Tonic ErbB signaling underlies TGFβ-induced activation of ERK and is required for lens cell epithelial to myofibroblast transition

Abstract

Fibrosis is a major, but incompletely understood, component of many diseases. The most common vision-disrupting complication of cataract surgery involves differentiation of residual lens cells into myofibroblasts. In serum-free primary cultures of lens epithelial cells (DCDMLs), inhibitors of either ERK or of ErbB signaling prevent TGFβ from upregulating both early (fibronectin) and late (αSMA) markers of myofibroblast differentiation. TGFβ stimulates ERK in DCDMLs within 1.5 h. Kinase inhibitors of ErbBs, but not of several other growth factor receptors in lens cells, reduce phospho ERK to below basal levels in the absence or presence of TGFβ. This effect is attributable to constitutive ErbB activity playing a major role in regulating the basal levels pERK. Additional studies support a model in which TGFβ-generated reactive oxygen species serve to indirectly amplify ERK signaling downstream of tonically active ErbBs to mediate myofibroblast differentiation. ERK activity is in turn essential for expression of ErbB1 and ErbB2, major inducers of ERK signaling. By mechanistically linking TGFβ, ErbB, and ERK signaling to myofibroblast differentiation, our data elucidate a new role for ErbBs in fibrosis and reveal a novel mode by which TGFβ directs lens cell fate.

• It is not understood how TGFβ contributes to secondary cataract (PCO), the most common vision-disrupting complication of cataract surgery.

• Using our serum-free, primary lens epithelial cell culture system, we show that TGFβ leverages basal ERK activity downstream of tonic ErbB signaling to induce differentiation of lens epithelial cells into myofibroblasts.

• By mechanistically linking TGFβ, oxidative stress, ErbB, and ERK signaling, our data elucidate a new role for ErbBs in fibrosis and support the use of ErbB inhibitors as therapeutics against PCO.

INTRODUCTION

Our vision is critically dependent on the proper development of lens cells. Epithelial cells on the anterior face of the organ differentiate into highly elongated, crystallin-rich lens fibers cells at what is referred to as the equator of the lens (Cvekl and Ashery-Padan, 2014). Defects in this process can result in lens opacity, although more common causes of cataract include older age, injury, and excessive UV exposure (Liu et al., 2017). The only treatment for the vast majority of cataracts regardless of etiology is surgery, in which the cells of the cloudy lens are removed from their encasing lens capsule and replaced with an artificial lens. Lens epithelial cells that survive this operation can undergo two types of abnormal differentiation (Apple et al., 2000; Awasthi et al., 2009; Wormstone et al., 2009; Findl et al., 2010; Fișuș and Findl, 2020). Whereas some of these residual lens cells attempt to become lens fiber cells, others appear to transform into contractile, extracellular matrix-producing myofibroblasts. If the latter accumulate at the rear of the lens capsule, they can interfere with transmission of light to the retina and lead to fibrotic posterior capsule opacification (PCO). PCO is the most common cause of long-term visual disruption after cataract surgery (Konopińska et al., 2021). Fibrosis of lens epithelial cells that postoperatively remain at the anterior of the capsule can result in anterior capsule opacification (ACO), which also adversely affects eye health (Trivedi et al., 2002). In addition, epithelial cells in the intact lens can also become fibrotic due to injury or disease, forming vision-impairing anterior subcapsular cataracts (Novotny and Pau, 1984; Shu et al., 2017).

It has been shown in in vitro and in vivo mammalian (Srinivasan et al., 1998; Lovicu et al., 2004; Robertson et al. 2007; West-Mays et al., 2010; Wormstone and Eldred, 2016; Tiwari et al., 2016) and avian (Boswell et al., 2017) systems that TGFβ induces lens epithelial cells to undergo an epithelial to mesenchymal transition (EMT) to myofibroblastic cells (EMyT; epithelial to myofibroblast transition; Masszi et al., 2010), in keeping with the major role of this growth factor in fibrosis in other organs (Eldred et al., 2011; Meng et al., 2016; Di Gregorio et al., 2020; Frangogiannis, 2020). TGFβ signaling in lens cells is known to be elevated after cataract surgery (Meacock et al., 2000; Saika et al., 2002; Wormstone et al., 2002; de Iongh et al., 2005), and is thought to first rise as part of a wounding response and then be maintained by chronic production and activation of endogenous TGFβ (Wormstone et al., 2006; Mamuya and Duncan, 2012; VanSlyke et al., 2018). In rat (Wojciechowski et al., 2017), human (Tiwari et al., 2016), and chick (Boswell et al., 2017) primary lens cell systems, TGFβ upregulation of EMyT is blocked by inhibitors of ERK activation. Conversely, conditional knock-out in the lens of two genes (sprouty 1and 2) that encode proteins that inhibit ERK signaling induce the formation of myofibroblasts (Shin et al., 2012).

More recently, it has been demonstrated that the ability of TGFβ to upregulate EMyT in primary rat (Shu et al., 2019) and chick (VanSlyke et al., 2023) lens cells is blocked by small molecule inhibitors of the ErbB family of receptor tyrosine kinases. ErbBs are widely expressed and regulate several essential processes including differentiation, proliferation, growth, survival, and migration in a cell type-specific manner (Wieduwilt and Moasser, 2008; Chen et al., 2016). We recently reported (VanSlyke et al., 2023) that serum-free, nonpassaged primary cultures of embryonic chick lens epithelial cells (DCDMLs) express on the cell surface all three functional tyrosine kinases in the ErbB family (namely, ErbB1/EGFR1, ErbB2/EGFR2/HER2, and ErbB4/EGFR4) and release ErbB-activating ligands into the medium. Our findings further indicated that most ErbB signaling under basal conditions is mediated by ErbB1/B2 heterodimers, and that upregulation of ErbB1 in response to TGFβ is likely to be important for myofibroblast differentiation. In contrast, TGFβ greatly reduces the total and cell surface levels of ErbB4, and there is no evidence that this species plays a causative role in TGFβ-induced EMyT (VanSlyke et al., 2023). ErbB inhibitors have previously been reported to have anti-PCO activity in human lens capsular bags (Maidment et al., 2004; Wertheimer et al., 2015, 2018), and siRNA knockdown of ErbB1 has been reported to reduce PCO in an in vivo rat model (Huang et al., 2011). In DCDMLs, TGFβ induces not only the formation of myofibroblasts, but also stimulates other lens cells in the same cultures to become lens fiber cells by an as yet incompletely characterized mechanism that is pharmacologically distinct from that involved in EMyT (Boswell et al., 2017). Notably, ErbB inhibitors do not prevent lens fiber cell differentiation downstream of TGFβ (VanSlyke et al., 2023).

Despite its relevance for PCO, we do not understand why ErbB activity is essential for TGFβ-induced EMyT. Here, we show that ongoing (i.e., tonic; basal) ErbB signaling is the major source of the ERK activity required for TGFβ to promote EMyT in primary cultures of lens epithelial cells. Our data support a model in which TGFβ indirectly increases the level of phospho ERK by altering a redox-dependent process downstream of ErbB autophosphorylation, but before activation of Raf-1. ERK activity is reciprocally required for expression of ErbB1 and ErbB2 protein. Together, these results elucidate how TGFβ, ErbB, and ERK signaling interact to direct the formation of myofibroblasts, a cell type central to the etiology of PCO and ACO. Given the widespread role of TGFβ, oxidative stress, ERK, and myofibroblasts in fibrosis throughout the body, our findings are also relevant to nonlenticular organs including lung, kidney, and liver.

RESULTS

Role of ERK and ErbB signaling in TGFβ-induced myofibroblast formation in DCDMLs

After 6 d of treatment with TGFβ (beginning on the day after plating, with the day of plating designated as d 0 of culture), some cells in DCDML primary lens epithelial cells undergo EMyT, including expression of the canonical myofibroblast marker αSMA in stress fibers and fibronectin in extracellular fibrils (Boswell et al., 2010, 2017). The highly selective ErbB inhibitor lapatinib blocks TGFβ-induced expression of αSMA and FN to near the level obtained when ERK is inactivated by the MEK inhibitor UO126, or if cells are cocultured with the TGFβR inhibitor SB-431542 (Figure 1A). In contrast, lapatinib does not inhibit the ability of TGFβ to promote lens fiber cell differentiation (Figure 1B; VanSlyke et al., 2023). Similar results were obtained with other mechanistically and structurally distinct ErbB inhibitors tested including erlotinib, and the irreversible blocker afatinib (Supplemental Figure S1). As expected (Liu et al., 2007), ErbB inhibitors do not prevent TGFβ from activating the downstream regulator Smad3, as assessed by pSer423/425 Smad3 immunoreactivity and expression of a Smad3 transcriptional reporter (VanSlyke et al., 2023). Taken together, these results support a specific, non-Smad3 role for ErbB signaling in TGFβ-induced EMyT.

FIGURE 1:

Effect of inhibition of ERK or ErbB on TGFβ stimulation of myofibroblast differentiation in DCDMLs. DCDML primary lens epithelial cells were treated starting on d 1 of culture with or without 4 ng/ml TGFβ1, in the presence or absence of the following kinase inhibitors: lapatinib (lap), UO126 (UO), or SB-431542 (SB4). Cells were lysed on d 7 and whole cell lysates analyzed for the EMyT markers FN and αSMA (A), or for FN, αSMA, and the lens fiber cell markers δ-crystallin, CP115, and CP49 (B) by Western blot. (C) Quantitation of EMyT marker data, expressed as the percent inhibition induced by lapatinib, UO126, or SB-431542 relative to cells treated with TGFβ and vehicle (0.1% DMSO) only. All data are normalized to tubulin in the same sample. For all, n ≥ 6; p = 0.000.

In DCDMLs, αSMA levels do not start to rise until ∼ 96 h after addition of TGFβ (Boswell et al., 2010, 2017). In contrast, expression of FN protein is stimulated within 48 h, consistent with initiation of protomyofibroblast differentiation (Tomasek et al., 2002; Eyden, 2008). Lapatinib and UO126 both prevented TGFβ from upregulating FN expression at this time (Figure 2). The finding that the activities of both ErbB and ERK are critical during the early stage of TGFβ-induced myofibroblast differentiation raises the possibility of a causal relationship between the two kinases in this process.

FIGURE 2:

Both ERK and ErbB activity are required for TGFβ to induce the early stage of EMyT. DCDMLs were cultured from d 1–3 in the presence or absence of TGFβ with or without lapatinib, UO126, or SB-431542 as indicated. Cells were lysed on d 3 and whole cell lysates analyzed for expression of FN. For all, p ≤ 0.004.

ErbBs are the major source of basal ERK activity in lens cells

To determine the extent to which growth factors produced by lens epithelial cells contribute to basal ERK activity in DCDMLs, we interrogated the ability of inhibitors of these pathways to reduce the level of phospho ERK. Addition of lapatinib to otherwise untreated DCDMLs elicited a striking decrease in the level of pERK within 90 min without affecting the amount of total ERK (Figure 3A) or the basal phosphorylation of p38 MAP kinase (Figure 3B) or the basal phosphorylation of AKT (Figure 3C). Avian cells express ERK2, but not ERK1, accounting for the single phospho ERK immunoreactive band (Lefloch et al., 2008). Notably, the reduction in pERK attained with lapatinib or erlotinib was greater than that observed after inhibition of FGFRs (Figure 3, A and D), a previously identified positive effector of ERK activity in lens cells in vivo (Govindarajan and Overbeek, 2001) and in culture (Wang et al., 2010; Boswell et al., 2017). Combining maximally effective concentrations of lapatinib with the FGFR inhibitor PD173074 resulted in an additive decrease in pERK to very near the level obtained with U0126 treatment, without affecting total ERK levels (Figure 3, A and D). Blockers of receptors for PDGF (crenolanib) or TGFβ (SB-431542) did not reduce pERK when tested under the same conditions (Figure 3D). These findings indicate that ErbB and FGFR signaling are together responsible for virtually all basal ERK activity in DCDMLs, with ErbB playing the major role. The finding that ErbBs are critical for the basal activity of ERK, but not of p38 or AKT, is indicative of a highly selective role for ErbBs in the regulation of lens cell signaling.

FIGURE 3:

Effect of blocking ErbB and/or other growth factor receptors on the basal level of phospho ERK, p38, and AKT in DCDMLs. (A–C) DCDMLs were incubated for 90 min with DMSO, lapatinib, the selective FGFR inhibitor PD173074, lapatinib plus PD173074, or UO126 before Western blot analysis of phospho and total forms of ERK, p38, AKT, or of tubulin as indicated. (D) Results graphed as fold controls treated with DMSO only. Also included are data demonstrating that the ErbB inhibitor erlotinib (erlot), but not inhibitors of the receptors for TGFβ (SB-431542) or PDGF (crenolanib), reduces basal levels of pERK. Asterisks indicate p ≤ 0.001; for all other conditions, p ≥ 0.337. (E) DCDMLs were incubated with DMSO or lapatinib and lysed after a 5 min treatment with 10 ng/ml HB-EGF or medium only to demonstrate that the anti-pAKT antibody recognizes activated (pS475) AKT in DCDMLs.

ErbB signaling is specifically required for TGFβ-induced activation of ERK

We show here that, as expected, activation of ERK in response to the ErbB1/4 ligand HB-EGF is abolished by preincubation with UO126 or lapatinib, but not by the FGFR inhibitor PD173074 (Figure 4A). Notably, pERK levels drop to basal (e.g., no added HB-EGF) levels within 90 min of HB-EGF addition (Figure 4A), indicating that in DCDMLs, as in many other cell types (Traverse et al., 1992; Wang et al., 2010; Lemmon et al., 2016), activation of ERK by ErbBs is rapidly attenuated.

FIGURE 4:

ErbB and TGFβ signaling in lens cells. DCDMLs were pretreated for 1 h with DMSO or the indicated inhibitors before incubation for 5 or 90 min with either 10 ng/ml HB-EGF, 4 ng/ml TGFβ, 1 or 10 ng/ml FGF, or an equal volume of M199 medium (-). Whole cell lysates were analyzed for phospho and total forms of ERK, Smad3, and p38, or for tubulin by Western blot. (A) ErbB ligand induces a rapid and transient activation of ERK compared with TGFβ. Results graphed as fold pERK relative to DMSO only control. p < 0.001, except NS (p >0.325). Activation by phosphorylation decreases the apparent mobility of total (t) ERK(2) on SDS–PAGE, resulting in a doublet under conditions in which ERK is partially activated. (B) Inhibitor sensitivity of activation of ERK by TGFβ. Results graphed as fold pERK relative to DMSO only (no TGFβ) control. p < 0.001, except NS (p = 0.22). Compared to cells treated with TGFβ plus DMSO, the level of pERK in cells treated with TGFβ plus PD173074 is significantly lower (p = 0.000), whereas crenolanib (0.5 μM) and noggin (1 μg/ml) have no effect on activation of pERK by TGFβ (p > 0.628). Also graphed are data demonstrating that the ErbB inhibitor erlotinib and the TGFβR inhibitor SB-431542 reduce upregulation of pERK by TGFβ. (C) Lapatinib does not block activation of Smad3 by TGFβ. (D) Lapatinib does not block activation of p38 by TGFβ. (E) Quantitation of the data as compared with treatment with TGFβ and DMSO. In all cases, p ≥ 0.133. (F) Lapatinib does not block activation of pERK by FGF, as quantitated as percent inhibition of pERK in cells treated with FGF + lapatinib compared with cells treated with FGF + DMSO.

TGFβ also increases activation of ERK, but this response is slower than that elicited by ErbB ligand (Figure 4A). As previously reported (Boswell et al., 2017), UO126 pretreatment abolishes TGFβ-mediated elevation of pERK (Figure 4B). We show here that TGFβ upregulation of pERK in DCDMLs is also prevented by lapatinib or erlotinib (Figure 4B). Notably, the level of pERK in cells pretreated with either the MEK or ErbB inhibitor before the addition of TGFβ was below that of untreated controls. In contrast, blockers of two other growth factor receptors active in lens cells, namely PDGFR (crenolanib) and BMPRs (noggin), were ineffective in preventing TGFβ from increasing pERK (Figure 4B). Blocking FGF signaling partially (by ∼ 30%) reduced, but did not eliminate, ERK activation in response to TGFβ (Figure 4B). Notably, ErbB inhibition did not decrease the expression of total ERK (Figure 4, C and E), or diminish the ability of TGFβ to activate Smad3 or p38 (Figure 4, D and E) under the same treatment conditions. ErbB inhibitors also failed to reduce activation of ERK in response to physiologically relevant levels of FGF (Figure 4F; p ≥ 0.211; n = 10). Taken together, these results indicate that ErbB inhibitors specifically block activation of ERK downstream of TGFβ, and not in response to FGF.

In the intact lens and shortly after cataract surgery, central lens epithelial cells have access to the aqueous humor. FGF at the concentration thought to be present in aqueous humor (∼ 1 ng/ml) increases the level of pERK in mammalian (Iyengar et al., 2009) and avian (Le and Musil, 2001) primary lens cells more than fivefold within 30 min. This is considerably greater than the ∼2X increase in pERK observed after addition of TGFβ in either system (Boswell et al. 2017; Wojciechowski et al., 2017; Figure 4A). Given this more potent (and ErbB-independent; Figure 4F) source of ERK activity, how could stimulation of ERK by TGFβ be important for initiation of EMyT of lens epithelial cells under physiologically relevant conditions? The elevation of pERK induced by 1 ng/ml FGF in lens cells is transient, subsiding to basal levels within 6–12 h in cultured primary lens cells (Le and Musil, 2001; Iyengar et al., 2009). Such kinetics are consistent with the very low levels of active ERK detected in central lens epithelial cells in vivo (Zhao et al., 2008; Li et al., 2014) despite their expression of ligand-responsive FGFRs. Importantly, DCDMLs chronically exposed to 1 ng/ml FGF remain capable of increasing pERK in response to TGFβ in a process blocked by lapatinib (Figure 5A). After cataract surgery, this would allow wounding-induced TGFβ to upregulate pERK and initiate myofibroblast differentiation in an ErbB-dependent manner.

FIGURE 5:

Effect of long-term exposure to FGF or TGFβ on ERK. (A) TGFβ continues to stimulate ERK in DCDMLs chronically exposed to physiological levels of FGF. DCDMLs were cultured for 6 d in either the absence or presence of 1 ng/ml FGF2. Cells were then incubated for 60 min with DMSO or lapatinib, followed by a 90 min treatment with no additions (–), 1 ng/ml FGF2, or 4 ng/ml TGFβ. Whole cell lysates were assessed for phospho ERK, total ERK, or tubulin by Western blot. All results shown are from the same blot of a single experiment. Levels of pERK obtained under the indicated conditions are graphed relative to controls treated with DMSO without inhibitor or growth factor. NS, p = 0.593; for all other conditions, p < 0.012. (B) Prolonged exposure to TGFβ increases both phospho and total levels of ERK. DCDMLs were cultured for 6 d in either the absence or presence of TGFβ. Whole cell lysates were assessed for pERK, tERK, or tubulin by Western blot. pERK and tERK values were normalized to tubulin in the same sample.

A related physiologically important question is the effect on ERK of long-term exposure to TGFβ. We found that DCDMLs cultured for 6 d with TGFβ show a twofold elevation of pERK relative to untreated DCDMLs (Figure 5B). Unlike after a 1.5 h treatment with TGFβ (Figure 4C), this is accompanied by a similar increase in the amount of total ERK when normalized to tubulin (Figure 5B), demonstrating that prolonged exposure to TGFβ promotes the levels of both phospho and total ERK.

Lack of evidence for transactivation of ErbB signaling by TGFβ 

An obvious mechanism by which TGFβ could stimulate ERK in a ErbB-dependent manner would be if TGFβ transactivated ErbBs, as has previously been implicated in EMT in some cell types (Uchiyama-Tanaka et al., 2002; Samarakoon et al., 2013a). In one mode of transactivation, TGFβ stimulates the posttranslational processing of transmembrane ErbB ligand proforms into soluble, active species, thereby allowing them to bind and stimulate ErbB1 and then pERK within 1–2 h of exposure to TGFβ (Wang et al., 2008; Ebi et al., 2010). To investigate if enhanced ligand shedding could be responsible for the observed elevation of pERK in DCDMLs in response to TGFβ, we transiently transfected DCDMLs with a plasmid encoding the full-length (i.e., transmembrane) proform of the ErbB 1/4 ligand HB-EGF (Figure 6). Transcripts for HB-EGF have been reported in primary rat lens epithelial cells (Shu et al., 2019). Cells were cultured for 2 d with or without TGFβ, after which the conditioned medium was collected and added to HEK293 cells at 4°C, a temperature that allows ligand-induced tyrosine autophosphorylation of endogenously expressed ErbBs but no other signaling events (Longva et al., 2002; Tong et al., 2014). Active ErbBs were detected by Western blot using the 4G10 antiphosphotyrosine antibody as previously reported (Samarakoon et al., 2013a; VanSlyke et al., 2023). This assay measures the ability of mature, soluble ligands in the DCDML conditioned medium to stimulate the autophosphorylation of HEK cell ErbB 1, 2, and 4, which comigrate under our SDS–PAGE conditions. Control experiments demonstrated that incubation of HEK cells with exogenous, recombinant soluble HB-EGF resulted in a dose-dependent increase in a ∼170 kD, phosphotyrosine-positive ErbB band (Figure 6, lanes 1–4). Conditioned medium from proHB-EGF-transfected DCDMLs cultured with TGFβ failed to activate ErbBs in HEK cells to a greater extent than conditioned medium from no TGFβ transfectants (compare lanes 6 and 7). These data indicate that TGFβ does not increase the generation of soluble HB-EGF ligand in DCDMLs and are in keeping with the reported constitutive expression in lens cells of ADAM-17 (Hodgkinson et al., 2010; Shu et al., 2019), the enzyme primarily responsible for maturation of HB-EGF and a subset of other ErbB ligands (Sahin et al., 2004).

FIGURE 6:

TGFβ does not stimulate the release of HB-EGF from DCDMLs. (A) DCDMLs were transiently transfected with plasmids encoding either a full-length, transmembrane proform of HB-EGF or an irrelevant control transmembrane protein (E208K Cx32; VanSlyke et al., 2023; Ctl) and cultured for 48 h in either the absence or presence of TGFβ. Untransfected HEK293 cells were then incubated at 4°C for 15 min with either 10–0 ng/ml mature, soluble recombinant HB-EGF (lanes 1–4), or the indicated DCDML transfectant conditioned medium (lanes 5–9). HEK whole cell lysates were analyzed by Western blot using antibodies against phosphotyrosine. To confirm the identity of the bracketed 170 kD pY band as autophosphorylated ErbB, recipient HEK cells were incubated at 37°C for 1 h with lapatinib before addition of conditioned medium (lanes 8 and 9); asterisk indicates a nonspecific band. Uniform expression of ErbB1, a major endogenous ErbB in HEKs, is also shown. All data shown are from the same blot of a single experiment. Phospho-Y values (in arbitrary units, normalized to β actin in the same samples) for the HB-EGF concentration curve are provided. (B) ErbB-associated 4G10 signal from three independent experiments was normalized to β actin and showed that conditioned medium from proHB-EGF transfectants cultured with TGFβ induced 0.893 ± 0.115 times the ErbB-associated 4G10 signal as was elicited by conditioned medium from cells cultured in the absence of TGFβ. Individual data points are plotted as black squares.

Next, we directly examined the activation state of ErbBs by subjecting DCDMLs to cell surface biotinylation, after which all active ErbBs on the plasma membrane were detected by Western blotting using the 4G10 antiphosphotyrosine antibody. In untreated DCDMLs, a ∼170 kD, lapatinib-sensitive band was detected that represents cell surface ErbBs basally activated by endogenous ligand (VanSlyke et al., 2023) (Figure 7A, lane 3). Pretreatment with TGFβ at 37°C for 5, 45 min, or 1.5 h did not significantly increase the amount of tyrosine autophosphorylated ErbB recovered (Figure 7A, lanes 5–10).

FIGURE 7:

Short-term treatment with TGFβ does not enhance activation of ErbBs in DCDMLs. (A) The level of tyrosine phosphorylated ErbBs on the cell surface is unaffected by TGFβ. DCDMLs were incubated at 37°C for 1 h with DMSO or lapatinib, or for 5, 45, or 90 min with or without TGFβ as indicated. In lanes 1 and 2, cells were then incubated at 4°C for 15 min with 100 ng/ml HB-EGF. All cultures were then subjected to cell surface biotinylation at 4°C, and plasma membrane pools of autoactivated ErbB analyzed by Western blotting using the antiphosphotyrosine 4G10 antibody. To demonstrate equal loading, biotinylated samples were reprobed with ErbB1 antibody and whole cell lysates were analyzed for β actin. (B) TGFβ does not increase autophosphorylation of ErbB1. DCDMLs were incubated at 37°C for 1 h with DMSO or lapatinib, or for 5, 45, or 90 min with or without TGFβ as indicated. In lanes 13-16, cells were then treated at 37°C for 5 min with 0.1, 1, or 10 ng/ml HB-EGF. All cultures were then lysed. Western blotting of whole cell lysates with a rabbit antibody specific to pY1068ErbB1 demonstrated that treatment with HB-EGF, but not with TGFβ, increased the activation of ErbB1. Blots were reprobed with a rat antibody against total ErbB1. (C). Results from (A) and (B) were graphed as fold value from cells incubated with TGFβ versus without TGFβ for the indicated time. In all cases, p > 0.2. Individual data points are plotted from experiments in which n = 3.

ErbB1 is a potent activator of ERK (Albeck et al., 2013; Pinilla-Macua et al., 2017). To specifically examine the activation state of this receptor, we used a rabbit polyclonal antibody specific for the tyrosine 1068 phosphorylated form of ErbB1 (Figure 7B). Autophosphorylation of this residue plays an essential role in downstream activation of many signaling pathways including ERK and AKT (Sebastian et al., 2006; Roskoski, 2014). Unlike the mouse reagent used in Figure 11A, this antibody readily detects endogenous active ErbB1 in whole cell lysates as demonstrated by its recognition of a Mr = 170 kD species (Figure 7B, lane 12) that was blocked by incubation with lapatinib (lane 11) and that is dose-dependently enhanced by exogenously added HB-EGF (lanes 13–15). The level of pY1068R1 immunoreactivity recovered from DCDMLs was unaffected by short-term treatment with pERK-inducing levels of TGFβ (lanes 17–22).

FIGURE 11:

ERK is required for expression of ErbB1 and ErbB2 protein. (A) DCDMLs were cultured for 6 d with DMSO, UO126, TGFβ, or both (DMSO or UO126 added 1 h before TGFβ). The cells were then incubated at 4°C for 15 min in either the absence or presence of 60 ng/ml TGFα as indicated. Whole cell lysates were then analyzed by Western blot with mouse anti-pY1068ErbB1 antibodies, followed by reprobing for total ErbB1. Data are graphed as fold obtained in cultures treated with DMSO instead of UO126 in the same experiment. For all conditions, p = 0.000. Note that the anti-pY1068ErbB1 antibody (#2236) employed in these experiments is not as sensitive as the rabbit reagent used in Figure 7B. (B) DCDMLs were cultured for 6 d with DMSO or the p38 inhibitor SB203580 before analysis of whole cell lysates for total ErbB1. (C) DCDMLs cultured for 6 d with DMSO or UO126 were lysed and analyzed for ErbB2 or ErbB4. (D) Effect of SB203580 on the level of ErbB1 (p = 0.012), or UO126 on the levels of ErbB2 (p = 0.000) and ErbB4 (p = 0.123) detected by Western blot expressed as fold DMSO only cultures in the same experiment.

A new model for TGFβ-induced activation of ERK

If transactivation of ErbBs is not the mechanism by which TGFβ stimulates ERK, then why do ErbB inhibitors prevent TGFβ from upregulating pERK levels? In addition to ErbBs and ERK itself, the core canonical ErbB-to-ERK pathway requires the sequential activity of two intermediate kinases, namely Raf-1 and MEK (Roskoski, 2014). We used specific inhibitors of each kinase in the core pathway to test their effect on pERK levels after stimulation of DCDMLs with either TGFβ or with sodium orthovanadate (NaV). The latter is a general inhibitor of tyrosine phosphatases (Swarup et al., 1982) that can directly block the activity of the phosphatases that inactive ERK (Caunt and Keyse, 2013). We found that sodium orthovanadate was capable of increasing the level of pERK in DCDMLs in the presence of the Raf inhibitor sorafinib, the MEK inhibitor UO126, or the ErbB inhibitor lapatinib (Figure 8A). In contrast, each of these compounds abolished the ability of TGFβ to activate ERK without affecting stimulation of Smad3 (Figure 8B). We conclude that upregulation of pERK by TGFβ requires the canonical ErbB-Raf-MEK-ERK cascade.

FIGURE 8:

Stimulation of ERK by TGFβ requires the canonical ErbB-to-ERK pathway and is not phenocopied by a tyrosine phosphatase inhibitor. DCDMLs were preincubated for 1 h with either DMSO, 10 uM sorafenib, UO126, or lapatinib before exposure to either the general tyrosine phosphatase inhibitor sodium orthovanadate (NaV; 0.2 mM) for 30 min (A), or to TGFβ for 1.5 h (B). The levels of pERK, tERK, pSmad3, and tubulin were assessed by Western blot. (C) The level of pERK in cells treated with either sodium orthovanadate or TGFβ in the presence of each kinase inhibitor was graphed relative to pERK in cells exposed to kinase inhibitor only. n = 5.

In DCDMLs, as in other cell types, activation of ERK by exogenously added ErbB ligand is short-lived (Figure 4A). This cannot be attributed to ErbB degradation, given that even very high concentrations of HB-EGF (10 nM; 100 ng/ml) cause a ∼25% decrease in ErbB1, ErbB2, and ErbB4 levels after a 4-h incubation (VanSlyke et al., 2023). The duration of ERK activation in other cell types is known to be limited by several negative feedback mechanisms (Lake et al., 2016; Muta et al., 2019). Some of these processes are mediated by effectors that are regulated by physiologically attainable levels of reactive oxygen species (ROS) (Son et al., 2013), and oxidative stress can prolong and/or enhance ERK signaling downstream of TGFβ (Rhyu et al., 2005; Liu et al., 2010). Given that ErbBs are the major source of basal ERK activity in lens cells (Figure 3), and that TGFβ is known to rapidly induce ROS in many cell types (Ohba et al., 1994; Junn et al., 2000; Rao et al., 2004; Rhyu et al., 2005: Yao et al., 2007; Joo et al., 2008; Liu et al. 2010; Michaeloudes et al., 2011; Samarakoon et al., 2013a; Raghavan and Nagaraj, 2016; Shin et al., 2017), we hypothesized that TGFβ may be inactivating or otherwise negating one or more negative feedback pathways in a ROS-dependent manner. This could lead to the relatively slow, ErbB-dependent increase in pERK we have observed. If such a model were correct, then TGFβ upregulation of pERK in DCDMLs should be blocked by an antioxidant. We, therefore, preincubated DCDMLs for 1 h with the commonly used cysteine source N-acetyl cysteine (NAC; Zhitkovich, 2019). Control experiments demonstrated that, as expected, this treatment prevented activation of pERK in response to the classical oxidant stressor hydrogen peroxide (H₂O₂; Figure 9A). We found that NAC pretreatment also prevented TGFβ from increasing the level of pERK (Figure 9A). In contrast, NAC had no effect on the ability of TGFβ to activate Smad3 under the same conditions. A 1-h preincubation with NAC also did not block activation of pERK in response to either FGF (Figure 9A) or HB-EGF (Figure 9B), further demonstrating a specific effect on stimulation of ERK by TGFβ. Similar results were obtained when NAC was replaced with the mechanistically distinct (Ezerina et al., 2018) antioxidant glutathione monoethyl ester (GEE; Supplemental Figure S2A).

FIGURE 9:

TGFβ upregulation of pERK is specifically inhibited by the antioxidant NAC. DCDMLs were preincubated for 1 h with or without 20 mM NAC before treatment with: (A) 4 ng/ml TGFβ, 1 ng/ml FGF, or 250 μM H₂O₂ for 90 min, or (B) 10 ng/ml HB-EGF for 5 min. Whole cell lysates were then analyzed for pERK, tERK, pSmad3, and tubulin. (C) Results were graphed as percent inhibition by NAC of upregulation of pERK or pSmad3 by the indicated treatment. Similar results were obtained with 10 mM NAC.

If EMyT of lens epithelial cells requires activation of ERK downstream of TGFβ-generated ROS, then NAC would be expected to prevent TGFβ from inducing myofibroblast differentiation. We found this to be that case as assessed by Western blot analysis (Figure 10A) and immunofluorescence microscopy (Figure 10B). The antioxidant GEE had a similar effect on EMyT marker expression (Supplemental Figure S2B). NAC did not inhibit the expression of the fiber differentiation marker δ-crystallin, demonstrating the specificity of the effect (Figure 10A). As after short-term treatments (Figure 9A), coculture with NAC for 6 d prevented TGFβ from increasing the levels of pERK (Figure 10A). Reduced glutathione was previously shown to block EMT in mammalian lens cells (Chamberlain et al., 2009).

FIGURE 10:

Oxidative stress is required for upregulation of EMyT by TGFβ and is enhanced by TGFβ. (A) EMyT in response to TGFβ is blocked by the antioxidant NAC. DCDMLs were cultured for 6 d with no additions (-), 20 mM NAC, TGFβ, or both (NAC added 1 h before TGFβ). Whole cell lysates were then analyzed by Western blot for FN, αSMA, δ-crystallin, phospho and total ERK, and tubulin. Results graphed as fold values obtained in cells cultured with NAC plus TGFβ relative to values from cells cultured with TGFβ only. (B) DCDMLs cultured as in (A) were fixed and immunostained to demonstrate the lack of αSMA-containing stress fibers in NAC treated cells. NAC also prevented TGFβ from inducing the expression of the fibrotic marker collagen I as assessed by an antibody specific to procollagen I. (C) Multi-day exposure to TGFβ increases pERK levels in response to oxidative stress. DCDMLs cultured for 6 d in the presence or absence of TGFβ were preincubated with or without 20 mM NAC for 1 h before a 1.5 h incubation in the absence or presence of 250 μM H₂O₂. Cell lysates were analyzed for phospho and total ERK, or tubulin. NAC inhibited the activation of ERK by H₂O₂ by 100% ± 0 in cells cultured in the absence of TGFβ, and by 84% ±10.8 in cells cultured with TGFβ (n = 9; p = 0.00).

In many cell types, TGFβ decreases the expression of multiple antioxidant enzymes (Liu and Desai, 2015). This also appears to be the case in lens epithelial cells. DCDMLs cultured with TGFβ for 2–6 d had ∼8.5-fold (8.56 ± 3.06; n = 8; p = 0.000) higher levels of pERK after exposure to H₂O₂ than no TGFβ control cells, an increase that was blocked by preincubation with NAC (Figure 10C). This effect was not observed if cells were exposed to TGFβ for only 90 min (unpublished data).

ErbB1 and ErbB2 are targets of ERK signaling

We have previously shown that TGFβ increases the expression of total as well as cell surface, ligand-activatable pools of ErbB1, a process that appears to be involved in TGFβ upregulation of EMyT (VanSlyke et al., 2023). Given the importance of ERK and ErbBs for TGFβ-induced EMyT, we considered the possibility that ERK activity could be required for TGFβ to enhance ErbB1 levels. Indeed, we found that inactivating ERK in DCDMLs by addition of UO126 prevented TGFβ from increasing ErbB1 levels (Figure 11A). Decreased expression of ErbB1 was associated with reduced function as assessed by autoactivation of ErbB1 in response to exposure to TGFα ligand (Figure 11A). Notably, UO126 also reduced basal levels of ErbB1 protein and function in the absence of TGFβ (Figure 11A). Culturing DCDMLs with an inhibitor of another class of MAP kinases, namely p38, increased instead of reduced ErbB1 levels (Figure 11B), demonstrating the specificity of this effect for ERK. UO126 also decreased the level of ErbB2, the favored dimerization partner of ErbB1 in the activation of ERK (Graus-Porta et al.,1995; Pinkas-Kramarski et al., 1996; Olayioye et al., 2000; Figure 11C). In contrast, blocking ERK had no significant effect on the level of ErbB4, the other functional ErbB kinase expressed in lens epithelial cells (Figure 11C) that does not appear to play a causative role in in TGFβ-induced EMyT (VanSlyke et al., 2023).

DISCUSSION

DCDMLs are maintained under serum-free conditions to mimic the avascular environment of the lens, allowing us to carry out studies on ERK and its growth factor signaling effectors without the confounding effects of exogenous serum or serum withdrawal. Multiple ERK-stimulating growth factors including PDGF (Potts et al., 1994; Reneker and Overbeek, 1996; Wang et al., 2010), FGF (Le and Musil, 2001; Lovicu and McAvoy, 2001; Robinson, 2006), and ErbB ligands (VanSlyke et al., 2023) are active in lens epithelial cells, likely a prerequisite for their ability to survive and differentiate in the avascular environment of the lens. The absence of ERK- or AKT- activating ErbB ligands in the aqueous humor (Iyengar et al., 2009) implies that ErbB signaling in lens central epithelial cells in vivo is controlled by their endogenously produced ligands. The level of ErbB-dependent pERK we observe in DCDMLs is therefore likely to reflect physiologically relevant conditions. Basal ERK activity downstream of tonic ErbB signaling has previously been shown to be required for TGFβ to upregulate the production of prostaglandin E2 in human bronchial cells (Liu et al., 2007). Our studies extend such a role to a new cell type (lens epithelial cells) and downstream process (EMyT).

Mechanism of ERK activation by TGFβ

In chick (Boswell et al., 2017), rodent (Wojciechowski et al., 2017), and human (Tiwari et al., 2016) primary lens epithelial cells, TGFβ increases the level of pERK within 1–2 h. In the same systems, blocking ERK activity prevents TGFβ from inducing EMyT. A causal relationship between TGFβ upregulation of ERK activity and of EMT has been established in some nonlenticular cell types (Xie et al., 2004; Iea-Flores et al., 2019). Derynck and colleagues (Lee et al., 2007; Muthusamy et al., 2015) have described a pathway by which ligand-stimulated TGFβRs phosphorylate the adaptor protein ShcA, which (via a ShcA/Grb2/SOS complex) directly activates the canonical Ras/Raf/MEK/ERK cascade. This is not the mechanism by which TGFβ increases pERK in lens cells, because: (1) induction of pERK mediated by ShcA is detectable within 5–10 min of TGFβ addition, a much shorter period than is required in lens cells, and (2) the aforementioned ShcA pathway would not be blocked by ErbB inhibitors. A major means by which TGFβ can less directly stimulate ERK is via transactivation of receptor tyrosine kinases upstream of the canonical Ras-to-ERK pathway. A simple explanation for the ability of ErbB inhibitors to block TGFβ-induced ERK activation in lens cells would be if TGFβ transactivated ErbBs (Uchiyama-Tanaka et al., 2002; Wang et al., 2008; Ebi et al., 2010; Samarakoon et al., 2013a).

However, we failed to obtain any evidence that TGFβ increases the level of autophosphorylated, activated ErbBs before or at the time at which it upregulates pERK levels (Figure 7). Addition of exogenous ErbB ligands to DCDMLs stimulates the phosphorylation of AKT within 5 min (Figure 3C). If TGFβ transactivated ErbBs, then we would expect to observe an increase in pAKT levels within 1.5 h of TGFβ addition. Such rapid, concomitant activation of AKT and ERK due to transactivation of ErbB has been reported in other cell types (Eguchi et al., 1999; Chiu et al., 2005). Instead, upregulation of pAKT by TGFβ in DCDMLs is only detectable after a 24-h treatment (Boswell et al., 2017). Our data also indicate that TGFβ does not stimulate ERK by transactivating either FGFRs or PDGFRs, given that TGFβ can still increase the level of pERK in the presence of inhibitors of these receptors (Figure 4). If TGFβ induced the transactivation of some other major class of receptor tyrosine kinase, it might be expected to lead to an increase in a [gt] 80 kD phosphotyrosine-positive species on the plasma membrane. No such band was detectable by cell surface biotinylation (Figure 7A). TGFβ-induced transactivation of a non-ErbB receptor also cannot explain why TGFβ stimulation of ERK is abolished by ErbB inhibitors.

The canonical core ErbB-to-ERK pathway involves multiple kinase-dependent steps (Figure 12). First, ligand activation of ErbBs leads to their lapatinib-inhibitable autophosphorylation on multiple tyrosine residues. These phosphotyrosines form binding sites for various molecules that initiate the assembly of a pY ErbB scaffold/effector complex that ultimately acts to increase the activation of the serine/threonine Raf-1 kinase. Raf-1 then activates MEK. The latter dual-specific kinase in turn phosphorylates ERK on threonine and tyrosine residues, thereby activating its kinase activity. In principle, one mechanism by which TGFβ could enhance pERK levels would be by inhibiting tyrosine phosphatases that directly or indirectly reduce the activity of one or more of these kinases. Several enzymes with such dephosphorylating activity have been reported to be inhibited by ROS (e.g., MKP1/2/3; Cdc25; Xia et al., 1999; Liu et al., 2010; Ostman et al., 2011; Tonks, 2013; Okoh et al., 2015). If this were the case, then TGFβ should phenocopy the ability of the general tyrosine phosphatase inhibitor sodium orthovanadate to increase pERK levels even if an upstream kinase is blocked. We found instead that inhibitors of ErbB (lapatinib), Raf-1 (sorafinib), or MEK (U0126) each completely abolished upregulation of pERK levels in response to TGFβ (Figure 8). Moreover, sodium orthovanadate did not prolong or increase activation of pERK when added immediately after a 5-min treatment with HB-EGF (unpublished data), implying that protein tyrosine phosphatases are not major negative determinants of ErbB-to-ERK signaling strength in DCDMLs. This contention is supported by evidence that lens epithelial cells have low levels of endogenous tyrosine phosphatase activity (Bozulic et al., 2004).

FIGURE 12:

Kinases in the core canonical ErbB-to-ERK pathway, and proposed step at which TGFβ/ROS acts to stimulate ERK in lens epithelial cells in an ErbB-dependent manner.

Another potential means by which TGFβ could increase pERK levels in lens cells would be by positively upregulating the activity of either Raf-1, MEK, or ERK in a ROS-dependent manner. Such a mechanism is, however, also inconsistent with our finding that blocking an upstream kinase prevents TGFβ from enhancing pERK levels (Figure 8). A stimulatory effect of TGFβ on the Raf/MEK/ERK signaling module would also not explain why upregulation of pERK by TGFβ is more dependent on ErbB than on the other lens growth factor receptors (e.g., FGFR; PDGFR) that use the same pathway to activate ERK. We therefore, favor the possibility that TGFβ acts at the level of the pY ErbB scaffold/effector complex. An attractive mechanism would be if TGFβ oxidatively activates Src, which then activates a positive effector (e.g., Gab1) that acts upstream of Raf-1 in the ErbB-to-ERK pathway (Furcht et al., 2015). However, Src is not detectably activated (i.e., phosphorylated on Y416) in DCDMLs in response to a 1.5-h treatment with TGFβ, nor does the Src inhibitor dasatinib reduce the level of pERK in the absence or presence of TGFβ (unpublished data). More than a dozen proteins have been identified to bind directly to tyrosine autophosphorylated ErbB1, which then recruit multiple other molecules to the complex in a cell type-specific manner. Consequently, over a hundred proteins have been reported to become associated with activated ErbB1 based on biochemical evidence, with many more proposed as putative ErbB-interacting proteins (Morandell et al., 2008; Tong et al., 2014). Determining the precise mechanism by which TGFβ upregulates pERK levels downstream of ErbB autoactivation is therefore likely to be challenging, especially in a limited through-put primary cell system.

Regardless of the underlying signaling pathways, our results support a model in which oxidative stress plays a central role in the upregulation of ERK and EMyT by TGFβ in lens epithelial cells (Figures 9, 10; Supplemental Figure S2). Although we favor TGFβ as a key source of such stress (discussed below), even low-level activation of ErbBs is well known to transiently generate ROS (Rao et al., 2004; Koseska and Bastiaens, 2020; Joshi et al., 2023). It has been proposed that isolated lens epithelial cells are constitutively in a state of relative oxidative stress, in part due to the absence of antioxidants provided by lens fiber cells in the intact organ (discussed by Wei et al., 2017). Our data suggest that in DCDMLs, as in many other cell types (Liu and Desai, 2015), TGFβ reduces the expression of antioxidants (Figure 10C). After cataract surgery, the loss of lens fiber cells and the TGFβ-induced decrease in antioxidants in lens epithelial cells could therefore combine with the generation of ROS downstream of TGFβ and ongoing basal ErbB signaling to create an environment particularly conducive to TGFβ-induced, ErbB-dependent activation of ERK and subsequent induction of EMyT.

Comparison with previous results

As previously reviewed (Musil, 2012), growth factor signaling is conserved between DCDMLs and primary mammalian lens epithelial cell systems. Blocking ErbB or ERK activation at the time of initial addition of TGFβ prevents TGFβ from inducing EMyT in both rat central lens epithelial explants (Wojciechowski et al., 2017) and in chick DCDMLs (Boswell et al., 2017). Lovicu and coworkers (Shu et al., 2019) have reported that an 18-h treatment of rat lens central epithelial explants with TGFβ increased the intensity of an anti-pY1068 ErbB1 immunoreactive species. The significance of this finding, which we have failed to reproduce in DCDMLs, is difficult to assess and could be a consequence of the time-dependent increase in ErbB1 transcripts they observed 24 h after addition of TGFβ in this system. We have recently reported (VanSlyke et al., 2023) that a 6-d treatment with TGFβ also does not increase anti-pY1068 ErbB1 immunoreactivity in otherwise untreated DCDMLs. More relevant to our current study is the observation of Shu et al. (2019) that a 15-min or 2-h treatment of rat lens explants with TGFβ failed to increase pY1068 ErbB1 levels to a statistically significant extent. Thus in rat, as in chick, lens epithelial cells (Figure 7B), there is no evidence that TGFβ stimulates autophosphorylation of ErbB1 on a major site of receptor activation at or before the time at which TGFβ increases pERK levels. Our finding that a 5–90 min exposure to TGFβ also fails to increase ErbB immunoreactivity detected with a pan-phosphotyrosine antibody (Figure 7A) rules out the possibility that Y1068 of ErbB1 is somehow insensitive to TGFβ-induced modification. Instead, our results indicate that TGFβ has no short-term effect on the overall activation state of ErbBs in lens cells.

The ability of TGFβ to induce oxidative stress has previously been examined in rodent lens epithelial explants using the ROS probe dihydroethidium (DHE). It was reported that a minimum of 6 h of exposure to TGFβ was required for a detectable increase in DHE signal (Das et al., 2016). Such a delay is unlikely to be attributable to a lack of enzymes capable of generating ROS in untreated cells, given the basal expression of NADPH oxidase subunits in lens epithelium (Rao et al., 2004; Wang and Lou, 2009), including Nox4 protein (Das et al., 2020).

As is true for many assays used to measure ROS, DHE has limitations, including a tendency to autoxidize, rapidly photobleach, and because of the nonstoichiometric yield of ethidium. It has, therefore, been considered to provide a qualitative instead of quantitative readout of superoxide generation (Ortega-Villasante et al., 2018). It may be that under the conditions used by Das et al., (2016), the DHE assay is only sufficiently sensitive to detect relatively large increases in ROS secondary to TGFβ enhancement of Nox4 levels. This possibility is supported by studies using other probes that demonstrate that induction of ROS by TGFβ is detectable within 1–2 h in many other systems (Ohba et al., 1994; Junn et al., 2000; Rhyu et al., 2005; Yao et al., 2007; Joo et al., 2008; Liu et al., 2010; Michaeloudes et al., 2011; Samarakoon et al., 2013a; Shin et al., 2017). This includes human lens epithelial cell lines in which TGFβ has been reported to increase ROS within 2 min (Rao et al., 2004) or 2 h (Raghavan and Nagaraj, 2016) as assessed using 2′,7′-dichlorofluorescein diacetate (DCFDA). Our attempts to use DCFDA in DCDMLs were unsuccessful due to limited retention of the deacetylated carboxydichlorofluorescein (CDCF) indicator dye within the cell, a known limitation of this assay in some systems (Ng and Ooi, 2021). Of note, primary lens epithelial cells have been reported to express functional MRP5 (multidrug resistance-associated protein 5), a major cellular exporter of CDCF (Umapathy et al. 2015; Li et al., 2020). Because of these limitations, we attempted to address the role of Nox activity in TGFβ upregulation of ERK activity using a Nox inhibitor (VAS2870) previously reported to reduce both ROS and EMyT in rodent primary lens epithelial cells (Das et al., 2016). These studies were uninformative because the drug increased pERK levels in the absence of TGFβ within 90 min; genetic knockout of NOX4 has recently been reported to have a similar effect in rodent lens (Das et al., 2020). As previously proposed in human lung fibroblasts (Junn et al., 2000), it is possible that generation of fibrosis-supporting levels of ROS is initiated by TGFβ stimulation of basally expressed ROS-generating enzymes, after which TGFβ upregulation of NOX4 transcription may play an important (but not in all cases essential; Das et al., 2020) role in EMyT progression.

ErbBs as major determinants of basal ERK signaling

Given the plethora of effectors that can stimulate downstream activation of ERK, why should ErbB signaling play such a major role in determining the level of ERK in lens epithelial cells? Partially occupied ErbB can signal and activate unliganded receptors (Koseska and Bastiaens, 2020). Moreover, ErbBs has been proposed to be subject to less negative regulation than FGFRs (Schlessinger, 2004; Raina et al., 2022). These properties make ErbBs particularly powerful activators of ERK, the functional levels of which are then restrained by multiple negative feedback pathways that prevent kinase overstimulation. It has been demonstrated in multiple cell types that terminating ErbB kinase activity with specific ErbB inhibitors results in a very rapid drop in pERK levels (Kleiman et al., 2011; Sparta et al., 2015). This underscores the importance of continuous ErbB activity for the maintenance of ERK function, and explains why exploiting ErbB signaling would be an effective means for TGFβ to stimulate ERK and promote ERK-dependent processes such as EMyT. Notably, we have reported evidence for the predominance of ErbB1/ErbB2 heterodimers in DCDMLs under basal conditions (VanSlyke et al., 2023). This species has been reported to be a more potent inducer of ERK activation than ErbB1 homodimers (Graus-Porta et al., 1995; Pinkas-Kramarski et al., 1996). Our finding (Figure 11) that ERK activity is required for basal levels of both ErbB1 and ErbB2 (but not of ErbB4) creates a reciprocal relationship to maintain ERK activity in which ErbB1 and ErbB2 are required for basal levels of pERK, and ERK activity is in turn required for expression of these receptors. Our results therefore establish a previously unrecognized TGFβ/ErbB/ERK axis in lens cells.

Therapeutic implications

As in other primary lens cell culture systems, the MEK inhibitor UO126 is an effective means to block TGFβ-induced EMyT in DCDMLs (Figure 1). Given the plethora of important roles of ERK in a wide variety of cell types, using such a global inhibitor of ERK signaling in the eye in vivo could, however, have unintended consequences in other ocular tissues (Méndez-Martínez et al., 2019). In this paper, we have shown that ErbBs are the main source of ERK activity in isolated lens epithelial cells. This finding raises the possibility that ErbB inhibitors would be as effective as UO126 in blocking the initiation of fibrotic PCO after cataract surgery without deleteriously reducing ERK activity in other ocular tissues in which ERK levels are driven by non-ErbB pathways. Notably, no ocular toxicity has been reported in humans after systemic administration of lapatinib despite its likely ability to enter the eye based on its permeability to the blood-brain barrier (Huillard et al., 2014).

Outside of the lens, fibrotic disorders are a leading cause of human death (Wynn, 2004). In at least some non-lenticular cell types, ERK, oxidative stress, TGFβ, and/or ErbB1 have been implicated in the development or progression of fibrosis (Leask, 2012; Samarakoon et al., 2013b; Vallath et al., 2014). Interestingly, these same factors have also been linked to cancer, consistent with the concept that tumors are “wounds that do not heal” (Rybinski et al., 2014).

Cooperation between TGFβ and ErbB1 in promoting fibrosis has been established in both kidney (Zhuang and Lui, 2011; Chen et al., 2012) and lung (Zhou et al., 2012; Cheng et al., 2022). Moreover, it has been reported that blocking ERK reduces lung fibrosis induced by overactivation of ErbB1 in both in culture and in vivo (Madala et al., 2012). It is therefore important to investigate whether the TGFβ/ErbB/ERK axis we have discovered in lens cells is also operative in other tissues.

MATERIALS AND METHODS

Materials

Recombinant human TGFβ1, TGFα, HB-EGF, noggin, and bovine FGF2 were from R & D Systems (Minneapolis, MN). The following antibodies were purchased from Cell Signaling Technology, Danvers, MA: antiphospho-p44/42 MAP kinase (#9106), antitotal p44/42 MAPK (#9102), antiphospho-p38 (#9211), antitotal AKT (#9272), and antiphospho (Ser473) AKT (#9275). The following antibodies against the cytoplasmic tail domains were used to detect ErbBs in DCDMLs: for ErbB1, rabbit SC-03 from Santa Cruz Biotechnology (Santa Cruz, CA) and the rat monoclonal 20.3.6 (Hayman et al., 1986; a kind gift of Dr Michael Hayman, Stonybrook University); for ErbB4, C18 (Santa Cruz Biotechnology), and for ErbB2, A0485 from Agilent (Santa Clara, CA). Two antibodies specific for the tyrosine1068 autophosphorylated, activated form of ErbB1 were used, a rabbit polyclonal (#2234) and a mouse monoclonal (#2236), both from Cell Signaling Technology. Other commercial antibodies used in this study: for phosphoSmad3, ab51451 from Abcam (Cambridge, MA); for total-Smad3, ab84177 from Abcam; for total p38, sc-535 from Santa Cruz Biotechnology; for α-tubulin, T5168 from Sigma-Aldrich (St Louis, MO); for αSMA, clone 1A4 from Agilent; for β actin, clone C4 (MilliporeSigma; Burlington, MA); for phosphotyrosine, clone 4G10 (MilliporeSigma); for chick fibronectin, B3/D6 (from D. Fambrough, Johns Hopkins University; Developmental Studies Hybridoma Bank, University of Iowa), and for procollagen I, SP1.D8 (from H. Furthmayr, Stanford University; Developmental Studies Hybridoma Bank). Rabbit antimouse CP49 polyclonal serum (#899 or 900) was a generous gift of Paul FitzGerald, University of California, Davis, as was the rabbit anti-CP115 antiserum (#76). Sheep anti–δ-crystallin antibody was produced in the laboratory of Joram Piatigorsky (National Institutes of Health) and was a gracious gift of Steve Bassnett (Washington University School of Medicine). SB-431542, glutathione monoethyl ester (GEE), and N-acetyl-l- cysteine (NAC) were from Sigma-Aldrich. UO126, PD173074, and SB-203580 were from Millipore Sigma. Lapatinib, erlotinib, and sorafinib were purchased from LC Labs (Woburn, MA), sodium orthovanadate from New England BioLabs (Ipswich, MA), and afatinib and crenolanib from Selleck (Houston, TX).

DCDML cell culture and treatments

DCDML cultures were prepared from E10 chick lenses as previously described (Le and Musil, 1998). During this process, cells exterior to the lens capsule are removed and mature lens fiber cells die, leaving a preparation of purified lens epithelial cells. Cell were plated at subconfluent density (0.9 × 10⁵ cells/well) onto laminin-coated 96-well tissue culture plates, and cultured at 37°C in a 5% CO₂ incubator in the absence of serum in 100–200 μl M199 medium supplemented with penicillin G, streptomycin, and BOTS (2.5 μg/ml bovine serum albumin, 25 μg/ml ovotransferrin, 300 nM selenium). Cells were fed every 2 d with fresh medium. Inhibitors were used at the following final concentrations: 4 μM lapatinib, 5 μM erlotinib, 3 μM SB-431542, 15 μM UO126, 0.1 μM PD173074, 20 μM SB-203580, 0.5 μM afatinib, 0.5 μM crenolanib, 1 ug/ml noggin, and 10 μM sorafenib.

Cell line

Human kidney HEK 293 cells were from the American Type Culture Collection (ATCC; Manassas, VA, USA).

Immunofluorescence microscopy

DCDMLs grown on laminin-coated glass coverslips were fixed in 2% paraformaldehyde in phosphate-buffered saline and processed as previously described (Le and Musil, 1998; 2001). Images were captured using a Leica DM LD photomicrography system and Scion Image 1.60 software.

Cell surface biotinylation

DCDMLs were biotinylated on ice using freshly prepared 0.25 mg/ml EZ-Link Sulfo-NHS-SS-Biotin (Thermoscientific #21331) as previously described (VanSlyke et al., 2023).

Plasmids and transient transfection

One day after plating, DCDML cultures were transfected in M199 medium without BOTS or antibiotics using Lipofectamine 2000 (Invitrogen, Waltham, MA, USA) following the manufacturer’s suggested protocol. Control experiments confirmed that the efficiency of transient transfection of DCDMLs is consistently ∼70% (Boswell et al., 2009). The pro HB-EGF WT plasmid (Nguyen et al., 2000) was a gift from Rosalyn Adam (Addgene plasmid # 11600).

Detection of HB-EGF bioactivity in conditioned medium

Medium was collected from transfected DCDML cells cultured for 2 d with or without TGFβ and immediately centrifuged at 3000 rpm in a microfuge (0.8 rcf) for 5 min to remove any cell debris before incubation with HEK cell recipients at 4°C for 15 min. Medium was snap-frozen and stored at -80°C before use.

Immunoblot analysis

Whole cell lysates were solubilized directly into SDS–PAGE sample buffer and boiled. Equal volumes of total cell lysate were transferred to polyvinylidene fluoride membranes, and the blots probed with primary antibodies. Immunoreactive proteins were detected using secondary antibodies conjugated to either IRDye800 conjugated to either IRDye800 (Rockland Immunochemicals, Pottstown, PA) or Alexa Fluor 680 (Invitrogen, Waltham, MA) and directly quantified using the LI-COR Biosciences Odyssey infrared imaging system (Lincoln, NE) and associated software. The level of each protein was normalized to the level of β actin or α tubulin in the same sample. Data are graphed as the means ± SD obtained in the number of experiments indicated in each figure. Data were analyzed for significance using the two-tailed paired Student’s t test.

Abbreviations used:

αSMA

α−smooth muscle actin

DCDML

dissociated cell-derived monolayer

DHE

dihydroethidium

EMT

epithelial–mesenchymal transition

EMyT

epithelial–myofibroblast transition

ERK

extracellular signal–regulated kinase

FGF

fibroblast growth factor

FN

fibronectin

MEK

mitogen-activated protein kinase kinase

NAC

N-acetyl cysteine

NaV

sodium orthovandate

PCO

posterior capsule opacification

PDGF

platelet-derived growth factor

ROS

reactive oxygen species

TGFβ

transforming growth factor β

TGFβR

transforming growth factor β receptor.