Protein Never in Mitosis Gene A Interacting-1 regulates calpain activity and the degradation of cyclooxygenase-2 in endothelial cells

Background The peptidyl-proline isomerase, Protein Never in Mitosis Gene A Interacting-1 (PIN1), regulates turnover of inducible nitric oxide synthase (iNOS) in murine aortic endothelial cells (MAEC) stimulated with E. coli endotoxin (LPS) and interferon-γ (IFN). Degradation of iNOS was reduced by a calpain inhibitor, suggesting that PIN1 may affect induction of other calpain-sensitive inflammatory proteins, such as cyclooxygenase (COX)-2, in MAEC. Methods MAEC, transduced with lentivirus encoding an inactive control short hairpin (sh) RNA or one targeting PIN1 that reduced PIN1 by 85%, were used. Cells were treated with LPS/IFN, calpain inhibitors (carbobenzoxy-valinyl-phenylalaninal (zVF), PD150606), cycloheximide and COX inhibitors to determine the effect of PIN1 depletion on COX-2 and calpain. Results LPS or IFN alone did not induce COX-2. However, treatment with 10 μg LPS plus 20 ng IFN per ml induced COX-2 protein 10-fold in Control shRNA MAEC. Induction was significantly greater (47-fold) in PIN1 shRNA cells. COX-2-dependent prostaglandin E2 production increased 3-fold in KD MAEC, but did not increase in Control cells. The additional increase in COX-2 protein due to PIN1 depletion was post-transcriptional, as induction of COX-2 mRNA by LPS/IFN was the same in cells containing or lacking PIN1. Instead, the loss of COX-2 protein, after treatment with cycloheximide to block protein synthesis, was reduced in cells lacking PIN1 in comparison with Control cells, indicating that degradation of the enzyme was reduced. zVF and PD150606 each enhanced the induction of COX-2 by LPS/IFN. zVF also slowed the loss of COX-2 after treatment with cycloheximide, and COX-2 was degraded by exogenous μ-calpain in vitro. In contrast to iNOS, physical interaction between COX-2 and PIN1 was not detected, suggesting that effects of PIN1 on calpain, rather than COX-2 itself, affect COX-2 degradation. While cathepsin activity was unaltered, depletion of PIN1 reduced calpain activity by 55% in comparison with Control shRNA cells. Conclusion PIN1 reduced calpain activity and slowed the degradation of COX-2 in MAEC, an effect recapitulated by an inhibitor of calpain. Given the sensitivity of COX-2 and iNOS to calpain, PIN1 may normally limit induction of these and other calpain substrates by maintaining calpain activity in endothelial cells.


Background
Protein Never in Mitosis Gene A Interacting-1 (PIN1) is an enzyme that regulates transcription, and turnover of mRNA and proteins. PIN1 is a cis-trans peptidyl-prolyl isomerase that contains an amino-terminal domain, the tryptophan-tryptophan (WW) domain, which is characterized by two tryptophan residues separated by 22 amino acids that can bind to phosphorylated serine-or threonine-proline sequences in substrate proteins. PIN1 also isomerizes this motif with its carboxy-terminal catalytic domain [1]. Isomerization of the phosphorylated serine-or threonineproline motif has a significant effect on conformation of many phospho-proteins. The conformational switching catalyzed by PIN1 allows it to regulate transcription factors, mRNA stabilization factors, and the susceptibility of a growing list of proteins to post-translational modifications and proteases [1][2][3][4][5].
Previously, we found that depletion of PIN1 and treatment with a calpain inhibitor each reduced the degradation of inducible nitric oxide synthase (iNOS) in murine aortic endothelial cells (MAEC) stimulated with E. coli endotoxin (LPS) and interferon-γ (IFN). PIN1 bound to iNOS suggesting that it might directly regulate the sensitivity of iNOS to calpain [6]. PIN1 may also regulate expression of inflammatory proteins by an effect on calpain.

Cells
MAEC were cultured from aortas of mice in accordance with the Guide for the Care and Use of Laboratory Animals from the U.S. National Institutes of Health [21]. As described previously, cells were transduced with short hairpin RNA (shRNA) to knockdown (KD) PIN1 or with an inactive mutant sequence (Control), and selected for stable modification. This produced KD MAEC with approximately 15% of the level of PIN1 protein found in Control and non-transduced MAEC [6].

Treatments
KD and Control MAEC were incubated in Dulbecco's minimum essential medium/0.5% fetal bovine serum for 18 h, and then treated with medium or 10 μg LPS and 20 ng IFN per ml, and other agents for various times. zVF or PD150606 were added 1 h before LPS/IFN to inhibit calpain [22]. Ninety μg cycloheximide/ml was used to inhibit protein synthesis after induction of COX-2 with LPS/IFN [6]. COX-2-dependent prostaglandin E2 production was measured after incubating cells with LPS/IFN for 24 h. Cells were then incubated in fresh medium containing 20 μM arachidonic acid, LPS, IFN and the COX-1 selective antagonist, SC-560 (1 μM) [23], with or without the COX-2 selective antagonist, NS-398 (10 μM) [24]. The medium was collected after 2 h and stored at -80 degrees C. Prostaglandin E2 was measured by competition enzyme-linked immunosorbent assay in comparison with prostaglandin E2 standard by the manufacturer's instructions.
mRNA Levels RNA was extracted with Trizol, precipitated, and dissolved in water. cDNA was produced from 3 μg of RNA. cDNA was amplified by polymerase chain reaction for β-actin as described previously [25], and for COX-2. COX-2 primers were sense, 5'-CCG GAC TGG ATT CTA TGG TG, and antisense, 5'-AGG AGA GGT TGG AGA AGG CT from Genbank accession BC052900, producing a 263 base pair product. Half of each reaction was electrophoresed in 1% agarose. Gels were imaged and analyzed after ethidium bromide staining [25].

Immunoprecipitation, glutathione s-transferase pulldown, and Western blotting
As previously described [6], cells were washed, sonicated in lysis buffer, and protein concentration was measured. For western blotting, 12 μg of sample protein were denatured and separated on 4-20% Tris-gylcine, SDS-polyacrylamide gels and transferred to nitrocellulose. For immunoprecipitation, 500 μg of cell lysate protein was incubated with 5 μg anti-PIN1 antibody and protein G agarose. For pulldown, glutathione S-transferase or glutathione S-transferase-PIN1 fusion protein was added to 500 μg of cell lysate protein and glutathione-sepharose. Samples were then denatured for electrophoresis and western blotting. Blots were immunostained and imaged on X-ray film by enhanced chemiluminescence. Films were scanned and digital images of proteins were analyzed.
The susceptibility of COX-2 to degradation by exogenous calpain in vitro was also determined. Extracts were incubated in calpain reaction buffer as above in the presence of porcine μ-calpain for 30 min at 30°C. Reactions were stopped by addition of denaturing sample buffer and subjected to western blotting as described above.

Cathepsin activity
Cells were washed three times with PBS and scraped in 1 ml of ice-cold PBS. Cells were collected by centrifugation at 1500 × g for 2 min at 4°C. The pellet was resuspended in reaction buffer (50 mM Na-acetate, 1 mM EDTA and 2 mM dithioerythritol pH 5.5), and sonicated four times for 10 s with 1 min breaks. The lysate was centrifuged at 1500 × g for 5 min at 4°C to remove debris. The reaction was started by mixing 20 μM cathepsin substrate, carboxybenzyl-phenylalanine-arginine-7-amido-4-methylcoumarin, in the reaction containing 2 μg supernatant protein, as described by Werle et al. [27]. 7-amido-4-methylcoumarin release was monitored at 37°C for 30 min by fluorescence, with excitation at 380 nm and emission at 460 nm, and the initial velocity was determined.

Data analysis
Bands in images of polymerase chain reaction gels and scanned western blots were measured with Image J 1.34 s (NIH). Prostaglandin E2 concentrations were estimated from a standard curve and calpain activity was indicated by the fluorescence increase per minute. Data were analyzed by Student's t test or analysis of variance with Bonferroni correction for multiple comparisons [28].

Results
Previously, KD shRNA was shown to reduce PIN1 by 85% compared with Control shRNA in MAEC [6]. COX-2 protein was very low in vehicle-treated KD and Control MAEC (figure 1), and incubation with either LPS or IFN alone did not induce it (data not shown). However, stimulation with 10 μg LPS plus 20 ng IFN per ml increased COX-2 expression. The protein appeared to increase as early as 1 h after treatment, and induction persisted through 24 h. Differences between KD and Control cells were qualitatively noticeable by 4 h after treatment and became greater with time ( figure 1A). After 24 h, the signal for COX-2 protein was increased 10-fold in Control shRNA MAEC (figure 1B). The COX-2 signal was significantly more induced in PIN1 KD cells (47-fold). Similar results were obtained in 2 other independent pairs of cul-tures selected for the KD and Control shRNA (data not shown). COX-2-mediated prostaglandin E2 production increased 3-fold in KD MAEC, but not in the Control cells PIN1 KD and Control shRNA MAEC were pretreated with vehicle or the calpain inhibitors, zVF or PD150606, for 1 h, and then treated with LPS and IFN for 24 h. Again, COX-2 increased more in KD than Control cells ( figure 4). In Control cells, COX-2 was induced 5.5-fold more in the presence of zVF than in its absence. zVF also increased the induction of COX-2 from its elevated level in KD MAEC by a factor of two (figure 4A). PD 150606, which is more selective than zVF for calpain relative to cathepsin activities [29,30], also increased the induction of COX-2 in KD and Control cells. zVF did not increase the induction of COX-2 mRNA in cells treated with LPS/IFN for 1 or 24 h (figure 5).
The effect of PIN1 depletion and zVF on degradation of COX-2 was assessed. Cells were induced with LPS/IFN and then treated with 90 μg cycloheximide/ml to block translation. The level of COX-2 protein fell to 44% of its initial value 2 h after addition of cycloheximide to Control shRNA cells ( figure 6). However, a similar decrease to 47% was delayed until 4 h in KD MAEC. COX-2 protein fell only to 78% of initial 2 h after cycloheximide in zVFtreated Control cells. zVF also inhibited the loss of COX-2 in KD cells at 4 h.
To confirm that COX-2 is a potential substrate for calpain, its digestion in vitro was examined. Addition of porcine μcalpain to extracts of LPS/IFN-treated Control cells caused a concentration-dependent loss of COX-2 signal (figure 7).
Since PIN1 is known to bind its substrate proteins, interaction with COX-2 was investigated. Immunoprecipitation of PIN1 from extracts of vehicle-or LPS/IFN-treated Control cells did not produce any COX-2 detectable on western blots. COX-2 was not pulled down with glutathione-S-transferase-PIN1 fusion protein or glutathione-Stransferase (not shown).
Given the effect of calpain inhibitors on COX-2, calpain and cathepsin activities were measured. LPS/IFN increased calpain activity 6.0-fold in KD cells, and 5.  Here, suppression of PIN1 in endothelial cells increased the induction of COX-2, and COX-2-dependent production of prostaglandin E2 by LPS/IFN (figures 1 and 2). Despite a nearly 5-fold greater induction of COX-2 protein in KD compared with Control MAEC, there was no difference in the induction of COX-2 mRNA ( figure 3). This suggests that PIN1 regulates COX-2 by a post-transcriptional mechanism. Consistent with a post-transcriptional effect, PIN1 depletion reduced the turnover of COX-2 ( figure 6). Since COX-2 has a relatively short halflife, inhibition of turnover could lead to large, cumulative, post-transcriptional increases after induction with LPS/ IFN [8].
One prior study revealed that cleavage of COX-2 was reduced by the inhibitor, E-64d, in human synovial fibroblasts [18]. Here, the calpain inhibitor, zVF, increased the induction of COX-2 (figure 4) and reduced its degradation (figure 6), without increasing its mRNA (figure 5). As for E-64d, zVF can also inhibit cathepsin activity at concentrations similar to those that inhibit calpain [29,37,38]. Therefore, PD 150606, which is more selective for calpain compared with cathepsin [30], was Effect of PIN1 knockdown on COX-2 mRNA Figure 3 Effect of PIN1 knockdown on COX-2 mRNA. A, KD and Control (Con) shRNA MAEC were treated with LPS and IFN for 0-8 h. B, Cells were treated with medium or LPS and IFN for 24 h. mRNAs encoding COX-2 and β-actin were determined by RT-PCR and agarose gel electrophoresis. Representative ethidium bromide-stained gels are shown. Replicate COX-2 or β-actin products in a single gel were imaged for analysis. Bars represent mean + S.E. ratio of COX-2: β-actin products from densitometric analysis of images from 3 independent cultures. *:p < 0.05 for comparison with cells treated for 0 h in A, or with medium in B.  In support of this idea, it was shown here for the first time that PIN1 depletion reduced calpain activity in endothelial cells. In contrast, cathepsin activity was not affected by PIN1 depletion ( figure 8). This result, combined with the effects of zVF and PD150606 on COX-2 induction and turnover, suggests again that calpain limits the expression of COX-2 in MAEC. Indeed, COX-2 was degraded by μcalpain in vitro, indicating that it is a potential substrate in cells ( figure 7). The reduced calpain activity in KD extracts (figure 8) could be due to an increase in expression or function of calpastatin or other unrecognized endogenous calpain inhibitors in KD cells, or to a reduction in expression or function of calpains [39]. Nevertheless, the results suggest that PIN1 depletion reduces calpain activity, consequently reducing the turnover of COX-2 in MAEC. zVF further reduced the loss of COX-2 in cycloheximidetreated KD cells ( figure 6). This may be due to the partial 55% reduction of calpain activity in KD MAEC (figure 8). The partially reduced calpain activity could account for the intermediate loss of COX-2 in the cycloheximidetreated KD cells, allowing zVF to further suppress turnover. It may also explain the ability of calpain inhibitors to increase induction of COX-2 in both KD and Control MAEC ( figure 4). The partial reduction of calpain activity may be due to the incomplete (85%) suppression of PIN1 by the shRNA [6]. PIN1 may also function as a modulator of calpain activity and not as an absolute requirement.

Effect of calpain inhibitors on COX-2 protein
Previously, we observed that PIN1 depletion and zVF each increased the induction of iNOS, and reduced its degrada-Effect of zVF on COX-2 mRNA Representative agarose electrophoresis of PCR products are shown. Replicate COX-2 or β-actin products in a single ethidium bromide-stained gel were imaged for analysis. Bars represent the mean + SE of ratio of COX-2/β-actin signal intensity + SE of 3 independent cultures. *: p < 0.05 for comparison between LPS/IFN-and medium-treated cells. +: p < 0.05 for comparison between KD and Control cells treated with LPS/IFN and zVF.
tion. PIN1 physically interacted with iNOS. The WW and catalytic domains of PIN1 appeared to contribute to the association [6]. This suggested that PIN1 depletion might alter the susceptibility of its targets to digestion by calpain. For example, PIN1 could associate with these substrates and catalyze proline isomerization, affecting protease sensitivity. In contrast to iNOS, however, interaction between COX-2 and PIN1 was not detected here. A role for direct interaction between COX-2 and PIN1 cannot be completely excluded, however, since association of the proline isomerase with its putative substrate may be weak or transient. PIN1 could also affect association of COX-2, or iNOS, with other proteins that may indirectly regulate proteolysis. Thus, effects of PIN1 on calpain activity and/ or COX-2, or associated factors, could affect the sensitivity of COX-2 to digestion with calpain.
Overall, the results indicate that PIN1 regulates the induction of COX-2, and iNOS, by a previously unknown effect on calpain-mediated turnover in MAEC. The mechanisms by which PIN1 regulates calpain activity are under investigation. In particular, PIN1 could affect the expression or activity of calpain subunits, and the endogenous inhibitor of heterodimeric calpains, calpastatin [39].
The effects of COX-2 in acute and chronic inflammatory responses in the vasculature are complicated by multiple primary and secondary stimuli that may be present, and by cellular factors, such as supply of arachidonic acid, complement of various prostaglandin synthases, and expression of prostaglandin receptors [16]. Here, depletion of PIN1 and inhibition of calpain each caused overinduction of both COX-2 and iNOS. The consequences of co-induction of these two particular enzymes may be significant. Peroxynitrite from NO increases prostaglandin synthesis [40], and S-nitrosylation of COX-2 activates the enzyme and contributes to cell injury [41,42]. The impact of the co-induction of iNOS and COX-2 in endothelium requires further investigation.
The role of calpain activity may also be complex. The most well-studied calpains, heterodimeric μand m-calpain, can cleave numerous protein substrates, and enhance or down-regulate different signal transduction processes. Excessive calpain activity can also cause cell injury and death in several organs, which can be reduced with calpain inhibitors [39,43]. Thus, it remains to be determined Effect of PIN1 knockdown and calpain inhibition on COX-2 stability Figure 6 Effect of PIN1 knockdown and calpain inhibition on COX-2 stability. KD and Control shRNA MAEC were treated with vehicle (DMSO) or 25 μM zVF for 1 h, then with LPS and IFN for 24 h. Cycloheximide (90 μg/ml) was added, and cell extracts were collected at the indicated times and western blotted for COX-2 and α-tubulin. A, Representative images of COX-2 and α-tubulin. Blots were processed in the same reagents for each protein, and exposed on one film for all samples. B, The average COX-2/α-tubulin signal intensity ratio ± SE of 4 independent cultures for each point, as a percent of the value at 0 h after cycloheximide treatment, is shown. The dashed line marks the 50% value. *, p < 0.05 for comparison between vehicle-and zVF-treated KD cells or Control cells at the indicated time. +, p < 0.05 for comparison between similarly treated KD and Control cells at the indicated time. Extracts were then mixed with the indicated units of porcine μ-calpain, in calpain reaction buffer, incubated 30 min, and then denatured for western blotting. A representative blot of COX-2 is shown. Bars represent the average COX-2 signal intensity ± SE of 3 independent cultures for each point, as a percent of the value without added calpain.
whether PIN1 or specific calpains in endothelial cells can be exploited to manipulate inflammatory activation in a therapeutically useful manner. In any case, the results here indicate that COX-2 is degraded by calpain, and that PIN1 regulates its expression via effects on calpain activity in MAEC.

Conclusion
Depletion of PIN1 increased induction of COX-2 by LPS/ IFN by a post-transcriptional mechanism associated with reduced calpain activity. Consistent with the short lifespan of COX-2 in MAEC, suppression of PIN1 and calpain inhibitors increased its induction. This previously unknown connection suggests that PIN1 may normally function to maintain calpain activity and, consequently, restrain the induction COX-2, iNOS, and perhaps other substrates in MAEC. PIN1 is likely to regulate a range of calpain-dependent endothelial activities.

List of abbreviations
COX: cyclooxygenase; LPS: E. coli endotoxin; iNOS: inducible nitric oxide synthase; IFN: interferon-γ; KD: knockdown; MAEC: murine aortic endothelial cells; PIN1: Protein Never in Mitosis Gene A Interacting-1; shRNA: short hairpin RNA; zVF: carbobenzoxy-valinyl-phenylalaninal Effect of PIN1 knockdown on calpain and cathepsin activity Figure 8 Effect of PIN1 knockdown on calpain and cathepsin activity. Activities were determined from the initial rate of cleavage of fluorogenic substrates for calpain (A), or cathepsin (B). These were assessed in extracts of KD and Control cells treated with medium or LPS/IFN for 24 h. Bars represent mean + SE fluorescence increase/min for 4 independent cultures in each group. *: p < 0.05 for comparison with medium-treated cells. +: p < 0.05 for comparison between KD and Control MAEC treated in the same way.