Animal and tissue preparation

Pdgfra^eGFP/+ mice [19] were obtained from Jackson Laboratory. Pdgfra^eGFP/+ mice were crossbred with Myh11^Cre-ERT2/+ [21] and Rosa26^LacZ/LacZ (Jackson Laboratory) to generate Pdgfra^eGFP/+;Myh11^Cre-ERT2/+;Rosa26^LacZ/+. Intestinal partial obstruction surgeries were performed on one month old Pdgfra^eGFP/+ mice and one month old Pdgfra^eGFP/+;Myh11^Cre-ERT2/+;Rosa26^LacZ/+ mice after five consecutive days of intraperitoneal tamoxifen injection as previously described [22]. Jejunal and colonic tunica muscularis, from mice that underwent partial obstruction or sham surgery, were used to isolate PαC through flow cytometry. The animal protocol was approved by the Institutional Animal Care and Use Committee at the University of Nevada-Reno Animal Resources.

Human tissue preparation

Segments of human colon were obtained from patients undergoing colonic resections due to neoplasm formation or diverticulitis at Renown Medical Center (Reno, NV). Paraffin embedded sections and/or frozen tissue samples of human small intestine, obtained from patients with Crohn’s disease and control normal colon, were obtained from Stanford University School of Medicine, Stanford, California. The Human Subjects Research Committees at Renown Regional Medical Center, and the Biomedical Institutional Review Board at University of Nevada, Reno approved the use of human tissues.

Flow cytometry and fluorescence-activated cell sorting (FACS)

Cells were dispersed from the tunica muscularis of mouse jejunum/colon and GFP⁺ PαC were sorted from dispersed cells using FACS as previously described [23]. Isolated PαC were lysed and pooled from approximately 30 mice (15 males and 15 females) and used to isolate total RNAs as one collective sample. GFP⁺ PαC were also isolated from partial obstruction surgery mice along with sham operation control mice in a similar manner.

Isolation of total RNAs

Total RNA was isolated from jejunal PαC (JPαC), and colonic PαC (CPαC) using the mirVana miRNA isolation kit (Life Technologies, Carlsbad, CA). Quality of total RNAs was analyzed via NanoDrop 2000 Spectrometer (Thermo Scientific, Waltham, MA) and 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA).

Real time PCR

cDNA libraries were constructed through reverse transcription of the total RNA isolated from FACS-purified PαC as previously described.[24] Reverse-transcription polymerase chain reaction (RT-PCR) and quantitative PCR (qPCR) analyses of cDNA were performed as previously described [24]. All primers used for RT-PCR are shown in S11 Table.

Construction of RNA-seq libraries and next-generation sequencing

Two RNA-seq libraries were generated and sequenced via Illumina HiSeq 2000 (Illumina, San Diego, CA) following the vendor’s instruction at LC Sciences (Houston, TX) as previously described.[17]

Bioinformatics data analysis

Paired-end sequencing reads were processed and analyzed as previously described [17]. A cutoff of FPKM = 0.025 generated equal false positive and false negative ratios of reliability. The expression levels of transcripts with FPKM values less than 0.025 was considered to be 0.

Confocal microscopy and immunohistochemical analysis

Jejunal tissue was analyzed by whole mount and cryostat section staining or GFP fluorescence using confocal microscopy as previously described [25, 26]. Primary antibodies against the following antigens were used: anti-CACNA1G-C (rabbit, 1:50, SantaCruz, TX), anti-CACNA1G-N (rabbit, 1:50, Alomone Labs, Jerusalem, Israel and goat, 1:50, SantaCruz, TX), anti-PDGFR-alpha for mice (goat, 1:100, R&D system, MN), anti-PDGFR-alpha for humans (goat, 1:50, R&D system, MN), anti-PDGFR-beta (goat, 1:200, R&D system, MN), and anti-ACTA2 (rabbit, 1:200, Abcam, MA). Images were collected using the Fluoview FV10-ASW 3.1 Viewer software (Olympus, Tokyo, Japan) with an Olympus FV1000 confocal laser scanning microscope. Cryostat sections were also stained with β-galactosidase using LacZ Tissue Staining Kit (InvivoGen, San Diego, CA).

Western blot

Protein was extracted from jejunal tissue samples of Pdgfra^eGFP/+ mice and human GI tissues. Western blotting was performed as previously described [27]. Primary antibodies against the following antigens were used: anti-CACNA1G-C (rabbit, 1:100, SantaCruz, TX), anti-CACNA1G-N (goat, 1:100, SantaCruz, TX), and GAPDH (rabbit, 1:5000, Cell Signaling, MA).

Availability of supporting data

The PαC transcriptome was added to the Smooth Muscle Genome Browser [17] in the custom track of the UCSC genome database (UCSC Smooth Muscle Genome Browser) [20]. The browser is available at http://med.unr.edu/physio/transcriptome

(It requires Google Chrome and takes approximately a few minutes to upload the large files). The genome browser contains the transcriptome menus on the “Custom Tracks.” Each menu has different display options.

The abbreviated instructions are as follows: 1) To search transcriptional variants of a gene, type in the gene symbol, and click “go.” 2) Under “Custom Tracks,” select the view option (e.g., “full”) for type of sample (e.g., “PαC Jejunum”), and click “refresh.” 3) Select the bioinformatics data of interest (e.g., click on “full” under “RefSeq Genes” in “Genes and Gene Predictions”), and then click “refresh.” 4) Click “configure” to optimize views (change image width and text size).

The RNA-seq data from this study have been also submitted to the NCBI: jejunal PDGFRαC, GSM1388410 and colonic PDGFRαC, GSM1388411.

Statistical analysis

qPCR data obtained in the present study was compared using the student’s t test in order to determine whether the differences were statistically significant. Measured variables were expressed as mean ± SEM. The differences in mean values between the two animal groups (sham and hypertrophy) were evaluated and considered significantly different when p ≤ 0.05 or p ≤ 0.01.

Identification and isolation of PDGFRα⁺ cells

PDGFRα⁺ cells (PαC) were identified by eGFP expression in GI smooth muscle tissues of Pdgfra^eGFP/+ mice [15, 19] (left panel, Fig 1A). Nuclear eGFP-labeled PαC were confirmed through immunohistochemistry utilizing an anti-PDGFRA antibody (middle panel, Fig 1A). Nuclear eGFP coincided with antibody labeling, as previously demonstrated work [15] (right panel, Fig 1A). Primary PαC from jejunum and colon were further analyzed by flow cytometry. Distinct populations of eGFP⁺ PαC were identified in smooth muscles from the jejunum and colon (Fig 1B). There were at least two distinct groups of jejunal and colonic PαC, which expressed eGFP at relatively high or low levels. eGFP⁺ PαC from jejunal and colonic smooth muscle were only 4–6% of the total events observed by flow cytometry. Therefore, we sorted PαC from 30 mice on the basis of eGFP by FACS, pooled them according to tissue of origin, and isolated mRNAs from each of the two groups of cells.

Download:

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

Fig 1. Anaysis of transcriptomes obtained from jejunal and colonic PDGFRα⁺ cells.

(A) Identification of PDGFRα⁺ cells in the intestinal smooth muscle with eGFP and a PDGFRA antibody. A z-stack image, obtained through confocal microscopy, of whole-mount jejunum muscularis showing PDGFRα⁺ cells expressing eGFP in the nucleus (left). Immunohistochemistry of PDGFRα⁺ cells using a PDGFRA antibody (middle). Merged images of eGFP and PDGFRA (right). (B) Primary eGFP⁺ PDGFRα⁺ cells from jejunum (left) and colon (right) identified (circled) by flow cytometry. (C) Venn diagram showing the number of genes identifed in jejunal and colonic PDGFRα⁺ cells (JPαC anc CPαC) by RNA-seq. (D) Comparison of expression levels of genes in JPαC and CPαC. (E) Gene ontologies reported in JPαC and CPαC. The gene ontology (GO: function, process, and component) of PαC-specific genes was analyzed, and key GO terms were compared using normalized expression (FPKM) percentile.

https://doi.org/10.1371/journal.pone.0182265.g001

Comparison and analysis of transcriptomes in PDGFRα⁺ cells

To identify all genes expressed in PαC in the jejunum and colon, we obtained the PαC transcriptome by performing RNA-seq on the previously mentioned pooled samples. The transcriptome consisted of 16,252 (jejunal PαC) and 15,735 known genes (colonic PαC) (S1 Table). We obtained 137–149 million reads of which 90–91% were mapped to the genome. By gene annotation, we found 48,974 and 48,894 unique gene isoforms in jejunal and colonic PαC, respectively. Complete lists of all the isoforms identified in this study are shown along with tracking ID, gene ID/name, chromosome location, isoform length, and expression levels in both jejunal and colonic PαC (S2 Table). PαC expressed an average of 3 isoforms per gene, produced from different transcription start sites, and/or subject to alternative splicing (S1 Table). Most genes (15,249) were expressed in both jejunal and colonic PαC, but a few hundred genes within each cell-type were specific to the tissue of origin. (Fig 1C). More cell-specific genes were resolved in jejunal PαC (1,003) than in colonic PαC (486). A complete list of the genes expressed in jejunal and colonic PαC are shown with their respective, combined expression levels of all splice variants and numbers of splice variants can be found in S3 Table. Expression levels of all genes in the two PαC groups were compared to each other. Several hundred genes were expressed at high levels (>100 of FKKM) in both PαC groups while genes expressed at low levels (<10 of FKKM) showed a more divergent expression pattern (Fig 1D). Overall expression profiles were similar in PαC from jejunum and colon (correlation coefficient = 0.87). To validate the identity of the cells, markers specific for each cell type (Pdgfra for PαC, Kit for ICC, and Myh11 for SMC) were examined. Jejunal and colonic PαC dominantly expressed Pdgfra over ICC and SMC (S1A & S1B Fig). Conversely, Kit expression was minimal in jejunal and colonic PαC (S1C & S1D Fig). However, jejunal PαC, but not colonic PαC, expressed Myh11, but levels were lower than levels shown in SMC (S1E & S1F Fig), suggesting that jejunal PαC may include PDGFRα⁺ SMC precursors [28]. The expression levels of the cell-specific markers validate the authenticity of the RNA-seq data obtained from jejunal and colonic PαC.

To further investigate cell identity and function from our transcriptome data, we analyzed gene ontology (GO) terms of cell specific genes abundantly expressed in each cell type. This analysis revealed key GO terms that distinguish PαC from SMC [17] and ICC [18] (Fig 1E). Key GO terms obtained from the two types of PαC were similar, suggesting they play similar functions in different regions of the GI tract. The most common terms related to structural and extracellular matrix function including plasma membrane, as well as the extracellular region, matrix, and space (Fig 1E).

Addition to UCSC Smooth Muscle Genome Browser

In an effort to have an interactive transcriptomic profile for each cell type found within the SIP cell types, we previously built a smooth muscle genome browser with jejunal and colonic SMC [17] as well as ICC [18] using the UCSC genome browser (UCSC Smooth Muscle Genome Browser) [20]. We have updated the browser with the PαC transcriptome data, which now contains all three cell types of the SIP (SMC, ICC, and PαC) in jejunal and colonic smooth muscle tissue. This transcriptome browser provides not only the genomic structure of each splice variant (promoter region, exons, and introns) for all genes expressed in PαC (S2 and S3 Tables) as well as SMC and ICC, but also allows for analysis of our transcriptome data using the gene expression and regulation data (ENCODE) [29] that other groups have deposited in the genome database.

Comparison and analysis of ion channels and transporters expressed in PDGFRα⁺ cells

PαC are coupled via gap junctions to SMC, as are ICC to SMC, creating an electrical syncytium (SIP syncytium) that contributes to the regulation of GI motility [1]. Thus, electrophysiological events in one type of SIP cell can affect the excitability of the other cells in the syncytium. From the PαC transcriptome data we identified various ion channels and transporters that are expressed in jejunal and colonic PαC. We identified 513 and 469 ion channel and transporter isoforms expressed in jejunal and colonic PαC, respectively (S4 Table). The most highly expressed type of channels in both PαC is Ca²⁺ channels (Fig 2A). Expression of these channels makes sense in that the major functional conductance observed thus far in PαC is due to small conductance Ca²⁺-activated K⁺ channels encoded by Kcnn3 (SK3) [6, 15]. Sources of Ca²⁺ are therefore required for activation of SK3. However, to date, no functional voltage-dependent Ca²⁺ currents have been recorded from PαC under the voltage-clamp conditions of the experiments performed on these cells. In addition, hydrogen transporters are the most dominantly expressed transporters in PαCs of jejunum and colon (Fig 2B). The main channel and transporter classes were further analyzed to identify the most abundantly expressed and cell-specific isoforms.

Download:

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

Fig 2. Comparison of ion channel and transporter isoform genes expressed in PDGFRα⁺ cells.

(A) Comparison of expression levels of ion channel isoforms in JPαC and CPαC. (B) Comparison of expression levels of ion transproter isoforms in JPαC and CPαC. (C) Ca²⁺ channel isoforms enriched in JPαC and CPαC. (D) PαC-specific Ca²⁺ channel isoforms. Cell specificity was determined by comparative analysis of gene expression profiles among PαC, SMC, and ICC: PαC^{expression level (FPKM)}/[SMC^{expression level (FPKM)} + ICC^{expression level (FPKM)}]. (E and F) Voltage-dependent Ca²⁺ channel isoforms (L-type: Cacna1c & d, T-type: Cacna1h & g, R-type: Cacna1e, P/Q-type: Cacna11) expressed in JPαC (E) and CPαC (F).

https://doi.org/10.1371/journal.pone.0182265.g002

Identification of a new T-type Ca²⁺ channel Cacna1g specifically expressed in PDGFRα⁺ cells

Each type of PαC differentially expresses Ca²⁺ channel isoforms. The cells abundantly express voltage dependent Ca²⁺ channels Cacna1g and Cacna1h (T-type, Cav3.1 and Cav3.2), Cacna1d (L-type, Cav1.2), and Ca²⁺ channel regulators including Gas6 and Jph2 (Fig 2C). Gas6 appears to be the most abundantly expressed in both PαC. However, the L-type and T-type channels are differentially expressed in jejunal and colonic PαC (Fig 2E and 2F). Cacna1g and Cacna1d are predominantly expressed in colonic PαC while Cacna1h is mainly expressed in jejunal PαC (Fig 2C). Jph2 is also expressed predominantly in jejunal PαC (Fig 2C). Interestingly, Cacna1g appears to be the most cell specific to PαC (Fig 2D), thus we selected it for further study.

The Cacna1g gene encodes a low voltage-dependent Ca²⁺ channel subunit alpha1 G. The Ca²⁺ channel regulates a variety of Ca²⁺-dependent processes including muscle contraction, secretion, neurotransmission, cell motility, cell division, and cell death [30]. There are 11 transcriptional variants of Cacna1g expressed in PαC (Fig 3A and S2 Table). The gene is large and complex, consisting of 37 exons. It contains transcription start sites at seven different exons (V1-7) and alternative splicing occurs at exons 8, 14, 25, 30, 34, and 37 (Fig 3A). Cacna1g is expressed in PαC at significantly higher levels than seen in either SMC or ICC from both jejunum and colon (Fig 3B). Major variants expressed in jejunal and colonic PαC appear to be TCONS_00035784 (V1, 8,133 bp), TCONS_00042465 (V2, 5,230 bp), and TCONS_00039865 (V6, 4,201 bp) (Fig 3C and S2 Table). Ten variant cDNAs were retrieved from the browser, an open reading frame for each variant was identified, and all subsequent amino acid sequences were aligned and analyzed (S2 Fig). The channel consists of 4 pore and 24 transmembrane regions (4 S1-S6), each S4 contains 5–6 voltage sensor residues (Fig 3D). The four variant transcripts starting at exon 1 (TCONS_00035784, TCONS_00035783, TCONS_00035785, TCONS_00040966) appear to contain all the pore regions and transmembrane domains, while the other variants are truncated at either the N- or C-terminus (S2 Fig). The four variants contain the differentially spliced exons E8, E14, and E25. Among the four full-length variants, the most dominant variant expressed in both PαC is TCONS_00035784, in which E14 and E25 are missing (Fig 3D).

Download:

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

Fig 3. Identification of mutiple Cacna1g trancritional variants.

(A) A genomic map view of Cacna1g variants expressed in JPαC and CPαC. Seven alternative initial exons (V1-7) are circled in red and six differentially spliced exons (E8, E14, E25, E30, E34, and E37) are boxed in blue. (B) Expression (FPKM) levels of total Cacna1g mRNAs in JPαC and CPαC. (C) Expression levels of Cacna1g transcriptional vaiants in JPαC and CPαC. (D) A topological map of CACNA1G variants. Each circle denotes a single amino acid. Colors on amino acid sequence show distinct regions and domains. Red represents missing, or inserted, peptides from differentially spliced exons. Green represents start codons found in differentially spliced variants. Six transmembrane domains (S1-6) and a pore region are shown.

https://doi.org/10.1371/journal.pone.0182265.g003

PDGFRα^low and PDGFRα^high cells are proliferative in hypertrophy

We have previously reported that PαC are highly proliferative during intestinal smooth muscle hypertrophy [22]. Partial obstruction (PO) surgery on the small intestine of mice induced jejunal smooth muscle hypertrophy (Fig 4A). Both circular and longitudinal muscle layers were significantly thicker in PO than in sham operations (SO) (Fig 4B). eGFP⁺ PαC were obviously increased in the muscle layers (Fig 4C). Cytometric analysis identified two distinct populations of cells, eGFP^low and eGFP^high (Fig 4D). Although both populations were increased in partial obstruction, increases in the number of eGFP^low PαC was more dramatic. Increase in the amount of eGFP^low and eGFP^high PαC in hypertrophic muscle layers was also confirmed in whole mount images (Fig 4E). Since the amount of eGFP correlates with the expression of Pdgfra, our data indicates the existence of two populations of PDGFRα^low (Pα^lowC) and PDGFRα^high cells (Pα^highC) in hypertrophic smooth muscle where they are highly proliferative.

Download:

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

Fig 4. Increased PDGFRα⁺ cells in hypertrophic smooth muscle.

Hypertrophic tissue was surgically induced for ~2 weeks by placing a small silicone ring on the distal ileum of transgenic PDGFRα-eGFP mice to partially obstruct normal peristaltic movement. (A) Gross images of GI tract in sham and obstruction surgeries. (B) Representative H&E staining of jejunal cross sections from sham control and partially obstructed mice. Hypertrophied jejunum contained significantly thicker circular and longitudinal muscle layers compared to a sham control. Scale bar: 50 μm. (C) Representative confocal laser scanning images of jejunal cross sections from sham operation control and partial obstruction mice showing nuclear eGFP expression in PDGFRα⁺ cells and DAPI (blue) counterstained in the cells. Scale bar: 50 μm. (D) Two populations (eGFP^high and eGFP^low) of primary PDGFRα⁺ cells from hypertrophic jejunum identified by flow cytometry. Note that eGFP^high PDGFRα⁺ cells are significantly increased in partial obstruction smooth muscle. (E) A z-stack image, obtained through confocal microscopy, of whole-mount jejunum muscularis from the partial obstruction (bottom) and sham operation control (top) showing eGFP^high (arrow heads) and eGFP^low (arrows) PDGFRα⁺ cells. Scale bar: 50 μm.

https://doi.org/10.1371/journal.pone.0182265.g004

All subpopulations of differentially localized PDGFRα⁺ cells are highly proliferative in hypertrophic tissue

PαC are differentially localized within the small intestine. Main populations were readily detected in the subserosal (PαC-SS), and myenteric region (PαC-MY), as well as the deep muscular plexus (PαC-DMP) by Pdgfra-eGFP expression levels and immunohistochemical analysis (Fig 5A–5C). The three populations were distinctive in shape and organization. Cross section images revealed three subpopulations localized in distinct regions (Fig 5D). They also showed muscular, submucosal, and mucosal PαC. These subpopulations were well organized in smooth muscle layers in the SO mice, but they were disorganized in hypertrophic tissue induced by PO (Fig 5E). All PαC subpopulations were expanded in the subserosal layer, muscle layers, myenteric region, and deep muscular plexus in PO models. PαC-SS were most expanded, and three distinct subpopulations of cells were detected in the outside layer, middle layer, and inside boundary of the serosal epithelium. Most proliferating PαC were smaller while some cells were hypertrophic. These data suggest that the phenotypes of PαC are dynamic, and they appear to be a major cell type contributing to the remodeling of the tunica muscularis in the hypertrophic response to partial obstruction injury.

Download:

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

Fig 5. Identification of PDGFRα⁺ cell subpopulations dedifferentiated in hypertrophic smooth muscle.

(A-C) Confocal section images of PDGFRα⁺ (Pdgfra-eGFP⁺) cell subpopulations identified in the subserosal layer (SS), myenteric region (MY), and deep muscular plexus (DMP) with PDGFRA antibody (A), eGFP (B) and merged (C) in jejunum. (D and E) Cross section images of PDGFRα⁺ cell subpopulations (green) in sham operation (SO, D) and partial obstruction (PO, E). Proliferating PDGFRα⁺ cells are marked by arrows. LM, lonitudinal muscle; CM, circular muscle; SM, submucosa; M, mucosa. All scale bars are 50 μm.

https://doi.org/10.1371/journal.pone.0182265.g005

Expression of Cacna1g is induced in serosal PDGFRα⁺ cells and dedifferentiated SMC in hypertrophic tissue

Induced expression of the transcriptional variants of Cacna1g in PO-induced hypertrophic tissue was examined by RT-PCR. All the variant exons with four alternative transcriptional start sites and/or three alternative spliced sites were increased in jejunal smooth muscle tissue of PO compared to SO (Fig 6A). Next, we examined the gene induction in isolated Pα^lowC and Pα^highC by RT-PCR. The expression of the three main variants (see Fig 3C: V1, V2, and V6) was induced in both types of isolated PαC (Fig 6B). Then we measured expression levels of the three variants in the cells by qPCR. mRNA expression levels of all three variants of Cacna1g in Pα^lowC and Pα^highC in PO were significantly higher than in those cells in SO (Fig 6C). Furthermore, the gene was induced at much higher levels in PO Pα^lowC than PO Pα^highC. Increased expression of the mRNA in PO Pα^lowC also mirrored the CACNA1G protein expression in PO-induced hypertrophic tissue. Both C-terminal (CACNA1G-C) and N-terminal (CACNA1G-N) antibodies detected two bands at ~250 kDa, showing that CACNA1G was expressed at significantly higher levels in PO as compared to SO (Fig 6D). Expression of CACNA1G was examined within the subpopulations of PαC. CACNA1G (detected by CACNA1G-C antibody) appeared to be expressed only in serosal PαC, in which the protein was increased in hypertrophic tissue induced by PO (Fig 6E). Expression and induction in serosal PαC was confirmed through immunohistochemical analysis of cross-sectioned tissue utilizing the CACNA1G-C antibody (Fig 6F). CACNA1G was localized to PαC in the serosal layer. The protein was also co-localized with PDGFRB (PDGFRA partner β subunit) suggesting that CACNA1G is expressed in PDGFRα⁺/β⁺ in the serosal layer. The protein expressing cells were changed in shape and number. In the PO-induced hypertrophic serosal layer, the cells become round and hyperplasic in the outside layer of the enlarged epithelium, as compared to a long spindle shape in SO. Furthermore, we examined localization of the Ca²⁺ channel by CACNA1G-N antibody. However, the N-terminal antibody detected cells in different layers of the muscle. In SO, cells in the serosal, circular muscle, and longitudinal muscle layers were weakly stained by CACNA1G-N antibody (Fig 6F). The N-terminal antibody immunoreactivity was greatly increased in cells within the serosal, circular muscle, and longitudinal muscle layers in PO. CACNA1G-N⁺ cells were colocalized with PDGFRα⁺/β⁺ within the serosal layer (Fig 6F). In addition, CACNA1G-N antibody robustly detected hypertrophic SMC in the circular and longitudinal muscle layers in PO.

Download:

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

Fig 6. Induced expression of Cacna1g mRNAs and protein in hypertrophic smooth muscle.

(A) Expression of Cacna1g exons with alternative transcriptional start sites (ATSS) and/or alternative spliced sites (ASS) in sham operation (SO) and partial obstruction (PO) examined by RT-PCR. NTC is non-template control. (B) Detection of Cacna1g mRNAs in isolated eGFP^high and eGFP^low PDGFRα⁺ cells from sham operation and partial obstruction opertaion by RT-PCR. PCR products were analyzed with a DNA size marker on 1.5% agarose gels. Note expression of Cacna1g is increased in eGFP^high and eGFP^low PDGFRα⁺ cells. (C) Quantification of Cacna1g mRNAs by qPCR. * p ≤ 0.05 and ** p ≤ 0.01, SO versus PO. (D) Western blot analysis using N-terminal and C-terminal CACNA1G antibodies (CACNA1G-N and CACNA1G-C), showing that the protein has significantly higher expression levels in hypertrophic tissue induced by partial obstruction. (E) Detection of serosal PDGFRα⁺ cells expressing CACNA1G in partial obstruction models by CACNA1G-C antibody. Scale bars are 50 μm (F) Confocal cross section images of serosal PDGFRα⁺ cells in sham operation and partial obstruction screened with CACNA1G-N and CACNA1G-C antibodies co-labeled with PDGFRA and PDGFRB antibodies. Scale bars are 50 μm.

https://doi.org/10.1371/journal.pone.0182265.g006

The expression levels of CACNA1G were also examined in diseased human GI tissues (colorectal cancer, Crohn’s disease small intestine, and diverticulitis colon). The normal colon tissue contained distinct circular and longitudinal smooth muscle layers and healthy mucosa layer while the colon cancer tissue contained hypertrophied muscle layers and cancerous mucosa (Fig 7A). The Crohn’s disease and diverticulitis tissues also showed hypertrophied muscle layers. In addition, the diverticulitis tissue showed a degenerated mucosa layer.

Furthermore, all three disease tissues including colorectal cancer had severe inflammation (Fig 7B). RT-PCR confirmed expression of CACNA1G in the human tissues with two primer sets amplifying PCR products spanning the two independent regions (Fig 7C). The CACNA1G protein was detected, with a mass of 250 kDa, and the expression was robustly increased in diseased tissues (Fig 7D). Immunohistochemical analysis (CACNA1G-N) of cross sectioned tissues showed CACNA1G protein localized in PDGFRα⁺/β⁺ cells at low level in healthy colon, but the protein was increased in PDGFRα⁺/β⁺ cells within the hypertrophied smooth muscle of colorectal cancer (Fig 7E). Expression of CACNA1G was also increased in ACTA2⁺ SMC in the hypertrophied smooth muscle. In agreement with the Western blot data, CACNA1G was increased within PDGFRα⁺/β⁺ cells and SMC in the hypertrophied smooth muscle of colorectal cancer.

Download:

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

Fig 7. Induced expression of CACNA1G in human GI diseases.

(A) Anatomical cross-sections of colorectal cancer (CRC), Crohn’s disease (CD) small intestine and diverticulitis (DIV) colon tissue. Marginal colon tissue away from diverticula was used as healthy colon tissue. LM: longitudinal smooth muscle, CM: circular smooth muscle, Mu: mucosa. Scale bars are 100 μm. (B) Inflamed CRC compared to healthy colon tissue. (C) Detection of human CACNA1G mRNAs in diseased GI tissues (n = 2). Two independent regions of CACNA1G exons (7–8, and 28–29) were amplified by RT-PCR and PCR products were analyzed with a DNA size marker on 1.5% agarose gels. GAPDH gene was used as an endogenous control. (D) Western blot analysis of diseased GI tissues (n = 2) using CACNA1G antibody (N-terminal). (E) Confocal cross section images of PDGFRα⁺ cells screened with CACNA1G antibody (N-terminal) co-labeled with PDGFRA, PDGFRB, and ACTA2 antibodies in CRC and healthy tissues. Scale bars are 50 μm.

https://doi.org/10.1371/journal.pone.0182265.g007

Comparative analysis of potassium, cation channel, chloride, and sodium channels

PαC express as many as 92 K⁺ channel subunits (S4 Table). K⁺ inwardly-rectifying channel subfamily J member 8 (Kcnj8), intermediate/small conductance Ca²⁺-activated K⁺ channel subfamily N member 3 (Kcnn3), ATP-binding cassette sub-family C (CFTR/MRP) member 9 (Abcc9), and K⁺ voltage-gated channel subfamily G member 4 (Kcng4) were predominantly, but differentially, expressed in JPαC and CPαC. Kcnn3, Abcc9, and Kcng4 were more highly expressed in CPαC while Kcnj8 was more highly expressed in JPαC (S3A Fig). Kcnn3 was also the most PαC–specific gene (S3B Fig). Cation channel subunits enriched in JPαC and CPαC include aquaporin 1 (Aqp1), amine oxidase copper containing 3 (Aoc3), and cholinergic receptor, nicotinic beta polypeptide 1 (muscle) (Chrnb1). Aqp1 and Aoc3 were expressed more in JPαC, but Chrnb1 was predominantly expressed in CPαC (S3C Fig). PαC–specific cation channel subunits include amiloride binding protein 1 (Abp1) and peroxisomal biogenesis factor 5-like (Pex5l) (S3D Fig). Both JPαC and CPαC had FXYD domain-containing ion transport regulator 1 (Fxyd1) as the most highly expressed cation channel subunit (S3E Fig), as well as Cl^- channel Ca²⁺ activated 3 (Clca3) and Cl^- channel Ca²⁺ activated 2 (Clca2) being the most PαC–specific among Cl^- channel subunits (S3F Fig). As it pertains to Na⁺ channel expression, Na⁺ channel voltage-gated type VII alpha (Scn7a) and Na⁺ channel voltage-gated type I beta (Scn1b) were the most predominantly expressed but were differentially expressed between JPαC and CPαC: Scn7a was JPαC-dominant and Scn1b was CPαC-dominant (S3G Fig). Additionally, Scn7a was expressed in the most PαC-specific manner (S3H Fig).

Comparative analysis of hydrogen transporters

Hydrogen transporter isoforms are the main class of transporters expressed in JPαC and CPαC (Fig 2B). Dominant isoforms expressed in both PαC are H⁺ transporting ATP synthase subunits that consist of mitochondrial F1 and F0 complexes (S4 Table). Among them, H⁺ ATP synthase mitochondrial F1 complex β (Atp5b) and H⁺ Transporting, Mitochondrial Fo Complex Subunit F6 (Atp5j) were the most abundantly expressed in JPαC and CPαC, respectively (S4A Fig). PαC–specific hydrogen transporter isoforms include ATPase H⁺/K⁺ transporting non-gastric alpha polypeptide (Atp12a) and ATPase H+ transporting lysosomal V0 subunit A4 (Atp6v0a4) (S4B Fig).

Comparative analysis of growth factors, receptors, and transcription factors

PαC expressed 52 growth factors (S5 Table).The dominant growth factors expressed include Ptn, Gpi1, Nenf, Ogn, and Gdf10 which were differentially expressed in JPαC and CPαC (S5A Fig). Ptn and Gdf10 were expressed much higher in CPαC while Gpi1 and Ogn were expressed more in JPαC. However, Efemp1 and Ngf were the most PαC–specific growth factors for both cell types (S5B Fig). PαC expressed as many as 436 receptors (S6 Table). Some receptors that were highly expressed in either JPαC or CPαC were also differentially expressed between the two cell types. Atp5b was the most highly and predominantly expressed in JPαC (S5C Fig). Moreover, App and Lgals3bp were all expressed predominantly in CPαC as compared to JPαC. Epor and Pparg appeared to be the most PαC–specific receptors in both cell types (S5D Fig). PαC express as many as 134 transcription factors (S7 Table). The most predominantly expressed transcription factors include Tceb2, Ctnnb1, Dazap2, Jun, and Fos (S5E Fig). Interestingly, Tceb2, Ctnnb1, and Dazap2 were expressed higher in JPαC, but Jun, and Fos were expressed much higher in CPαC (S5E Fig). The most PαC–specific growth factor was Heyl (S5F Fig).

Comparative analysis of epigenetic enzymes and regulators

PαC expressed two DNA methyltransferases (Dnmt1, Dnmt3a), three Tet methylcytosine dioxygenases (Tet1, Tet2, Tet3), aka DNA demethylation enzymes, and a DNA oxidative demethylase (Alkbh1) (S8 Table). Dnmt1 and Tet2 appeared to be the most predominantly expressed enzymes involved in DNA methylation and demethylation, respectively (S6A Fig). Interestingly all the isoforms were expressed higher in CPαC than in JPαC. Dnmt1 was also the most cell specific in both cell types (S6B Fig). Furthermore, methyl-CpG-binding domain (MBD) proteins Mbd3 and Mbd2 appeared to be the most abundantly expressed MBD among the nine MBD protein isoforms and specifically expressed in both cell types (S6C and S6D Fig). A total of 39 genes expressed in PαC are associated with histone acetyltransferases (S8 Table). Dominant isoforms of histone acetyltransferases include Taf10, Taf9, and Ogt which were differentially expressed in JPαC and CPαC. Taf10 and Taf9 were expressed more in JPαC while Ogt was expressed at much higher levels in CPαC. (S7A Fig). Taf10 also appeared to be the most specific to both cell types (S7B Fig). A total of 16 genes with histone deacetyltransferase activity were expressed in PαC (S8 Table). The dominant isoforms of histone deacetyltransferases expressed in PαC include Hdac5, Hdac1, Hdac3, and Mta2 which were differentially expressed between JPαC and CPαC. Hdac5 was expressed at a higher level in JPαC while Hdac1, Hdac3, and Mta2 were more highly expressed in CPαC (S7C Fig). Hdac1 also appeared to be the most specific to both cell types (S7D Fig). A total of 24 genes expressed in PαC encode histone methyltransferase isoforms (S8 Table). The most predominantly expressed isoforms of histone methyltransferases include Setd3, Prmt1, Ehmt2, and Suv420h2 (Kmt5c) which were differentially expressed between JPαC and CPαC. Setd3 and Prmt1 were expressed at higher levels in JPαC while Prmt1 and Suv420h2 were expressed at much higher levels in CPαC (S7E Fig). Eth1 was the most specific to both cell types (S7F Fig). A total of 15 genes encoding histone demethyltransferases were found to be expressed in PαC (S8 Table). Generally, CPαC expressed higher levels of histone demethyltransferase genes when compared to JPαC. The predominantly expressed isoforms of histone demethyltransferase genes in PαC include Kdm1b, Jmjd6, Kdm2a, and Phf2 which were differentially expressed between JPαC and CPαC. Jmjd6, Kdm2a, and Phf2 had much higher expression in CPαC while Kdm1b was expressed at higher levels in JPαC. (S7G Fig). 2410016O06Rik (Riox1, bifunctional lysine-specific demethylase) and Jmjd6 were the most specific to both cell types (S7H Fig).

Comparative analysis of protein kinases and phosphatases

PαC expressed 354 protein kinase isoforms (S9 Table) and 105 phosphatase isoforms (S10 Table). The most prominently expressed protein kinases in JPαC include Mylk and Dmpk, which expressed at much lower levels in CPαC (S8A Fig). The predominantly expressed kinases in CPαC included Fgfr1 and Pdgfra (S9 Table). Pdgfra and Lyn were the most specific to both cell types (S8B Fig). The most prominently expressed isoforms of phosphatases in PαC include Ppp1cb, Ppp1r12a, Ctdsp2, and Ppp1cc and they were also differentially expressed between JPαC and CPαC. Ppp1cb and Ppp1r12a were expressed more in JPαC while Ctdsp2 and Ppp1cc were expressed much higher in CPαC (S8C Fig). Ptprn and Ptprn2 were the most specific to both cell types (S8D Fig).