similar pattern of expression ( Figure 4a). The miRNA expression patterns of KO upper and lower jejunum samples were also similar with a correlation coefficient of 0.962. However, the lower correlation coefficients between WT and KO upper jejunum miRNAs (0.950) and between WT and KO lower jejunum miRNAs (0.954) signified significant expression pattern changes in the KO samples.
After excluding minimally expressed miRNAs with o10 normalized reads, the expression profile analysis revealed that in Srf KO samples, approximately one-third of miRNA species (151; 37.9%) were upregulated and one-third (124; 31.2%) were downregulated (Figure 4b). When total amounts of miRNAs were considered, however, KO samples had significantly more downregulated miRNAs (6.9 million reads) than upregulated miRNAs (0.3 million reads) relative to WT samples (Figure 4c). The IPA software analysis showed that of the 124 miRNAs that were downregulated in Srf KO samples, 11 were known miRNAs (including SMC-specific miR-143 and miR-145), 8 were predicted miRNAs and 105 were unknown miRNAs that may have been directly or indirectly induced by SRF (Figure 4d). Thus, the in vivo miRNA data from Srf KO mice confirmed that SRF regulates the expression of most miRNAs predominantly expressed in jejunum SMCs (Supplementary Table 2) and that KO of Srf in SMCs results in a significant reduction in the expression levels of SRF-induced miRNAs.
SRF-induced miRNAs target apoptotic proteins. To identify the proteins modulated by SRF in GI smooth muscle, a proteomic analysis of Srf KO and WT jejunum smooth muscle at PT15 was performed using multidimensional protein identification technology. The analysis identified 1531 proteins, which were filtered by a statistical cutoff (95% probability) into 1035 proteins (Table 2). We anticipated a significant decrease in number of expressed proteins with the Srf KO because many smooth muscle contractile genes, including smooth muscle actin and myosin heavy chain, are direct targets of SRF. 1 As expected, we found 82 downregulated proteins (Figure 5a), including several smooth muscle contractile proteins in Srf KO tissue (unpublished data). Interestingly, however, there were significantly more upregulated (141) than downregulated (82) proteins (Figure 5a), and overall, the total amount of upregulated proteins was greater than that of downregulated proteins (Figure 5b). This protein expression profile was consistent with loss in miRNA-mediated repression of target mRNA/proteins. Based on our experimental results, we hypothesized that the downregulated proteins are directly associated with the early synthetic state phenotypic changes of SMCs, whereas the upregulated proteins are linked to the late degeneration of SMCs in Srf KO mice. The IPA software analysis of the upregulated proteins (Supplementary Table S3) in relation to cell death pathways revealed that most of these proteins were indeed linked to known apoptotic pathways in cells such as cardiomyocytes, neuronal cells, hematopoietic cells and cancer cells (Figure 5c). The identified apoptosis-associated proteins that may be responsible for SMC death in Srf KO mice included CAST, CRYAB, HSPA5 and ATG7.

Discussion
Although several animal studies have consistently demonstrated that SRF is a key regulator in the development and maintenance of both embryonic and adult muscle cells, SRF-deficient phenotypes in the heart and GI tract differ  14 and heart development (cardiac muscle-specific Myh6 promoter-driven KO 15 ; vascular muscle-specific SM22α promoter-driven KO 16 ) as well as in GI tract development (SMC-specific Myh11 promoter-driven KO). 13 Similarly, inducible Srf gene mutageneses in adult mice     18 and SM22α (Tagln) has higher levels of expression than Myh11 in jejunal and colon SMCs. 19 Furthermore, Myh11 is expressed only in differentiated SMCs, 20 whereas SM22α is expressed in both proliferating and differentiated SMCs. 21 Although proliferating SMCs may represent mature SMC precursors, such as myofibroblasts, Figure 6 Interactions of SRF-dependent miRNAs and upregulated proteins. Interactions between SRF-dependent miRNAs and upregulated proteins were analyzed using IPA. Note that the SRF-dependent miRNAs target the upregulated proteins in Srf KO SMCs microRNA-mediated apoptosis C Park et al these cells are not true SMCs phenotypically. Therefore, the phenotypes from the congenital and inducible Myh11-Credriven Srf KO mice shown in this study more precisely reflect the role of SRF in embryonic and adult SMCs. We previously reported that GI smooth muscle of mice expressed 312 miRNAs, of which 36 were SRF dependent as evidenced by in vitro Srf knock-down. 10 Using miRNA-seq analysis, we have presented here more comprehensive miRNA profile of the GI smooth muscle from an in vivo Srf KO mouse model. The genome-scale miRNA profile obtained from this model consisted of 891 miRNAs, of which 124 were induced by SRF. Our results were further validated by the finding that 12 of the 36 SRF-dependent miRNAs, including miR-143 and miR-145 that were discovered through in vitro Srf knock-down were also found to be SRF dependent in the in vivo Srf KO model. Surprisingly, over 95% of miRNAs expressed in SMCs were affected by SRF in our in vivo model further alluding to the complexity of SRF influence on epigenetic regulation.
Among the miRNAs predominately expressed in SMCs, miR-143 and miR-145 are the most well established, and not surprisingly, these two miRNAs account for 78% of all miRNAs expressed in the smooth muscle of our in vivo Srf KO model (Supplementary Table 1). This finding can be partly explained by the fact that miR-143 and miR-145 are generated from the same primary transcript, and binding of SRF to a conserved CArG box located in the distal promoter region modulates their expression. 22 As SRF-induced miR-143 and miR-145 expression promotes GI SMC differentiation and suppression of proliferation, 10 deficiency of Dicer, which prevents generation of mature miRNAs, can lead to degeneration of SMCs in GI smooth muscle. 9 This SMC degeneration resembles that of Srf deficiency suggesting that both SRF and Dicer may be on the same regulatory pathway of SMC differentiation and survival.
Collectively, our data indicate that SRF-induced miRNAs have a critical role in SMC apoptosis and offers an explanation for why SMCs could not survive without miRNAs in the Dicer deficiency model. 9 Mechanistically, our findings suggest that SRF is the major regulator of predominantly expressed miRNAs in SMCs, which account for 95% of miRNAs expressed in SMCs. This includes miR-143/145 that function to target and inhibit apoptotic proteins ( Figure 6).
Remarkably, 63 of 141 proteins that were overexpressed with loss of SRF-induced miRNAs were associated with the massive apoptotic activity in Srf KO muscle at PT21 (late stage). The cell death occurred without necrotic disruption of plasma membrane (Figure 2b), which was consistent with apoptosis. DNA fragmentation, which is a hallmark of apoptosis, was detected in our studies (Figure 3h). Apoptosis naturally occurs in cells during development and aging, and apoptotic bodies are immediately phagocytosed by macrophages or adjacent normal cells. 23 SMC apoptosis in Srf KO muscle occurs in a narrow time window around PT21-P22 ( Figure 3). The process of phagocytosis or removal of apoptotic bodies follows instantly to avoid an inflammatory immune response. 23 The presence of inflammation in the smooth muscle as early as PT15-P18 suggests that there may have been SRF-independent changes before the induction of apoptosis. Our data, therefore, provides evidence that loss of SRF leads to apoptosis through loss of inhibitory miRNAs and offers insight into a new mechanism for SMC degeneration, which may occur under certain pathophysiological conditions.
Although the role of SRF in SMC growth, differentiation and phenotypic maintenance has been well established, 1-3 the evidence of SRF in apoptosis has only recently begun to emerge. Parlakian et al. 24 first observed that the number of apoptotic cells increase within the embryonic heart of the cardiac muscle-restricted (Myh7) Srf KO mice. Next, using a mammary epithelial cell model, Wiese et al. 25 showed that restoration of SRF antagonizes Myc repression of SRF target genes, attenuates Myc-induced apoptosis, and reverts a Myc-dependent decrease in Akt phosphorylation and activity. The authors also noted that there was no evidence for direct interaction of Myc or Miz1 with SRF and that Myc/Miz1 complexes most likely repressed SRF target genes by recruiting histone deacetylases to joint target promoters because transcriptional activation by SRF requires histone acetylation. 25 Furthermore, Sisson et al. 6 also recently demonstrated that the in vivo administration of CCG-203971, a SRF/MYOCD-related transcription factor (MRTF) pathway inhibitor, into two mouse lung fibrosis models promotes myofibroblast apoptosis, decreases alveolar plasminogen activator inhibitor-1 and leads to significantly reduced lung collagen content, thereby decreasing lung fibrosis. In another study examining the role of SRF in hepatocellular carcinoma resistance to sorafenib, an oral multi-kinase inhibitor, Bae et al. 5 showed that antisense inhibition of SRF expression in SH-J1 cells significantly enhances the apoptotic effects of sorafenib while reducing expression of mesenchymal markers and restoring expression of E-cadherin. In regards to miRNA involvement in apoptosis, Chen et al. 26 also recently reported their discovery of SRF as a potential target for miR-320a through bioinformatics analysis, which was validated by in vitro and in vivo studies. In the latter, the authors noted that miR-320a contributes to atherogenesis by downregulating SRF, inhibiting human-derived endothelium cell proliferation and inducing apoptosis. 26 Consistent with their findings, we also found that miR-320a was increased twofold in our miRNA-seq analysis of the Srf KO GI smooth muscle compared with WT control (Supplementary Table 1).
Interestingly, our ultrastructural and TUNEL assay findings in the GI smooth muscle of Srf KO mice were strikingly similar to that of the aortic smooth muscle in SMC-specific Myocd (myocardin) KO mice generated by Huang et al. 7 Mirroring our results, Huang et al. 7 noted increased TUNEL, caspase 3, caspase 9 and p53 activities at 4 months after tamoxifen exposure in their conditional KO of Myocd in SMCs. Consistent with programed cell death, the authors also noted nuclear chromatin aggregation, nuclear fragmentation and cytoplasmic apoptotic body formation on electron micrographs. Furthermore, the aorta of Myocd KO mice displayed autophagy markers at 14 days after tamoxifen exposure indicating that loss of MYOCD activated a stress-induced autophagy program. Grossly, Huang et al. 7 also noted significant dilation of the stomach, small intestine, bladder and ureters in the Myocd KO mice at 4 months following tamoxifen administration. The remarkable similarities in the gross, microscopic and molecular findings of their Myocd KO and our Srf KO mice models are not surprising given that  3 and interestingly, MRTF-B/MKL2 KO mice have a phenotype that is different from that of mice lacking MYOCD, which underscores their distinctive roles in SMCs. 28 Importantly, the combination of our results and that of Huang et al. indicates that both SRF transcription factor and the MYOCD cofactor are equally necessary for the development and maintenance of the contractile phenotype, as well as suppression of apoptotic activity in GI SMCs.
In absence of SRF, the GI smooth muscle temporarily hypertrophied at PT15 (Figure 1d). This transient growth was supported by evidence of increased numbers of proliferating cells in the muscle layers at PT10 and PT15 (Figure 3g). We recently found that SMCs dedifferentiate into myofibroblast-like cells that express low levels of PDGFRα (PDGFRα low cells) and SRF in a partial bowel obstruction injury model. 13 The PDGFRα low cells represent actively proliferating cells that are responsible for the hypertrophic response of the smooth muscle. 13 In the SRF-deficient mice of this study, SMCs may transform into these proliferative PDGFRα low cells during the early period following SRF ablation. The rapid and global loss of SRF may later trigger apoptosis through the SRF-induced miRNA pathway, whereas the slow and gradual loss of SRF in SMCs that ensues with partial obstruction may enable dedifferentiation and continual growth of SMCs without apoptotic induction.
In conclusion, we propose that SRF has a new functional role in the suppression of apoptosis via SRF-dependent miRNAs in SMCs. Rapid ablation of SRF expression in Srf KO SMCs reduces expression of SRF-dependent miRNAs that in turn induce abnormal overexpression of apoptotic proteins of the miRNA targets. The latter ultimately leads to SMC death. These findings provide new insight into how pathophysiological conditions of SMC degeneration may develop within the GI tract.

Materials and Methods
Generation of inducible Srf KO mice. The SMC-specific inducible Srf KO mouse line Tg(Myh11-Cre-ERT2);Srf lox/lox was generated by cross-breeding a Srf lox/lox female homozygote mouse (The Jackson Laboratory, Bar Harbor, ME, USA) with Tg(Myh11-Cre-ERT2) male mouse 29 according to procedures approved by the Institutional Animal Care and Use Committee at the University of Nevada, Reno.
Tissue preparation. The transgenic KO and WT mice were anesthetized by isoflurane inhalation after multiple time points following tamoxifen injection. Stomach, small intestine and colon were dissected from the mice. Whole tissue was used for hematoxylin and eosin (H&E) staining, electron microscopic examination, immunohistochemistry, proliferation and apoptosis assays. Smooth muscle stripped free of mucosa and submucous plexus was used for RNA-seq, DNA fragmentation analysis, proteomics and confocal microscopic examination.  ), was used for sequencing data analysis. All miRNAs were annotated from pre-miRNA and mature miRNA sequences listed in the latest release of miRBase. 30 miRNA expression profiles for each sample were generated for comparison by normalizing with the total number of miRNAs.
DNA fragmentation analysis. Genomic DNA was extracted from jejunal smooth muscle with AllPrep DNA/RNA/Protein Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. DNA electrophoresis was carried out on 1% agarose gel in TAE buffer at 50 V for 2 h. DNA was stained with ethidium bromide before viewing and imaging by GelDoc XR (Bio-Rad Laboratories, Milan, Italy).
Proteomics analysis. Soluble proteins were prepared from jejunum tissues of WT and Srf KO mice (n = 6). Each extract contained proteins from two mice. Proteins in the extracts were trypsin digested and analyzed by two-dimensional LC-MS/MS on a Thermo Finnigan LTQ-Orbitrap (Thermo Scientific, Fremont, CA, USA). Data were analyzed using Sequest (The Scripps Research Institute, La Jolla, CA, USA) and validated with Scaffold (Proteome Software, Portland, OR, USA). Proteins were quantified by spectral counting and potentially perturbed molecular pathways were found with Ingenuity Computational Pathway Analysis (IPA; QIAGEN, Redwood City, CA, USA).
Bioinformatics analysis. Genes of miRNAs and mRNAs regulated by SRF were analyzed using IPA. The IPA software analyzed relationships (1) among downregulated miRNAs and SRF (Figure 4d), (2) upregulated proteins and cell death or apoptosis ( Figure 5c) and (3) among downregulated miRNAs, upregulated proteins and SRF ( Figure 6). Genomic location of exons, CArG boxes (collectively termed the 'CArGome') and SRF-binding sites (ChIP data available in UCSC ENCODE) in the genes of transcriptional variants were analyzed using Smooth Muscle Cell Genome Browser, 19 which was custom-built on the UCSC Genome Browser. 31 The CArGome Search Browser within the Smooth Muscle Cell Genome Browser was updated with the transcriptome data of primary SMCs isolated from colon and jejunum. 19 Confocal microscopy and histological analysis. Jejunum and colon tissues were analyzed by whole mount and cryostat section staining using confocal microscopy as previously described. 32 Cell-specific primary antibodies against the following antigens were used: anti-MYH11 (1 : 800, Alfa Aesar, Ward Hill, MA, USA) for SMCs, CD117/c-kit (1 : 20, R&D Systems, Minneapolis, MN, USA) for ICCs or PGP9.5 (1 : 1000, UltraClone Limited, Wellow, UK) for enteric nerve system. For histological analysis, jejunum tissues were dehydrated, embedded in paraffin, cut into 4 μm-thick coronal sections, rehydrated and stained with H&E. Images were collected using an Olympus FV1000 confocal laser scanning microscope with Fluoview FV10-ASW 3.1 Viewer software (Olympus, Tokyo, Japan) or the iScan Coreo scanner (Ventana Medical Systems, Tucson, AZ, USA).
Electron microscopy. Whole jejunum and colon tissues were fixed in a fresh fixation buffer (3% glutaraldehyde and 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4) for several days at room temperature. Specimens were then postfixed in 1% osmium tetroxide for 2 h at 4°C, rinsed in distilled water, block stained with saturated aqueous uranyl acetate solution for 3 h, dehydrated in a graded series of ethyl alcohol and embedded in EPON 812. Ultrathin sections were stained with uranyl acetate and lead citrate and examined under microscope (Hitachi H-7650, Tokyo, Japan).
Statistical analysis. Protein quantification was done by spectral counting and the data were analyzed using the in program statistical package provided with Scaffold. Scaffold was used to generate P-values for proteins based on quantitative microRNA-mediated apoptosis C Park et al