A novel mechanism of ERK1/2 regulation in smooth muscle involving acetylation of the ERK1/2 scaffold IQGAP1

Ceramide, a bioactive lipid and signaling molecule associated with cardiovascular disease, is known to activate extracellular signal regulated kinases 1 and 2 (ERK1/2). Here, we determined that the effect of ceramide on ERK1/2 is mediated by ceramide signaling on an ERK scaffold protein, IQ motif containing GTPase activating protein 1 (IQGAP1). Experiments were performed with aortic smooth muscle cells using inhibitor screening, small interfering RNA (siRNA), immunoprecipitation (IP), immunoblots and bioinformatics. We report here that C6 ceramide increases serum-stimulated ERK1/2 activation in a manner dependent on the ERK1/2 scaffold IQGAP1. C6 ceramide increases IQGAP1 protein levels by preventing its cleavage. Bioinformatic analysis of the IQGAP1 amino acid sequence revealed potential cleavage sites for proteases of the proprotein convertase family that match the cleavage products. These potential cleavage sites overlap with known motifs for lysine acetylation. Deacetylase inhibitor treatment increased IQGAP1 acetylation and reduced IQGAP1 cleavage. These data are consistent with a model in which IQGAP1 cleavage is regulated by acetylation of the cleavage sites. Activation of ERK1/2 by ceramide, known to increase lysine acetylation, appears to be mediated by acetylation-dependent stabilization of IQGAP1. This novel mechanism could open new possibilities for therapeutic intervention in cardiovascular diseases.

muscle cells with the cell permeable C6 ceramide, and analyzed ERK1/2 phosphorylation as a measure of ERK1/2 activation in response to either a phorbol ester, 12-deoxyphorbol 13-isobutylate 20-acetate (DPBA), a stimulus that activates contractile pathways in vascular smooth muscle, or fetal bovine serum (FBS), a stimulus that activates proliferative pathways. Whereas ceramide treatment had no significant effect on phorbol ester mediated activation of ERK1/2, the relative amount of phosphorylated ERK1/2 in response to serum was significantly increased after ceramide treatment compared to control treated cells (Fig. 1).

IQGAP1, but not KSR1, is involved in ceramide-mediated amplification of ERK1/2 activation.
Based on our previous studies 1, 3 , demonstrating stimulus-specific and scaffold-specific ERK1/2 signaling pathways in aortic smooth muscle cells, we speculated that the stimulus-specific effect of ceramide on ERK1/2 activation might be mediated by an ERK1/2 scaffold.
KSR1 is an ERK1/2 scaffold reported to be involved in ceramide signaling pathways 9,11,12 in other systems. Thus, we investigated a possible role of KSR in the effect of ceramide in aortic smooth muscle cells (Fig. 2). However, siRNA against KSR1 decreased KSR1 levels ( Fig. 2A,D) but did not significantly affect ERK1/2 phosphorylation in the presence of C6 plus FBS (Fig. 2B).
IQGAP1 is another well known ERK scaffolding protein that we have previously shown to play a role in these aortic smooth muscle cells 3 . Knockdown of IQGAP protein levels with siRNA ( Fig. 2A,C), in contrast to the results for KSR1, significantly reduced C6 plus FBS-induced ERK1/2 phosphorylation (Fig. 2B). This indicated a role for IQGAP1, but not KSR1, in the ceramide mediated increase in ERK1/2 phosphorylation in aortic smooth muscle cells and raised the question of the mechanism involved. Ceramide stabilizes IQGAP1 by preventing its proteolytic processing. After immunoblotting with an IQGAP1-specific antibody that recognizes an epitope in the N-terminal region of IQGAP1 (rabbit polyclonal anti-IQGAP1, H-109, Santa Cruz), we observed additional bands of 185 kDa(faintly), 165 kDa and 89 kDa in addition to full length IQGAP1 (190 kDa) (Fig. 3A). The intensities of these bands are reduced along with the intensity of full length IQGAP1 after siRNA knock down of IQGAP1 as shown for some of the bands in Fig. 2A, indicating that these bands are specifically stained by the IQGAP1 antibody and might represent splice variants or cleavage products of IQGAP1. However, we were unable to find any reports on IQGAP1 splice variants in the literature. The additional bands were detected by using a different antibody that is raised against amino acids 3-37 (mouse monoclonal anti-IQGAP1, clone C-9, Santa Cruz), but not by an antibody that was raised against the C-terminus of IQGAP1 (goat polyclonal anti-IQGAP1, clone C-17, Santa Cruz) (Fig. 3A), consistent with the additional bands being products that arise from cleavage in the C-terminal region of the protein.
Interestingly, when we analyzed the amount of full length IQGAP1 in FBS-stimulated samples (Fig. 3B), we detected (with H-109) a significant increase in IQGAP protein expression in lysates from cells that had been pretreated with ceramide, whereas the ceramide synthase inhibitor fuminisin B1 (FB1) had the opposite effect and led to reduced full length IQGAP1 levels. Conversely, when we quantitated the total amount of the putative IQGAP1 cleavage fragments and expressed this as a percentage of the total IQGAP1 band staining (Fig. 3C), we found decreased IQGAP1 cleavage in samples that had been treated with ceramide, but increased cleavage of IQGAP1 in cells treated with the ceramide synthesis inhibitor FB1. Thus ceramide appears to stabilize IQGAP1 by preventing its cleavage. IQGAP1 sequence analysis reveals acetylation motifs close to predicted cleavage sites. To narrow down the possible proteases responsible for IQGAP1 cleavage, we screened the IQGAP1 amino acid sequence for possible cleavage sites using protease cleavage prediction software (Expasy PeptideCutter). Predicted cleavage sites were then compared to the experimentally found cleavage products, but none of the proteases considered by this tool produced a cleavage pattern similar to the observed band pattern in our experiments.However, analysis of the IQGAP1 amino acid sequence for recurring motifs did reveal the motif K/R-X-K/R-K/R, that is present in rat IQGAP1 with 9 repeats. This motif is similar to the consensus motif for Subtilisin/kexin Proprotein Convertases (SPCs), also named Paired basic Amino acid Cleaving Enzymes (PACEs), namely K/R-X n -K/R↓, with the downward arrow indicating the position of the cleavage site within or relative to the motif 13 . Some copies of this motif also conform with the more specific consensus cleavage site recognized by furin (Fig. 4B), a member of the SPC family, which consists of the sequence R-X-K/R-R↓ 14 . The positioning of those SPC motif copies that have a lysine in position P1 (the residue closest to the predicted cleavage site) matches closely with the expected cleavage sites derived from calculated fragment sizes of IQGAP1 (black arrows in Fig. 4B). While they match the general consensus for SPCs, these sites are unlikely to be cleaved by furin, which requires an arginine in position P1 14,15 .

Figure 2.
SiRNA knockdown demonstrates a role for IQGAP1 in ceramide-induced ERK1/2 activation. (A) Aortic smooth muscle cells were treated with control, IQGAP1, or KSR1 siRNA. 5 days after transfection, cells were treated with ceramide (6 hours, 50 µg/ml). Additionally, cells were stimulated with FBS (10%) for 5 minutes before preparation of cell extracts. Cell lysates were analyzed for expression of IQGAP1, KSR1, phospho-ERK1/2 and total ERK1/2 using specific antibodies. GAPDH staining is shown as reference. Cropped gels are shown. (B) The statistical analysis of 5 independent experiments shows a significant decrease of relative ERK1/2 phosphorylation after knock down of IQGAP, but not KSR1. Statistical analysis of the degree of siRNAinduced knock down are shown in (C) for IQGAP1 and in (D) for KSR1.  Interestingly, in addition to being putative cleavage sites, these motifs also resemble a type of lysine acetylation motif with the consensus K-X-K-K 16 (Fig. 4A). Lysine acetylation has been shown to regulate protein stability, for example in the cases of beta secretase BACE1 17 , FoxA2 18 and glutamase carboxypeptidase II 19 . The overlap of acetylation motifs in close vicinity of the putative cleavage sites therefore raises the possibility that IQGAP1 cleavage could be regulated by acetylation.
Acetylation regulates IQGAP1 protein stability. To determine whether IQGAP1 is acetylated/ deacetylated, we treated aortic smooth muscle cells with the histone deacetylase (HDAC) inhibitors sodium phenylbutyrate (PB), which inhibits class I and class II HDACs, or nicotinamide (NAM), which inhibits the class III HDACs, also called the Sirtuins. Cell lysates were then analyzed for lysine acetylation using a lysine acetylation-specific antibody. After HDAC inhibitor treatment, several bands showed increased signal intensity. Staining of the same membranes with an anti-IQGAP1 antibody revealed an overlap of the 190 kDa acetylated lysine band with IQGAP1 (Fig. 5A). The bar graph in Fig. 5B shows the statistical analysis of p190 IQGAP1 acetylation in untreated, NAM treated and PB treated cells. Statistically significant increases were seen with both deacetylase inhibitors. Moreover, IQGAP1 was pulled down from NAM-treated aortic smooth muscle lysates in IP experiments with an acetylated lysine-specific antibody (Fig. 5C).
Cleavage of IQGAP was noted to be less prominent in the presence of the deacetylase inhibitors NAM and PB (Fig. 5D). When analyzed by densitometry, quantitating the amount of full length IQGAP, under control conditions, or in the presence of NAM, or PB, and expressing those values as a percentage of the control demonstrated a statically significant inhibition of cleavage, and a relative increase in full length IQGAP was seen with both HDAC inhibitors (Fig. 5E).

Does ceramide increase IQGAP1 acetylation?
Ceramide has been shown to promote lysine acetylation 17,20 . Since HDAC inhibitor treatment and ceramide have similar effects on IQGAP1 stability, the question arises as to whether the effect of ceramide on IQGAP1 is mediated by increased acetylation of IQGAP1. If so, we should be able to detect an increased acetylation in the p190 band. As is shown in Fig. 6A, ceramide does indeed increase acetylation of a 190 kDa protein consistent with acetylation of IQGAP1.
The increased acetylation could be due to either inhibition of HDACs or to activation of histone acetyltransferases (HATs). To determine which mechanism is involved, we used the HDAC inhibitor, NAM to determine if blocking HDACs can produce any additional effect to that of ceramide. If ceramide acts through inhibition of HDACs no additional effect is expected. Alternatively, if ceramide increases acetylation by activating HATs, a HAT inhibitor should prevent the effect of ceramide. As is shown in Fig. 6B, the HDAC inhibitor NAM did not produce a statistically significant additional effect on acetylation. In contrast, the HAT inhibitor had a dramatic effect to decrease acetylation. Thus, the data are consistent with ceramide increasing acetylation by activating HATs.

HDAC inhibitor treatment increases ERK1/2 activation. Since IQGAP is an ERK scaffold, and
we have shown that ceramide increases ERK activation (Fig. 1), the question arises as to whether increasing acetylation of IQGAP with NAM would also alter ERK phosphorylation levels. Thus, we used FBS to trigger a proliferative ERK activation pathway and found (Fig. 7A,B) that a significant augmentation of ERK activation occurred. Interestingly, phorbol ester, a contractile rather than a proliferative stimulus, also increased ERK activity but this was not significantly augmented by NAM, consistent with a difference in the scaffolding.

Discussion
Our data are consistent with ceramide, which we show here increases lysine acetylation, stabilizing IQGAP1 by increasing its acetylation, thereby regulating ERK1/2 activation.
Furthermore, based on our data, we suggest a model (Fig. 8) in which cleavage of IQGAP1 is regulated by its acetylation at or in close proximity to the cleavage sites. Others have reported similar ceramide-induced acetylation and stabilization of tubulin 20 as well as the secretase BACE1 which affects stability of the Alzheimer's disease (APP) 17 . In the case of BACE1, the stabilizing effect of acetylation has been attributed to ceramide-induced transcriptional upregulation of two acetyltransferases 21 . Our findings that ceramide-induced IQGAP1 acetylation is caused by HAT activation rather than HDAC inhibition, and that the effect of ceramide required a treatment duration of six hours, indicate the possibility that IQGAP1 stabilization, too, is based on ceramide-induced transcriptional upregulation of acetyltransferase genes.
Because of the modular structure of IQGAP1, cleavage could also generate interesting IQGAP1 fragments that may have specific functions in signaling. Cleavage of IQGAP1 may lead to the generation of smaller IQGAP1 fragments with specific function due to specific subdomain composition (e.g. removal of C-terminal cadherin binding site). Interestingly, the potential SPC cleavage sites are all outside of the IQGAP1 functional domains (see domain map in Fig. 4), with the exception of one site in the IQ domain. Cleavage of this site could generate a cleavage products with either the first or the last two IQ motifs only. Of note, the four IQ motifs show selective binding to myosin essential light chain, the glial calcium binding protein S100B, apocalmodulin and Ca 2+ /calmodulin 13,22 . Furthermore, smaller IQGAP1 fragments containing only one binding domain might by themselves modulate cellular signaling by acting as decoy peptides.
The acetylation motif KXKK, that appears to play a role in IQGAP1 stability, is typical for steroid receptors, e.g. the glucocorticoid receptor. It may seem surprising that IQGAP1 harbors this motif although it is itself not a steroid receptor; however, it was recently shown to interact with a steroid receptor 23 . None of the previously reported IQGAP1 acetylation sites overlap with the KXKK motifs. 10 acetylation sites were identified by Chen et al. 24 , and 8 sites were shown to be acetylated by Lundby et al. 25 . However, between the two studies, only 3 sites were detected in both screens, indicating that in each screen many acetylation sites might be missed.
Our results also highlight the importance of carefully considering experimental conditions in the interpretation of scaffold function. For example, we have shown previously that IQGAP1 is required for phorbol ester induced ERK1/2 activation 3 . However, no effect of HDAC treatment on phorbol ester induced ERK1/2 activation was observed in the present study. This apparent contradiction can be explained by the fact that although IQGAP1 is required, its acetylation does not play a role in phorbol ester-induced ERK activation In conclusion, our data are consistent with a model where activation of ERK1/2 by ceramide is mediated by ceramide-induced lysine acetylation close to proteolytic cleavage sites of the ERK1/2 scaffold IQGAP1 and consequent stabilization IQGAP. The data also indicate that ceramide-induced acetylation of IQGAP1 occurs via the activation of Histone acetyl transferases. This novel mechanism could open new possibilities for selective therapeutic intervention in cardiovascular diseases. For example, if ceramide-dependent activation of ERK through acetylation of the scaffold IQGAP leads to cardiovascular disease, the present results suggest that it would be possible to leave intact other ERK scaffolds while inhibiting the acetylation pathway to selectively inhibit the ceramide pathway. A similar strategy has been used in the cancer field to cause apoptosis but not necrosis by selective inhibition of p38 MAPK but not JNK or ERK leading to selective cancer cell cytoxicity 26 .
Cell culture and siRNA transfection. A7r5 rat aorta cells (ATCC, Manassas, VA) were cultured as described previously 3 . For all experiments, cells were subjected to serum starvation (0% serum for 24 hours) to ensure differentiation to the smooth muscle-like phenotype 27,28 . SiRNA knock down of IQGAP1 and KSR1 as well as control siRNA treatment was performed as described previously 3  Cell extracts and IP experiments. For preparation of whole cell lysates, plates were washed three times with ice-cold phosphate-buffered saline (pH 7.2), drained on ice and then scraped off in lysis buffer (140 mmol/L NaCl, 3 mmol/L MgCl 2 , 1 mmol/L dithiothreitiol and 0.5% Nonidet-P40 in a 20 mmol/L sodium phosphate buffer, pH 8.0) or, for IP experiments, in IP lysis buffer (50 mmol/L NaCl, 10% glycerol, 1% Nonidet-P40 in a 10 mmol/L sodium phosphate buffer, pH 8.0). Lysis buffers were supplemented with protease inhibitors. Cell lysis was performed on ice for 30 minutes. Lysates were subsequently cleared by centrifugation (16,000 rcf for 10 minutes at 4 °C). For IP experiments, lysates in IP lysis buffer were incubated rotating at 4 °C over night in the presence of anti-acetylated lysine antibody (5 μl antibody for 500 mg total protein) crosslinked to Protein G-dynabeads (R) (Invitrogen). For control IPs, rabbit anti-GFP antibody (Clontech, Mountain View, CA) crosslinked to protein G-dynabeads (R) was used. The immobilized antigen-antibody complexes were washed three times with IP lysis buffer. Bound proteins were eluted from the immobilized antibodies with sodium dodecylsulfate (SDS) sample buffer and further processed for western blotting.
Western blot. Proteins in IP samples or whole cell lysates were separated by SDS-polyacrylamide gel electrophoresis (PAGE) according to standard procedures. For immunoblots, proteins were transferred from SDS gels to nitrocellulose membranes (Whatman, Florham Park, NJ) and stained with specific primary antibodies and appropriate secondary antibodies. Bands were detected using an Odyssey(R) infrared scanner a system that is highly linear in the range of intensities used. Odyssey 2.1 software was used for densitometric analysis of the raw data. For comparison of protein expression levels, bands were normalized to either GAPDH or tubulin signals on the same membrane. To determine relative ERK1/2 phosphorylation, pERK1/2 signals were normalized to the total ERK1/2 signal on the same membrane.
Statistics and sequence analysis. All values given in the text are means ± standard error. Statistical significance was evaluated using two-tailed Student's t-tests, and differences with p values below 0.05 were considered significant. Data from at least 5 independent experiments were used for statistical analyses. For IQGAP1 sequence analysis, the PeptideCutter software 29 , which can be found at http://web.expasy.org/peptide_cutter, was used to identify potential caspase cleavage sites. Potential proprotein convertase cleavage sites were located by sequence comparison with the known proprotein converatase motifs 15 .
Data availability. All data generated or analysed during this study are included in this published article.