ABSTRACT

Glomerular matrix accumulation is the hallmark of diabetic nephropathy. The metalloprotease ADAM17 mediates high glucose (HG)-induced matrix production by kidney mesangial cells through release of ligands for the epidermal growth factor receptor. Here, we study the mechanism by which HG activates ADAM17. We find that the C-terminus is essential for ADAM17 activation and the profibrotic response to HG. In the C-terminus, Src-mediated Y702 phosphorylation and PI3K–MEK–Erk-mediated T735 phosphorylation are crucial for ADAM17 activation, both are also required for the HG-induced increase in cell surface mature ADAM17. The non-receptor tyrosine kinase FAK is a central mediator of these processes. These data not only support a crucial role for the C-terminus in ADAM17 activation and downstream profibrotic responses to HG, but also highlight FAK as a potential alternative therapeutic target for diabetic nephropathy.

INTRODUCTION

Diabetic nephropathy is a major complication of diabetes and the leading cause of kidney failure in North America. Patients with diabetic nephropathy suffer the highest morbidity and mortality of any kidney failure patient group (Johnson and Spurney, 2015). Prevention and treatment of diabetic nephropathy is thus a major unmet clinical need. Although pathological changes occur in all kidney compartments, glomerular sclerosis initiated by deposition of extracellular matrix (ECM) protein is the hallmark and earliest manifestation. Multifactorial interventions including glucose control at best only delay disease progression (Lewis et al., 1993, 2001). There is thus a major need to better understand its pathogenesis to enable identification of new therapeutic targets.

High glucose (HG) plays a central role in the pathogenesis of diabetic nephropathy by increasing ECM production in glomerular mesangial cells. The prosclerotic cytokine transforming growth factor β1 (TGFβ1) is a major mediator of the HG-induced MC fibrotic response (Li et al., 2003). We and others have shown that the metalloprotease A disintegrin and metalloprotease 17 (ADAM17), through release of ligands for the epidermal growth factor receptor (EGFR), is an important mediator of HG-induced TGFβ1 upregulation and ECM protein production in kidney cells (Ford et al., 2013; Uttarwar et al., 2011). Of the several EGFR ligands that can be cleaved by ADAM17, TGFα, heparin binding (HB)-EGF and amphiregulin have been implicated in the pathogenesis of diabetic kidney disease (Kefaloyianni et al., 2016; Melenhorst et al., 2009; Uttarwar et al., 2011). Cleavage of the renoprotective exopeptidase ACE2 by ADAM17 is also thought to contribute to diabetic kidney tubular injury (Salem et al., 2014). In diabetic mice, a nonselective metalloprotease inhibitor decreased glomerular sclerosis, although effects cannot be specifically attributed to inhibition of ADAM17 (Ford et al., 2013). More specific ADAM17 inhibitors, however, have been associated with adverse effects, preventing their long-term use in the treatment of diabetic nephropathy (Liu and Kurzrock, 2014; Rossello et al., 2016). Toxicity may be due to the ubiquitous expression of the enzyme and its multitude of substrates including proinflammatory cytokines and adhesion molecules in addition to growth factor receptor ligands (Gooz, 2010). Since ADAM17 activation shows stimulus specificity, this provides an opportunity for developing targeted context-dependent ADAM17 inhibition. We thus focused our studies on understanding the key mediators of ADAM17 regulation by HG.

ADAM17 is a transmembrane metalloprotease with an inhibitory N-terminal prodomain that is cleaved by proprotein convertases, primarily furin, in the Golgi. The mature enzyme comprises a metalloprotease domain followed by a disintegrin domain, a cysteine-rich domain important for substrate recognition, an EGF-like domain, a transmembrane domain and a cytoplasmic domain. The transmembrane domain is required for enzyme activation (Maretzky et al., 2011; Rossello et al., 2016).

The role of the cytoplasmic domain (amino acids 695–824) in regulating ADAM17 activation, however, is as yet not fully clear, and appears to depend on the type of stimulus-inducing activity (Gooz, 2010). For example, phorbol esters and several growth factors can induce early ADAM17 activation in the absence of a C-terminus (Doedens et al., 2003; Hall and Blobel, 2012; Le Gall et al., 2010; Maretzky et al., 2011; Mendelson et al., 2010), whereas angiotensin II requires the C-terminus to be present for induction of enzyme activity (Elliott et al., 2013). The C-terminus may allow interaction with other proteins through its SH2 and SH3 binding domains, with phosphorylation at both serine/threonine and tyrosine residues shown to be functionally important in different settings (Gooz, 2010). Whether this is relevant to HG-induced ADAM17 activation is as yet undetermined.

Indeed, little is known of the mechanism by which HG enables activation of ADAM17. We thus began by investigating the relevance of the C-terminus in this setting, and found it to be essential for activation of ADAM17 by HG. We identified two phosphorylation sites in the C-terminus – Src-mediated Y702 phosphorylation and PI3K–MEK–Erk-mediated T735 phosphorylation – to be crucial for the activation and downstream profibrotic responses of ADAM17. HG also induced increases in the amount of mature ADAM17 at the cell surface, which is dependent on furin activity and phosphorylation at both T735 and Y702. We further identified the non-receptor tyrosine kinase FAK as a central upstream regulator of HG-induced ADAM17 activation. These data identify a crucial role for the C-terminus of ADAM17 in HG-induced activation, but also demonstrate regulation of ADAM17 by HG at multiple levels.

RESULTS

The C-terminus is required for ADAM17 activation by HG and downstream profibrotic responses

To determine the role of the C-terminus in HG responses in MC, either WT ADAM17 or ADAM17 with its C-terminus deleted (ΔC) were transfected into mesangial cells, with ADAM17 activity assessed after 1 h of HG (Fig. 1A). The HG-induced ADAM17 activation seen in cells transfected with the empty vector pcDNA was augmented after overexpression of WT ADAM17. This augmentation is similar to the response seen after prolonged HG exposure (24 h), which results in ADAM17 upregulation (Li et al., 2015a). In contrast, no HG response occurred after transfection with ΔC ADAM17. To confirm these results, we next used ADAM17-knockout (KO) mouse embryonic fibroblasts (MEFs). Absence of ADAM17 in these cells was confirmed by immunoblotting (not shown). We first established that HG increases ADAM17 activity in MEF cells (Fig. 1B). In ADAM17 KO cells, baseline values measured by the assay were significantly lower, and no HG response was seen. ADAM17 activity was next tested after transfection of empty vector, WT ADAM17 or ΔC-ADAM17 into KO cells. Construct expression was confirmed by immunoblotting for the HA tag (Fig. 1C). While transfection of either ADAM17 construct increased basal activity, only cells transfected with WT ADAM17 showed a HG response (Fig. 1D). Since PMA is well known to induce ADAM17 activity independently of its C-terminus (Doedens et al., 2003), we next tested responsiveness of WT or ΔC ADAM17 to PMA in KO cells (Fig. 1E). ADAM17 activity was induced to the same degree with both constructs, confirming previous findings. Finally, to determine whether the catalytic activity of ADAM17 was required, we determined the HG responsiveness of a catalytically inactive mutant ADAM17 E406A. As shown in Fig. 1F, ADAM17 KO MEFs transfected with the catalytically inactive mutant (with expression shown in the inset) had neither an increase in basal activity, nor did they show increased activity in response to HG, indicating that both the C-terminus and ADAM17 catalytic activity are required for ADAM17 activation by HG.

Fig. 1.

ADAM17 C-terminus is required for its activation by HG and for the profibrotic response to HG. (A) ADAM17 (WT or ΔC) or empty vector was expressed in mesangial cells. Absence of the C-terminus prevented HG-induced ADAM17 activation (n=6, P<0.001). (B) ADAM17 activity was tested in ADAM17 KO MEFs. As expected, no response to HG was observed (n=6, P<0.001 versus all others; *P<0.001 versus WT control). (C) ADAM17 WT or ΔC, or empty vector, were expressed in ADAM17 KO MEFs. Immunoblotting for HA confirmed construct expression. It should be noted that with overexposure of the film, two bands can also be seen, although that of the pro-form predominates (not shown). (D) ADAM17 activation was induced by HG only in KO MEFs cells expressing WT ADAM17 (n=4, P<0.01 versus all others). (E) ADAM17 WT or ΔC was expressed in KO MEFs. PMA induced ADAM17 activation equally with both constructs (n=2, *P<0.05 versus either control). (F) ADAM17 activation was induced by HG only in KO MEFs cells expressing WT ADAM17, but not the catalytically inactive mutant of ADAM17, E406A. This mutant also did not show basal activity, unlike that seen with WT ADAM17 (n=4, P<0.001 versus all others). The inset shows successful overexpression of ADAM17 E460A. (G,H) ADAM17 KO MEFs were transfected with either WT or ΔC ADAM17. (G) Activation of a TGFβ1 promoter luciferase by HG (48 h) was seen only with WT ADAM17 (n=6, P<0.001 versus all others). (H) Total TGFβ1 in the medium was also increased by HG only in cells expressing WT ADAM17 (n=5, P<0.01 versus all others).

Fig. 1.

ADAM17 C-terminus is required for its activation by HG and for the profibrotic response to HG. (A) ADAM17 (WT or ΔC) or empty vector was expressed in mesangial cells. Absence of the C-terminus prevented HG-induced ADAM17 activation (n=6, P<0.001). (B) ADAM17 activity was tested in ADAM17 KO MEFs. As expected, no response to HG was observed (n=6, P<0.001 versus all others; *P<0.001 versus WT control). (C) ADAM17 WT or ΔC, or empty vector, were expressed in ADAM17 KO MEFs. Immunoblotting for HA confirmed construct expression. It should be noted that with overexposure of the film, two bands can also be seen, although that of the pro-form predominates (not shown). (D) ADAM17 activation was induced by HG only in KO MEFs cells expressing WT ADAM17 (n=4, P<0.01 versus all others). (E) ADAM17 WT or ΔC was expressed in KO MEFs. PMA induced ADAM17 activation equally with both constructs (n=2, *P<0.05 versus either control). (F) ADAM17 activation was induced by HG only in KO MEFs cells expressing WT ADAM17, but not the catalytically inactive mutant of ADAM17, E406A. This mutant also did not show basal activity, unlike that seen with WT ADAM17 (n=4, P<0.001 versus all others). The inset shows successful overexpression of ADAM17 E460A. (G,H) ADAM17 KO MEFs were transfected with either WT or ΔC ADAM17. (G) Activation of a TGFβ1 promoter luciferase by HG (48 h) was seen only with WT ADAM17 (n=6, P<0.001 versus all others). (H) Total TGFβ1 in the medium was also increased by HG only in cells expressing WT ADAM17 (n=5, P<0.01 versus all others).

We previously showed that ADAM17 activation was required for HG-induced upregulation of the profibrotic cytokine TGFβ1 (Uttarwar et al., 2011), a key pathogenic factor in diabetic nephropathy (Li et al., 2003). To confirm the importance of the ADAM17 C-terminus in this response, we used ADAM17 KO MEFs transfected with WT or ΔC ADAM17. Only in cells expressing WT ADAM17 was HG able to increase TGFβ1 promoter luciferase activity (Fig. 1G) and protein secretion into the medium (Fig. 1H).

Src phosphorylation of the C-terminus is required for ADAM17 activation by HG

Src associates with ADAM17 in response to hormonal stimulation (Zhang et al., 2006) and was shown to phosphorylate ADAM17 at Y702 in response to mechanical stress (Niu et al., 2013, 2015). Src is known to mediate ADAM17 activation after long-term HG exposure (24–48 h) (Taniguchi et al., 2013). To determine whether acute activation of ADAM17 is mediated by Src, the effect of its inhibitors on ADAM17 activity after 1 h of HG was assessed. Two structurally distinct inhibitors, SU6656 and PP2, prevented early ADAM17 activation (Fig. 2A). HG increased the interaction between ADAM17 and Src as assessed by coimmunoprecipitation (Fig. 2B). Phosphorylation of ADAM17 at Y702 in response to HG was observed in a similar timeframe (Fig. 2C). This phosphorylation was confirmed to be mediated by Src (Fig. 2D). Antibody specificity was confirmed in ADAM17 KO MEFs transfected with WT or Y702A ADAM17 (Fig. S1).

Fig. 2.

Src phosphorylation of the ADAM17 C-terminus is required for its activation by HG. (A) Src inhibitors SU6656 or PP2 blocked HG (1 h)-induced ADAM17 activation (n=7, P<0.001 versus all others). (B) Mesangial cells were treated with HG for the indicated times. ADAM17 was then immunoprecipitated. Immunoblotting shows HG induces ADAM17–Src association (n=3). (C) HG induces a time-dependent increase in ADAM17 Y702 phosphorylation (pADAM17 Y702) (n=2). (D) This was blocked by the Src inhibitors PP2 and SU6656 after 1 h of HG (n=6, *P<0.05 versus all others). (E) ADAM17 KO MEFs were transfected with WT or ΔC ADAM17, treated with HG for 1 h, then ADAM17 was immunoprecipitated using its HA tag. HG-induced Src/ADAM17 association was seen only in cells expressing WT ADAM17 (n=3, P<0.01 versus all others). Note that the HA blot is the same as that used in Fig. 3I as this originates from the same experiment, with reprobing done to detect FAK and Src. (F) ADAM17 KO MEFs were transfected with empty vector pcDNA, WT ADAM17 or Y702A ADAM17. HG induced ADAM17 activation only in cells expressing WT ADAM17 (n=4; *P<0.05, P<0.01, P<0.001, #P<0.001 versus both pcDNA groups).

Fig. 2.

Src phosphorylation of the ADAM17 C-terminus is required for its activation by HG. (A) Src inhibitors SU6656 or PP2 blocked HG (1 h)-induced ADAM17 activation (n=7, P<0.001 versus all others). (B) Mesangial cells were treated with HG for the indicated times. ADAM17 was then immunoprecipitated. Immunoblotting shows HG induces ADAM17–Src association (n=3). (C) HG induces a time-dependent increase in ADAM17 Y702 phosphorylation (pADAM17 Y702) (n=2). (D) This was blocked by the Src inhibitors PP2 and SU6656 after 1 h of HG (n=6, *P<0.05 versus all others). (E) ADAM17 KO MEFs were transfected with WT or ΔC ADAM17, treated with HG for 1 h, then ADAM17 was immunoprecipitated using its HA tag. HG-induced Src/ADAM17 association was seen only in cells expressing WT ADAM17 (n=3, P<0.01 versus all others). Note that the HA blot is the same as that used in Fig. 3I as this originates from the same experiment, with reprobing done to detect FAK and Src. (F) ADAM17 KO MEFs were transfected with empty vector pcDNA, WT ADAM17 or Y702A ADAM17. HG induced ADAM17 activation only in cells expressing WT ADAM17 (n=4; *P<0.05, P<0.01, P<0.001, #P<0.001 versus both pcDNA groups).

Fig. 3.

FAK is required for ADAM17 activation and Src enables FAK–ADAM17 interaction in response to HG. (A) Mesangial cells were treated for the indicated times with HG. FAK activation, assessed by its phosphorylation on Y397 (pFAK), was increased by HG (n=2). (B) FAK was immunoprecipitated from whole cell lysate after treatment with HG. Immunoblotting shows Src association with FAK increases in response to HG (n=3). (C) ADAM17 activation by HG (1 h) is prevented by the FAK inhibitor PF573228 (n=6, P<0.001 versus others). (D) Association of ADAM17 and Src in response to HG (1 h) was prevented by PF573228, as assessed by coimmunoprecipitation (n=2). (E) ADAM17 phosphorylation at Y702 was also prevented by FAK inhibition with PF573228 (n=2). (F) ADAM17 was immunoprecipitated from whole cell lysate after treatment with HG for the indicated times. Immunoblotting shows FAK association with ADAM17 increases in response to HG (n=4). (G) Inhibition of FAK activity with PF573228 prevents ADAM17/FAK association in response to HG (1 h) (n=4). (H) ADAM17 interaction with FAK in response to HG requires Src activity, since this was prevented by the two Src inhibitors SU6656 and PP2 (n=3). (I) ADAM17 KO MEFs were transfected with WT or ΔC ADAM17, treated with HG for 1 h, then ADAM17 was immunoprecipitated using its HA tag. HG-induced FAK–ADAM17 association was seen only in cells expressing WT ADAM17 (n=3, *P<0.05 versus other groups). The HA blot is the same as that used in Fig. 2E as this originates from the same experiment, with reprobing done to detect FAK. (J) ADAM17 KO MEFs were transfected with WT or Y702A ADAM17, treated with HG for 1 h, then ADAM17 was immunoprecipitated using its HA tag. No HG-induced association between ADAM17 and FAK was seen in cells expressing Y702A (n=3). IgG control immunoprecipitation of HG-treated cells with WT ADAM17 shows some nonspecific pull-down of FAK. HC, antibody heavy chain.

Fig. 3.

FAK is required for ADAM17 activation and Src enables FAK–ADAM17 interaction in response to HG. (A) Mesangial cells were treated for the indicated times with HG. FAK activation, assessed by its phosphorylation on Y397 (pFAK), was increased by HG (n=2). (B) FAK was immunoprecipitated from whole cell lysate after treatment with HG. Immunoblotting shows Src association with FAK increases in response to HG (n=3). (C) ADAM17 activation by HG (1 h) is prevented by the FAK inhibitor PF573228 (n=6, P<0.001 versus others). (D) Association of ADAM17 and Src in response to HG (1 h) was prevented by PF573228, as assessed by coimmunoprecipitation (n=2). (E) ADAM17 phosphorylation at Y702 was also prevented by FAK inhibition with PF573228 (n=2). (F) ADAM17 was immunoprecipitated from whole cell lysate after treatment with HG for the indicated times. Immunoblotting shows FAK association with ADAM17 increases in response to HG (n=4). (G) Inhibition of FAK activity with PF573228 prevents ADAM17/FAK association in response to HG (1 h) (n=4). (H) ADAM17 interaction with FAK in response to HG requires Src activity, since this was prevented by the two Src inhibitors SU6656 and PP2 (n=3). (I) ADAM17 KO MEFs were transfected with WT or ΔC ADAM17, treated with HG for 1 h, then ADAM17 was immunoprecipitated using its HA tag. HG-induced FAK–ADAM17 association was seen only in cells expressing WT ADAM17 (n=3, *P<0.05 versus other groups). The HA blot is the same as that used in Fig. 2E as this originates from the same experiment, with reprobing done to detect FAK. (J) ADAM17 KO MEFs were transfected with WT or Y702A ADAM17, treated with HG for 1 h, then ADAM17 was immunoprecipitated using its HA tag. No HG-induced association between ADAM17 and FAK was seen in cells expressing Y702A (n=3). IgG control immunoprecipitation of HG-treated cells with WT ADAM17 shows some nonspecific pull-down of FAK. HC, antibody heavy chain.

Tyrosine 702 is located in the C-terminus of ADAM17 (aa 695–824). Using ADAM17 KO MEFs transfected with either WT or ΔC ADAM17, a requirement for the C-terminus in the association between ADAM17 and Src was confirmed (Fig. 2E). To determine whether Y702 phosphorylation is required for ADAM17 activation by HG, ADAM17 KO MEFs were transfected with either WT or nonphosphorylatable Y702A ADAM17. Fig. 2F shows that although the baseline activity remains higher in Y702A-transfected cells, consistent with that seen with expression of ΔC ADAM17, HG fails to induce ADAM17 activation. Together, these data show that Src interacts with and phosphorylates ADAM17 at its C-terminal Y702, and this is required for ADAM17 activation by HG.

FAK mediates Src phosphorylation and activation of ADAM17

We next sought to determine the upstream factors mediating Src–ADAM17 activation by HG. Activation of the nonreceptor tyrosine kinase FAK was seen in diabetic kidney lysates (Regoli and Bendayan, 1999), but whether FAK is activated by HG in mesangial cells or regulates ADAM17 activation is as yet unknown. In other settings, activated FAK is known to recruit Src, enabling its activation (Bolos et al., 2010). We first assessed whether HG activates FAK in mesangial cells. Fig. 3A shows that HG induced early activation of FAK, determined by immunoblotting for its autophosphorylation at Y397. HG induced the association of FAK with Src, as assessed by coimmunoprecipitation (Fig. 3B). Using the FAK inhibitor PF573228, Fig. 3C shows that FAK is required for ADAM17 activation by HG. FAK activation by HG is also required for Src association with ADAM17 (Fig. 3D), as well as HG-induced phosphorylation of ADAM17 at Y702 by Src (Fig. 3E). Thus, FAK is required as an upstream effector of Src-mediated ADAM17 phosphorylation and ADAM17 activation.

Activated Src is known to enable the full catalytic activity of FAK through phosphorylation at additional tyrosine residues on FAK. This expands the ability of FAK to interact with other proteins, further enhancing its scaffolding function (Bolos et al., 2010). As shown in Fig. 3F, we found that FAK associated with ADAM17 in response to HG. This was not only dependent on FAK activity, as it was blocked by the inhibitor PF573228 (Fig. 3G), but also required Src activity since it was prevented by two independent Src inhibitors (Fig. 3H). To determine whether the C-terminus of ADAM17 is needed for FAK–ADAM17 interaction, ADAM17 KO MEFs were transfected with either WT or ΔC ADAM17. After HG treatment for 1 h, FAK associated with only WT ADAM17, showing the requirement of the C-terminus of ADAM17 for this interaction (Fig. 3I). Interestingly, expression of Y702A ADAM17 in ADAM17 KO MEFs also prevented the HG-induced association between FAK and ADAM17 (Fig. 3J). Here, IgG control immunoprecipitates of cells transfected with WT ADAM17 showed minor nonspecific pull-down of FAK (Fig. 3J). Taken together, FAK-induced Src activation and subsequent phosphorylation of ADAM17 at Y702 promotes interaction between FAK and ADAM17.

PI3K mediates FAK–ADAM17 association and ADAM17 activation by HG

We previously showed an important role for PI3K–Akt signaling in HG-induced matrix upregulation (Wu et al., 2007, 2009). Interestingly, PI3K was shown to associate with both ADAM17 and Src in cancer cells (Zhang et al., 2006). Furthermore, after its full activation, FAK is able to recruit PI3K by binding to its SH2 domain (Bolos et al., 2010). This enables a conformational change that releases the catalytic from the inhibitory subunit of PI3K, enabling its activation (Cantley, 2002). To assess whether PI3K is required for HG-induced ADAM17 activation, we used two distinct PI3K inhibitors, wortmannin and LY294002. Fig. 4A shows that PI3K inhibition prevented HG-induced ADAM17 activation. As expected, since FAK is known to recruit PI3K, FAK activation (assessed by its Y397 autophosphorylation) was unaffected by PI3K inhibitors (Fig. 4B). Src activation by HG was similarly unaffected by PI3K inhibitors (Fig. 4C). However, PI3K activity was required for FAK–ADAM17 association in response to HG (Fig. 4D). These data suggest that FAK and Src form a complex that associates with ADAM17. PI3K activity is required for this, as part of the complex and/or by facilitating complex formation through its generation of signaling intermediates.

Fig. 4.

PI3K mediates FAK–ADAM17 association and ADAM17 activation by HG. (A) PI3K inhibitors LY294002 or wortmannin inhibited HG (1 h)-induced ADAM17 activation (n=4, P<0.001 versus other groups, P<0.01, #P<0.01 versus wort). (B) FAK activation, as assessed by its Y397 autophosphorylation, was unaffected by PI3K inhibitors (n=3). (C) Similarly, Src activation, assessed by its Y416 phosphorylation, was also unaffected by PI3K inhibitors (n=3). (D) Mesangial cells were treated with both PI3K inhibitors prior to HG for 1 h, after which ADAM17 was immunoprecipitated. FAK–ADAM17 association was prevented by both inhibitors (n=3).

Fig. 4.

PI3K mediates FAK–ADAM17 association and ADAM17 activation by HG. (A) PI3K inhibitors LY294002 or wortmannin inhibited HG (1 h)-induced ADAM17 activation (n=4, P<0.001 versus other groups, P<0.01, #P<0.01 versus wort). (B) FAK activation, as assessed by its Y397 autophosphorylation, was unaffected by PI3K inhibitors (n=3). (C) Similarly, Src activation, assessed by its Y416 phosphorylation, was also unaffected by PI3K inhibitors (n=3). (D) Mesangial cells were treated with both PI3K inhibitors prior to HG for 1 h, after which ADAM17 was immunoprecipitated. FAK–ADAM17 association was prevented by both inhibitors (n=3).

ADAM17 phosphorylation at T735 is required for its activation by HG

PI3K was shown to mediate the serine/threonine phosphorylation of ADAM17 through PDK1 in cancer cells, although the phosphorylated residue was not identified (Zhang et al., 2006). Several serine/threonine residues of the ADAM17 C-terminus are known to be phosphorylated, with phosphorylation at T735 important for ADAM17 activation and/or trafficking and maturation in response to specific stimuli (Diaz-Rodriguez et al., 2002; Soond et al., 2005; Xu and Derynck, 2010). We first determined whether HG induces ADAM17 phosphorylation on T735. Fig. 5A shows phosphorylation at this site within 30 min of HG exposure. The mitogen-activated protein kinases p38 and Erk are known to phosphorylate ADAM17 at T735 in different settings, and PI3K–PDK1 can also regulate Erk activation (Ha et al., 2013). We thus first tested the effects of p38 or Erk inhibition on ADAM17 activation by HG. Fig. 5B,C shows that Erk inhibition using the MEK inhibitor U0126, but not p38 inhibition, prevented HG-induced ADAM17 activation. We next assessed the requirement of PI3K and Erk for T735 phosphorylation. Fig. 5D shows that inhibitors of either PI3K or MEK, the immediate upstream activator of Erk, prevented HG-induced T735 phosphorylation. To confirm that PI3K functions upstream of the MEK–Erk signaling pathway, we determined whether PI3K inhibition could block Erk activation by HG. Fig. 5E shows that both PI3K inhibitors prevented HG-induced Erk activation. Inhibition of FAK, upstream of PI3K, also prevented HG-induced Erk activation. None of the inhibitors affected basal Erk activity (Fig. S2).

Fig. 5.

Phosphorylation of ADAM17 at T735 is required for its activation by HG. (A) HG induces a time-dependent increase in ADAM17 T735 phosphorylation (pADAM17 T735) (n=6; *P<0.05 versus control, P<0.01 versus control). (B) The MEK inhibitor U0126, which inhibits Erk activation, prevented ADAM17 activation by HG (1 h) (n=4, P<0.01 versus others). (C) The p38 inhibitor SB203580 had no effect on ADAM17 activation by HG (n=3, P<0.01 versus control). (D) HG (1 h)-induced ADAM17 T735 phosphorylation was inhibited by the PI3K inhibitors LY294002 (LY) and wortmannin (wort), and by the MEK inhibitors PD98059 (PD) and U0126 (n=3, P<0.01 versus control and LY, and *P<0.05 versus wortmannin, PD98059 and U0126). (E) PI3K inhibitors LY294002 and wortmannin, and the FAK inhibitor PF5732008 (PF) inhibit HG (1 h)-induced Erk activation (n=3, P<0.01 versus other groups). (F) ADAM17 KO MEFs were transfected with empty vector pcDNA, WT ADAM17 or T735A ADAM17. HG induced ADAM17 activation only in cells expressing WT ADAM17 (n=4, *P<0.05, P<0.01, #P<0.001 versus both pcDNA groups). (G) ADAM17 KO MEFs were transfected with WT or T735A ADAM17, treated with HG for 1 h, then ADAM17 immunoprecipitated using its HA tag. No HG-induced association between ADAM17 and FAK was seen in cells expressing T735A (n=3, P<0.01 versus other groups). IgG control immunoprecipitation of HG-treated cells with WT ADAM17 shows some nonspecific pull-down of FAK. LC, antibody light chain.

Fig. 5.

Phosphorylation of ADAM17 at T735 is required for its activation by HG. (A) HG induces a time-dependent increase in ADAM17 T735 phosphorylation (pADAM17 T735) (n=6; *P<0.05 versus control, P<0.01 versus control). (B) The MEK inhibitor U0126, which inhibits Erk activation, prevented ADAM17 activation by HG (1 h) (n=4, P<0.01 versus others). (C) The p38 inhibitor SB203580 had no effect on ADAM17 activation by HG (n=3, P<0.01 versus control). (D) HG (1 h)-induced ADAM17 T735 phosphorylation was inhibited by the PI3K inhibitors LY294002 (LY) and wortmannin (wort), and by the MEK inhibitors PD98059 (PD) and U0126 (n=3, P<0.01 versus control and LY, and *P<0.05 versus wortmannin, PD98059 and U0126). (E) PI3K inhibitors LY294002 and wortmannin, and the FAK inhibitor PF5732008 (PF) inhibit HG (1 h)-induced Erk activation (n=3, P<0.01 versus other groups). (F) ADAM17 KO MEFs were transfected with empty vector pcDNA, WT ADAM17 or T735A ADAM17. HG induced ADAM17 activation only in cells expressing WT ADAM17 (n=4, *P<0.05, P<0.01, #P<0.001 versus both pcDNA groups). (G) ADAM17 KO MEFs were transfected with WT or T735A ADAM17, treated with HG for 1 h, then ADAM17 immunoprecipitated using its HA tag. No HG-induced association between ADAM17 and FAK was seen in cells expressing T735A (n=3, P<0.01 versus other groups). IgG control immunoprecipitation of HG-treated cells with WT ADAM17 shows some nonspecific pull-down of FAK. LC, antibody light chain.

To verify the importance of T735 phosphorylation in ADAM17 activation by HG, ADAM17 KO MEFs were transfected with either WT ADAM17 or the nonphosphorylatable mutant T735A. Fig. 5F shows that compared with empty vector, both WT and T735A ADAM17 demonstrated basal activity, but the HG response was observed only in cells expressing WT ADAM17. Finally, since we had observed that PI3K inhibition prevented ADAM17–FAK association (Fig. 4D), we assessed the role of T735 phosphorylation in this interaction. Using ADAM17 KO MEFs transfected with either WT or T735A ADAM17, Fig. 5G shows that expression of ADAM17 T735A prevented HG-induced association between FAK and ADAM17. These data suggest that PI3K–MEK–Erk induction of T735 phosphorylation is required for HG-induced ADAM17 association with FAK and ADAM17 activation.

HG-induced increase of mature ADAM17 at the cell surface is regulated by T735 and Y702 phosphorylation

Presence of ADAM17 at the cell surface has been correlated in some cases with ligand cleavage (Lorenzen et al., 2016). We first assessed whether HG increases cell surface presence of the mature enzyme using biotinylation and pull down of cell surface proteins. Fig. 6A shows a time-dependent increase in the mature form of ADAM17 (∼100 kDa) in response to HG. PDGFR served as the loading control for cell surface proteins. In contrast, the pro-form of ADAM17 was not seen at the cell surface. The increase in the mature form of ADAM17 at the cell surface suggests that cleavage of its prodomain in response to HG is important for its activation. We thus used two inhibitors of furin, decanoyl-RVKR-CMK and the more specific inhibitor hexa-D-arginine, to assess their effects on ADAM17 maturation and activity in HG. Mesangial cells were first treated with inhibitors to assess their effect on cell surface ADAM17 after 1 h of HG. As seen in Fig. 6B, both inhibitors prevented the increase of mature ADAM17 at the cell surface. We next determined whether this increase in mature ADAM17 was required for activity in response to HG. As seen in Fig. 6C,D, there was a dose-dependent decrease in HG-induced ADAM17 activation with both inhibitors, confirming that the increased level of mature ADAM17 is functionally relevant.

Fig. 6.

HG increases mature ADAM17 at the cell surface through furin-mediated processing. (A) Mesangial cells were treated with HG for the indicated times. Cell surface proteins were biotinylated and immunoprecipitated, and ADAM17 identified by immunoblotting. Two bands are seen in the input (cell lysate), at around 120 kDa and 100 kDa, representing the proform of ADAM17 (pro) at the higher molecular weight, and the mature form at the lower molecular weight. Only the mature form appears at the cell surface, which is increased by HG (n=4). PDGFR serves as the loading control for cell surface proteins. (B) Mesangial cells were treated with the furin inhibitors hexa-D-arginine (hexa) or decanoyl-RVKR-CMK (CMK) prior to HG for 1 h. Cell surface proteins were biotinylated and pulled down, and ADAM17 assessed by immunoblotting. Both inhibitors decrease the HG-induced increase in mature ADAM17 at the cell surface, with a concomitant increase in the pro-form (n=4; P<0.01, *P<0.05). (C,D) Mesangial cells were incubated with increasing concentrations of either hexa-D-arginine or decanoyl-RVKR-CMK. There was a dose-dependent decrease in ADAM17 activation by HG (1 h) with both inhibitors (for both, n=4; P<0.01, P<0.001).

Fig. 6.

HG increases mature ADAM17 at the cell surface through furin-mediated processing. (A) Mesangial cells were treated with HG for the indicated times. Cell surface proteins were biotinylated and immunoprecipitated, and ADAM17 identified by immunoblotting. Two bands are seen in the input (cell lysate), at around 120 kDa and 100 kDa, representing the proform of ADAM17 (pro) at the higher molecular weight, and the mature form at the lower molecular weight. Only the mature form appears at the cell surface, which is increased by HG (n=4). PDGFR serves as the loading control for cell surface proteins. (B) Mesangial cells were treated with the furin inhibitors hexa-D-arginine (hexa) or decanoyl-RVKR-CMK (CMK) prior to HG for 1 h. Cell surface proteins were biotinylated and pulled down, and ADAM17 assessed by immunoblotting. Both inhibitors decrease the HG-induced increase in mature ADAM17 at the cell surface, with a concomitant increase in the pro-form (n=4; P<0.01, *P<0.05). (C,D) Mesangial cells were incubated with increasing concentrations of either hexa-D-arginine or decanoyl-RVKR-CMK. There was a dose-dependent decrease in ADAM17 activation by HG (1 h) with both inhibitors (for both, n=4; P<0.01, P<0.001).

Since we showed that the C-terminus is required for ADAM17 activation by HG, we next determined whether it was necessary for the increased cell surface levels. ADAM17 KO MEFs were transfected with either WT ADAM17 or ΔC ADAM17, treated with HG for 1 h, cell surface proteins biotinylated and pulled down and transfected ADAM17 identified by its HA tag. As seen in Fig. 7A, HG increased cell surface presence of only WT ADAM17. Both Src and Erk have been shown to regulate inducible trafficking of ADAM17 to the cell surface (Soond et al., 2005; Taniguchi et al., 2013). Having identified phosphorylation of Y702 by Src and T735 by Erk as important to ADAM17 activation by HG, we next determined if phosphorylation at these sites regulated the cell surface increase of ADAM17. We used ADAM17 KO MEFs expressing WT, T735A or Y702A ADAM17 to assess their effect on cell surface localization in response to HG. As seen in Fig. 7A, only WT ADAM17 trafficked to the cell surface, while this was completely prevented in the absence of T735 or Y702 phosphorylation.

Fig. 7.

HG-induced increase of mature ADAM17 at the cell surface is differentially regulated by T735 and Y702 phosphorylation. (A) ADAM17 KO MEFs were transfected with WT ADAM17, T735A ADAM17 or Y702A ADAM17. After HG for 1 h, cell surface proteins were isolated and constructs identified by their HA tag. HG-induced ADAM17 cell surface localization required both T735 and Y702 phosphorylation (n=4, P<0.001 versus its control). (B) PI3K inhibitors LY294002 (LY) and wortmannin (wort) prevented cell surface localization of mature ADAM17 (n=6, P<0.01 versus other groups). (C) Src inhibitors SU6656 and PP2 also prevented cell surface localization of mature ADAM17 (n=4; P<0.01 versus control, *P<0.05 versus HG).

Fig. 7.

HG-induced increase of mature ADAM17 at the cell surface is differentially regulated by T735 and Y702 phosphorylation. (A) ADAM17 KO MEFs were transfected with WT ADAM17, T735A ADAM17 or Y702A ADAM17. After HG for 1 h, cell surface proteins were isolated and constructs identified by their HA tag. HG-induced ADAM17 cell surface localization required both T735 and Y702 phosphorylation (n=4, P<0.001 versus its control). (B) PI3K inhibitors LY294002 (LY) and wortmannin (wort) prevented cell surface localization of mature ADAM17 (n=6, P<0.01 versus other groups). (C) Src inhibitors SU6656 and PP2 also prevented cell surface localization of mature ADAM17 (n=4; P<0.01 versus control, *P<0.05 versus HG).

Since ADAM17 T735 phosphorylation is mediated by PI3K–Erk (Fig. 5), we next tested the effect of PI3K inhibition on the increase in cell surface mature ADAM17 seen with HG. Fig. 7B shows that both PI3K inhibitors LY294002 and wortmannin effectively prevented this. Src inhibitors SU6656 and PP2 also prevented the cell surface increase of mature ADAM17 in response to HG (Fig. 7C). These data suggest that both PI3K and Src mediate mature ADAM17 cell surface localization in response to HG, with their downstream phosphorylation of ADAM17 at T735 and Y702 respectively being required. However, the precise mechanism underlying this increase is as yet unknown.

Phosphorylation at T735 and Y702 are both required for ADAM17-mediated TGFβ1 upregulation in response to HG

Since we identified the necessity of both T735 and Y702 phosphorylation for ADAM17 activation by HG (Fig. 2F, Fig. 4F), and we previously showed that ADAM17 is required for HG-induced TGFβ1 upregulation (Uttarwar et al., 2011), we tested the effects of these two mutants on TGFβ1 production. Either WT ADAM17 or one of the mutants was transfected into ADAM17 KO MEFs, and TGFβ1 promoter upregulation and secretion into the medium were assessed. Fig. 8 shows that phosphorylation at both sites is required for this profibrotic response.

Fig. 8.

Phosphorylation at T735 and Y702 are both required for ADAM17-mediated TGFβ1 upregulation in response to HG. ADAM17 KO MEFs were transfected with WT ADAM17 or T735A ADAM17 (A,C) and Y702A ADAM17 (B,D). (A,B) Activation of a TGFβ1 promoter luciferase by HG (24 h) was seen only with WT ADAM17 (A, n=6; P<0.001 versus other groups; B, n=6, *P<0.05 versus WT controls). (C,D) Total TGFβ1 secretion into the medium, as assessed by ELISA, was increased only by WT ADAM17 (C, n=6; P<0.01 versus other groups; D, n=6, P<0.001 versus other groups).

Fig. 8.

Phosphorylation at T735 and Y702 are both required for ADAM17-mediated TGFβ1 upregulation in response to HG. ADAM17 KO MEFs were transfected with WT ADAM17 or T735A ADAM17 (A,C) and Y702A ADAM17 (B,D). (A,B) Activation of a TGFβ1 promoter luciferase by HG (24 h) was seen only with WT ADAM17 (A, n=6; P<0.001 versus other groups; B, n=6, *P<0.05 versus WT controls). (C,D) Total TGFβ1 secretion into the medium, as assessed by ELISA, was increased only by WT ADAM17 (C, n=6; P<0.01 versus other groups; D, n=6, P<0.001 versus other groups).

DISCUSSION

ADAM17 is becoming increasingly recognized as important to the pathogenesis of diabetic nephropathy. How it is activated by HG, however, has not been identified. We propose the following model for ADAM17 activation by HG in kidney mesangial cells: FAK is an upstream key mediator of ADAM17 activation through its recruitment of both Src and PI3K, with subsequent phosphorylation of ADAM17 at two sites in its C-terminus. Y702 is phosphorylated by Src and T735 by PI3K–MEK-activated Erk. Phosphorylation at both sites enhances association of ADAM17 with FAK and is required for downstream profibrotic effects. HG also leads to increased furin-mediated processing of ADAM17 to its mature form and increased translocation of mature ADAM17 to the membrane. These studies underscore a crucial role for the C-terminus of ADAM17 in its ability to interact with upstream signaling mediators, and highlight the central role of FAK in ADAM17 activation by HG. Inhibition of ADAM17 activation through targeting HG-specific activators such as FAK, or their interaction with ADAM17, may offer an alternative novel approach to the treatment of diabetic nephropathy.

The importance of the cytoplasmic domain of ADAM17 to its activation has been controversial. It is well established that this region is not required for constitutive activity (Le Gall et al., 2010), and our data confirm this since there was no difference between the basal activity of WT and ΔC ADAM17. We further confirmed that phorbol esters do not require the C-terminus for induction of ADAM17 activity. However, the requirement for the C-terminus in ADAM17 activation appears to be stimulus specific. For example, while phorbol esters, IL-1, thrombin, EGF, LPA, TNFα, FGF and PDGF can activate ADAM17 independently of its C-terminus (Doedens et al., 2003; Hall and Blobel, 2012; Le Gall et al., 2010; Maretzky et al., 2011; Mendelson et al., 2010), activation by angiotensin II requires the presence of the C-terminus (Elliott et al., 2013). The cytoplasmic domain of ADAM17 is also required for shedding of TNFα (Schwarz et al., 2013), although interestingly, it has been suggested that only the six most membrane-proximal amino acids of the cytoplasmic tail are required (Schwarz et al., 2013). We now show that HG-induced ADAM17 activation is also dependent on an intact C-terminus, and confirm similar dependence on the C-terminus of HG-induced shedding of the membrane-anchored substrate HB-EGF (Fig. S3). This is likely to be due to a dependence on interaction of this region with upstream signaling mediators, namely a FAK–PI3K–Src complex. Indeed, this region of ADAM17 has both a phosphotyrosine site which can bind to SH2 domains, as well as a proline-rich region which can bind to SH3 domains, enabling interaction with various signaling molecules (Arribas and Esselens, 2009; Kleino et al., 2015). Thus, Src and PI3K can directly interact with the ADAM17 C-terminus through their SH2/SH3 and SH2 domains, respectively. While FAK does not possess either of these domains, it is able to bind both Src and PI3K. Thus, the interaction between FAK and ADAM17 seen in our coimmunoprecipitation studies is likely to be indirect. Interestingly, however, phosphorylation of ADAM17 at both its Y702 and T735 appear to be required for mediating interaction with FAK. The precise nature of the molecular interactions in this complex remain to be more fully defined.

Several phosphorylation sites have been identified in ADAM17, including T735, S791, S819 and Y702. Phosphorylation at T735 is the most studied and is induced by either Erk or p38, depending on the stimulus (Diaz-Rodriguez et al., 2002; Rousseau et al., 2008; Xu and Derynck, 2010). Here, we showed that phosphorylation on T735 is essential for ADAM17 activation by HG, and that Erk, but not p38, was required for this. How phosphorylation at this site enables ADAM17 activation is, as yet, not fully understood. Interestingly, however, T735 phosphorylation has recently been shown to prevent the dimerization of ADAM17, releasing it from inhibition by TIMP3 and thereby allowing activation (Xu et al., 2012). It has also been shown to regulate ADAM17 maturation and trafficking to the cell surface in COS-7 cells (Soond et al., 2005; Xu and Derynck, 2010), potentially increasing the pool of available mature ADAM17 for activation. Our data support a role for T735 phosphorylation in increasing membrane-localized mature ADAM17 in response to HG. Interestingly, phosphorylation at this site is also important in allowing interaction with FAK, suggesting multiple roles in enabling ADAM17 activation.

Although Src is known to contribute to ADAM17 activation in various settings, its phosphorylation of ADAM17 at Y702 has been appreciated more recently. Mechanical stress in myoblasts and cardiomyocytes induces Src-mediated ADAM17 phosphorylation at Y702, mediating myogenesis and TNFα shedding, respectively (Niu et al., 2013, 2015). Y702 phosphorylation is also needed for angiotensin-II-mediated hypertrophy in vascular smooth muscle cells (Elliott et al., 2013). We now show that HG induces Src-mediated ADAM17 Y702 phosphorylation. While not affecting basal ADAM17 activity, phosphorylation at this site is essential to HG-induced ADAM17 activation and downstream matrix upregulation. Interestingly, Y702 phosphorylation is also required for HG-induced association between ADAM17 and FAK, likely through intermediates including Src and/or PI3K. Like T735 phosphorylation, Y702 phosphorylation is also required for increased membrane localization of ADAM17 in response to HG. These data are consistent with other studies in which Src was found to be required for ADAM17 translocation to the membrane after prolonged (48 h) exposure to HG in mesangial cells (Taniguchi et al., 2013). We confirmed a requirement for Src also in early (1 h) increases in the level of cell surface mature ADAM17 in response to HG. These data suggest an important role for Src-induced Y702 phosphorylation in mature ADAM17 membrane translocation and activation. The precise mechanism by which this occurs has yet to be determined.

Our data show a novel key role for the focal adhesion protein FAK in ADAM17 activation. FAK is a non-receptor tyrosine kinase known to link integrin stimulation to intracellular signals. Upon integrin engagement, FAK is activated by autophosphorylation on Y397 (pY397), enabling its interaction with, and activation of, Src. Phosphorylation at additional sites by Src leads to the full catalytic activity of FAK and enables interaction with additional kinases, including PI3K (Bolos et al., 2010). FAK thus appears to provide a scaffold for the assembly of a Src–PI3K–ADAM17 complex, as discussed above. Of relevance, ADAM17 has also been shown to associate with integrin α5β1, but whether this interaction is direct or mediated by FAK is unknown. Functionally, this integrin either inhibits ADAM17 activation in unstimulated cells (Gooz et al., 2012; Saha et al., 2010), improves cell adhesion (Bax et al., 2004) or inhibits cell migration (Gooz, 2010). Integrin β1 was shown to be activated by HG to regulate ECM assembly in mesangial cells with longer term HG exposure (Miller et al., 2014). Whether it is also required for early FAK-mediated activation of ADAM17 will be addressed in future studies.

ADAM17 is produced as a proenzyme synthesized in the endoplasmic reticulum. This intracellular pool can undergo rapid processing to the mature form, increasing the availability of mature enzyme at the membrane (Lorenzen et al., 2016). While some stimuli do not increase the availability of membrane mature ADAM17 (Le Gall et al., 2010; Lorenzen et al., 2016), our studies show its induction in response to HG in a short time frame (1 h) which was required for its activation. Proprotein convertases, primarily furin, cleave the N-terminal prodomain of ADAM17 in the Golgi prior to transport to the cell surface (Srour et al., 2003). Our data show that HG increases furin-mediated cleavage of ADAM17, and this is an important step in the increase in levels of the mature form at the cell surface and in HG-induced increases in ADAM17 activity. Recently, the endoplasmic reticulum-resident proteins iRhoms1 and 2 were shown to be important chaperones for escort of ADAM17 both to the Golgi for processing and to the cell membrane (Adrain et al., 2012; Christova et al., 2013; Li et al., 2015b; McIlwain et al., 2012). PACS-2 is another newly described regulator of ADAM17 trafficking to the cell membrane (Dombernowsky et al., 2015). Whether HG regulates iRhom and/or PACS2 function is as yet unknown.

An important role for phosphatidylserine exposure at the outer leaflet of the cell membrane is now proposed as a key common event in ADAM17 activation by diverse stimuli (Sommer et al., 2016). Phosphatidylserine binds the membrane-proximal domain of ADAM17, facilitating ADAM17 membrane binding through its short juxtamembrane segment CANDIS. This brings the catalytic site in close proximity to the cleavage site of membrane-bound substrates, enabling substrate cleavage. Whether this process also occurs in response to HG requires confirmation.

In an effort to prevent adverse effects seen with compounds which more broadly inhibit metalloproteases, ongoing efforts are underway for the development of inhibitors with increased specificity for ADAM17. Yet even ADAM17-specific inhibitors are associated with unwanted adverse effects, probably because of the multiple substrates and ubiquitous nature of the enzyme (Rossello et al., 2016). An alternative approach to ADAM17 inhibition is the identification of stimulus-specific regulation of its activation. Our data provide novel insight into the mechanism of HG-induced ADAM17 activation, showing the importance of C-terminus interaction with a FAK–PI3K–Src complex and phosphorylation of ADAM17 at both Y702 and T735 in regulating ADAM17 activation through complementary mechanisms. Thus, targeting one or a combination of these upstream regulators can be explored as novel approaches to the treatment of diabetic nephropathy. Whether this may be more broadly extended to include other complications of diabetes bears investigation.

MATERIALS AND METHODS

Cell culture

Primary rat mesangial cells (passages 6–15) were isolated from Sprague–Dawley rats. They were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 20% fetal calf serum (Invitrogen), streptomycin (100 μg/ml) and penicillin (100 units/ml) at 37°C in 95% air, 5% CO2. These are periodically checked for mycoplasma contamination. SV40-immortalized MEFs isolated from ADAM17 KO and WT mice were generously provided by Carl Blobel (Cornell University, NY, USA), and their expression of ADAM17 was confirmed by immunoblotting. They were cultured in DMEM supplemented with 10% fetal calf serum. Medium for both cell types contained 5.6 mM glucose. Glucose at 24.4 mM (final concentration 30 mM, Sigma) was added for high glucose (HG) conditions. Cells were made quiescent by serum deprivation for 24 h prior to treatment.

Pharmacological inhibitors were added prior to glucose as follows: LY294002 (10 μM, 30 min, Sigma), wortmannin (100 nM, 1 h, Sigma), PF573228 (1 μM, 1 h, Tocris), SU6656 (10 μM, 30 min, Calbiochem), PP2 (10 μM, 30 min, Calbiochem), PD98059 (10 μM, 30 min, Alexis), U0126 (10 μM, 30 min, Promega), SB203580 (50 μM, 30 min, Sigma), PMA (500 ng/ml, 1.5 h, Sigma), decanoyl-RVKR-CMK (100 μM, 24 h, Tocris) and hexa-D-arginine (100μM, 24 h, Tocris).

Transfection and constructs

Constructs for WT ADAM17 or ADAM17 with the C-terminus deleted (ΔC) and containing a 3′ HA tag were kindly provided by Carl Blobel (Cornell University), with both originally provided by Gillian Murphy, University of Cambridge, UK). A 3′ HA tag was added to the WT ADAM17 construct. ADAM17 T735A (3′ HA) and ADAM17 E406A (5′ HA) were also provided by Carl Blobel, and ADAM17 Y702A (5′ HA) was provided by Yi-Ping Li (University of Texas, TX, USA). All constructs were based on mouse ADAM17.

ADAM17 activity assay

Where indicated, ADAM17 KO MEFs or mesangial cells were transfected with 7.5 μg of ADAM17 construct in 6-well plates, coated with poly-D-lysine for MEFs, at 40% confluence using X-fect (Clontech). Cells were serum-deprived for 24 h, then treated with HG for 1 h. Protein was extracted in activity assay buffer (50 mM Tris-HCl, pH 7.4, 25 mM NaCl, 4%glycerol, 10 mM ZnCl2). ADAM17 activity was measured in duplicate for each sample using 20 μg protein and the TACE Substrate IV (Calbiochem). Cleavage of this substrate was measured in a fluorometer at 420 nm.

HB-EGF cleavage was assessed in cells transfected with pRC/CMV-HBEGF-AP, kindly provided by Michael Freeman (Harvard University, Cambridge, MA, USA). ADAM17 KO MEFs were transfected with 3.5 μg of this vector in conjunction with 3.5 μg ADAM17 construct as above. After serum deprivation, they were incubated in 1 ml of 0% FBS medium to assess baseline shedding. This was replaced with 1 ml of new medium with or without HG which was collected after 1 h. Shedding was assessed using an alkaline phosphatase activity (ALP) assay (AnaSpec). Shedding induced by treatment was assessed by comparing to baseline shedding for each well.

Luciferase assay

ADAM17 KO MEFs plated to 30% confluence were transfected with 2.5 μg in 12-well plates with the ADAM17 construct as indicated using X-fect (Clontech), followed the next day by transfection with 0.5 μg of TGFβ1 promoter-luciferase construct (kindly provided by Naoya Kato, University of Tokyo, Japan) and 0.05 μg pCMV-β-galactosidase (β-gal) (Clontech) using Lipofectamine (Qiagen). Cells were serum-deprived for 24 h after transfection, then exposed to HG for 48 h. Lysis was achieved with Reporter Lysis Buffer (Promega) using one freeze–thaw cycle, and luciferase and β-gal activities measured on clarified lysate using specific kits (Promega) with a Berthold luminometer and a plate reader (420 nm), respectively. β-gal activity was used to adjust for transfection efficiency.

TGFβ1 ELISA

ADAM17 KO MEFs were transfected with 2.5 μg ADAM17 construct as above in 12-well plates. After HG (48 h), medium was harvested and debris removed by centrifugation at 4°C, 4000 rpm. Medium was stored at −80°C until processing. Total TGFβ1 was measured after acid activation of samples according to the manufacturer's instructions (R&D Systems).

Protein extraction and analysis

Cells were lysed and protein extracted as we described previously (Krepinsky et al., 2003), with the addition of BB-94 (1 μM, Tocris) to the lysis buffer. Cell lysates were centrifuged at 4°C, 14,000 rpm for 10 min to pellet cell debris. Supernatant (50 g) was separated on SDS-PAGE and western blotting was performed. Antibodies were goat ADAM17 (1:1000, Santa Cruz, sc-6416), rabbit pADAM17 Y702, kindly provided by Yi-Ping Li (University of Texas) (Niu et al., 2013) (1:1000), rabbit pADAM17 T735 (1:1000, Sigma, SAB4504073), mouse HA (1:1000, Abcam, Ab18181), rabbit Src (1:1000, Cell Signaling, 2108), rabbit pSrc Y416 (1:1000, Cell Signaling, 2101), rabbit FAK (1:1000, Santa Cruz, sc-558), rabbit pFAK Y397 (1:1000, Upstate, 07-012), mouse pErk T202/Y204 (1:1000, Cell Signaling, 9106), total Erk (1:1000, Cell Signaling, 9102) and mouse tubulin (1:10,000, Sigma, T6074).

For immunoprecipitation experiments, cells were lysed, clarified and equal amounts of lysate incubated overnight with 2 μg primary antibody rotating at 4°C, followed by 25 μl of protein-G–agarose slurry for 1.5 h at 4°C. Immunoprecipitates were extensively washed, resuspended in 2× sample buffer, boiled and analyzed by immunoblotting.

Biotinylation for isolation of cell surface proteins

Cells plated in 100 mm plates were washed with ice-cold PBS and incubated with EZ-link Sulfo-NHSLC-Biotin (0.5 mg/ml in PBS, Fisher) for 20 min. Biotinylation was stopped with 0.1 M glycine in PBS. Cells were lysed in IP lysis buffer (PBS pH 7.4, 5 mM EDTA, 5 mM EGTA, 10 mM sodium pyrophosphate, 50 mM NaF, 1 mM NaVO3, 1% Triton X-100, 1 μM BB-94, protease inhibitors). Biotinylated proteins were precipitated with 50% neutravidin slurry (Fisher) overnight, after which the beads were washed, boiled in PSB and proteins assessed by immunoblotting. When used, ADAM17 KO MEFs were transfected with 30 μg of the indicated ADAM17 plasmid as above prior to treatment.

Statistical analyses

Statistical analyses were performed with SPSS 22 (IBM) for Windows using one-way ANOVA, with Tukey's HSD for post hoc analysis. For experiments with two conditions being analysed, a t-test was used and a test for equal variances used to interpret significance. A P-value <0.05 (two-tailed) was considered significant. Data are presented as the mean±s.e.m. The number of experimental repetitions (n) is indicated.

Acknowledgements

J.C.K. gratefully acknowledges the support of St Joseph's Healthcare for nephrology research. We thank Naoya Kato (University of Tokyo, Japan) for providing the TGFβ1 promoter luciferase construct, Carl Blobel (Cornell University) for providing the ADAM17 KO MEFs, ADAM17 T735A and ADAM17 E406A; Carl Blobel and Gillian Murphy (University of Cambridge, UK) for providing WT ADAM17 and ΔC ADAM17; Dr Yi-Ping Li (University of Texas) for providing ADAM17 Y702A; and Michael Freeman (Harvard University) for providing pRC/CMV-HBEGF-AP.

Footnotes

Author contributions

Conceptualization: J.C.K.; Methodology: R.L.; Formal analysis: J.C.K.; Data curation: R.L., T.W., K.W., B.G.; Writing - original draft: J.C.K.; Writing - review & editing: J.C.K.; Supervision: J.C.K.; Project administration: J.C.K.

Funding

This work was supported by the Institute of Nutrition, Metabolism and Diabetes of the Canadian Institutes of Health Research (CIHR) (MOP119308 to J.C.K.).

References

Adrain
,
C.
,
Zettl
,
M.
,
Christova
,
Y.
,
Taylor
,
N.
and
Freeman
,
M.
(
2012
).
Tumor necrosis factor signaling requires iRhom2 to promote trafficking and activation of TACE
.
Science
335
,
225
-
228
.
Arribas
,
J.
and
Esselens
,
C.
(
2009
).
ADAM17 as a therapeutic target in multiple diseases
.
Curr. Pharm. Des
15
,
2319
-
2335
.
Bax
,
D. V.
,
Messent
,
A. J.
,
Tart
,
J.
,
van Hoang
,
M.
,
Kott
,
J.
,
Maciewicz
,
R. A.
and
Humphries
,
M. J.
(
2004
).
Integrin alpha5beta1 and ADAM-17 interact in vitro and co-localize in migrating HeLa cells
.
J. Biol. Chem.
279
,
22377
-
22386
.
Bolos
,
V.
,
Gasent
,
J. M.
,
Lopez-Tarruella
,
S.
and
Grande
,
E.
(
2010
).
The dual kinase complex FAK-Src as a promising therapeutic target in cancer
.
Onco. Targets. Ther.
3
,
83
-
97
.
Cantley
,
L. C.
(
2002
).
The phosphoinositide 3-kinase pathway
.
Science
296
,
1655
-
1657
.
Christova
,
Y.
,
Adrain
,
C.
,
Bambrough
,
P.
,
Ibrahim
,
A.
and
Freeman
,
M.
(
2013
).
Mammalian iRhoms have distinct physiological functions including an essential role in TACE regulation
.
EMBO Rep.
14
,
884
-
890
.
Diaz-Rodriguez
,
E.
,
Montero
,
J. C.
,
Esparis-Ogando
,
A.
,
Yuste
,
L.
and
Pandiella
,
A.
(
2002
).
Extracellular signal-regulated kinase phosphorylates tumor necrosis factor alpha-converting enzyme at threonine 735: a potential role in regulated shedding
.
Mol. Biol. Cell
13
,
2031
-
2044
.
Doedens
,
J. R.
,
Mahimkar
,
R. M.
and
Black
,
R. A.
(
2003
).
TACE/ADAM-17 enzymatic activity is increased in response to cellular stimulation
.
Biochem. Biophys. Res. Commun.
308
,
331
-
338
.
Dombernowsky
,
S. L.
,
Samsøe-Petersen
,
J.
,
Petersen
,
C. H.
,
Instrell
,
R.
,
Hedegaard
,
A. M. B.
,
Thomas
,
L.
,
Atkins
,
K. M.
,
Auclair
,
S.
,
Albrechtsen
,
R.
,
Mygind
,
K. J.
, et al. 
(
2015
).
The sorting protein PACS-2 promotes ErbB signalling by regulating recycling of the metalloproteinase ADAM17
.
Nat. Commun.
6
,
7518
.
Elliott
,
K. J.
,
Bourne
,
A. M.
,
Takayanagi
,
T.
,
Takaguri
,
A.
,
Kobayashi
,
T.
,
Eguchi
,
K.
and
Eguchi
,
S.
(
2013
).
ADAM17 silencing by adenovirus encoding miRNA-embedded siRNA revealed essential signal transduction by angiotensin II in vascular smooth muscle cells
.
J. Mol. Cell Cardiol.
62C
,
1
-
7
.
Ford
,
B. M.
,
Eid
,
A. A.
,
Göőz
,
M.
,
Barnes
,
J. L.
,
Gorin
,
Y. C.
and
Abboud
,
H. E.
(
2013
).
ADAM17 mediates Nox4 expression and NADPH oxidase activity in the kidney cortex of OVE26 mice
.
Am. J. Physiol. Renal Physiol.
305
,
F323
-
F332
.
Gooz
,
M.
(
2010
).
ADAM-17: the enzyme that does it all
.
Crit Rev. Biochem. Mol. Biol.
45
,
146
-
169
.
Gooz
,
P.
,
Dang
,
Y.
,
Higashiyama
,
S.
,
Twal
,
W. O.
,
Haycraft
,
C. J.
and
Gooz
,
M.
(
2012
).
A disintegrin and metalloenzyme (ADAM) 17 activation is regulated by alpha5beta1 integrin in kidney mesangial cells
.
PLoS One
7
,
e33350
.
Ha
,
J. S.
,
Kwon
,
K. S.
and
Park
,
S. S.
(
2013
).
PI3Kgamma contributes to MEK1/2 activation in oxidative glutamate toxicity via PDK1
.
J. Neurochem.
127
,
139
-
148
.
Hall
,
K. C.
and
Blobel
,
C. P.
(
2012
).
Interleukin-1 stimulates ADAM17 through a mechanism independent of its cytoplasmic domain or phosphorylation at threonine 735
.
PLoS One
7
,
e31600
.
Johnson
,
S. A.
and
Spurney
,
R. F.
(
2015
).
Twenty years after ACEIs and ARBs: emerging treatment strategies for diabetic nephropathy
.
Am. J. Physiol. Renal Physiol.
309
,
F807
-
F820
.
Kefaloyianni
,
E.
,
Muthu
,
M. L.
,
Kaeppler
,
J.
,
Sun
,
X.
,
Sabbisetti
,
V.
,
Chalaris
,
A.
,
Rose-John
,
S.
,
Wong
,
E.
,
Sagi
,
I.
,
Waikar
,
S. S.
, et al. 
(
2016
).
ADAM17 substrate release in proximal tubule drives kidney fibrosis
.
JCI. Insight.
1
,
e87023
.
Kleino
,
I.
,
Järviluoma
,
A.
,
Hepojoki
,
J.
,
Huovila
,
A. P.
and
Saksela
,
K.
(
2015
).
Preferred SH3 domain partners of ADAM metalloproteases include shared and ADAM-specific SH3 interactions
.
PLoS One
10
,
e0121301
.
Krepinsky
,
J. C.
,
Ingram
,
A. J.
,
Tang
,
D.
,
Wu
,
D.
,
Liu
,
L.
and
Scholey
,
J. W.
(
2003
).
Nitric oxide inhibits stretch-induced MAPK activation in mesangial cells through RhoA inactivation
.
J. Am. Soc. Nephrol.
14
,
2790
-
2800
.
Le Gall
,
S. M.
,
Maretzky
,
T.
,
Issuree
,
P. D. A.
,
Niu
,
X.-D.
,
Reiss
,
K.
,
Saftig
,
P.
,
Khokha
,
R.
,
Lundell
,
D.
and
Blobel
,
C. P.
(
2010
).
ADAM17 is regulated by a rapid and reversible mechanism that controls access to its catalytic site
.
J. Cell Sci.
123
,
3913
-
3922
.
Lewis
,
E. J.
,
Hunsicker
,
L. G.
,
Bain
,
R. P.
and
Rohde
,
R. D.
(
1993
).
The effect of angiotensin-converting-enzyme inhibition on diabetic nephropathy. The Collaborative Study Group
.
N. Engl. J. Med.
329
,
1456
-
1462
.
Lewis
,
E. J.
,
Hunsicker
,
L. G.
,
Clarke
,
W. R.
,
Berl
,
T.
,
Pohl
,
M. A.
,
Lewis
,
J. B.
,
Ritz
,
E.
,
Atkins
,
R. C.
,
Rohde
,
R.
and
Raz
,
I.
(
2001
).
Renoprotective effect of the angiotensin-receptor antagonist irbesartan in patients with nephropathy due to type 2 diabetes
.
N. Engl. J. Med.
345
,
851
-
860
.
Li
,
J. H.
,
Huang
,
X. R.
,
Zhu
,
H.-J.
,
Johnson
,
R.
and
Lan
,
H. Y.
(
2003
).
Role of TGF-beta signaling in extracellular matrix production under high glucose conditions
.
Kidney Int.
63
,
2010
-
2019
.
Li
,
R.
,
Uttarwar
,
L.
,
Gao
,
B.
,
Charbonneau
,
M.
,
Shi
,
Y.
,
Chan
,
J. S. D.
,
Dubois
,
C. M.
and
Krepinsky
,
J. C.
(
2015a
).
High glucose up-regulates ADAM17 through HIF-1alpha in mesangial cells
.
J. Biol. Chem.
290
,
21603
-
21614
.
Li
,
X.
,
Maretzky
,
T.
,
Weskamp
,
G.
,
Monette
,
S.
,
Qing
,
X.
,
Issuree
,
P. D. A.
,
Crawford
,
H. C.
,
McIlwain
,
D. R.
,
Mak
,
T. W.
,
Salmon
,
J. E.
, et al. 
(
2015b
).
iRhoms 1 and 2 are essential upstream regulators of ADAM17-dependent EGFR signaling
.
Proc. Natl. Acad. Sci. USA
112
,
6080
-
6085
.
Liu
,
S.
and
Kurzrock
,
R.
(
2014
).
Toxicity of targeted therapy: Implications for response and impact of genetic polymorphisms
.
Cancer Treat. Rev.
40
,
883
-
891
.
Lorenzen
,
I.
,
Lokau
,
J.
,
Korpys
,
Y.
,
Oldefest
,
M.
,
Flynn
,
C. M.
,
Kunzel
,
U.
,
Garbers
,
C.
,
Freeman
,
M.
,
Grotzinger
,
J.
and
Dusterhoft
,
S.
(
2016
).
Control of ADAM17 activity by regulation of its cellular localisation
.
Sci. Rep.
6
,
35067
.
Maretzky
,
T.
,
Evers
,
A.
,
Zhou
,
W.
,
Swendeman
,
S. L.
,
Wong
,
P.-M.
,
Rafii
,
S.
,
Reiss
,
K.
and
Blobel
,
C. P.
(
2011
).
Migration of growth factor-stimulated epithelial and endothelial cells depends on EGFR transactivation by ADAM17
.
Nat. Commun.
2
,
229
.
McIlwain
,
D. R.
,
Lang
,
P. A.
,
Maretzky
,
T.
,
Hamada
,
K.
,
Ohishi
,
K.
,
Maney
,
S. K.
,
Berger
,
T.
,
Murthy
,
A.
,
Duncan
,
G.
,
Xu
,
H. C.
, et al. 
(
2012
).
iRhom2 regulation of TACE controls TNF-mediated protection against Listeria and responses to LPS
.
Science
335
,
229
-
232
.
Melenhorst
,
W. B.
,
Visser
,
L.
,
Timmer
,
A.
,
van den Heuvel
,
M. C.
,
Stegeman
,
C. A.
and
van Goor
,
H.
(
2009
).
ADAM17 upregulation in human renal disease: a role in modulating TGF-alpha availability?
Am. J. Physiol. Renal Physiol.
297
,
F781
-
F790
.
Mendelson
,
K.
,
Swendeman
,
S.
,
Saftig
,
P.
and
Blobel
,
C. P.
(
2010
).
Stimulation of platelet-derived growth factor receptor {beta} (PDGFR{beta}) activates ADAM17 and promotes metalloproteinase-dependent cross-talk between the PDGFR{beta} and epidermal growth factor receptor (EGFR) signaling pathways
.
J. Biol. Chem.
285
,
25024
-
25032
.
Miller
,
C. G.
,
Pozzi
,
A.
,
Zent
,
R.
and
Schwarzbauer
,
J. E.
(
2014
).
Effects of high glucose on integrin activity and fibronectin matrix assembly by mesangial cells
.
Mol. Biol. Cell
25
,
2342
-
2350
.
Niu
,
A.
,
Wen
,
Y.
,
Liu
,
H.
,
Zhan
,
M.
,
Jin
,
B.
and
Li
,
Y.-P.
(
2013
).
Src mediates the mechanical activation of myogenesis by activating TNFalpha-converting enzyme
.
J. Cell Sci.
126
,
4349
-
4357
.
Niu
,
A.
,
Wang
,
B.
and
Li
,
Y.-P.
(
2015
).
TNFalpha shedding in mechanically stressed cardiomyocytes is mediated by Src activation of TACE
.
J. Cell Biochem.
116
,
559
-
565
.
Regoli
,
M.
and
Bendayan
,
M.
(
1999
).
Expression of beta1 integrins in glomerular tissue of Streptozotocin-induced diabetic rats
.
Biochem. Cell Biol.
77
,
71
-
78
.
Rossello
,
A.
,
Nuti
,
E.
,
Ferrini
,
S.
and
Fabbi
,
M.
(
2016
).
Targeting ADAM17 sheddase activity in cancer
.
Curr. Drug Targets.
17
,
1908
-
1927
.
Rousseau
,
S.
,
Papoutsopoulou
,
M.
,
Symons
,
A.
,
Cook
,
D.
,
Lucocq
,
J. M.
,
Prescott
,
A. R.
,
O'Garra
,
A.
,
Ley
,
S. C.
and
Cohen
,
P.
(
2008
).
TPL2-mediated activation of ERK1 and ERK2 regulates the processing of pre-TNF alpha in LPS-stimulated macrophages
.
J. Cell Sci.
121
,
149
-
154
.
Saha
,
A.
,
Backert
,
S.
,
Hammond
,
C. E.
,
Gooz
,
M.
and
Smolka
,
A. J.
(
2010
).
Helicobacter pylori CagL activates ADAM17 to induce repression of the gastric H, K-ATPase alpha subunit
.
Gastroenterology
139
,
239
-
248
.
Salem
,
E. S. B.
,
Grobe
,
N.
and
Elased
,
K. M.
(
2014
).
Insulin treatment attenuates renal ADAM17 and ACE2 shedding in diabetic Akita mice
.
Am. J. Physiol. Renal Physiol.
306
,
F629
-
F639
.
Schwarz
,
J.
,
Broder
,
C.
,
Helmstetter
,
A.
,
Schmidt
,
S.
,
Yan
,
I.
,
Müller
,
M.
,
Schmidt-Arras
,
D.
,
Becker-Pauly
,
C.
,
Koch-Nolte
,
F.
,
Mittrücker
,
H.-W.
, et al. 
(
2013
).
Short-term TNFalpha shedding is independent of cytoplasmic phosphorylation or furin cleavage of ADAM17
.
Biochim. Biophys. Acta
1833
,
3355
-
3367
.
Sommer
,
A.
,
Kordowski
,
F.
,
Büch
,
J.
,
Maretzky
,
T.
,
Evers
,
A.
,
Andrä
,
J.
,
Düsterhöft
,
S.
,
Michalek
,
M.
,
Lorenzen
,
I.
,
Somasundaram
,
P.
, et al. 
(
2016
).
Phosphatidylserine exposure is required for ADAM17 sheddase function
.
Nat. Commun.
7
,
11523
.
Soond
,
S. M.
,
Everson
,
B.
,
Riches
,
D. W. H.
and
Murphy
,
G.
(
2005
).
ERK-mediated phosphorylation of Thr735 in TNFalpha-converting enzyme and its potential role in TACE protein trafficking
.
J. Cell Sci.
118
,
2371
-
2380
.
Srour
,
N.
,
Lebel
,
A.
,
McMahon
,
S.
,
Fournier
,
I.
,
Fugère
,
M.
,
Day
,
R.
and
Dubois
,
C. M.
(
2003
).
TACE/ADAM-17 maturation and activation of sheddase activity require proprotein convertase activity
.
FEBS Lett.
554
,
275
-
283
.
Taniguchi
,
K.
,
Xia
,
L.
,
Goldberg
,
H. J.
,
Lee
,
K. W. K.
,
Shah
,
A.
,
Stavar
,
L.
,
Masson
,
E. A. Y.
,
Momen
,
A.
,
Shikatani
,
E. A.
,
John
,
R.
, et al. 
(
2013
).
Inhibition of Src kinase blocks high glucose-induced EGFR transactivation and collagen synthesis in mesangial cells and prevents diabetic nephropathy in mice
.
Diabetes
62
,
3874
-
3886
.
Uttarwar
,
L.
,
Peng
,
F.
,
Wu
,
D.
,
Kumar
,
S.
,
Gao
,
B.
,
Ingram
,
A. J.
and
Krepinsky
,
J. C.
(
2011
).
HB-EGF release mediates glucose-induced activation of the epidermal growth factor receptor in mesangial cells
.
Am. J. Physiol. Renal Physiol.
300
,
F921
-
F931
.
Wu
,
D.
,
Peng
,
F.
,
Zhang
,
B.
,
Ingram
,
A. J.
,
Gao
,
B.
and
Krepinsky
,
J. C.
(
2007
).
Collagen I induction by high glucose levels is mediated by epidermal growth factor receptor and phosphoinositide 3-kinase/Akt signalling in mesangial cells
.
Diabetologia
50
,
2008
-
2018
.
Wu
,
D.
,
Peng
,
F.
,
Zhang
,
B.
,
Ingram
,
A. J.
,
Kelly
,
D. J.
,
Gilbert
,
R. E.
,
Gao
,
B.
and
Krepinsky
,
J. C.
(
2009
).
PKC-beta1 mediates glucose-induced Akt activation and TGF-beta1 upregulation in mesangial cells
.
J. Am. Soc. Nephrol.
20
,
554
-
566
.
Xu
,
P.
and
Derynck
,
R.
(
2010
).
Direct activation of TACE-mediated ectodomain shedding by p38 MAP kinase regulates EGF receptor-dependent cell proliferation
.
Mol. Cell
37
,
551
-
566
.
Xu
,
P.
,
Liu
,
J.
,
Sakaki-Yumoto
,
M.
and
Derynck
,
R.
(
2012
).
TACE activation by MAPK-mediated regulation of cell surface dimerization and TIMP3 association
.
Sci. Signal.
5
,
ra34
.
Zhang
,
Q.
,
Thomas
,
S. M.
,
Lui
,
V. W. Y.
,
Xi
,
S.
,
Siegfried
,
J. M.
,
Fan
,
H.
,
Smithgall
,
T. E.
,
Mills
,
G. B.
and
Grandis
,
J. R.
(
2006
).
Phosphorylation of TNF-alpha converting enzyme by gastrin-releasing peptide induces amphiregulin release and EGF receptor activation
.
Proc. Natl. Acad. Sci. USA
103
,
6901
-
6906
.

Competing interests

The authors declare no competing or financial interests.

Supplementary information