Local presentation of Steel factor increases expression of c-kit immunoreactive interstitial cells of Cajal in culture

Rich, Adam, Steven M. Miller, Simon J. Gibbons, John Malysz, Joseph H. Szurszewski, and G. Farrugia. Local presentation of Steel factor increases expression of c-kit immunoreactive interstitial cells of Cajal in culture. Am J Physiol Gastrointest Liver Physiol 284: G313–G320, 2003. First published October 9, 2002; 10.1152/ajpgi.00093.2002.— The binding of Steel factor (SF) to c-kit initiates a signaling pathway essential for development and maintenance of interstitial cells of Cajal (ICC). Soluble and membrane-bound isoforms of SF are expressed in the gastrointestinal tract, but the role for either isoform in supporting ICC development is unknown. The aim of this study was to determine the role of SF in supporting ICC in culture. ICC were cultured from dissociated mouse jejunum and grown with fibroblast cell lines that produced either soluble, membrane-bound or membrane-restricted SF. ICC were identified and counted by c-kit immunoreactivity. The number of c-kit immunoreactive cells was greater in the coculture system compared with cultures grown without SF-producing fibroblasts. All forms of SFproducing fibroblasts increased ICC number in culture but physical separation of the fibroblasts from the c-kit immunoreactive cells, the addition of exogenous SF to the culture medium, or fibroblast-conditioned media did not. These results are consistent with the hypothesis that the membranebound form of SF preferentially contributes to expression of c-kit-positive ICC under cell culture conditions.

c-kit is also necessary to maintain ICC, because the administration of neutralizing c-kit antibody results in the subsequent disappearance of ICC (21,30).At the mRNA level, SF and c-kit expression are temporally correlated with SF expression preceding c-kit expression and with both peaking between embryonic days 13 and 15 in the murine small intestine (37).Thus SF/ c-kit signaling is required for both ICC development and for normal gastrointestinal function.
The KITLG gene encodes two distinct isoforms of SF, and both are synthesized as transmembrane proteins expressed at the cell surface (12).Proteolytic processing produces biologically active soluble SF, but the rate of cleavage of SF from the two isoforms differs.One isoform, known as the soluble SF, is characterized by rapid proteolytic cleavage and release from the plasma membrane.The other isoform is known as the membrane-bound SF isoform, because it lacks the proteolytic cleavage site encoded by exon 6 in the KITLG gene and releases soluble SF much more slowly (27).Therefore, both isoforms initially produce membrane-bound SF and the main difference is the rate of release of soluble SF.The membrane-bound SF isoform contributes higher steady-state levels of membrane-bound SF, whereas the soluble SF isoform contributes more significantly to steady-state levels of soluble SF.Each isoform is thought to play a specific physiological role, because the ratio of the membrane-bound-to-soluble SF varies in different tissues (12).Separation of these isoforms occurs in the spontaneous Steel-Dickie (Sl d ) mutant mouse that exclusively expresses soluble SF, possibly providing an indication for the role of membrane-bound SF (2).The Sl/Sl d mouse does not display a spontaneous, rhythmic electrical slow wave, and ICC are not present in the myenteric plexus (34).These observations suggest that membrane-bound SF provides an essential role for the development and/or maintenance of ICC-MY.Interestingly, ICC populations in the colon and the distal stomach appear normal in Sl d mice, which indicates that membrane-bound SF is not required for the development of these ICC populations (34).
The costs of publication of this article were defrayed in part by the payment of page charges.The article must therefore be hereby marked ''advertisement'' in accordance with 18 U.S.C.Section 1734 solely to indicate this fact.
The physiologically relevant form of SF supporting ICC in humans is not clear.SF is present in low concentration in serum but is thought to act locally, close to the site of production where the concentration is likely much higher (2).Soluble SF circulates as a dimer and a monomer, but the dimeric form is more biologically active (11).Membrane-bound SF is a more effective agonist for the c-kit receptor compared with soluble SF (25).Stimulation of c-kit with membranebound SF in a SF-dependent myeloid cell line MO7e resulted in more persistent activation of c-kit kinase compared with stimulation with soluble SF (25).Therefore, it is possible that membrane-bound SF may activate c-kit on ICC more effectively than soluble SF.
Elucidating the role of membrane-bound vs. soluble SF in the SF/c-kit signaling pathway in ICC is important, because loss of ICC is observed and may be involved in the pathophysiology of several motility disorders, such as slow transit constipation (8), diabetic gastroenteropathy (9), and pseudoobstruction (38).Loss of ICC may be a primary event or secondary to loss of a signaling molecule, such as SF, from a specific cell type in the gut.Fibroblasts genetically engineered to produce soluble SF or only membranerestricted SF have been used to differentiate the role of membrane-bound SF in hematopoietic tissue.With the use of these modified fibroblasts, Miyazawa et al. (25) demonstrated that, compared with soluble SF, membrane-bound SF induced more persistent activation of c-kit and increased the lifetime of activated c-kit complex at the plasma membrane.In addition, membranebound SF induced greater proliferation of an erythrocytic progenitor cell line compared with the soluble isoform of SF (16).These results suggest that soluble SF and membrane-bound SF play different physiological roles in hematopoiesis.
The aim of this study was to test the hypothesis that the membrane-bound form of SF is required for ICC development and survival.Primary cell cultures from the murine jejunum were used as a source of ICC and cocultured with murine fibroblasts that express soluble SF, membrane-bound SF, or membrane-restricted SF.The results suggest that local expression of SF is required for successful culture of ICC.

MATERIALS AND METHODS
Primary cultures were derived from freshly dispersed cells from the murine jejunum.BALB/c mice (Harlan Sprague-Dawley) of either sex between 9 and 15 days old were killed with CO 2 inhalation.The small intestine was removed and pinned out in a Sylgard-lined dish containing Hanks calciumfree buffer and 1% antibiotic-antimycotic (GIBCO).The muscularis propria was gently peeled from the mucosa and submucosa and placed in a collagenase-based dissociation cocktail.The cocktail contained 8 mg collagenase (model 4176; Worthington), 20 mg bovine serum albumin (model A-7511; Sigma), 20 mg trypsin inhibitor (model T-9128; Sigma), 5 mg adenosine triphosphate (model A2620-9; Aldrich) and 10 ml of calcium-containing Hank's balanced salt solution.The pH was adjusted to 7.0 with 0.1 M NaOH.After 15 min of incubation at 32°C in a gently shaken water bath, the tissue was washed twice with fresh calcium-free Hank's balanced salt solution and returned to 32°C.The tissue was then gently triturated every 3 min until single cells were obtained, for ϳ10 min.Cells were washed and resuspended in 12 ml of smooth muscle basal medium (Clonetics).This preparation likely contains ICC-MY, ICC in the deep muscular plexus region (ICC-DMP), and ICC distributed throughout the muscle layers (ICC-IM).The majority of ICC in culture is thought to be ICC-MY, because this region contains the largest number of ICC.Also, the majority of c-kit-positive cells in culture exhibits branching processes, similar to ICC-MY in situ.However, the exact proportions of ICC-MY, ICC-DMP, and ICC in the smooth muscle in the cell dispersions or the primary cultures cannot be determined at present, because all cells studied were c-kit positive.It is currently unclear whether the local environment and presentation of SF favor a particular ICC class or influence morphology and physiology.
Murine jejunal cell culture and murine cells/fibroblast coculture.Freshly dispersed cells obtained from the murine small intestine were cultured on 25-mm glass coverslips at a cell density of ϳ5 ϫ 10 4 cells/ml.Coverslips with established murine fibroblasts covering approximately one-third of the surface were used for cocultures with freshly dissociated murine jejunal cells.Control murine fibroblasts and fibroblasts genetically modified to express one of the SF isoforms were generously provided by David Williams (Indiana University School of Medicine, Indianapolis, IN).Four different fibroblast cell lines were used to provide soluble, membranebound, membrane-restricted, or no SF, respectively.A schematic of the different forms of SF and the potential role for each type is shown in Fig. 1.SF exists in several forms in vivo including soluble monomeric and dimeric forms, as well as a membrane-bound form (Fig. 1A).Each form is capable of activating c-kit, but the efficacy and the specific effects are varied (25,26).Membrane-bound SF persistently activates c-kit, compared with soluble SF, and the dimeric form of soluble SF preferentially stimulates mast cell growth compared with monomeric soluble SF (25,26).In the present study, the role of membrane-bound SF on ICC cultures was assessed utilizing three types of fibroblasts genetically engineered to produce soluble SF, membrane-bound SF, or membrane-restricted SF (Fig. 1B).A fibroblast line that does not express SF was used as a control.Soluble and membranebound SF have different proteolytic cleavage sites.However, both soluble SF-producing fibroblasts and membrane-bound SF-producing fibroblasts contribute to soluble SF levels.Membrane-restricted SF will not contribute to soluble SF, because it lacks the sequence where the proteolytic cleavage sites are located.Murine fibroblasts were immunostained with an antibody that recognizes SF to show that the fibroblasts do produce SF (Fig. 1C).
Fibroblasts were plated on 25-mm plain glass coverslips (Fisherbrand) at 3 ϫ 10 4 cells/ml in high-glucose Dulbecco's modified Eagle medium containing 10% fetal bovine serum, 1% sodium pyruvate, and 1% antibiotic-antimycotic (all from GIBCO).After 12 h, fibroblast cell division was arrested by treatment with mitomycin C (5 g/ml) for 90 min to control the total number of fibroblasts per coverslip.After a 24-h recovery period, freshly dissociated jejunal murine cell suspensions (1.5 ml) were plated onto fibroblasts.The cell culture medium was changed after 24 h.Cultures appeared to be robust, i.e., spontaneous contractions of smooth muscle cells were observed.These contractions were maintained for at least 72 h.For these studies, all cell counts were performed 48 h after the murine cell suspension was plated.This time point was optimal, because ICC and c-kit expression appeared to be most stable.More time in culture allowed other undifferentiated fibroblast-like cells to overpopulate the cultures, and at shorter times, cell attachment was variable.
Immunohistochemistry. ICC were identified with the use of the rat monoclonal anti-c-kit antibody ACK2 (GIBCO) after cultures were briefly fixed in acetone (4°C for 10 min).Immunostaining using the antisera SCF G-19 (Santa Cruz Biotechnology, Santa Cruz, CA) or anti-mouse SCF (R&D Systems, Minneapolis, MN) verified SF expression.Acetonefixed cultures were washed with phosphate-buffered saline (0.1 M; pH 7.4), incubated with blocking solution (10% normal donkey serum) for 1 h to minimize nonspecific antibody binding, and then incubated with primary antibody (5 g/ml) for 24 h at 4°C.Cultures were then rinsed in phosphatebuffered saline and incubated for 2 h at 4°C with donkey anti-goat IgG, donkey anti-mouse Ig, or donkey anti-rat IgG conjugated to CY3 or fluorescein (1:200 dilution; Chemicon).Nonspecific immunoreactivity was assessed by immunostaining cultures in an identical manner except that the primary antibody was omitted.Immunostained cell cultures were examined with the use of a laser scanning confocal microscope (model LSM 510; Zeiss).A 40ϫ (numerical aperature ϭ 1.2) water immersion objective was used, with additional electronic zoom, when necessary.The full width at halfmaximum signal intensity was ϳ1.3 m and, therefore, outof-plane fluorescence was negligible.An excitation dichroic mirror was used with a bandpass emission filter of 530 Ϯ 15 nm and a 590-nm long-pass filter.Images were reconstructed from confocal stacks of Z-series scans of 10-30 optical sections through a depth of 5-15 m.The immunostained cultures were surveyed for c-kit-positive cells with the use of an algorithm to automatically move the stage in the X-Y plane in a 1 ϫ 1-mm grid pattern.The number of c-kit-positive ICC per field was recorded at 50 intersections under direct observation.Number values reported in the text refer to the individual experiments.Each experiment was carried out with the use of tissue obtained from two mice; Three cover slips were used for each experiment and for each control.Fifty high power fields were counted for each coverslip.Occasionally, clumps of ICC containing more than six overlap-ping cells were found.However, no difference was noted with clump frequency in cultures grown with fibroblasts expressing soluble, membrane-bound, or membrane-restricted SF.The larger networks (Ͼ6 cells) were omitted from the total ICC cell count to be certain that increased counts of ICC did not result from undigested tissues.Parallel cultures, grown under identical cell culture conditions but without fibroblasts, were used to normalize ICC counts.A parallel culture was grown for each jejunal cell dispersion to allow normalization between cell cultures derived from different animals.
Data are expressed as means Ϯ SE.Differences in data were evaluated by Student's t-test.P values Ͻ0.05 were taken as statistically significant.

RESULTS
Primary cultures yielded single cells and small clumps of cells.Spontaneous contractions of smooth muscle cells were observed between days 1 and 3 in small networks containing cells with ICC-like appearance (e.g., triangular cell bodies and multiple processes) and with smooth muscle-like appearance (7,13,17).Positive immunoreactivity with the anti-c-kit antibody ACK2 confirmed that the cultures contained ICC.However, the density of ACK2-positive cells was low.The mean number of ACK2-positive cells (0.23 by 0.23 mm; expressed as number of ICC per high-powered field) under these control-culture conditions was 0.85 Ϯ 0.2.One typical cell culture containing ICC is shown in Fig. 2. Transmitted light images at low magnification (Fig. 2A) show many cells with triangular cell bodies and several processes.Immunohistochemical staining revealed a much smaller subset of ACK2positive cells, thus verifying ICC presence (Fig. 2B).
Freshly dispersed murine jejunal cells grown with fibroblasts expressing membrane-bound SF contained more c-kit-positive cells compared with controls grown without fibroblasts producing membrane-bound SF.One typical cell culture is shown in Fig. 3. Single ACK2-positive cells were distributed throughout the cultures (Fig. 3A), and ICC networks were also common (Fig. 3B).In the regions with ICC, the ICC density was much higher compared with control cultures grown without fibroblasts producing SF.Overall quantification of ICC showed an increase of 119 Ϯ 14% (P Ͻ 0.05, n ϭ 5) in coculture with fibroblasts expressing membrane-bound SF compared with control cultures grown without fibroblasts.ACK-2 immunoreactivity was not observed in pure fibroblast cultures (data not shown).
Both soluble and membrane-bound SF can stimulate c-kit in vitro, but soluble SF does not support long-term hematopoietic stem cell growth (11).To determine whether soluble SF was sufficient to support ICC, freshly dispersed murine jejunal cells were cocultured with fibroblasts expressing soluble SF.Appearance of these cultures was similar to those grown with membrane-bound SF with many single ICC and ICC networks identified by ACK2 immunoreactivity (Fig. 4).Freshly dispersed murine jejunal cells grown with soluble SF-secreting fibroblasts yielded an 80 Ϯ 16% increase in ICC (Fig. 4B) compared with control cultures (P Ͻ 0.05, n ϭ 6).
Both isoforms of SF initially produce membranebound SF, and, to different degrees, both produce soluble SF (11).To distinguish the effects of soluble from membrane-bound SF on ICC in culture, we used a genetically engineered form of SF known as membrane restricted.This form of SF lacks the amino acid sequence to allow proteolytic processing and, therefore, does not release soluble SF.Freshly dispersed murine jejunal cells grown with fibroblasts expressing membrane-restricted SF are shown in Fig. 5.The appearance of these cultures was qualitatively similar to those grown with either soluble or membrane-bound SF.The effect of membrane-restricted SF on the number of cells expressing the c-kit receptor is shown in Fig. 5B.Freshly dispersed murine jejunal cells cocultured with fibroblasts expressing membrane-restricted SF resulted in a 180 Ϯ 43% increase in c-kit-positive ICC compared with control cultures (P Ͻ 0.05, n ϭ 3).To determine whether one isoform of SF more effectively increases the number of c-kit-positive ICC, membrane-restricted or soluble presentation of SF was tested in parallel cultures.Membrane-restricted SF  did not increase c-kit-positive ICC expression more than that observed with fibroblasts expressing soluble SF (Ϫ9 Ϯ 14%, P Ͼ 0.05, n ϭ 3; Fig. 5C).
Soluble SF is presented in membrane-bound form before proteolytic cleavage and, therefore, it was possible that the stimulation of ICC by soluble SF-producing fibroblasts results from membrane-bound SF.To test this hypothesis, fibroblasts were grown on permeable supports to physically separate the freshly dispersed murine jejunal cells from fibroblasts expressing soluble SF.Separation of the cocultures with the Transwell barrier prevented stimulation of ICC (Fig. 6A).The number of c-kit-positive ICC grown in cocultures separated by a semipermeable barrier (pore size ϭ 0.45 m) was unchanged from control cultures (Ϫ16 Ϯ 11%, P Ͼ 0.05, n ϭ 3) grown without fibroblasts.To test whether an increased concentration of SF could enhance the number of c-kit-positive ICC recombinant soluble murine, SF was added to the culture medium.The addition of 20 or 100 ng/ml did not stimulate expression of c-kit-positive ICC compared with control cultures (37 Ϯ 19% and 17 Ϯ 14%, respectively, P Ͼ 0.05, n ϭ 3; Fig. 6B).This suggests that enhancement of c-kit-positive ICC required close contact between the fibroblasts and the freshly dispersed murine jejunal cells.
Fibroblasts are support cells that enhance cell proliferation and survival by several mechanisms, and it is possible that stimulation of c-kit-positive ICC cell counts in these cultures could be independent of SF.This was tested with the use of fibroblasts that do not produce SF for coculture with freshly dispersed murine jejunal cells.Under these conditions, expression density of c-kit-positive was very low (Fig. 7A) and the   total number of c-kit-positive ICC observed under coculture conditions was not different from control cultures grown without fibroblasts (Ϫ11 Ϯ 11%, P Ͼ 0.05, n ϭ 3; Fig. 7B).The specificity of SF-stimulation through the c-kit pathway was also tested with the use of ACK2 to block c-kit activation.The addition of ACK2 to the culture medium of freshly dispersed murine jejunal cells with fibroblasts expressing soluble SF resulted in decreased expression of c-kit-positive ICC (Fig. 8A).The number of c-kit-positive ICC was reduced in a dose-dependent manner with a 54 Ϯ 10% decrease after the addition of 5 ng/ml ACK2 and 74 Ϯ 6% (P Ͻ 0.05, n ϭ 4) decrease after the addition of 10 ng/ml (Fig. 8B).

DISCUSSION
We have shown an increase in the number of c-kitpositive ICC when mouse jejunal cells were cocultured with SF-producing fibroblasts.Increased c-kit-positive ICC specifically resulted from fibroblast SF expression, because few c-kit-positive ICC were found in cocultures with fibroblasts that did not express SF.The c-kitspecific antibody ACK2, which interferes with SF activation of the c-kit receptor, also reduced the number of c-kit-positive ICC in culture, further suggesting a specific SF-c-kit interaction needed for survival of ICC.Effective culture of ICC required local production of SF, because physical separation of the jejunal cells from the fibroblasts or the addition of recombinant SF to the cell culture media reduced the relative number of c-kit-positive ICC in culture.In addition, conditioned media from fibroblasts synthesizing soluble SF did not support ICC in culture (data not shown).These results showed that the location of SF production is critical, with close contact between the c-kit receptor expressed on ICC and SF expressed on fibroblasts necessary for ICC expression.Data also suggest that in the gastrointestinal tract, required SF needs to be provided by membrane-associated SF on closely apposed supporting cells rather than by SF secreted from more distant cells or in the blood plasma.One alternative explanation is that presentation of membrane-bound SF merely enhances expression of c-kit by ICC but doesn't increase the number of ICC.For example, ICC with c-kit expression below detectable limits would not be identified under control conditions and coculture with SF-expressing fibroblasts would enhance c-kit expression in those cells to detectable levels.
Stromal cells, fibroblasts, and endothelial cells express SF as a membrane-bound protein (2,10,19).Soluble SF results from the proteolytic cleavage of an extracellular portion of SF (5,22).Binding of SF to the tyrosine kinase receptor c-kit results in rapid receptor autophosphorylation, endocytosis, and degradation, thus regulating surface c-kit density and the duration of c-kit signaling (1).Although soluble and membranebound SF activate c-kit, membrane-bound SF stabilizes the receptor at the cell surface slowing receptor internalization and prolonging c-kit signaling (18,25).Thus the membrane-bound isoform of SF increases the life span and the downstream signaling activity of the c-kit receptor.In contrast, soluble SF induces rapid downregulation of cell surface c-kit expression and a correspondingly rapid window of kinase activity (25).Recent studies have shown that the duration of c-kit activation may also function to differentiate specific downstream signaling elements (16).These data suggest that membrane-bound SF may persistently activate c-kit expressed on the ICC cell surface and stimulate specific downstream signaling pathways that ultimately enhance ICC differentiation and proliferation.
Development of ICC-MY in the murine small intestine depends on functional SF/c-kit signaling, appear-  ing at embryonic day 13 (17,31,34,37).However, ICC expression may not be solely dependent on this signaling pathway, because mutant mice with impaired c-kit or SF signaling exhibit ICC-DMP and ICC in the colon and stomach (14,35).This suggests that ICC-MY may be more dependent on SF signaling compared with ICC-IM or ICC-DMP.In this study, the number of ICC was much higher in cocultures with fibroblasts presenting SF and was thought to be mainly ICC-MY on the basis of percent and morphological observations.Complete disruption of the SF/c-kit signaling pathway results in the abolishment of ICC suggesting that the spontaneous mutant mice have developed compensatory mutations enabling ICC development by another pathway, or that a small residual c-kit activity is sufficient for certain classes of ICC (30).
The source for SF, as well as the relevant isoforms necessary for the development and maintenance of ICC in mature mice, is currently unknown.Very low numbers of ICC were observed in these studies in cultures without the addition of recombinant SF or fibroblast cocultures that supply SF.Survival of c-kit-positive ICC in these cultures may depend on SF production by another cell type that survives the enzymatic dissociation procedure.This could explain why increasing plating density in primary cultures improves ICC expression (37).One potential source for SF production is enteric neurons.These cells are ideal candidates for providing SF and stimulating ICC development and survival, because enteric neurons form close anatomical relationships with many ICC (33).Enteric neurons have been shown to produce mRNA for SF (32), but it is not known whether SF is expressed on the neuronal processes that intercalate with ICC.Murine knockout models suggest that enteric neurons are not required for ICC development.For example, the glial cell-linederived neurotrophic factor (GDNF) knockout mice lack the enteric nervous system but exhibit normal ICC networks and mRNA expression levels of SF similar to wild-type mice (36).ICC are also normal in the c-ret knockout mouse, which lacks the GDNF receptor and the enteric nervous system (37).Another potential source of SF is smooth muscle.Recent data (4) showed that smooth muscle cells from the murine small intestine express mRNA for the soluble isoform of SF, and smooth muscle cells isolated from the gastric fundus express the membrane-bound isoform of SF mRNA (4).The authors (4) also found that single ICC, selected from the murine gastric fundus or small intestine, express the soluble isoform of SF mRNA.Thus it is possible that smooth muscle provides SF necessary for ICC development and maintenance.Because ICC develop from the same precursor cell as smooth muscle, ICC may produce SF that acts in an autocrine manner for the SF/c-kit signaling pathway.Such a mechanism has been reported for mast cells (3) and for neural crest cell development (6).Finally, it is well known that fibroblasts and endothelial cells express SF to support hematopoiesis (1,2), and therefore it is also possible that fibroblasts within the tunica muscularis express SF and support ICC.
Proximity between the cell type producing SF and ICC may affect both ICC density and morphology.ICC development in cell culture of dispersed murine intestine improves as plating density is increased (37), suggesting that the enzymatically dispersed cells are the likely source of SF.The authors (37) hypothesize that membrane-bound SF expressed in the myenteric plexus region by enteric neurons or fibroblasts could promote ICC proliferation and survival in this region, resulting in the dense network of ICC vital for spontaneous electrical activity in the gastrointestinal tract.This hypothesis also predicts the expression of bipolar ICC with smaller total volume and the lack of ICC network formation within the longitudinal muscle layer where a nonneuronal source of SF supports ICC development (37).Smooth muscle has been shown to preferentially express soluble SF, which is consistent with development of intramuscular ICC due to a brief stimulation of c-kit (4).
In summary, the present results show that coculture of freshly dispersed jejunal cells from the mouse with SF-expressing fibroblasts enhances the expression of ICC.Increase in ICC number in these cocultures is consistent with enhanced activation of c-kit, resulting from an increase in the local concentration of SF. Results suggest that membrane-bound SF potently stimulates the c-kit receptor and maintains the ICC phenotype.Soluble SF may promote very transient stimulation of c-kit, resulting in rapid receptor turnover and subsequent downregulation, whereas membrane-bound SF may persistently stimulate c-kit, promoting more robust ICC growth, proliferation, and survival.Identification of the specific expression pattern during development and later in life for each isoform of SF, as well as the cell type that expresses SF and supports ICC development, proliferation, and growth, will further determine the functional role for each SF isoform.

Fig. 1 .
Fig.1.Schematic diagram of fibroblast (fb) cell lines expressing genetically modified Steel factor (SF) is shown.Two forms of SF, soluble and membrane-bound, activate c-kit.A: membrane-bound SF is the best stimulator of c-kit and dimeric SF stimulates c-kit more effectively compared with monomeric SF.B: three types of fibroblasts were used to determine the effects of membrane-bound and soluble SF on interstitial cells of Cajal (ICC) grown in culture.Soluble SF, resulting from rapid proteolysis, is expressed by fbsol.Membranebound SF, expressed by fbmb, may also contribute to soluble SF by slow proteolysis.Membrane-restricted (fbmr) SF does not contain a proteolytic cleavage site and, therefore, does not contribute to soluble SF levels.A fourth type of fibroblast that does not express SF was used for control experiments.C: immunohistochemical staining for SF shows expression by fbsol that appears on the plasma membrane.

Fig. 2 .
Fig. 2. A: 2-day culture of freshly dispersed murine jejunal cells.Cells in these cultures exhibited spontaneous rhythmic contractions and had the appearance associated with ICC.B: ACK2-positive immunohistochemical staining revealed ICC.

Fig. 3 .
Fig. 3. Coculture of freshly dispersed murine jejunal cells and fibroblasts expressing murine membrane-bound SF resulted in increased c-kit-positive ICC expression.A: low-power image of c-kit-positive cells with single ICC distributed throughout the field.ICC were not distributed homogeneously throughout the coverslip, and these images show regions containing ICC.Some regions revealed many ICC and others were blank.B: one example of an ICC network is shown at higher magnification.ICC networks were commonly observed under coculture conditions.C: counting the number of c-kit-positive cells per field and normalizing with control cultures grown without fibroblasts showed an overall increase of 119 Ϯ 14% (n ϭ 5, *P Ͻ 0.05).Data are the means Ϯ SE of 5 independent experiments.Error bars on the control data represent the variation in the raw data scaled by the normalization denominator.

Fig. 4 .
Fig. 4. Fibroblasts expressing soluble SF also enhance c-kit-positive ICC expression.A: typical field showing a single ICC and an ICC network is shown.Overall c-kit-positive ICC morphology was similar to those grown with membrane-bound SF.B: counting the number of c-kit-positive cells per field and normalizing with control cultures grown without fibroblasts showed an overall increase of 80 Ϯ 16% compared with control cultures (n ϭ 6, *P Ͻ 0.05).Data are the means Ϯ SE of 6 independent experiments.Error bars on the control data represent the variation in the raw data scaled by the normalization denominator.

Fig. 5 .
Fig. 5. Fibroblasts expressing membrane-restricted SF enhance c-kit-positive ICC expression under coculture conditions.A: c-kit-positive cells and networks grown with membrane-restricted SF.B: counting the number of c-kit-positive cells per field and normalizing with control cultures grown without fibroblasts showed an overall increase in c-kit-positive ICC expression by 180 Ϯ 43% (n ϭ 3, *P Ͻ 0.05).C: membrane-restricted SF did not appear to increase overall c-kit-positive ICC expression compared with fibroblasts expressing soluble SF when grown in parallel cultures (Ϫ9 Ϯ 14%, P Ͼ 0.05).Data are the means Ϯ SE of the number of independent experiments.Error bars on the control data represent the variation in the raw data scaled by the normalization denominator.

Fig. 6 .
Fig. 6.Fibroblasts expressing soluble SF separated from the murine jejunal cells with semipermeable supports did not increase c-kitpositive ICC cell counts.A: counting the number of c-kit-positive cells per field and normalizing with control cultures grown without fibroblasts showed that physical separation blocked the fibroblast effects on c-kit-positive ICC expression (Ϫ16 Ϯ 11%; n ϭ 3, P Ͼ 0.05).TW, Transwell.B: recombinant SF added to the culture media also did not substitute for fibroblasts secreting soluble SF.The addition of 20 or 100 ng/ml of soluble SF did not effect c-kit-positive ICC expression (37 Ϯ 19 and 17 Ϯ 14%, respectively; n ϭ 3, P Ͼ 0.05).Data are the means Ϯ SE of 3 independent experiments.Error bars on the control data represent variation in raw data scaled by the normalization denominator.

Fig. 8 .
Fig. 8.The addition of the c-kit-inactivating antibody to the culture medium blocked c-kit-positive ICC expression under coculture conditions with fibroblasts secreting soluble SF.A: coculture grown with 5 ng/ml of c-kit inactivating antibody (ACK2) in the culture medium.ICC expression was reduced.B: counting the number of c-kit-positive cells per field and normalizing with control cultures grown with fibroblasts expressing soluble SF showed that the addition of 5 ng/ml inactivating antibody resulted in a 54 Ϯ 10% reduction in c-kitpositive ICC expression and 10 ng/ml reduced c-kit-positive ICC expression by 74 Ϯ 6% (n ϭ 4, *P Ͻ 0.05).Data are the means Ϯ SE of 4 independent experiments.Error bars on the control data represent the variation in the raw data scaled by the normalization denominator.

Fig. 7 .
Fig. 7. Fibroblasts that do not express soluble or membrane-bound SF did not enhance c-kit-positive ICC expression.A: c-kit-positive ICC cell density was very low after coculture with fibroblasts that do not express SF.B: a count of the number of c-kit-positive cells per field and normalizing with control cultures grown without fibroblasts showed that fibroblasts not expressing SF do not increase c-kitpositive ICC expression (Ϫ11 Ϯ 11%; P Ͼ 0.05).Data are the means Ϯ SE of 3 independent experiments.Error bars on control data represent variation in raw data scaled by the normalization denominator.