c-Abl regulates gastrointestinal muscularis propria homeostasis via ERKs

The gastrointestinal tract is responsible for food digestion and absorption. The muscularis propria propels the foodstuff through the GI tract and defects in intestine motility may cause obstruction disorders. Our present genetic studies identified non-receptor tyrosine kinase c-Abl as an important regulator of the muscularis propria homeostasis and a risk factor for rectal prolapse. Mouse deficient for c-Abl showed defects in the muscularis propria of gastrointestinal tract and older c-Abl−/− mice developed megaesophagus and rectal prolapse. Inhibition of c-Abl with imatinib mesylate, an anti-CML drug, or ablation of c-Abl using Prx1-Cre, which marks smooth muscle cells, recapitulated most of the muscularis propria phenotypes. The pathogenesis of rectal prolapse was attributable to overproliferation of smooth muscle cells, which was caused by enhanced ERK1/2 activation. Administration of ERK inhibitor U0126 impeded the development of rectal prolapse in c-Abl deficient mice. These results reveal a role for c-Abl-regulated smooth muscle proliferation in the pathogenesis of rectal prolapse, and imply that long-term use of imatinib mesylate may cause gastrointestinal problems in patients while ERK inhibitor may be effective in treating rectal prolapse.

. c-Abl −/− mice showed progressive development of megaesophagus phenotype. (a) The whole GI tracts of wild type and c-Abl −/− mice. The GI tracts were dissected out and rinsed with cold PBS, from which the pictures were taken. N = 5. (b) c-Abl expression was detectable in the submucosa and muscular externa but not epithelial cells of esophagus and colorectal tissues. N = 3. (c) Representative histological sections of esophagus of day 1, 2 month, and 5 month-old c-Abl −/− and wild type mice. The organs were paraffin-embedded and the section slides were stained with hematoxylin and eosin. N = 3. (d) High magnification of muscles in the esophagus of 5 month-old c-Abl −/− and wild type mice. The organs were paraffin-embedded and the section slides were stained with hematoxylin and eosin. N = 3. was detectable in esophagus and colorectal tissues, mainly in the submucosa and muscular externa but not in epithelial cells (Fig. 1b). The dilation of the esophagus was associated with peristalsis problems as food was often seen stuck in the middle of the esophagus. However, the esophagus-stomach junction of c-Abl −/− mice did not show the "bird's beak" morphology characteristic of achalasia ( Fig. 1a and Supplementary Fig. S1) 27 .
We then sectioned the esophagus of c-Abl −/− and wild type mice of various ages and stained them with hematoxylin and eosin (Fig. 1c). All the six c-Abl −/− mice older than 2 months showed an increase in the diameter, although to various degrees. This phenotype was not obvious in newborn or 1-month old c-Abl −/− mice ( Fig. 1c and data not shown). Under higher magnifications, while the circular and the longitudinal muscle layers were tidily organized and well separated from each other in normal esophagus, the muscle fibers were not properly aligned or separated into distinct layers in c-Abl −/− mice. In older mice, the muscularis propria layer is thinner ( Fig. 1c and d), although the cross-section area of the muscle fibers even showed a modest increase (1.43 ± 0.35 fold, p < 0.05) and the total number of smooth muscle cells was increased by 1.0 fold in c-Abl −/− mouse esophagus ( Supplementary Fig. S2). The morphology of the epithelium layer also looked slightly altered (Fig. 1d), which might be secondary to muscularis propria defects as c-Abl is not expressed in epithelial cells (Fig. 1b). In sum, c-Abl −/− mice presents a megaesophagus model that is likely caused by muscularis propria defects. c-Abl −/− mice showed anomalies in the muscularis propria of stomach, colon and rectum and developed rectal prolapse. We then compared other parts of the GI tracts of c-Abl −/− and wild type mice and observed an obvious alteration in the forestomach of c-Abl −/− mice, with an increase in the thickness of the muscle layers and the secretory mucosa ( Supplementary Fig. S3a). However, in glandular stomach and intestine, the difference is minimal (Supplementary Fig. S3b). The number and the size of the intestine villi were not altered by c-Abl deficiency, nor was the muscle layer, which is rather thin in the intestine ( Supplementary Fig. S3c). The colon showed a thicker muscle layer in c-Abl −/− mice ( Supplementary Fig. S3d). The differential effects of c-Abl deficiency on esophagus, stomach, and intestines may be caused by the different nature of the muscle tissues in these organs. However, the distribution of food, chime, and metabolic waste along the digestive tract looked normal in the mutant mice ( Supplementary Fig. S1), suggesting that the gut motility was not impaired by c-Abl ablation. Mouse esophagus muscularis propria contains both skeletal muscle and smooth muscle while stomach and intestines muscularis propria is composed of solely smooth muscle.
Another obvious problem of c-Abl −/− mouse is rectal prolapse. While 2 month-old c-Abl −/− mice seldom showed rectal prolapse, all 4-5 month-old-mice showed this phenotype (Fig. 2a). In human, rectal prolapse is a disorder that affects mainly elderly women and its etiology is still unknown 28 . It is estimated that the annual incidence of rectal prolapse is 2.5 per 100, 000 people 29,30 . Histological analysis showed that c-Abl −/− mice had an increase in the thickness of the smooth muscle layer and the number of smooth muscle cells in the rectal prolapse tissues ( Fig. 2b and c). The muscle layers were also thicker in the colon-rectum junction region of c-Abl deficient mice (Fig. 2d). Thus, c-Abl −/− mice showed a thickened muscularis propria in the stomach, colon, and rectum and represents a model of atypical rectal prolapse.
Imatinib mesylate treatment also resulted in muscularis propria defects in mouse GI tract. We then determined whether long-term administration of imatinib mesylate showed any effects on the GI tract in mice, with comparison to c-Abl −/− mice. Two-month-old normal mice were injected with 50 mg/kg imatinib mesylate every other day for 14 weeks. This dose was shown to inhibit c-Abl activation in esophagus and rectal tissues, manifested by a decrease in the levels of phosphorylation of CRKL, a substrate of Abl kinases (Fig. 3a, Supplementary Fig. S4 and data not shown). We found that compared to control mice that received saline injection, imatinib mesylate led to a slight reduction of body weight in mice (data not shown). However, imatinib mesylate seemed not to cause obvious morphological change in the whole GI tract (data not shown).
Careful histological analysis of the esophagus, stomach, and rectum revealed that imatinib mesylate treatment showed an effect on the muscularis propria of all three organs. In esophagus, the muscularis propria showed a thinner muscle layer and distorted muscle fibers alignment ( Fig. 3b and c), which were similar to c-Abl −/− mice ( Fig. 1c and d). On the other hand, the forestomach and the large intestines showed an increase in the thickness of the muscularis propria in imatinib mesylate-treated mice ( Fig. 3d and e), which were in general consistent with the GI alteration observed in c-Abl −/− mice. However, imatinib mesylate administration showed no further effect on the phenotypes of c-Abl −/− mice (data not shown), suggesting there is no additive effect between c-Abl deficiency and inhibition.
Unlike c-Abl −/− mice, mice receiving imatinib mesylate treatment did not develop megaesophagus or rectal prolapse. The reason can be that c-Abl plays a role in the early development, c-Abl has kinase-independent functions, c-Abl is not completely inhibited by 50 mg/kg imatinib mesylate administrated every 48 hours, or the combination of these possibilities.
Ablation of c-Abl in smooth muscle cells led to muscularis propria defects and development of rectal prolapse. The above findings suggest that c-Abl may play critical roles in GI muscularis propria homeostasis and c-Abl deficiency leads to the development of megaesophagus and rectal prolapse. To test whether c-Abl plays a cell-autonomous role in smooth muscle cells, we crossed c-Abl f/f mice to Prx1-Cre mice, which is a marker for mesenchymal stem cells (MSCs) 31 . It is well established MSCs can differentiate into smooth muscle cells in addition to osteoblasts and chondrocytes 32,33 . Lineage tracing experiment using Prx1-Cre; Rosa-tdTomato mice revealed that Prx1 labeled the smooth muscle cells of esophagus and colon muscularis propria (Fig. 4a). Prx1-Cre; c-Abl f/f mice showed a decrease in the protein level of c-Abl in the colorectal smooth muscle tissues ( Fig. 4b and Supplementary Fig. S5). These mutant mice developed prolapse and a modest megaesophagus ( Fig. 4c and d). The rectum of Prx1-Cre; c-Abl f/f mice also showed increased thickness of smooth muscle layer and  aging-related phenotypes including lymphopenia, senile osteoporosis, and shortened lifespan [16][17][18] . One of the important regulators of aging is p16INK4a, a CDK inhibitor that is up-regulated in senescent cells and tissues 34,35 . We then generated c-Abl −/− p16Ink4a −/− mice and we found that p16Ink4a deficiency partially rescued the neonatal lethality of c-Abl −/− mice (Supplementary results Fig. S6a), as well as the reduction in fertility, without affecting the body weight and the organ weight of liver, spleen, kidney, or testes (Supplementary results Fig. S6b,c and data not shown) 36 . While none of the 6 c-Abl −/− male mice tested was fertile, all 6 c-Abl −/− p16Ink4a −/− mice could give rise to offspring when crossed to wild type female mice 37 . However, p16INK4a deficiency failed to rescue the megaesophagus phenotype of c-Abl −/− mice and only slightly rescued the rectal prolapse phenotype of c-Abl −/− mice (Supplementary results Fig. S6d and e). These results, taken together, suggest that c-Abl deficiency-induced onset of megaesophagus and rectal prolapse is not likely to be mediated by the pro-aging protein p16INK4a. Scientific RepoRts | 7: 3563 | DOI:10.1038/s41598-017-03569-0 c-Abl deficiency or inhibition promoted smooth muscle cell proliferation. We found that the rectal smooth muscle layer was thickened and the number of smooth muscle cells in the esophagus was also increased in c-Abl −/− mouse ( Supplementary Fig. S2), suggesting c-Abl deficiency caused overgrowth of smooth muscle in these two tissues. On the other hand, the structure of the skeletal muscle in the esophagus of c-Abl −/− mice appeared normal (Fig. 1d). Staining for ganglia in the esophagus and rectum revealed no alteration in c-Abl deficient mice ( Supplementary Fig. S7). Moreover, the colorectal crypts showed normal cell proliferation ( Supplementary Fig. S8a). Lastly, c-Abl deficiency did not cause infiltration of CD3 + cells into the esophagus or rectum ( Supplementary Fig. S9), nor did it affect the expression of inflammatory cytokines, e.g., TNFα, MPO, and IL6, in the intestines (data not shown), excluding possible involvement of inflammation in the development of the atypical rectal lapse. These results, together with the findings that ablation of c-Abl in Prx1+ mesenchymal cells reproduced the esophagus and rectal phenotypes, suggest that megaesophagus and rectal prolapse may be caused by defects in smooth muscle cells.
We found that c-Abl deficient mouse rectum showed an increased number of Ki67-positive cells (Fig. 5a). Yet the esophagus showed an insignificant change in cells positive for Ki67 ( Supplementary Fig. S8b). This could be due to slow progress of megaesophagus (a 1-fold-increase in the number of smooth muscle cells over a period of 5 months). These results suggest that that c-Abl deficiency results in overproliferation of smooth muscle cells. Previous studies have shown that knockdown of c-Abl affects myoblast and smooth muscle cell adhesion and proliferation in vitro [38][39][40][41] . We then tested whether imatinib mesylate show any effects on the proliferation, cell death, and morphology of smooth muscle cells. It was found that imatinib mesylate led to a modest but significant increase in the proliferation rates of smooth muscle cells, but not primary skeletal muscle cells (Fig. 5b and Supplementary Fig. S8c). Moreover, imatinib mesylate treatment showed no effect on death rates or morphology of all three cell types (data not shown).
Moreover, siRNA-mediated c-Abl knockdown led to accelerated proliferation of αSMA-expressing primary esophagus smooth muscle cells, manifested by a decrease in doubling time and an increase in BrdU-labeled S phase cells, without affecting the cell cycle profiles (Fig. 5c,d, Supplementary Fig. S10, and data not shown). c-Abl deficiency also led to an increase in the number of proliferating cells in primary rectal smooth muscle cells ( Supplementary Fig. S8f). c-Abl knockdown smooth muscle cells also showed an increase in proliferation in response to PDGF-AA ( Supplementary Fig. S8e). This is not consistent with a previous report showing that c-Abl is required for PDGF-AA or ET-1-induced proliferation in rat vascular smooth muscle cells 38 . This discrepancy may be caused by differences in cell types and/or cell culture conditions and warrants further investigation.

c-Abl deficiency increased smooth muscle cell proliferation via ERK. Smooth muscle overpro-
liferation in the absence of c-Abl activity suggests that c-Abl has a negative effect on the mitogenic pathways. Consistently, we found that imatinib mesylate treatment led to enhanced ERK1/2 activation on rectum sections (Fig. 6a). Consistent with previous finding that imatinib mesylate treatment led to activation of ERK1/2 in osteoblasts 18 , imatinib mesylate also led to an increase in ERK activation in smooth muscle cells (data not shown). Functionally, inhibition of ERK1/2 activation with U0126 or PD25901 could impede imatinib mesylate-induced increase in smooth muscle cell proliferation (Fig. 6b). More importantly, long-term administration of U0126 could prevent the development of rectal prolapse but not megaesophagus (Fig. 6c and data not shown), which were confirmed by HE staining of the colorectum (Fig. 6d). These data, taken together, suggest that c-Abl deficiency or inhibition leads to smooth muscle cell overproliferation via activating ERK1/2, which contributes to thickening of the muscularis propria layer in the gut and the development of rectal prolapse.

Discussion
The gastrointestinal tract is responsible for food digestion and absorption and about 10% of the population experience GI tract disease at some time in their lives, including intestinal immobility and rectal prolpase. However, the pathogenesis of these disorders is largely unknown 42 . Our present genetic studies indicate that non-receptor tyrosine kinase c-Abl plays an important role in the homeostasis of the muscularis propria of the GI tract. c-Abl deficiency or inhibition led to defects in muscularis propria homeostasis and mice with c-Abl ablated in the whole body or in Prx1+ cells develop megaesophagus and rectal prolapse. Megaesophagus can be a symptom of achalsia due to the loss of ganglion cells in Auerbach's plexus, or a symptom of myasthenia gravis, a neuromuscular disease 27 , or following the infection of the parasite Trypanosoma Cruzi 27 , while rectal prolapse mainly affects elder women and is believed to be caused by the weakening of the ligaments or muscle that holds rectum in place. Our findings suggest that pathogenesis of rectal prolapse may involve overproliferation of rectal smooth muscle cells, which can be a target for prevention and treatment of this disorder.
We found in the present study that megaesophagus and rectal prolapse develop in a progressive manner, with older mice showing more severe phenotypes. This is in line with the previous findings that c-Abl −/− mice show premature aging-related phenotypes such as osteoporosis [16][17][18] . Two critical regulators of aging at the cell and the organism levels are p16INK4a and p53 43 . By analyzing c-Abl −/− p16Ink4a −/− mice, we found that deletion of p16INK4a was able to rescue some of the defects including reduced fertility and shortened lifespan of c-Abl −/− mice, as well as defects in bone formation 21 , suggesting that p16INK4a is an important downstream effector of c-Abl. However, p16INK4a deficiency showed no effect on c-Abl deficiency-caused megaesophagus and only a minor effect on rectal prolapse, suggesting that these two aging-related disorders may not involve aging-promoting gene p16Ink4a. This is in contrast to the pro-aging roles for p16INK4a in bone formation, pancreases, and other organs 21,[44][45][46] . One explanation for these organ-specific effects of p16 could be that the smooth muscle cells in adult mice may have limited potential to proliferate, a process that is inhibited by p16INK4a.
How does c-Abl regulate the homeostasis of muscularis propria and how does c-Abl deficiency cause megaesophagus and rectal prolapse? c-Abl −/− mice, like imatinib mesylate-treated mice, seem to have problems in the proliferation and alignment of smooth muscle cells in most of the digestive organs. Recent studies show that c-Abl plays roles in myogenic differentiation and myoblast fusion 39,47,48 , smooth muscle cell proliferation in vitro 41,49 , and cardiomyocyte proliferation 36 . We showed that c-Abl plays a negative role in smooth muscle cells proliferation. c-Abl deficiency causes rectal prolapse by enhancing smooth muscle cell proliferation via Erks. The reason why c-Abl deficiency fails to cause a significant phenotype in small intestines may be that the small intestine contains much fewer microbes than the rectum. It is known that c-Abl can be activated by bacterial products such as lipopolysaccharide (LPS) 14,22 , which may affect smooth muscle proliferation via c-Abl. Development of megaesophagus is a complex process 9, 50, 51 , whether c-Abl deficiency-caused megaesophagus is solely mediated by smooth muscle needs further investigation.
Our cell-based studies with c-Abl knockdown and inhibition and in vivo studies revealed that the anti-proliferation function for c-Abl in smooth muscle cells could be mediated by enhanced ERK1/2 activation, as ERK inhibitor U0126 not only suppressed smooth muscle cell proliferation but also the development of rectal prolapse in c-Abl deficient mice. This study thus reveals a link between c-Abl and ERK activation in smooth muscle cells, which appears to underlie the pathogenesis of rectal prolapse caused by c-Abl deficiency. Our findings also suggest ERK inhibitors may be drug candidates for the treatment of rectal prolapse.
The present study provides important information regarding imatinib mesylate, a drug used to treat CML and other cancers 10,11 . We found that imatinib mesylate administration, like c-Abl deficiency, results in anomaly in the muscle tissue of esophagus and other parts of the GI tract. Recent clinical studies have revealed that long term use of imatinib mesylate could alter bone remodeling 52 and toxic myopathy in mice and congestive heart failure in human 53 . On the other hand, a recent study suggests that imatinib mesylate can also reduce necrosis, inflammation, and fibrosis in a Duchenne Meryon muscular dystrophy mouse model 54 . Our present study uncovered another potential adverse effect of imatinib mesylate on smooth muscle maintenance in the GI tract, which suggests that in imatinib mesylate-treated patients, GI tract, especially the muscularis propria, should be monitored.
In summary, our studies uncover an important role for non-receptor tyrosine kinase c-Abl in the homeostasis of the muscularis propria of the GI tract, and suggest that c-Abl −/− mouse is a model for megaesophagus and (atypical) rectal prolapse, which are likely caused by increased proliferation of smooth muscle cells. Moreover, this study also implies that imatinib mesylate might have an adverse effect on the maintenance of the GI muscularis propria in patients and that ERK inhibitors may be effective to treat rectal prolapse.

Materials and Methods
Mice. c-Abl −/− and c-Abl f/f mice were generated at S. P. Goff 's lab of Columbia University and p16Ink4a −/− mice were from R. DePinho, Harvard University, USA. Prx1-Cre and ROSA-tdTomato mice were purchased from The Jackson Laboratory. All mice were maintained at Shanghai Jiao Tong University. All animal experiments were approved in accordance with the University of Shanghai Jiao Tong's institutional guidelines on animal welfare [SYXK(SH)2011-0112]. All experimental methods were performed in accordance with the approved guidelines.
Tissue preparation and histological analysis. Mice with the age from 1 day to 5 months were used in this study. The whole digestive tract were dissected out and rinsed in cold PBS. They were directly frozen in liquid nitrogen for RNA or protein isolation, or for cryosections. Slides of 8 μmthickness were sectioned, fixed overnight, equilibrated in 30% sucrose, and embedded in OCT. Sections were stained with hematoxylin and eosin. For western blot analysis of c-Abl expression, these tissue samples were thawed on ice, weighted, and homogenated. Imatinib mesylate treatment of mice. Two-month-old mice were weighed and imatinib mesylate (Cat. S1026, Sellecchem, dissolved in saline) was intraperitoneal (IP) injected every other day at the dose of 5 mg/kg. The same amount of saline was injected as control. After 14 weeks, the mice were weighed and sacrificed. The whole GI tract was dissected out and fixed for histological analysis.
RNA isolation and real-time PCR. Total RNAs were collected from esophagus of 3 pairs of mice using TRIzol (Invitrogen) following the product manual. RNA was subjected to reverse transcription following the manufacturer's instructions (Roche). A portion of the reaction was used in real-time PCR assays.
Cell culture and cell proliferation assay. Primary skeletal muscle cells were prepared from newborn mice, whereas primary mouse esophagus smooth muscle cells were purchased from Pricells (Shanghai, China). Log phase cells were treated with imatinib mesylate or transfected with c-Abl siRNA for three days. Trypan blue staining was used to count the cells every day.
Western blot analysis. Tissues or cultured cells were lysed with TNEN buffer (50mM Tris, 150mM NaCl, 5 mM EDTA, 0.5% NP-40, 0.1% Triton X-100, 1 mM Na2VO3, 1 mM PMSF, 1 μg/ml aprotonin, 1 μg/ml leupeptin, 1 μg/ml pepstatin). The protein lysates of each sample was measured by BCA protein assay kit. Equal amounts of total proteins were loaded into 7.5% SDS-PAGE gel and transferred onto nitrocellulose membranes. Blots were incubated with the specific primary antibodies overnight and followed by incubating with secondary antibody and visualized by chemiluminescence. The western blot results were collected using a CCD camera and quantitated using the software provided by FluorChem M system (Protein Simple FM0405). We used Photoshop to process the images. Statistical analysis. Statistical analysis was performed using two-side Student's test. P < 0.05 is considered a significant difference.