An In Vitro Culture System for Long-Term Expansion of Epithelial and Mesenchymal Salivary Gland Cells: Role of TGF-β1 in Salivary Gland Epithelial and Mesenchymal Differentiation

Despite a pivotal role in salivary gland development, homeostasis, and disease, the role of salivary gland mesenchyme is not well understood. In this study, we used the Col1a1-GFP mouse model to characterize the salivary gland mesenchyme in vitro and in vivo. The Col1a1-GFP transgene was exclusively expressed in the salivary gland mesenchyme. Ex vivo culture of mixed salivary gland cells in DMEM plus serum medium allowed long-term expansion of salivary gland epithelial and mesenchymal cells. The role of TGF-β1 in salivary gland development and disease is complex. Therefore, we used this in vitro culture system to study the effects of TGF-β1 on salivary gland cell differentiation. TGF-β1 induced the expression of collagen, and inhibited the formation of acini-like structures in close proximity to mesenchymal cells, which adapted a fibroblastic phenotype. In contrast, TGF-βR1 inhibition increased acini genes and fibroblast growth factors (Fgf-7 and Fgf-10), decreased collagen and induced formation of larger, mature acini-like structures. Thus, inhibition of TGF-β signaling may be beneficial for salivary gland differentiation; however, due to differential effects of TGF-β1 in salivary gland epithelial versus mesenchymal cells, selective inhibition is desirable. In conclusion, this mixed salivary gland cell culture system can be used to study epithelial-mesenchymal interactions and the effects of differentiating inducers and inhibitors.


Introduction
The mesenchymal component of salivary glands plays a pivotal role during development of the salivary gland tissue for induction of epithelial differentiation and branching [1][2][3][4][5][6][7][8][9]. However, the role of the mesenchymal cells in homeostasis of the adult salivary gland and during repair following injury is understudied. Many salivary gland diseases, if not all, involve and affect the mesenchymal component of the salivary glands, particularly diseases that result in fibrosis of the salivary gland tissue such as radiation-induced damage, Graft-versus-host disease (GVHD), Sjogren's syndrome, and aging [10][11][12][13][14]. In addition, certain salivary gland tumors originate from or involve the mesenchyme [15,16]. Nonetheless, little attention has been paid in characterizing the salivary gland mesenchymal cells and in developing in vitro systems to model the epithelial-mesenchymal interaction during adult homeostasis and disease/injury of the salivary gland tissue.
We analyzed a transgenic mouse expressing GFP driven by the procollagen promoter (Col1a1-GFP) to identify and characterize the mesenchymal cells in the salivary gland tissue. Histological analysis of the salivary glands revealed that only mesenchymal cells of the salivary gland tissue expressed the GFP transgene. We tested different culture conditions to ex vivo expand mixed cultures of salivary gland mesenchymal and epithelial cells. We determined that medium containing DMEM +10% serum allowed in vitro long-term expansion of a mixed culture containing both mesenchymal and epithelial cells. Upon in vitro induction on matrigel, this mixed culture differentiated into acinilike structures surrounded by GFP-positive mesenchymal cells.
TGF-1 signaling is important for development and maintenance of salivary gland tissue but its differential effects on salivary gland epithelium versus mesenchyme have yet not been dissected apart [17][18][19][20][21][22][23][24]. Therefore, we tested the effects of TGF-1 induction and its inhibition in our culture system. TGF-1 induction resulted in elongation of GFPpositive mesenchymal cells, increased collagen production and inhibition of acini-like structure formation. In contrast, inhibition of TGF-R1 resulted in decreased collagen production, increased expression of the mesenchymal fibroblast growth factors, Fgf-7 and Fgf-10, increased expression of mature acini markers, and formation of larger and more mature acini-like structures.
This new in vitro culture system can be used to expand salivary gland mesenchymal and epithelial cells for tissue regeneration and also to in vitro study the role of mesenchymal cells in salivary gland differentiation and alterations of the mesenchymal-epithelial interactions in disease.

Isolation of Submandibular Salivary Gland
Cells. Col1a1-GFP mice were a kind donation from Dr. Jeremy Duffield [25,26]. Submandibular salivary gland (SMG) tissues were dissected (one gland per mouse) from 3-month-old Col1a1-GFP mice ( = 3 different preparations) in accordance with approved Institutional Animal Care and Use Committee (IACUC) guidelines, University of Washington. The SMG was separated from the cervical fascia and connective tissue, then gently isolated and kept in phosphate buffer saline (PBS) (Corning Cellgro). The tissues were washed with PBS, mechanically minced with a pair of curved scissors, and enzymatically dissociated with a 1.2 units/mL dispase II, 2 mg/mL collagenase type IV (Worthington) supplemented with 2 mM CaCl 2 in PBS for 45 min at 37 ∘ C. The digested tissues were pipetted up and down several times every 15 min to break up clumps and release mononuclear cells. Subsequently, an equal volume of Dulbecco's modification of eagle's medium (DMEM) with 4.5 g/L glucose, L-glutamine, and sodium pyruvate (Cellgro) was added to the digest prior to filtering through 70 mm nylon cell strainers (BD Falcon) and then centrifuging at 300 g for 10 min at room temperature. The mononuclear cells were then resuspended in two types of growth media described below, and single cell suspensions were initially plated at 50,000 cells/cm 2 on plastic tissue culture dishes (BD Biosciences).

Culture of Submandibular Salivary Gland Cells.
Cells (50,000 cells/cm 2 ) were cultured at 37 ∘ C under 5% CO 2 in two kinds of culture media to determine their difference in cell growth, DMEM medium plus 10% heat-inactivated fetal calf serum (HyClone), 100 units/mL penicillin with 100 mg/mL streptomycin (HyClone), and N2 medium containing DMEM, penicillin, streptomycin, 20 ng/mL EGF (Sigma), 20 ng/mL bFGF (Shenandoah biotechnology), 1/100 N2 supplement (Gibco, Invitrogen), 10 g/mL insulin-transferrin-selenium (ITS) (Cellgro), and 1 M dexamethasone (Sigma). Fresh medium was added or changed every three days. Once adherent cells were more than 70% confluent, they were detached with 0.25% trypsin-EDTA (Cellgro) and replated at a 1 : 4 dilution under the same culture condition with fresh media. For cells cultured in N2 medium, since they formed spheres, before trypsinization, spheres were collected and then mixed with trypsinized cells before replating in the same condition.

In Vitro Differentiation of Submandibular Salivary Gland
Cells on Matrigel. Mixed SMG cells (line 1; passage 9; 5 × 10 4 cells per well) were seeded in either noncoated cells or matrigel-coated plastic surfaces as undifferentiated or differentiated cells, respectively, with 300 L of additional DMEM media plus serum. Growth factor-reduced matrigel (20 mg/mL; BD Biosciences) was thawed on ice and diluted in PBS at a final concentration of 2 mg/mL. To form threedimensional matrix in culture dishes, 150 L of matrigel was added to 48-well tissue culture plate (0.75 cm 2 per well), incubated at 37 ∘ C for 1 hour, and the matrigel excess was removed before cell seeding. Culture media were changed every three days. For differentiation, mixed SMG cells cultured on matrigel were divided into three treatment groups: (1) matrigel alone (control group), (2) transforming growth factor-beta1 (TGF-1; cell signaling; 10 ng/mL), and (3) TGF-1 plus TGF-receptor 1 inhibitor (SB525334; Selleck Chemicals; 1 M) (TGF-1 + SB525334). At day 3 and 5 of differentiation, both undifferentiated and differentiated cells were fixed with 4% formaldehyde in PBS for 30 min and washed with three times of PBS to preserve GFP expression. The stained cells were determined for acini-like structure formation and further stained for specific salivary gland epithelial and mesenchymal markers.

RT-PCR and Q-RT-PCR Analyses.
Undifferentiated and differentiated mixed SMG cells were extracted for total RNA by using the total RNA kit (Omega Bio-tek) according to the manufacturer's protocol. Quantity and purity of RNA was determined by 260/280 nm absorbance. First-strand cDNA was synthesized from 1 g of RNA using the high capacity cDNA synthesis kit from Applied Biosystems per manufacturer's protocols using a randomized primer. RT-PCR and Q-RT-PCR mouse-specific primers were included in Table 1. For RT-PCR, cDNA of undifferentiated cells cultured in different passages (20 ng) was diluted in a final volume of 20 L per reaction using the Immomix PCR Mastermix from Bioline. PCR was performed using the following thermal cycling conditions; 95 ∘ C 7 min for initial activation followed  Reactions were processed by the ABI 7900HT PCR system with the following parameters: 50 ∘ C/2 min and 95 ∘ C/10 min, followed by 40 cycles of 95 ∘ C/15 s and 60 ∘ C/1 min. Results were analyzed using SDS 2.3 software, and relative expression was calculated using the comparative Ct method. Each sample was run in triplicate reactions for each gene.

Histology and Staining of Submandibular Salivary Gland
Tissues. Submandibular salivary glands (SMGs) were isolated from 3-month-old Col1a1-GFP mice ( = 3) and removed surrounding connective tissues. To preserve GFP, Col1a1-GFP derived SMG was fixed with 4% formaldehyde/PBS for 2 h at RT and washed. The first wash was 30 min followed by 20-min and 10-min washes, respectively. After washing, the fixed SMG was immersed through a gradient of sucrose solutions (10% for 20 min, 20% for 20 min, and 30% for overnight) at 4 ∘ C to preserve tissue morphology before embedding in OCT media (VWR) and frozen with liquid nitrogen cooled isobutane. The frozen SMG tissues were cut into 10 m thickness to get a good morphology of tissue section. Fixed sections were rehydrated and permeabilized with 1% BSA in 0.1% Triton X-100 (Sigma)/PBS for 10 min. Then, sections were blocked with 10% normal goat serum for 1 h at RT and incubated overnight at 4 ∘ C with primary antibodies listed in Table 2, following three wash steps. Stained sections were subsequently incubated with goatderived Alexa 594-conjugated secondary antibody (Invitrogen) at 1 : 1000 dilutions for 1 h at RT, following three times of washing. The tissues were stained with 4 , 6-diamine-2phenylindol (DAPI) (Life Technologies) at 1 : 1000 to visualize the nuclei.

Staining of Submandibular Salivary Gland Cells.
For immunocytochemistry, undifferentiated and differentiated cells were fixed with 4% formaldehyde in PBS for 30 min and washed three times of PBS. Fixed cells were permeabilized with 1% BSA in 0.1% Triton-X 100/PBS for 10 min and blocked nonspecific binding sites with 10% goat normal serum (Vector Burlingame, CA) for 1 h. All primary antibodies listed in Table 2 were used and incubated overnight at 4 ∘ C. Stained cells were incubated with goat-derived Alexa 594-conjugated secondary antibodies (Invitrogen) which were diluted at 1 : 1000 and incubated for 1 h. Cells were stained with DAPI at 1 : 1000 to visualize the nuclei. All antibodies were diluted in 1% BSA in 0.1% Triton-X 100/PBS. IgG isotype from the species made for the primary antibody (0.1 g/mL) (Vector Burlingame, CA) was used as negative control for all staining. All immunofluorescence images described in this manuscript was detected using either a Zeiss Axiovert 200 fluorescent microscope (Thornwood, NY) or a Nikon A1R Confocal microscope. Photomicrographs were taken with an onboard monochrome AxioCam MRm camera and colored using Adobe Photoshop (San Jose, CA). Background was reduced using brightness and contrast adjustments, and color balance was performed to enhance colors. All the modifications were applied to the whole image using Adobe Photoshop.

Statistical Analysis.
Data of Q-RT-PCR analyses were represented as ± the standard error of the mean (SEM) of results from three separated experiments. The data were analyzed by Student's -test where * * * value ≤ 0.001, * * value ≤ 0.005, or * value ≤ 0.05 represented significant differences between different culture media or treatments.

The Col1a1-GFP Transgenic Mouse Selectively Identifies
Mesenchymal Cells in the Salivary Glands. In this study, we analyzed GFP expression in the submandibular salivary glands of Col1a1-GFP transgenic mice. The Col1a1-GFP mice express enhanced green fluorescent protein gene under the control of the procollagen, type 1, alpha 1 (Col1a1) promoter. We hypothesized that salivary gland mesenchymal cells, but not epithelial cells, express Collagen type I (Col1a1) and drive the expression of GFP, resulting in labeled mesenchymal cells by green fluorescence. The histological analysis demonstrated that salivary gland mesenchymal stroma was GFPpositive whereas salivary gland parenchyma or epithelium was GFP-negative ( Figure 1). To confirm the specificity of the Col1a1-GFP mouse model and distinguish differences between salivary gland epithelium and mesenchyme, we stained for markers specific of salivary gland epithelium, CD44, E-cadherin (E-cad), amylase (AMY-1), aquaporin-5, and LAMP-1. CD44 is a cell surface glycoprotein found on basal epithelial cells including salivary gland epithelium [27,28]. CD44 staining was positive in basal and lateral membranes of salivary gland acini (AC) but not in ductal epithelium (DE) and mesenchyme (Figures 1(a) and 1(b)). Ecadherin (epithelial-calcium-dependent adhesion or E-cad) is a transmembrane protein which is crucial for cell-cell interaction in organ development including salivary gland formation, and expressed by salivary gland epithelium [29]. E-cadherin staining was positive for both salivary gland epithelial cells of acini and ducts (D) in particular at acinar cell-ductal cell contacts but not mesenchyme (Figures 1(c) and 1(d)). Salivary amylase catalyses the breakdown of starch into sugars and is found in the granular convoluted tubular cells and to a lesser extent in the acinar cells of submandibular gland (Figures 1(e) and 1(f)) [30]. Aquaporin 5 is important for fluid transport and saliva secretion and is found in luminal, lateral, and basal membrane of acinar cells (Figures 1(g) and 1(h)) [31]. LAMP-1 is a lysosomeassociated protein found in the ductal and acinar cells in salivary gland (Figures 1(i) and 1(j)) [32]. Smooth muscle actin (SMA) staining for myoepithelial cells [29] exhibited no colocalization between SMA and GFP found in mesenchymal cells, illustrating a differential staining pattern between mesenchymal and myoepithelial cells in normal salivary glands (Figures 2(a) and 2(b)). Additionally, staining for collagen type I mostly colocalized with GFP-positive mesenchymal cells, confirming the specificity of the Col1a1-GFP mouse model. As expected, we also observed positive collagen type I staining in salivary gland extracellular matrices labeling the basement membrane which was not positive for GFP as it was not cytoplasmic but extracellular collagen (Figures 2(c) and 2(d)).

Mixed Salivary Gland Cells Cultured in Different Media
Exhibited Differential Growth of Salivary Gland Epithelial and Mesenchymal Cells. To select a culture condition system capable of promoting proliferation of mixed cell populations containing salivary gland epithelium and mesenchyme, we used two different kinds of media to culture mixed salivary gland cells, N2 medium and DMEM plus 10% serum. N2 medium contains DMEM supplemented with EGF, bFGF, N2, and ITS and has been used as growth and differentiation media for salivary gland stem cells [33]. DMEM plus 10% serum has previously been used as medium to promote salivary gland epithelial and mesenchymal cell growth [34,35]. After 2 weeks in culture (cell passage 1), in N2 medium we observed sphere formation combined with monolayer of polyhedral-and spindle-shaped cells (Figure 3(a), arrowheads). The monolayer cells were mainly negative for GFP ( Figure 3(b)). Most of the spheres formed were negative for GFP but some of them were also GFP-positive ( Figure 3(c), arrowheads). This indicates that N2 medium induces both salivary gland epithelium and mesenchyme to form spheres, and a majority of cells proliferating in N2 medium were epithelial cells. However, mixed salivary gland cells in this N2 medium failed to proliferate beyond 2 weeks of culture and were not able to survive. In contrast, mixed cells cultured in DMEM plus serum did not form spheres but grew as polyhedral-shaped or round cells on top of spindle-shaped monolayer cells (Figure 3(d)). The major cell population was spindle-shaped cells expressing GFP (Figures 3(e) and 3(f)). This cell mixture was able to proliferate and graw beyond passage 10 (2 month in culture). Interestingly, after several cell passages, both spindle-shaped and polyhedral-shaped cells In contrast, DMEM plus serum seems to support salivary gland mesenchymal cell growth as evidence by increased gene expression levels of mesenchymal genes. Based on the culture morphological analysis and gene expression analysis the N2 medium enhanced more growth of salivary gland epithelium than mesenchyme but was only able to sustain these cells for a short period whereas the DMEM plus serum effectively promoted the proliferation of both salivary gland epithelium and mesenchyme in long-term culture. Therefore, we selected to use DMEM plus serum as the expanding culture medium for all the further experiments described herein. . Interestingly, we observed that some GFP+ cells are also SMA+ in late culture suggesting that some GFP+ mesenchymal cells can upregulate SMA expression (Figure 4(d)). S100 is another marker expressed by myoepithelial cells [37]. We detected S100+ cells in early and late cultures (Figures 4(e) and 4(f)). Interestingly, some GFP+ cells costained positive for S100 (Figure 4(f)

DMEM Plus Serum Medium Enhanced In Vitro Long
Salivary gland mesenchymal genes Salivary gland mesenchymal genes Salivary gland mesenchymal genes Late culture (P9)    (Figure 4(m)).

Expression of TGF-1 Ligand and Receptor in Salivary
Gland Tissues and Cultured Cells. We then examined the expression of TGF-1 ligand (TGF-1) and its receptors, TGF-1 receptor 1 (TGF-R1) and receptor 2 (TGF-R2), in both salivary gland tissues and cultured cells. Murine submandibular salivary gland tissues showed the expression of TGF-1, TGF-R1, and TGF-R2 in salivary gland epithelium, particularly in ductal epithelium (Figures 5(a)-5(f)). However, we observed very low expression of TGF-1 and TGF-R1 in the salivary gland mesenchyme. In contrast, some mesenchymal cells (GFP)+ in the salivary gland stained positive for TGF-R2 (Figures 5(e) and 5(f)). As negative control, staining of salivary gland tissues with rabbit IgG control exhibited completely negative staining result used to confirm the specificity of TGF-1, TGF-R1, and TGF-R2 antibodies (Figures 5(g) and 5(h)). We then studied the protein expression of TGF-1 and its receptors in early and late cultures. Salivary gland epithelial cells stained strongly positive for TGF-1 and TGF-R1 in early cultures. Interestingly, in early cultures mesenchymal cells, which were GFP-positive, showed weak positive staining for TGF-1 and its receptors (Figures 5(i), 5(k), and 5(m)). In contrast, in late cultures GFP+ mesenchymal cells stained strongly positive for TGF 1 and its receptors ( Figures 5(j), 5(l), and 5(n)). In particular, cultured salivary gland mesenchymal cells showed stronger TGF-R1 expression than that in epithelial cells ( Figure 5(l)). The TGF-R1 antibody we used recognizes the cytoplasmic domain which can be cleaved and can translocate to the nucleus [39]. This explains why we observed mainly nuclear staining. Q-RT-PCR analysis of mixed salivary gland cells from passage 2 demonstrated that the N2 medium which contained a majority of epithelial cells showed increased levels of Tgf-b1 and Tgf-br1, compared to DMEM plus serum ( ≤ 0.05) (Figures 5(o) and 5(p)).
The regulation TGF-signaling pathway at the receptor level is well understood. Briefly, TGF-ligands (TGF-1, 2, 3) bind TGF-R2 which recruits TGF-R1 to form a heterotetramer (two type I and two type II receptors) [40]. The formation of this heterotetramer is needed for TGFsignaling. TGF-R2 phosphorylates TGF-R1, activating   In the early culture, TGF-1 and TGF-R2 were strongly expressed in the cytoplasm of epithelial cells (i and m, resp.) whereas the strong TGF-R1 expression was found in cultured salivary gland epithelial cells in cytoplasmic and nuclear areas (k), but not in mesenchymal cells. In the late culture, the expression of TGF-1, TGF-R1, and TGF-R2 was seen in both salivary gland epithelial and mesenchymal cells (j, l, and n). The salivary gland mesenchymal cells increased the expression of TGF-1 (j), TGF-R1 (l, inset), and TGF-R2 (n) after late culture. The staining pattern of TGF-1 and TGF-R1 in late cell passage was similar to early cell passage (j and l) whereas that of TGF-R2 was shown in both membrane and nuclei (n). (o) and (p) Mixed salivary gland cells cultured in N2 media which contained a majority of salivary gland epithelium showed higher levels of both Tgf-b1 and Tgf-br1 expression, compared to that in DMEM media plus serum. Relative expression was normalized to the expression of Gapdh which was used as the reference gene. Values were represented as mean ± SEM from three independent experiments ( = 3). Student's -test was analyzed to compare between cells cultured in N2 and DMEM media plus serum, * ≤ 0.05. Scale bars = 100 m. TGF-R1 kinase which mediates the Smad pathway. We observed membranous and nuclear staining of TGF-R1 and TGF-R2 on GFP+ mesenchymal cells, indicating that the TGF-signaling is active in cultures, especially in late cultures. To confirm the staining pattern for TGF-and its receptors, we performed fluorescence image analysis by confocal microscopy. Similar to the results presented in Figure 5, in the salivary gland tissue, TGF-1 and TGF-R1 is expressed mainly by epithelial cells (Figures 6(a) and 6(b)), whereas TGF-R2 is expressed by epithelial cells but also strongly expressed by GFP+ mesenchymal cells (Figure 6(c), inset). In contrast, in late cultures TGF-1 and its receptors are expressed by both epithelial cells and GFP+ mesenchymal cells (Figures 6(d)-6(f)), indicating an up-regulation of TGF-1 and its receptors in cultured GFP+ cells, perhaps mediated by a positive feedback driven by the TGF-1 produced by the epithelial cells early in culture ( Figure 5(i)).

TGF-R1 Inhibitor Promoted the In Vitro Formation of Acinar-Like Structures.
Mixed salivary gland cells (from passage 9) induced on matrigel differentiated into acinilike structures demonstrating that cells maintained in longterm cultures still have differentiation capacity [6,7]. TGF-1 signaling is important for salivary gland formation during development [18,21,23]. To determine the effect of TGF-1 signaling on in vitro differentiation of salivary gland cells, we compared the differentiation capacity of matrigel-induced cells exposed to TGF-1 (TGF-1) or TGF-1 plus TGF-R1 inhibitor (TGF-1 + SB525334). SB5255334 is a potent inhibitor of TGF-R1 kinase activity [41,42]. The in vitro differentiation was conducted at two time points between day 3 and 5 and analyzed by specific antibody staining to salivary gland epithelium (at day 5) and the gene expression of salivary gland epithelial and mesenchymal genes (at day 3 and 5). The specific staining of epithelium, CD44 (Figures  7(a)-7(c)), E-cad (Figures 7(d), 7(e) inset, 7(f)), LAMP-1 (Figures 7(g), 7(h) inset, 7(i)), amylase (Figures 7(j)-7(l)), and aquaporin 5 (Figures 7(m)-7(o)) showed that salivary gland epithelial cells plated on matrigel in all three groups were able to differentiate based on the formation of acinar-like structures. As expected, all acinar-like structures found were GFP-negative, suggesting that only salivary gland epithelial cells differentiated into acini-like structures on matrigel (Figures 7(a)-7(c)). In addition, in all the differentiation cultures on matrigel we observed areas of mesenchymal clusters and areas free of mesenchymal cells (shown as figures with empty black backgrounds in Figures 7(a)-7(c), 7(k), and 7(n) and all insets). In control (matrigel only) and  -1), Occludin, and salivary gland mesenchymal genes, Fgf-7, Fgf-10, and Collagen type I, was determined after 3-and 5-day treatments. At day 3 (D3), cells treated with TGF-1 + SB525334 significantly upregulated some of salivary gland epithelial genes, Amylase-1 and Occludin, and all salivary gland mesenchymal genes except Collagen type I, compared to cells on matrigel alone and/or cells treated with TGF-1. The TGF-1-treated cells significantly expressed lower level of Amylase-1 expression (p), but higher level of Collagen type I (v), compared to untreated cells, and TGF-1 + SB525334-treated cells. At day 5 (D5), cells treated with TGF-1 + SB525334 expressed the highest levels of both salivary gland epithelial and mesenchymal genes and were significantly different compared to untreated and TGF-1 treated cells. The ZO-1 expression was comparable but insignificantly different between TGF-1 and TGF-1 + SB525334 treated cells (r). The expression of Fgf-7 and Fgf-10 was remarkably increased in the TGF-1 + SB525334 group (t and u). The highest level of Collagen type I was observed and shown a significantly statistical difference in TGF-1 treated cells at both day 3 and 5 compared to other groups (v). Relative expression was normalized to the expression of Gapdh which was used as the reference gene. Values were represented as mean ± SEM from three independent experiments ( = 3). Student's -test was analyzed to compare between Matrigel (untreated; black bar), TGF-1 treated (dark gray bar), and TGF-1 + SB525334 treated cells (light gray bar), * * * ≤ 0.001, * * ≤ 0.005, or * ≤ 0.05. Scale bars = 100 m.
TGF-1 + SB525334 cultures, abundant acinar-like structures were found in both mesenchymal rich and mesenchymal free areas whereas in TGF-1 exposed cultures acinar-like structures were observed only in the mesenchymal free areas.
The expression of Tgf-b1 and Tgf-br1 was also analyzed after in vitro differentiation and compared between the three different groups (Figures 8(m) and 8(n)). TGF-1 induced up-regulation of both Tgf-b1 and Tgf-br1 expression whereas inhibition of TGF-1 signaling in the TGF-1 + SB525334 resulted in significant downregulation of both Tgf-b1 and Tgf-br1 at day 3 and whereas at day 5, Tgf-b1 was still downregulated both not its receptor, Tgf-br1.

Discussion
Although the epithelium of the parotid glands is ectodermderived whereas the epithelium of the submandibular and sublingual glands is endoderm-derived, the salivary gland mesenchyme is neural crest-derived [43]. The interactions of epithelium and mesenchyme are essential for the branching morphogenesis of the salivary gland. Molecular cues such as secretion of fibroblast growth factors (FGF-10, FGF-7) by the ectomesenchyme and expression of FGF receptors (FGF-R1, FGF-R2) by the epithelium are important for the development salivary gland [5,[43][44][45][46][47][48]. Also other morphogens such as Shh and Wnt are also important for the saliva gland development [49][50][51]. Although the role of the mesenchyme during salivary gland development is well studied, the role of mesenchyme in adult salivary gland tissue homeostasis and its potential involvement in salivary diseases is understudied.
Therefore, we sought to better characterize the phenotype of salivary gland mesenchyme and to develop a culture system to in vitro study the interactions of adult salivary gland epithelial cells and mesenchymal cells. We used the transgenic Col1a1-GFP reporter mice to identify mesenchymal cells in the salivary gland [26]. Upon histological analysis of the Col1a1-GFP mice we can exclusively identify mesenchymal cells as GFP-positive cells in salivary glands.
Ex vivo cultures of the mixed salivary gland cells in N2 versus DMEM plus serum media revealed the presence of both epithelial and mesenchymal (GFP-positive) cells in culture. Cells in N2 medium stopped proliferating early in culture whereas the cells in DMEM plus serum medium continued proliferating for more than 10 passages. The levels of TGF-1 expression in early cultures are higher in N2 medium which may explain why cells cultured in this medium stopped proliferating early in culture. RT-PCR and immunohistochemical analysis of the DMEM plus serum cultures at several passages revealed the presence of both epithelial and mesenchymal cells even in late cultures. Upon differentiation on matrigel the mixed culture cells were able to differentiate into mature acini-like structures.
Therefore, this culture condition allows the long-term expansion of both epithelial and mesenchymal cells which in turn can be induced to differentiate in vitro. This culture system offers many advantages over existing culture conditions: (1) long-term expansion of mesenchymal cells, (2) long-term Tgf-b1  red) showed the formation of acinar-like structures, respectively. Acinar formation was observed adjacent to mesenchymal cells in the matrigel only (a and g) and TGF-1 + SB525334 (c and i) groups, but not in the TGF-1 (b and h) group. In the areas free of mesenchymal cells, acinar-like structures were also found in all three groups. TGF-1 + SB525334 treated cells formed larger acinar-like structures (c, f, i, and l) compared to untreated (a, d, g, and j) and TGF-1 treated cells (e and k). Acini-like structures in the matrigel (d and j) and TGF-1 + SB525334 (f and l), but not in the TGF-1 (e and k) showed GFP-positive mesenchymal cells located peripherally or adjacent to acini-like structures. Elongated and polarized salivary gland mesenchymal cells were found in the TGF-1 treated group (b and h), but not in the untreated (a and g) and TGF-1 + SB525334 groups (c and i). (m) and (n) The expression of TGF-1 ligand (Tgf-b1) and receptor (Tgf-br1) were also examined, respectively. The cells treated with TGF-1 significantly increased the level of Tgf-b1 and Tgf-br1 expression compared to untreated and TGF-1 + SB525334 treated cells both day 3 (D3) and similar pattern was observed at day 5 (D5). In the control, untreated cells on matrigel, the expression levels of both Tgf-b1 and receptor Tgf-br1 decreased at D5 compared to D3 but this pattern is not seen in the TGF-1 and TGF-1 + SB525344 groups. Relative expression was normalized to the expression of Gapdh which was used as the reference gene. Values were represented as mean ± SEM from three independent experiments ( = 3). Student's -test was analyzed to compare between matrigel (untreated; black bar), TGF-1 treated (dark gray bar), and TGF-1 + SB525334 treated cells (light gray bar), * * * ≤ 0.001, * * ≤ 0.005, or * ≤ 0.05. Scale bars = 100 m.

expansion of epithelial cells, and (3) in vitro differentiation of mixed cultured cells to study epithelial-mesenchymal interactions and effects of inducers and inhibitors.
The effects of TGF-1 on the salivary glands are complex and somewhat paradoxical. TGF-null mice developed multifocal infiltrates in heart, lungs and salivary glands [18,52]. These multifocal lesions and resulting damage to salivary glands were gender-specific and not only caused by defects in T-cell suppression but also by defects in TGFsignaling in salivary gland epithelial cells, as in another study it was shown that conditional deletion of TGF-R1 on salivary gland epithelial cells using the mammary tumor virus Cre mouse led to salivary gland inflammatory lesions and abnormal pattern of aquaporin-5 distribution, resulting in saliva secretion defects only in females but not in males mice [53]. On the other hand, conditional over-expression of TGF-1 in secretory cells (mammary and salivary gland epithelial cells) resulted in hyposalivation due to salivary gland fibrosis and atrophy [21]. We hypothesized these paradoxical effects may be explained by differential roles of TGF-signaling in different salivary gland cell types, namely epithelial cells versus mesenchymal cells. Therefore, we tested the utility of our mixed cell culture system to study the effects TGF-1 in vitro.
Supplementation of TGF-1 to mixed cell cultures induced expression of higher levels of acini markers; however, we also observed reduction of mesenchymal derived fibroblast growth factors (Fgf-7 and Fgf-10) and dramatic increased in the procollagen type 1 levels. FGF-7 (aka, Keratinocyte Growth Factor, KGF) in particular has been proven beneficial for salivary gland epithelial differentiation. In vitro salivary gland explants in the absence of mesenchyme can undergo differentiation with EGF and FGF-7. EGF induced lobule formation whereas FGF-7 induced stalk elongation morphogenesis [54]. FGF-7 has been shown beneficial for salivary gland restoration [55,56]. Transgenic mice expressing Fgf-7 under the keratin (K14) promoter exhibited excessive salivation [57].
Histological examination of the differentiation cultures revealed lack of acini-like structures formation and undifferentiated appearance of epithelial cells adjacent to mesenchymal cells which in turn adapted a fibroblasts morphology (elongated and polarized) in the TGF-1 induced group. In contrast, inhibition of TGF-R1 signaling with SB525334 results in the highest levels of acini markers, highest levels of Fgfs, and lowest levels of collagen type I, which corresponds with the most mature and largest acini-like structures, especially in the areas rich in mesenchymal cells which adapted a cobblestone morphology. Thus we conclude that inhibition of TGF-1 signaling in these in vitro differentiation cultures is beneficial, particularly in mesenchymal cells.
Alterations in the TGF-signaling have been associated with several salivary gland disorders. Sjogren's syndrome is an autoimmune disorder characterized primarily by T-cell but also B-cell infiltration in the salivary glands. Immunohistochemical staining showed TGF-1 was strongly expressed in ductal epithelial cells of normal and inflamed salivary glands but downregulated in Sjogren's salivary glands [22]. In another study, normal and Sjogren's salivary glands expressed TGF-in ductal and acinar epithelial cells, but TGFproduction was reduced in Sjogren's salivary gland cultures [58]. Another study showed all three isoforms of TGFexpressed in lymphocytes, endothelial cells and ductal cells of Sjogren's versus benign lymphoepithelial lesions. Interestingly, the expression of TGF-isoforms in ductal cells of Sjogren's was increased as compared to benign lesions [59]. These reports suggest that alterations of TGF-1 pathway may be involved in pathogenesis of Sjogren's disease Alterations of the TGF-pathway and abnormal expression of TGF-ligands and receptors have been reported in several salivary gland tumors. In pleomorphic adenomas (PA), the most common type of salivary gland tumors, TGF-2 was expressed in the inner ductal cells and TGF-3 was expressed in the myoepithelial cells of PA tumors [60]. Mucoepidermoid carcinoma (MEC) is another salivary gland tumor that exhibits differentiation in multiple lineages. TGF-1 was expressed in the salivary gland ducts, stroma and endothelial cells of the MEC tumors. Interestingly, TGF-R2 expression inversely correlated with tumor grade: all low grade tumors showed expression of TGF-R2 whereas none of the high-grade tumors, with greatest metastatic potential, showed TGF-R2 expression. TGF-R2 was expressed in surface epithelium, endothelial cells, nonneoplastic salivary gland ducts and stromal fibroblasts of the low-grade MEC [61]. Loss of TGF-R2 expression correlated with loss of tumor differentiation. In another study, it was shown that the Ms cell line derived from MEC, which highly expressed TGF-1, exhibited decreased invasion and migration capacity when TGF-1 was silenced by siRNA [24]. Likewise, another study showed that TGF-1 was highly expressed in a metastatic salivary adenoid cystic carcinoma cell line and exposure to TGF-1 in vitro activated the classical TGFpathway, suggesting that TGF-1 may promote migration and invasion of this tumor [19]. This suggests that the expression of TGF-1 on certain tumor salivary gland epithelial cells induces migration, invasion and metastasis, and that TGF-1 inhibition in these tumors may be beneficial.
In summary, given the complexity of TGF-signaling in salivary gland development, homeostasis and diseases, better tools are needed to understand the differential effects and role of TGF-signaling in different salivary gland cells (epithelial versus mesenchymal cells). The emergence of tissue specific conditional knockouts such as Cre-lox recombination [62], specific for salivary gland mesenchyme versus epithelium are necessary to dissect apart the differential roles of TGFsignaling.