Epithelial-mesenchymal transition and fibrosis are mutually exclusive reponses in shear-activated proximal tubular epithelial cells

Abstract

Renal fibrosis (RF) is thought to be a direct consequence of dedifferentiation of resident epithelial cells via an epithelial-mesenchymal transition (EMT). Increased glomerular flow is a critical initiator of fibrogenesis. Yet, the responses of proximal tubular epithelial cells (PTECs) to fluid flow remain uncharacterized. Here, we investigate the effects of pathological shear stresses on the development of fibrosis in PTECs. Our data reveal that type I collagen accumulation in shear-activated PTECs is accompanied by a ∼40–60% decrease in cell motility, thus excluding EMT as a relevant pathological process. In contrast, static incubation of PTECs with TGFβ1 increases cell motility by ∼50%, and induces stable expression of key mesenchymal markers, including Snail1, N-cadherin, and vimentin. Ectopic expression of TGFβ1 in shear-activated PTECs fails to induce EMT-associated changes but abrogates collagen accumulation via SMAD2-dependent mechanisms. Shear-mediated inhibition of EMT occurs via cyclic oscillations in both ERK2 activity and downstream expression of EMT genes. A constitutive ERK2 mutant induces stable expression of Snail1, N-cadherin, and vimentin, and increases cell motility in shear-activated PTECs by 250% without concomitant collagen deposition. Collectively, our data reveal that RF not only occurs without EMT but also that these two responses represent mutually exclusive cell fates.—Grabias, B. M., Konstantopoulos, K. Epithelial-mesenchymal transition and fibrosis are mutually exclusive reponses in shear-activated proximal tubular epithelial cells.

Renal hyperfiltration, or increased glomerular fluid flow, is the hallmark adaptive response to renal mass reduction. Following partial nephrectomy, remnant nephrons experience a dramatic increase in single-nephron glomerular filtration rate (1), thereby preventing a loss in overall filtration capacity. However, this physiological change is eventually accompanied by excessive fibrotic scarring and glomerular and tubular damage (2). The detrimental role of hyperfiltration in driving renal diseases has also been established in physiologically relevant models of hypertension (3) and diabetes (4, 5), often serving as the single best predictor of developing end-stage nephropathies. Despite relatively recent attempts at characterizing the effects of fluid shear stress on the renal epithelial cytoskeleton (6), inflammatory response (7), and fibrinolytic activity (8), our understanding of intratubular hydrodynamic forces as key stimuli in renal pathologies is woefully inadequate.

The predominant signaling molecule thought to mediate fibrosis in the kidney and other organs is transforming growth factor β1 (TGFβ1; refs. 9, 10). Acting primarily through downstream similar to mothers against decapentaplegic (SMAD) signaling molecules, TGFβ1 exhibits numerous activities across a wide array of cell types, potentially inducing collagen expression (11, 12), increasing inflammatory signaling (13), and both stimulating (14, 15) and inhibiting (16, 17) apoptosis. Furthermore, TGFβ1 is a potent initiator of epithelial-to-mesenchymal transitions (EMTs; refs. 18, 19), in which epithelial cells acquire a fibroblast-like phenotype, lose their characteristic intercellular adhesion molecules, such as E-cadherin, and become more motile.

In the progression of renal fibrosis, EMT-derived fibroblasts are thought to play a primary role in collagen deposition. A number of studies support this hypothesis by tracing the lineage of plaque-producing fibroblasts from their epithelial origins (20, 21). Targeting and preventing EMTs, in this context, would be a useful therapeutic approach to blocking kidney damage, and it is, indeed, a major focus of ongoing research. Still, this widely accepted premise is being contested by evidence identifying not only different sources of collagen-producing cells (22, 23) but also fibrosis occurring in the apparent absence of an EMT (24, 25). Likewise, additional data are also beginning to indicate that TGFβ1 itself may not be strictly fibrogenic (26–28). Thus, more work needs be done to fully decipher the roles of TGFβ1 and EMT in the progression of fibrotic diseases.

Utilizing supraphysiological levels of fluid shear stress (up to 4 dyn/cm²) to simulate the effects of hyperfiltration on kidney proximal tubular epithelial cells (PTECs), we investigate whether the mechanism of shear-induced matrix accumulation (i.e., fibrosis) requires phenotypic changes associated with an EMT. Our results reveal that fluid shear stimulates type I collagen deposition in PTECs not only in the absence of an EMT, but with coincident oscillatory modulation of the extracellular signal-regulated kinase 2 (ERK2)-dependent EMT transcription factor Snail1, as well as the downstream mesenchymal markers N-cadherin and vimentin. In contrast, static treatment of PTECs with TGFβ1 results in successful EMT-like changes and persistent expression of these characteristic proteins. Furthermore, while ectopic expression of TGFβ1 in sheared PTECs abrogates fibrosis via SMAD2-dependent mechanisms, it fails to overcome shear-mediated inhibition of EMT-signaling cascades during the fibrotic response. Only direct expression of a constitutively active ERK2 mutant results in persistent expression of both Snail1 and downstream mesenchymal markers and abrogation of collagen deposition by sheared PTECs. Taken together, these data imply that fibrosis and EMT may represent two distinct and mutually exclusive cellular fates. Moreover, these results provide novel evidence suggesting that dynamic temporal regulation of ERK2 and downstream signaling molecules are a key to maintenance of an epithelial phenotype in sheared PTECs.

MATERIALS AND METHODS

Reagents

TGFβ1 neutralizing antibody was purchased from GeneTex (Irvine, CA, USA). The TGFβ1 expression vector and empty vector control were acquired from OriGene Technologies (Rockville, MD, USA). The intrinsically active double mutant ERK2 constructs were gifts from Dr. David Engleberg (Department of Biological Chemistry, The Hebrew University of Jerusalem, Jerusalem, Israel). Antibodies specific for p-SMAD2/3, SMAD2/3 and dual specificity phosphatase 6 (DUSP6), and the SMAD2/SMAD3 and scramble control siRNA oligonucleotides were from Santa Cruz Biotechnology (Santa Cruz, CA, USA). The mitogen-activated protein kinase kinase 1/2 (MEK1/2) inhibitor U0126 and antibodies directed toward MEK1/2, p-MEK1/2 (Ser217/221), ERK1/2, p-ERK1/2 (Thr202/Tyr204), Snail1, and β-actin were obtained from Cell Signaling Technology (Danvers, MA, USA). Antibodies for vimentin and N-cadherin were acquired from BD Pharmingen (Sparks, MD, USA). All other reagents for sodium dodecyl sulfate–polyacrylimide gel electrophoresis (SDS-PAGE) experiments were purchased from Bio-Rad Laboratories (Hercules, CA, USA). Mouse monoclonal antibody (mAb) specific to native collagen I was from Abcam (Cambridge, MA, USA). The anti-mouse IgG (H+L) antibody (Alexa Fluor 488 conjugate), phalloidin (Alexa Fluor 568 conjugate), and Lipofectamine 2000 were acquired through Invitrogen (Carlsbad, CA, USA).

Cell culture and shear stress exposure

The immortalized human kidney 2 (HK-2) proximal tubular epithelial cells (CRL-2190; American Type Culture Collection, Manassas, VA, USA) were grown (37°C in 5% CO₂) on untreated glass slides (7.5×2.5 cm) in keratinocyte serum-free medium (Invitrogen) supplemented with epidermal growth factor (5 ng/ml) and bovine pituitary extract (50 μg/ml). Before shear exposure, cells were incubated for 18 h in medium free of exogenous growth factors. Cells were then subjected to a shear stress level of 0.3, 2, or 4 dyn/cm² for prescribed periods of time in medium lacking growth factor supplements using a streamer flow device (Flexcell International, Hillsborough, NC, USA; refs. 29–31). In select experiments, the pharmacological agent or neutralizing TGFβ1 antibody was added to the medium at the indicated concentrations just before the onset of shear exposure.

Transient transfection

For ectopic expression of TGFβ1 and mutant ERK2, HK-2 cells were transfected with 5 μg/slide of plasmid by using Lipofectamine 2000, according to the manufacturer's instructions. In control experiments, cells were transfected with 5 μg/slide of the empty vector pCMV6-XL (OriGene Technologies). In RNA interference assays, HK-2 cells were transfected with 100 nM siRNA specific to SMAD2 or a scramble control oligonucleotide sequence. Transfected cells were allowed to recover for ≥12 h in growth medium, and then incubated overnight in shear medium before their exposure to shear or static conditions.

Quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR)

qRT-PCR assays were performed on the iCycler iQ detection system (Bio-Rad) using total RNA (Qiagen RNEasy kit; Qiagen, Valencia, CA, USA), the iScript 1-step RT-PCR kit with SYBR green (Bio-Rad), and primers. The GenBank accession numbers and forward (F) and reverse (R) primers are as follows: GAPDH (NM_002046.3), F-GGCCTCCAAGGAGTAAGAC and R-AGGGGTCTACATGGCAACT; SNAIL1 (NM_005985.3), F-ATGCCGCGCTCTTTCCTCGTC and R-TCTCCGGAGGTGGGATGGCTG; α2 Collagen (I) (NM_000089.3), F-GATGCGGAGGGCGGAGGTATG and R-CGCAGAGGAGGGAGCGAATGG; and occludin (NM_001205254.1), F-GGAGACGTCCCCAGCCCAGT and R-CCAATCCGCTCGCCGCAAC. The specificity of primers was determined by dissociation curve and 1% agarose electrophoresis (data not shown). GAPDH was used as an internal control. Reaction mixtures were incubated at 50°C for 15 min, followed by 95°C for 5 min, and then 40 PCR cycles were performed with the following temperature profile: 95°C, 15 s; 58°C, 30 s; 68°C, 1 min; and 77°C, 20 s (29–31). Data were collected at the 77°C, 20-s step to remove possible fluorescent contribution from dimer primers. Gene expression values were normalized to GAPDH.

Western blot analysis

HK-2 cells from sheared and matched static control (SC) specimens were lysed in radioimmunoprecipitation assay (RIPA) buffer (25 mM Tris HCl, pH 7.6; 150 mM NaCl; 1% Nonidet P-40; 1% sodium deoxycholate; and 0.1% SDS) containing a cocktail of proteinase inhibitors (Pierce Chemical, Rockford, IL, USA). The protein content of the cell lysates was determined using bicinchoninic acid (BCA) protein assay reagent (Pierce Chemical). Total cell lysates (10 μg) were subjected to SDS-PAGE, transferred to a polyvinylidene fluoride membrane, and probed with a panel of specific antibodies. β-Actin was used as a loading control. All Western hybridizations were performed at least in triplicate with independent cell samples. Quantitation of protein band intensities was carried out utilizing the gel analysis software within ImageJ [U.S. National Institutes of Health (NIH), Bethesda, MD, USA].

Quantification of cell motility and transmigration

In wound-healing assays, confluent layers of human proximal tubule HK-2 cells were scratched with a P200 pipette tip to generate a wound ∼450 μm wide immediately prior to static or shear exposure in growth factor-/serum-free medium. The wound area was photographed using an inverted phase-contrast microscope (Nikon Eclipse TE300; Nikon, Tokyo, Japan) at the indicated time points and quantified using NIH ImageJ freeware. For the transwell analyses, sheared cells were detached from the slide surface with a brief treatment of 0.05% trypsin (Invitrogen) and 0.01% collagenase (Sigma, St. Louis, MO, USA), and centrifuged. Following a wash with cold phosphate-buffered saline (PBS), cells were counted using the Countess automated cell counter (Invitrogen). For each sample, 1 × 10⁵ cells suspended in growth factor-/serum-free medium were loaded into transwell filters with 8-μm pores (Fisher Scientific, Pittsburgh, PA, USA). Lower chambers were filled with fully supplemented medium containing EGF, and cells were allowed to migrate. After ∼16 h, cells were fixed with 3.7% formaldehyde and stained with 0.1% crystal violet dye (Sigma). Membranes were then photographed via phase microscopy, and the number of transmigrated cells was recorded.

Immunofluorescence staining

Following the indicated static or shear treatments, HK-2 cells were fixed with 3.7% formaldehyde for 10 min, permeabilized with 0.1% Triton X-100 for 10 min at 4°C, washed 2× with PBS, and incubated in buffer containing 5% BSA/PBS for 60 min at room temperature. Cells were then incubated with a mouse mAb to collagen I overnight at 4°C, washed 2× with PBS, and incubated in buffer containing Alexa Fluor 488-labeled goat anti-mouse IgG and phalloidin for 60 min at room temperature. Finally, the cells were washed 2× with PBS and photographed using a fluorescent microscope (Nikon Eclipse Ti).

Enzyme-linked immunosorbent assay (ELISA)

Quantification of secreted, active TGFβ1 was performed using a commercially available ELISA kit (eBiosciences, San Diego, CA, USA), according to the manufacturer's instructions. Following shear for the indicated periods of time, confluent cell monolayers were sheared, carefully scraped into PBS, and centrifuged before analysis of supernatant. Data were then normalized to the total cell lysate protein concentration of sample as measured via a BCA assay and corrected for background against SC samples. The experiments were repeated 3 times in triplicate and are shown as mean ± se.

Statistics

Data represent the means ± se of ≥3 independent experiments. Statistical significance of differences between means was determined by Student's t test or 1-way ANOVA, wherever appropriate.

RESULTS

Divergent effects of fluid shear stress and exogenous TGFβ1 on type I collagen accumulation and cell motility and transmigration of PTECs

Fluid shear stress is a physiological stimulus in kidney homeostasis (1, 4, 7). The physiological magnitudes of shear stress in the proximal tubule range from 0.06 to 0.3 dyn/cm² (32, 33). Following surgical renal mass reduction, flow in the remaining nephrons has been shown to increase ∼3-fold (1). Furthermore, systemic hypertension also increases hemodynamic forces in the kidney. To study the effects of pathological levels of fluid shear on the proximal tubule, the PTEC HK-2 cell line was subjected to prescribed levels of shear stress (2 and 4 dyn/cm²). Shear-activated vs. SC PTECs were then examined for differences in type I collagen content and cell motility. Grainy deposits appeared in PTECs following 72 h of shear exposure, which were absent in static, unsheared samples. Immunofluorescence identified these broad fibrous scars as type I collagen (Fig. 1A). In accordance with these immunofluorescence data, secondary analysis of type I collagen via qRT-PCR also confirmed expression of α2 collagen increasing with the magnitude of applied stress (Supplemental Fig. S1A, B). These data are consistent with the notion that renal hyperfiltration or increased glomerular flow is a critical initiator of fibrogenesis (1, 4, 7). Renal fibrosis is believed to be a direct consequence of dedifferentiation of tubular epithelial cells via an EMT. Despite the excess collagen accumulation, shear-activated PTECs maintained their typical cobblestone morphology, thereby suggesting minor, if any, phenotypic changes. A primary characteristic of mesenchymal cells is enhanced migratory capacity. To assess the potential involvement of EMT in shear-induced fibrosis, we quantified the transmigratory ability and motility of sheared PTECs relative to static controls in a transwell and wound-healing assay, respectively. Following shear exposure to 2 or 4 dyn/cm² for 72 h, PTECs exhibited a marked (40–60%) decrease in both transmigration through an 8-μm porous membrane (Fig. 1B) and wound closure (Fig. 1C). Collectively, these results do not support the acquisition of a motile mesenchymal phenotype in shear-activated PTECs and rather point to independence of EMT from shear-induced fibrosis.

Figure 1.

Shear stress increases type I collagen deposition and reduces cell motility and transmigration of PTECs. A) Confluent monolayers of human PTECs were subjected to either static (0 dyn/cm²) or shear stress (2 and 4 dyn/cm²) conditions for 72 h, and examined for type I collagen deposition by immunofluorescence. Immunofluorescence with a specific collagen I antibody reveals that type I collagen accumulation increases with increasing magnitudes of shear stress. Actin staining utilizing phalloidin was employed as an immunofluorescence control. Scale bars = 100 μm. B, C) In select experiments, cells exposed to either static conditions or shear stress for 72 h were then examined in transwell (B) or wound-healing (C) assays. Sheared PTECs display a decrease in transmigration and cell motility over the time period examined, thereby excluding EMT as a possible mechanism for collagen deposition. Data represent means ± se of n = 3 independent experiments. *P < 0.05.

Interestingly, static PTECs treated with the potent EMT inducer TGFβ1 (50 ng/ml) acquired a spindle-shaped, myofibroblast-like morphology, exhibited only intracellular staining for type I collagen (Fig. 2A), and displayed a drastic increase in transmigration (Fig. 2B) and cell motility (Fig. 2C). Intracellular staining of type I collagen was also observed after PTEC treatment with lower TGFβ1 concentrations (0.5 or 5 ng/ml), though negligible differences in transmigratory and cell motility potentials were noted relative to untreated control (UC) specimens (Fig. 2).

Figure 2.

TGFβ1 fails to induce type I collagen accumulation but increases cell motility in static PTECs. A) Following static incubation with 0.5, 5.0, or 50 ng/ml of recombinant TGFβ1, confluent monolayers of PTECs were examined by immunofluorescence for type I collagen accumulation. Following 72 h of TGFβ1 treatment, type I collagen staining remains intracellular. Actin staining utilizing phalloidin was employed as an immunofluorescence control. Scale bars = 100 μm. B, C) In select experiments, PTECs were incubated with recombinant TGFβ1 (0.5, 5.0, 10, and 50 ng/ml) and examined in transwell (B) and wound healing (C) assays. Data represent means ± se of n = 3 independent experiments. *P < 0.05, ^‡P < 0.01.

To determine whether shear-mediated inhibition of cell motility occurred via reduced expression of TGFβ1, confluent PTEC monolayers were transfected with a vector encoding full-length, wild-type TGFβ1. This genetic intervention resulted in increased TGFβ1 mRNA expression (Fig. 3A) and increased accumulation of mature TGFβ1 protein in sheared specimens relative to appropriate controls (Fig. 3B). While statically cultured PTECs overexpressing TGFβ1 exhibited ∼75% increased transmigration in transwell assays at 72 h (Fig. 3C), similarly transfected cells subjected to fluid shear stress (2 dyn/cm²) for 72 h failed to display any change in overall transmigratory potential compared to untransfected, sheared PTECs (Fig. 3C). Taken together, these data suggest that while exogenous or ectopically expressed TGFβ1 in statically cultured PTECs results in the acquisition of mesenchymal characteristics without subsequent type I collagen accumulation typical of fibrosis, fluid shear induces fibrosis while antagonizing TGFβ1-dependent EMT-associated increases in transmigration.

Figure 3.

Ectopic expression of TGFβ1 abrogates type I collagen accumulation in sheared PTECs but fails to reverse shear-dependent reduction of cell motility. A, B) PTECs transfected with a TGFβ1 expression vector were assessed for changes in TGFβ1 gene expression via qRT-PCR (A) and accumulation of active TGFβ1 protein via ELISA (B) relative to appropriate controls. C) Transfected PTECs were subsequently subjected to either static or shear stress (2 dyn/cm²) conditions for 72 h, and their transmigratory potential was assessed in a transwell assay. Data represent means ± se of n = 3 independent experiments. *P < 0.05 vs. SC; ^†P < 0.01 vs. static TGFβ1 overexpression samples. D) In select experiments, confluent monolayers of human PTECs were subjected to either static (0 dyn/cm²) or shear stress (2 dyn/cm²) conditions for 72 h, and examined for type I collagen deposition by immunofluorescence. Actin staining utilizing phalloidin was employed as an immunofluorescence control. In some experiments, a neutralizing TGFβ1 mAb (40 μg/ml) was added to medium just before the onset of shear. Scale bars = 100 μm.

In light of the divergent effects of fluid shear and exogenous (or ectopically expressed) TGFβ1 on cellular motility and transmigration, we next examined the effects of modulating TGFβ1 activity on collagen deposition in shear-activated PTECs. Our data reveal that blocking TGFβ1 activity via a neutralizing antibody did not alter the appearance of collagenous plaques in PTECs sheared for 72 h (Fig. 3D) nor abrogated type I collagen mRNA expression (Supplemental Fig. S1F). To the contrary, ectopic expression of TGFβ1 in sheared PTECs markedly suppressed extracellular collagen accumulation (Fig. 3D) and reduced type I collagen gene expression (Supplemental Fig. S1G); of note, the majority of type I collagen staining is intracellular (Fig. 3D). Thus, we postulate that fluid shear actively inhibits TGFβ1-dependent antifibrotic signaling in PTECs during the fibrotic response.

To delineate the signaling pathway through which TGFβ1 exerts its antifibrotic effects, we investigated the potential role of the canonical TGFβ1 targets SMAD2 and SMAD3 on type I collagen deposition. Targeted knockdown of SMAD2 resulted in recovery of collagen plaque formation in shear-activated PTECs overexpressing TGFβ1 (Fig. 4A). Similarly, targeted depletion of SMAD3 in shear-stimulated PTECs failed to attenuate collagen accumulation (Fig. 4A). These results are also substantiated by qRT-PCR data that demonstrate knockdown of SMAD2/3 signaling components results in overall increases in collagen gene expression (Supplemental Fig. S1H, I). Selective depletion of SMAD2 and SMAD3 via RNA interference (RNAi) was confirmed by Western blot analysis (Fig. 4B). It is noteworthy that incubation of PTECs with increasing concentrations of exogenous TGFβ1 under static conditions resulted in linear increases in both SMAD2 and SMAD3 phosphorylation (Fig. 4C). Similarly, ectopic expression of TGFβ1 led to direct, consistent phosphorylation of SMAD2 and SMAD3 in shear-activated PTECs (Fig. 4D). Notably, these results contrast with the oscillatory TGFβ1-dependent induction of type I collagen gene expression, and further suggest that collagen expression in shear-activated PTECs is SMAD-independent. Collectively, these data reveal that increasing concentrations of TGFβ1 readily activate SMAD2 and SMAD3 in PTECs and exert inherent antifibrotic pressures on renal tubular cells.

Figure 4.

Canonical TGFβ1 signaling in shear-activated PTECs is antifibrotic. A) PTECs, simultaneously transfected with the TGFβ1 expression vector and SMAD2 siRNA or empty vector and control oligonucleotides, were subjected to shear stress (2 dyn/cm²) for 72 h and examined by immunofluorescence for type I collagen accumulation. In select experiments, PTECs, transfected with either SMAD3 siRNA or scramble control oligonucleotides, were exposed to shear stress (2 dyn/cm²) for 72 h, and also examined by immunofluorescence. Actin staining using phalloidin was employed as an immunofluorescence control. Scale bars = 100 μm. B) SMAD2 and SMAD3 knockdown was confirmed via immunoblotting. C) In separate experiments, PTECs were incubated with increasing concentrations of recombinant TGFβ1 for 1 h, resulting in linear increases in SMAD2 and SMAD3 phosphorylation. D) In other experiments, PTECs transfected with the TGFβ1 expression vector were exposed to shear stress (2 dyn/cm²) for prescribed periods of time. Expression and activation of SMAD2 and SMAD3 in sheared PTECs were monitored via immunoblotting. SC was transfected with empty vector.

Exogenous TGFβ1 and fluid shear exhibit distinct temporal dynamics in the expression of EMT marker proteins

Prior observations reveal that ERK activation is required for EMT in multiple cell lines (34, 35). We herein confirm that blockade of ERK2 activation via the use of the specific pharmacological agent U0126 abolished TGFβ1-dependent changes in cell morphology (Fig. 5A) and transmigratory potential of PTECs (Fig. 5B). Treatment of statically incubated PTECs with increasing doses (0-5 ng/ml) of exogenous TGFβ1 for 1 h induced a linear increase in SMAD2/3 activation (Fig. 4C) but, in contrast, elicited oscillatory ERK2 phosphorylation (Fig. 5C). Quantitation of these Western blots is provided in Supplemental Fig. S2. Increasing TGFβ1 concentrations induced expression of DUSP6 (Fig. 5C), an ERK2 phosphatase. Therefore, we next chose to examine whether failure of ectopic TGFβ1 expression to induce EMT in shear-activated PTECs is due to inhibition of ERK2 signaling. Interestingly, immunoblotting assays reveal that, while ectopic expression of TGFβ1 resulted in nearly uniform SMAD2 phosphorylation in sheared PTECs (Fig. 4D), ERK2 activation still displays an oscillatory profile (Fig. 5D). Moreover, treatment of sheared PTECs overexpressing TGFβ1 with the MEK2 inhibitor U0126 resulted in complete inhibition of ERK2 phosphorylation and a concomitant increase in MEK2 phosphorylation (Fig. 5D). Together, these data establish that ERK2 exerts an antagonistic effect on its activating kinase MEK2. Because static treatment of PTECs with exogenous TGFβ1 resulted in transient ERK2 phosphorylation but successful EMT-associated responses, we hypothesized the existence of divergent regulatory mechanisms in exogenous TGFβ1- vs. shear- activation of PTECs. These data suggest that regulation of EMT and determination of cell phenotype may be more than just a binary switch process and instead involve complex temporal regulation of the relevant signaling molecules.

Figure 5.

The effects of TGFβ1 on PTEC morphology and motility are mediated via an ERK-dependent pathway. A) Confluent monolayers of human PTECs were incubated under static conditions for 72 h with recombinant TGFβ1 (50 mg/ml) in the absence or presence of the MEK1/2 inhibitor U0126 (10 μM). Cell morphology was examined by phase-contrast microscopy. Scale bar = 100 μm. B) Cell transmigratory potential was assessed in a transwell assay. C) In select experiments, PTECS were incubated under static conditions with increasing concentrations of recombinant TGFβ1 (0-5 ng/ml), and the activation of the ERK pathway was monitored via immunoblotting. D) In other experiments, PTECs transfected with the TGFβ1 expression vector were subjected to shear stress (2 dyn/cm²) for prescribed periods of shear exposure time in the absence or presence of U0126 (10 μM), and the activation of the ERK pathway was monitored via immunoblotting. SC was transfected with empty vector. Data represent means ± se of n = 3 independent experiments. *P < 0.05 vs. SC, ^†P < 0.05 vs. static TGFβ1 treatment without inhibitor.

Snail1 is a key transcription factor responsible for the initiation of the EMT program across multiple cell types (36). Though treatment of static PTECs with 50 ng/ml exogenous TGFβ1 led to transient ERK2 activation, the expression of downstream Snail1 remained stably up-regulated relative to UCs (Fig. 6A, B). More important, static TGFβ1 treatment of PTECs also resulted in steady expression of the mesenchymal marker proteins vimentin and N-cadherin. Increased expression of these mesenchymal markers was also confirmed via immunofluorescence (Supplemental Fig. S3C, D). Expression of the epithelial marker occludin remained relatively stable at early time points (up to 120 min; Fig. 6B) but was markedly suppressed after ≥6h of TGFβ1 stimulation (Supplemental Fig. S3A, B). Overall, these changes in EMT marker expression are consistent with the increased motility and transmigratory potentials of TGFβ1-treated PTECs.

Figure 6.

Divergent effects of exogenous TGFβ1 and shear stimulation on the persistent vs. oscillatory expression of EMT markers. Confluent monolayers of PTECs were either incubated with recombinant TGFβ1 (50 or 5 ng/ml) under no-flow conditions (A–D) or subjected to either static or fluid shear (2 dyn/cm²) conditions in the absence of exogenous TGFβ1 for the prescribed periods of time (E, F). Activation of the ERK2 signaling cascade and expression of EMT marker proteins was assessed via immunoblotting (B, D, F). Snail1 gene expression in TGFβ1-treated or sheared PTECs was quantified via qRT-PCR over the time period examined relative to UCs or SCs (A, C, E). UCs had no exogenous TGFβ1. Data represent means ± se of n = 3 independent experiments.

Treatment of confluent PTEC monolayers with lower concentrations of TGFβ1 (5.0 ng/ml), where EMT-associated changes in motility were less apparent (Fig. 2B, C), resulted in more sporadic bursts of ERK2 activity but still retained stable Snail1 gene expression (Fig. 6C). Snail1 protein expression, however, was delayed relative to specimens treated with 50 ng/ml of TGFβ1 (Fig. 6B, D). Similarly, N-cadherin protein expression in samples treated with 5.0 ng/ml TGFβ1 was oscillatory, whereas vimentin expression failed to rise above the basal levels of the SCs (Fig. 6D). Occludin protein expression was similarly unaffected (Fig. 6D). Cumulatively, these data link altered temporal activation patterns of ERK2 to divergent phenotypic outcomes and suggest that lower concentrations of TGFβ1 result in unstable or incomplete EMT in PTECs via less persistent ERK activation.

We next chose to examine the effects of fluid shear alone on ERK2 activation and expression of key EMT molecules. Our data show that PTECs exposed to 2 dyn/cm² exhibited oscillations in both ERK2 phosphorylation and Snail1 expression (Fig. 6E, F). Likewise, protein expression of vimentin and N-cadherin is also temporally regulated in sheared PTECs, potentially reflecting an incomplete EMT response. It is also noteworthy that ERK2 phosphorylation varies inversely with MEK2 activation in sheared PTECs (Fig. 6F), which further corroborates the antagonistic actions of ERK2 and MEK2 noted in Fig. 5D. In addition, the epithelial marker protein occludin, exhibited oscillatory albeit moderate up-regulation in shear-activated PTECs (Fig. 6F and Supplemental Fig. S4A), thereby precluding an EMT. Collectively, our data identify strong temporal regulation of key EMT molecules in sheared PTECs, and suggest that even their oscillatory, transient expression is insufficient to induce an EMT-like response.

To prove that the oscillatory expression profiles of phosphorylated ERK2 and its downstream EMT markers are critical to shear-mediated inhibition of the EMT response in PTECs, we transfected cells with a constitutively active, MEK-independent ERK2 autophosphorylation mutant or empty vector control. This mutant consists of full-length ERK protein with two critical mutations: R65S in the C helix, a residue near the active site, and D319N, which may reduce the affinity to phosphatases (37). Transfected PTECs were subsequently subjected to fluid shear (2 dyn/cm²) for 72 h, and analyzed for changes in cellular morphology via phase-contrast microscopy and transmigratory potential in transwell assays. Persistent activation of ERK2 in shear-activated PTECs, which was confirmed by immunoblotting (Fig. 7C), resulted in spindle-like cell morphology (Fig. 7A), which is typical of a mesenchymal phenotype. Moreover, this genetic intervention not only abolished the shear-mediated decrease in PTEC transmigratory potential but also increased it by ∼2-fold relative to SCs (Fig. 7B). Most important, the constitutive activation of ERK2 in sheared PTECs was accompanied by the persistent up-regulation of EMT markers, including Snail1, vimentin, and N-cadherin (Fig. 7C, D), as well as stable suppression of epithelial tight junction protein occludin (Fig. 7C and Supplemental Fig. S4A). Collectively, these data disclose that the constitutive ERK2 activation is sufficient to overcome the shear-mediated antagonism of EMT in PTECs. These data also show that novel temporal regulation of ERK2 activation and expression of downstream EMT molecules maintains an epithelial phenotype in PTECs during fibrogenesis.

Figure 7.

A constitutively active ERK2 mutant induces persistent expression of EMT markers and increases cell motility of shear-activated PTECs. A) Confluent monolayers of PTECs were transfected with either an MEK-independent, ERK2 autophosphorylation mutant or empty vector control, and subjected to either static or shear stress (2 dyn/cm²) conditions for 72 h. Cell morphology was examined by phase contrast microscopy. Scale bar = 100 μm. B) Cell transmigratory potential was assessed in a transwell assay. C) Stable ERK2 activation and expression of EMT marker proteins was confirmed via immunoblotting. D) Transcript levels of Snail1 were evaluated via qRT-PCR. Data represent means ± se of n = 3 independent experiments. *P < 0.05.

DISCUSSION

Our work offers a systematic examination of the role of an EMT in PTECs on the development of renal fibrosis. We report that pathological levels of fluid shear diminish the antifibrotic effect of TGFβ1-mediated canonical SMAD2/3 signaling in PTECs and further prevent induction of mesenchymal features via precise temporal control of downstream ERK signaling events. Our data firmly establish that the HK-2 proximal tubule cell line excretes type I collagen in response to fluid shear without an apparent phenotypic transformation. These results reinforce other potentially controversial reports of the independence of EMT from fibrosis, although we are the first to provide more than a simple phenomenological inquiry, and specifically identify a mechanism for inhibition of EMT during the fibrotic response. In particular, our results showing the absence of EMT in shear-activated PTECs are in accord with prior observations showing that moderate shear stress actually increases renal epithelial intercellular adhesion via the formation of tight and adherens junctions (6). Similarly, our data are consistent with observed increases in cellular E-cadherin resistant to TGFβ1 intervention in a study utilizing unilateral ureteral obstruction, a widely accepted animal model of kidney disease (38). Furthermore, our data suggest that progression of fibrosis and the initiation of the EMT machinery may be mutually exclusive events. This is corroborated by our observations that a constitutively active ERK mutant induces cell morphological and functional changes associated with an EMT response in sheared PTECs but simultaneously blocks shear-induced collagen deposition. Our findings are also consistent with work by Takeji et al. (39) in which elimination of α-smooth muscle actin, a marker of a mesenchymal phenotype, actually aggravated fibrosis in a mouse model. However, we cannot fully exclude the possibility that EMT may be a transient phase within a more complex temporal program whose cumulative effects result in a final fibrotic outcome. It is also noteworthy that our experimental approach accounts for the effects of a pathophysiologically relevant stimulus (i.e., shear stress) as opposed to focusing solely on the effects of exogenous growth factors on PTEC function.

Although several studies have established the fibrogenic potential of TGFβ1 (11, 12), more recent reports are starting to challenge the simplicity of this finding. For example, recent work has discovered that low doses of TGFβ1 stimulate expression of collagen, while high doses of TGFβ1 trigger inhibition of collagen I via the negative transcriptional regulator CUX1 (26). Even a phase I/II trial of anti-TGFβ1 therapy in systemic sclerosis failed to note any efficacy in halting the development of fibrotic lesions (28). Moreover, work in embryonic kidney cells suggests that the ability of TGFβ1 to initiate an EMT depends indirectly on the activity of GSK3-β, a kinase targeted by many different pathways (40). Cumulatively, our results reinforce these reports, and firmly establish TGFβ1 as a pleiotropic cytokine, exerting complex, context-dependent effects in concert with other pathways.

Although we do not observe negative feedback inhibition of TGFβ1-mediated SMAD2 signaling, we do identify oscillatory ERK activation via TGFβ1 induction of the DUSP6 phosphatase, as well as ERK2-mediated MEK2 inhibition. We further demonstrate that the network structures responsible for this dynamic ERK activation result in oscillations in downstream EMT signaling molecules in sheared PTECs. Oscillatory ERK dynamics has been noted before in several different systems activated with a wide range of different stimuli, and the potential kinetic architectures supporting this behavior have been modeled and discussed somewhat (41, 42). Given that our results, as well as work by other groups (35, 43), show ERK activation as a key step driving the EMT process, these oscillations are likely far more than a mere kinetic oddity and may act as a precise mechanism of control. Though oscillatory ERK2 activation is observed in both static and sheared cells, persistent expression of Snail1 and EMT marker proteins occurs only in static PTECs treated with high levels of TGFβ1 (50 ng/ml). This unique response may arise because these conditions result in an overall broader period of ERK2 activation, whereas ERK phosphorylation in shear-stimulated PTECs, as well as PTECs exposed to lower concentrations of TGFβ1, is more transitory (compare Fig. 6B, D). In addition, potential signal amplification from network components could result in an EMT stimulus more resistant to fluctuations in ERK2 activity in static PTECs. This effect could potentially be diminished in sheared cells due to crosstalk with other pathways. Important changes in the cellular phenotype, or lack thereof, are then encoded in the frequency and/or amplitude of the bursts of ERK activity or downstream genes due to these mechanistic and kinetic restraints (see Fig. 8). Such frequency-dependent transcriptional regulation has been established in nuclear factor-κB (NF-κB) signaling, also capable of undergoing cyclic oscillations (44, 45).

Figure 8.

Schematic showing that divergent temporal regulation of Snail1 in TGFβ1- vs. shear-activated PTECs distinguishes EMT from the fibrotic response. In both TGFβ1- and shear-activated PTECs, ERK2 activation oscillates because of feedback inhibition of MEK2 activation and induction of the ERK2 phosphatase DUSP6 (dashed lines). Expression of mesenchymal markers and repression of epithelial markers in static PTECs treated with excess exogenous TGFβ1 is less sensitive to these temporal fluctuations in ERK2 activation and results in phenotypic changes associated with an EMT. In contrast, oscillatory ERK activity in PTECs subjected to fluid shear (2 dyn/cm²), or low concentrations of TGFβ1, is transmitted to downstream EMT proteins, thereby allowing cells to maintain an epithelial phenotype during the fibrotic response. Use of a constitutive ERK2 mutant results in persistent expression of EMT molecules similar to treatment with high concentrations of TGFβ1 and recovers the EMT response. Successful EMT inversely correlates with excessive collagen deposition. Thus, EMT and renal fibrosis represent two mutually exclusive cell fates in shear-activated PTECs. TF represents an ERK2-activated transcription factor regulating expression of EMT-associated genes in our model; t represents time.

Taken together, our data indicate that the sole focus on TGFβ1 and reversal or inhibition of EMT as a therapeutic target in renal fibrosis may be inadequate. It is possible that some other growth factor or pathway occupies a more fundamental role in driving shear-induced fibrosis. The family of platelet-derived growth factor (PDGF) proteins, for example, are typically associated with stimulating growth of myofibroblast-like cells, but inhibition of the PDGF-D isoform, in particular, is capable of preventing tubulointerstitial fibrosis in a rat model (46). Similarly, inhibition of the Notch cascade has been shown to prevent renal injury (47) as well as inhibit folic acid-induced fibrosis (48). Finally, our observation that EMT and fibrosis may represent two divergent paths in the proximal tubular epithelial cell is surprising but reinforces our observation that TGFβ1, potentially the primary mediator of an EMT response, is itself antifibrotic. Our results not only represent a potentially new paradigm, but also advise extreme caution when interpreting data within seemingly time-invariant systems.