PDGF-D Expression Is Down-Regulated by TGFβ in Fibroblasts

Transforming growth factor-β (TGFβ) is a key mediator of fibrogenesis. TGFβ is overexpressed and activated in fibrotic diseases, regulates fibroblast differentiation into myofibroblasts and induces extracellular matrix deposition. Platelet-derived growth factor (PDGF) is also a regulator of fibrogenesis. Some studies showed a link between TGFβ and PDGF in certain fibrotic diseases. TGFβ induces PDGF receptor alpha expression in scleroderma fibroblasts. PDGF-C and -D are the most recently discovered ligands and also play a role in fibrosis. In this study, we report the first link between TGFβ and PDGF-D and -C ligands. In normal fibroblasts, TGFβ down-regulated PDGF-D expression and up-regulated PDGF-C expression at the mRNA and protein levels. This phenomenon is not limited to TGFβ since other growth factors implicated in fibrosis, such as FGF, EGF and PDGF-B, also regulated PDGF-D and PDGF-C expression. Among different kinase inhibitors, only TGFβ receptor inhibitors and the IκB kinase (IKK) inhibitor BMS-345541 blocked the effect of TGFβ. However, activation of the classical NF-κB pathway was not involved. Interestingly, in a model of lung fibrosis induced by either bleomycin or silica, PDGF-D was down-regulated, which correlates with the production of TGFβ and other fibrotic growth factors. In conclusion, the down-regulation of PDGF-D by TGFβ and other growth factors may serve as a negative feedback in the network of cytokines that control fibrosis.


Introduction
Transforming growth factor-b (TGFb) family members are cytokines that regulate numerous physiological processes such as embryonic development, cell differentiation, proliferation, migration and extracellular matrix production [1,2]. TGFb has also been linked to several human pathologies including fibrosis and carcinogenesis [3,4,5]. TGFb is a central mediator of several fibrotic diseases like renal fibrosis or liver fibrosis [6,7]. Smad proteins are the core components of the intracellular signaling cascade of TGFb receptors. The Smad proteins transduce signals from the cell surface directly to the nucleus where they regulate gene transcription in cooperation with co-activators and/or corepressors. In addition to the Smad pathway, also known as the conventional pathway, Smad-independent pathways can be activated by TGFb, including TGFb-associated kinase 1 (TAK1) [8], certain MAPKs such as extracellular signal-regulated kinase (ERK) 1 and 2, Jun N-terminal kinase (JNK) and p38 mitogenactivated protein kinase [8,9]. The PI3K-Akt-mTOR pathway has also been reported to be activated by TGFb [10]. The importance of these alternative pathways, NF-kB in particular, has been debated. NF-kB activation by TGFb seems to be cell type dependent [11,12].
Platelet-derived growth factors (PDGF) are also major players in fibrosis including intestinal, cardiac and liver fibrosis [13,14,15]. PDGFs are potent mitogens for myofibroblasts while TGFb stimulates collagen deposition by these cells [16,17]. PDGF family members are composed of 4 ligands A, B, C and D binding to two receptors, alpha and beta. PDGF-A and -B are well characterized and their role in fibrotic diseases is well established [13,14,15]. PDGF-C and -D were discovered more recently and are also implicated in fibrosis. PDGF-C is overexpressed and is a potent inducer of fibrosis in the heart [18], lung [19], kidney [20], pancreas [21] and liver [22]. Fewer studies have linked PDGF-D to fibrotic diseases. PDGF-D is overexpressed in fibrosis of the liver [23] and kidney [24]. By contrast, one study showed a negative correlation of PDGF-D expression in a mice model of bleomycin-induced lung fibrosis [19].
Crosstalk between TGFb and PDGF has been reported in several fibrotic conditions. In a model of hepatic fibrosis, one report showed that TGFb mediated the mitogenic effect of PDGF-BB, but not PDGF-AA, and this effect was due to the upregulation of PDGFRb mRNA and cell surface expression [25]. In scleroderma, TGFb also enhances PDGF-A-stimulated growth by increasing the expression of PDGFRa [26]. All together these are the most relevant studies that reported a regulation of PDGFs by TGFb in fibrotic disorders.
In the present study, we demonstrate that TGFb down-regulates PDGF-D expression in normal fibroblasts.

Ethics Statement
This study was approved by the local ethical committee for animal experiments (Comité d'Ethique pour l'Expérimentation Animale, Permit number LA1230312). Experiments were carried out in strict accordance with the Belgian and European regulations. All surgery was performed after euthanasia by intraperitoneal injection of 20 mg of sodium pentobarbital (Certa, Braine l'Alleud, Belgium), and all efforts were made to minimize suffering.

Reverse transcriptase-PCR and quantitative PCR analysis
Subconfluent AG01523 or MRC5 fibroblasts were plated at 600000 cells/plate, starved for 24 hours in the presence of 0.05% essentially fatty acid free BSA (Sigma) and treated for 30 minutes in presence or absence of the following inhibitors: TGFb receptor inhibitor (SB431542, 10 mM), MEK1/2 inhibitor (U0126, 10 mM), p38 inhibitors (SB203580 and SB202190, 10 mM), IKK inhibitor (BMS-345541, 10 mM), PI3K inhibitor (LY294002, 20 mM), HDAC inhibitor (TSA, 300 nM) or vehicle (DMSO). Cells were then incubated during 24 hours in the presence of TGFb (4 ng/ml). Total RNA was extracted from cells using the Qiagen RNeasy extraction kit according to the manufacturer's instructions. Complementary DNA synthesis from total RNA was performed following the M-MLV reverse transcriptase (Invitrogen) protocol. Real time quantitative PCR (q-PCR) was performed using SYBR-green (ABgene kit) on an iCycler instrument (Bio-Rad) as described [30]. The expression of the RPLP0 and b-Actin genes was used to normalize q-PCR experiments. All q-PCR primer sequences used in this study are provided in Table S1.

Protein extraction and Western blotting
Subconfluent MRC5 cells were plated at 200000 cells/well (six well plates), starved for 24 hours in the presence of 0.05% essentially fatty acid free BSA (Sigma) and treated with or without TGFb (4 ng/ml) during 24 or 48 hours. Cells were washed with phosphate-buffered saline (PBS) and then lysed as described [30]. Equal amounts of total protein extracts were analyzed by SDS-PAGE, then transferred on a polyvinylidene difluoride membrane (PVDF) and subjected to antibodies against PDGF-D (1 mg/ml, Santa Cruz), PDGF-C (1 mg/ml, monoclonal antibody) or b-actin (0.5 mg/ml, Sigma).

Enzyme-linked immunoabsorbent assays (ELISA)
An ELISA kit was obtained to measure concentrations of PDGF-DD (R&D systems) in supernatants of MRC5 cells cultured in the absence or presence of TGFb (4 ng/ml) for 24 h. The detection limit of this assay was 1.67 pg/ml.

Lung Fibrosis induced with crystalline silica particles
NMRI mice were instilled either with 2.5 mg of silica per mouse or with NaCl for control condition. Three mice were used for control group and four for the silica-treated group. Mice were sacrificed 60 or 120 days post instillation by intraperitoneal injection of 20 mg of sodium pentobarbital (Certa, Braine l'Alleud, Belgium) and total RNA was extracted from their lung samples [34].

Statistical analysis
Results were analyzed by Student's t-test. Multiple comparisons were carried out using analysis of variance (ANOVA test). A significance threshold was set on p-values #0.05. All experiments were performed in at least three independent replicates.

Results
PDGF-D and PDGF-C are regulated by TGFb at the mRNA and protein levels We analyzed the expression of PDGF-D and PDGF-C genes by RT-qPCR in MRC5 lung fibroblasts treated or not with TGFb for various periods of time. PDGF-D mRNA expression was significantly down-regulated by TGFb, whereas PDGF-C mRNA was up-regulated ( Figure 1A). We also observed a similar regulation of these genes by TGFb in AG01518 and AG01523 fibroblasts (data not shown). By reanalyzing a publicly available time course experiment, in which A549 epithelial cells were stimulated with TGFb during 72 hours (GEO, reference GDS3710), we observed that, starting from 16 hours of TGFb stimulation, PDGF-D and PDGF-C were strikingly regulated in an opposite manner, as in MRC5 cells ( Figure 1B). To investigate whether these observations were specific to TGFb we treated the MRC5 cells with a selective inhibitor of TGFb receptor activity (SB431542, Figures 2A and 2D). As expected, the inhibitor reversed the effects of TGFb on PDGF-C and PDGF-D expression levels.
PDGF-C and PDGF-D regulation by TGFb was confirmed at the protein level by western blotting (Figures 2B and 2E). We next measured PDGF-DD secretion by ELISA, using antibodies that recognize the active form of PDGF-DD, in which the CUB domain was removed. TGFb-induced down-regulation of PDGF-D mRNA expression had a significant effect on its secretion in the culture medium ( Figure 2C). This was reflected by the lower levels of active PDGF-DD protein in the supernatant following exposure of the cells to TGFb in the presence of serum. We did not see any effect of TGFb in the absence of serum, most likely because it provides plasmin, which cleaves latent PDGF-DD [35].

Role of Smad4 in the down-regulation of PDGF-D by TGFb
To further investigate the mechanism behind PDGF-D downregulation by TGFb stimulation, we tested the involvement of the Smad pathway. We infected MRC5 fibroblasts with two different shRNAs against Smad4 and compared PDGF-C and -D mRNA expression after 24 hours in TGFb-treated versus non-treated cells. TGFb strongly decreased PDGF-D expression, and this reduction was only marginally prevented by Smad4 knock-down (p,0.05; Figure S1A). The shRNA did not affect PDGF-C regulation by TGFb ( Figure S1B). As a control, we investigated Collagen1A1 (Col1A1) expression, a well-known target of the TGFb-Smad pathway [36]. TGFb treatment induced Col1A1 mRNA expression, and this induction was partially decreased by Smad4 shRNA (Figure S1C). This partial effect can be explained by the fact that Smad4 expression was not completely abolished ( Figure S1D).

Role of Smad-independent pathways in PDGF-D gene regulation by TGFb
We next tested whether TGFb regulates PDGF-D through a Smad-independent pathway. We used specific pharmacological inhibitors to inhibit the PI3K, MAPK and IKK pathways in the presence or absence of TGFb. Figure 3A shows the effect of PI3K (LY294002) and MEK1/2 (U0126) inhibitors, Figure 3B, the effect of two p38 MAPK inhibitors (SB202190 and SB203580) and Figure 3C, the effect of a JNK inhibitor (SP600125). None of these inhibitors influenced PDGF-D expression. These results suggested that neither the PI3K nor MAPK pathways mediated PDGF-D regulation by TGFb. We next investigated the NF-kB pathway, by treating fibroblasts with an inhibitor of IKK (BMS-345541). The IKK inhibitor alone had no effect on basal PDGF-D expression, but it abrogated the inhibition of PDGF-D expression observed in the presence of TGFb ( Figure 3D). Similar results were obtained in AG01523 fibroblasts (data not shown). These data suggest that TGFb regulates PDGF-D via IKK or another unknown BMS-345541 target.

PDGF-D is regulated at the transcriptional level by TGFb through IKK
We next evaluated whether TGFb affected the activity of the PDGF-D gene promoter in luciferase-reporter assays. We used gamma-2A fibrosarcoma cells because they are suitable for the transfection of luciferase reporters and because TGFb downregulated PDGF-D mRNA in these cells, as in fibroblasts (data not shown). Cells were transfected with a plasmid encoding luciferase under the control of the PDGF-D promoter, and treated or not with TGFb. The CAGA12 promoter was used as a control promoter, to verify the responsiveness of transfected cells to added TGFb ( Figure 4A). We observed a significant reduction of PDGF-D promoter activity in the presence of TGFb ( Figure 4A). PDGF-D promoter activity was also significantly down-regulated in cells co-transfected with a constitutively activated mutant of the TGFb receptor chain ALK5 (activin receptor-like kinase), in the presence or absence of TGFb ( Figure 4A). Moreover, IKK inhibition partially reversed the inhibitory effect of TGFb on PDGF-D promoter activity ( Figure 4B), confirming the results shown in Figure 3.

PDGF-D is down-regulated in a mouse model of lung fibrosis
TGFb is implicated in several fibrotic diseases. To analyze the expression of PDGF-D and PDGF-C genes in the context of a fibrotic disease, we performed RT-qPCR on lungs collected from NMRI mice that have received intra-tracheal instillations of silica. Mice developed lung fibrosis upon silica instillation, and Rabolli et al. showed that TGFb is highly expressed and released in fibrotic lungs by comparison to saline controls in this model [34]. PDGF-D and PDGF-C gene expression was analyzed 60 and 120 days post instillation by RT-qPCR. PDGF-D expression was significantly down-regulated in fibrotic lungs by comparison to controls ( Figure 5A). PDGF-C expression, however, was not different in fibrotic lungs by comparison to controls ( Figure 5B). Similar results were obtained with another strain of mice (C57/Bl6) instilled or not with silica (data not shown).

PDGF-D is down-regulated by other growth factors involved in fibrosis
We next tested whether, in addition to TGFb, other growth factors could also regulate PDGF-D and PDGF-C expression. In our previous microarray analysis, we found that PDGF-D expression was significantly down-regulated in fibroblasts 24 hours after stimulation with PDGF-BB or FGF-2 ( Figure S2). Using RT-qPCR, we observed that FGF, EGF and PDGF-BB down-regulated PDGF-D and up-regulated PDGF-C expression (Figures 6A and  6B). We also investigated whether inflammatory cytokines could regulate PDGF-D expression. Interleukin-1b, alone or in combination with TGFb, did not regulate PDGF-D expression ( Figure 6C).

Discussion
In this study we have shown that TGFb regulates PDGF-C and -D ligand expression at both the mRNA and the protein level in normal human fibroblasts. To our knowledge, this is the first study showing that the expression of PDGF-C and -D ligands is regulated by TGFb. Furthermore, a recent study reported that Bone morphogenetic proteins (BMPs), which are members of the of TGFb superfamily, are repressors of PDGF-A in ovarian granulosa cells and tumors [37]. Here we showed that TGFb induced PDGF-C in normal fibroblasts whereas PDGF-D was significantly down-regulated at the transcriptional level. Consistently, TGFb inhibited PDGF-D secretion by the cells.
Smad proteins mediate TGFb signaling. Down-regulation of Smad4 weakly restored PDGF-D expression in the presence of TGFb. These partial effects might be explained in part by the fact that Smad4 expression was not completely abolished but Smad-independent pathways are likely to play a role. Additional noncanonical pathways are known to transduce TGFb signals, and include ERK1/2, p38, JNK or PI3K pathways [10]. We showed that neither the PI3K nor the MAPK pathways are responsible for PDGF-D regulation by TGFb. Moreover, a study performed in renal fibrosis identified two non-canonical TGFb pathways, Abl and mTORC1, that play an important role in the development of fibrosis [38]. Our experiments with imatinib (an Abl inhibitor) and rapamycin (an mTOR inhibitor) did not support the hypothesis that these pathways played a role in PDGF-D regulation by TGFb (data not shown). BMS-3455541, an IKK inhibitor, rescued PDGF-D expression to the basal level in the presence of TGFb suggesting a regulation through IKK.
Whether TGFb activates NF-kB pathway is not fully understood and appears to be cell context dependent. Gingery et al. described a mechanism of NF-kB activation by TGFb through a TAK1/MEK1/2/AKT pathway, which promotes osteoclast survival in cooperation with Smad2/3 [39]. We showed that TGFb failed to induce p65 phosphorylation on Ser 536 in contrast to the anti-inflammatory cytokine IL-1b which strongly activated p65 (data not shown) but did not regulate PDGF-D in MRC5 fibroblasts. This demonstrates that activation of the classical NF-kB pathway alone does not regulate PDGF-D.
Moreover, an interesting study reported a non-canonical NF-kB activation pathway through TAK1 and NIK by silica in Rat2 fibroblasts [40]. Taken together with our results, it is likely that TGFb activates NF-kB in fibroblasts through a non-classical pathway. Alternatively, IKK may also signal independently of NF-kB downstream TGFb. Finally, we cannot rule out a non-specific effect of the IKK inhibitor.
Gene down-regulation frequently involves histone deacetylases (HDAC). Figure S3 showed that HDAC strongly down-regulated PDGF-D expression. Indeed, trichostatin A (TSA), an inhibitor of HDAC, induced PDGF-D expression. However, TGFb strongly repressed PDGF-D expression even in the presence of HDAC inhibitor, indicating that HDAC regulated PDGF-D independently of TGFb. Little information is available regarding PDGF-D promoter regulation. Some studies reported that PDGF-D is induced by Ets-1 and Sp1 transcription factors [41]. Sp1 is in general described as an activator rather than a repressor of transcription but some studies reported that Sp1 could act as an The authors reported that TGFb stimulation reduces Sp1 DNA binding [42]. Another study suggested also a negative regulation of the AT1 receptor by TGFb through Sp1.
Unpublished experiments obtained in the lab, demonstrated that the presence of serum down-regulated PDGF-D expression and that this effect was even stronger in the presence of saturating amounts of TGFb, suggesting that other growth factors may regulate PDGF-D and PDGF-C expression. We showed that EGF, FGF and PDGF-BB alone or combined reduced significantly PDGF-D expression and induced PDGF-C expression confirming our observation with serum. Interestingly, all these growth factors have been linked to fibrosis: PDGF-B in liver fibrosis [15], EGF in heart and renal fibrosis [43,44] and FGF-4 in lung and renal fibrosis [45,46]. These factors do not efficiently activate Smad or IKK proteins, thus further experiments will be required to identify the pathway responsible for this effect. Moreover, some studies have reported that FGF represses gene expression via the ERK/ Sp1 pathway. Bonello et al. showed in smooth muscle cells that FGF-2 represses PDGFRa expression in an ERK/Sp1 dependent manner. Phosphorylation of Sp1 upon FGF-2 stimulation switches Sp1 from an activator to a repressor of PDGFRa expression [47]. This pathway was also reported in another study in breast cancer cells where the authors showed that FGF-2 down-regulates TSP50 expression via the ERK/Sp1 pathway [48]. All together these studies suggest a mechanism of negative regulation via Sp1 upon FGF-2 activation. Our preliminary results do not go in the same  direction as these studies because U0126, an inhibitor of ERK1/2 pathway, did not have any effect on PDGF-D expression upon FGF-4 stimulation (data not shown).
Our observations in cultured fibroblasts may be relevant in pathophysiological conditions. Indeed, we showed in a mouse model of silica-induced lung fibrosis, in which TGFb expression is highly increased [34], that PDGF-D mRNA expression was downregulated whereas PDGF-C mRNA expression remained unchanged. These results were confirmed in two different strains of mice. Furthermore, Zhuo et al. have described in another mouse model of lung fibrosis induced by bleomycin that PDGF-D expression was decreased and PDGF-C expression was increased [19]. Our results shed a light on the potential mechanism of PDGF-D regulation in this model of lung fibrosis. However, it is unclear which cytokine regulates PDGF-D in vivo.
In summary, we showed that TGFb down-regulated PDGF-D and up-regulated PDGF-C expression. The regulation of PDGF-D expression was sensitive to BMS-345541. We speculate that this regulation may represent a negative feedback in the network of fibrotic cytokines.