A positive feedback loop of phosphodiesterase 3 (PDE3) and inducible cAMP early repressor (ICER) leads to cardiomyocyte apoptosis

Materials and Methods

For further details, see Supporting Text and Figs. 9-14, which are published as supporting information on the PNAS web site.

Animal Models. The rat Ang II hypertension model was induced in Sprague-Dawley rats by infusing Ang II (0.7 mg/kg per day) for 7 days via s.c. osmotic minipumps (model 2001; Alzet) as described (13). Control rats were infused with vehicle. Systolic and diastolic blood pressures and heart rate were measured in conscious rats with tail-cuff plethysmography (Visitech, Apex, NC).

Comparative Genomics and Luciferase Reporter Assay. Ten-kilobase regions surrounding the human and mouse PDE3A loci were downloaded by using the University of California, Santa Cruz, Genome Browser (http://genome.ucsc.edu). Identification of conserved regions was performed by using vista (www.gsd.lbl.gov/vista.index.shtml) as described (14). Promoter constructs for the human PDE3A promoter were designed to encompass ≈1, 2, or 3 kb of PDE3A upstream conserved regions identified by vista analysis. All constructs were PCR amplified from BAC human RP11 1023C8 (Invitrogen) and subcloned into pGL3-Basic (Promega) in the upstream of luciferase. Promoter activity of each construct was measured by luciferase activity in HEK293 cells as described (15).

Results

Sustained Effects of Ang II on PDE3A Down-Regulation and ICER Up-Regulation. We have previously found that Ang II treatment decreased PDE3A expression and concomitantly increased ICER expression in cultured rat neonatal cardiomyocytes, which is responsible for Ang II-induced cardiomyocyte apoptosis (12). It is critical to determine whether Ang II can regulate PDE3A and ICER expression as well as apoptosis in vivo in adult hearts. Therefore, PDE3A and ICER protein levels and cardiomyocyte apoptosis were evaluated in hearts from adult rats treated with Ang II (0.7 mg/kg per day) continuously for 7 days via osmotic minipump. Systolic blood pressure was significantly increased in Ang II-infused rats compared with control rats (182 ± 7 mmHg vs. 121 ± 6 mmHg, P < 0.01; 1 mmHg = 133 Pa) with no significant changes of body weight. We found that Ang II infusion induced PDE3A down-regulation and ICER up-regulation in rat hearts (Fig. 1A), similar to our observations in cultured rat neonatal cardiomyocytes (12). In addition, cardiomyocyte apoptosis measured by TUNEL-positive cells was also significantly increased in adult myocytes in vivo after Ang II infusion (Fig. 1B), consistent with previous findings in Ang II infused animals (16). We also observed a time-dependent down-regulation of PDE3A in rat heats with Ang II infusion for 1, 3, and 7 days (Fig. 9). These results suggest that Ang II promotes cardiomyocyte apoptosis via similar mechanisms in neonatal cardiomyocyte in vitro and adult myocardium in vivo.

Interestingly, we noticed that both PDE3A down-regulation and ICER up-regulation were sustained after Ang II stimulation in vitro (12) and in vivo (Fig. 1 A). To determine the possibility of an autoregulatory positive feedback mechanism responsible for the sustained change of PDE3A and ICER after Ang II stimulation, we first examined the effects of Ang II washout on sustained PDE3A reduction and ICER induction. As shown in Fig. 1C, Ang II treatment for 12 h resulted in PDE3A reduction and ICER induction and such changes were persistent for at least an additional 72 h even after Ang II washout. Furthermore, we found that blockade of the AT1R via the selective antagonist, losartan, before Ang II stimulation abolished the Ang II effects on PDE3A and ICER. In contrast, blockade of the AT1R 24 h after Ang II stimulation did not influence the Ang II effects on PDE3A and ICER (Fig. 1D). Because activation of the AT1R by Ang II is critical for the initiation but not the continuation of changes in PDE3A and ICER, these data strongly support the existence of an autoregulatory positive mechanism that maintains persistent changes of PDE3A and ICER expression by Ang II.

Role of ICER as a Repressor of PDE3A Expression. Based on our observation of concomitant and reciprocal changes of PDE3 and ICER expression by Ang II both in vitro and in vivo (Fig. 1) as well as in failing hearts (12), we hypothesize that PDE3A and ICER might mutually regulate each other. ICER might serve as a transcriptional repressor of the PDE3A gene, and PDE3A down-regulation by ICER could cause the activation of a signaling pathway that potentiates ICER induction; this constitutes an autoregulatory positive feedback loop between PDE3A and ICER (PDE3A-ICER feedback loop) (see Fig. 8).

To prove the existence of the PDE3-ICER feedback loop, we first determined whether ICER has a role in the regulation of PDE3A expression. As shown in Fig. 2, adenovirus-mediated overexpression of ICER-WT is sufficient to decrease PDE3A protein (Fig. 2 A) and mRNA (Fig. 2B), supporting the concept that ICER may function as a repressor of PDE3A gene expression.

To further determine whether ICER suppresses PDE3A gene transcription, we examined the effects of ICER on PDE3A gene promoter activity. First, the promoter and putative important regulatory regions were analyzed in the 10-kb 5′-flanking region of human PDE3A gene, which was determined based on the conserved regions from cross-species comparison by VISTA plot. As shown in Fig. 2C, multiple highly conserved regions between human and mouse were localized in the proximate 3-kb 5′ upstream flanking region of human PDE3A gene. Next, we constructed several luciferase reporters encompassing ≈1, 2, and 3 kb of the 5′-flanking region of human PDE3A gene (named as P-1kb, P-2kb, and P-3kb, respectively) (Fig. 2D). These constructs were tested in human HEK 293 cells to determine which region is regulated by ICER. As shown in Fig. 2D, the luciferase activities were modestly increased in P-2kb and P-3kb but increased to a greater degree in P-1kb, suggesting that there is a functional promoter region in the -1 kb region. Because the luciferase activities are similar for P-2kb and P-3kb but much lower than P-1kb, it suggests that there is negative regulatory element(s) in the region from -1 to -2 kb. Overexpression of ICER dose-dependently decreased luciferase activity in all three constructs to a similar extent, suggesting that the ICER regulatory element is localized within the -1 kb region.

To determine the role of ICER in regulating Ang II-mediated persistent PDE3 reduction, ICER induction was blocked by adenovirus-mediated overexpression of antisense ICER (ICER-AS). As shown in Fig. 3, overexpression of ICER-AS, either before Ang II treatment (Fig. 3A) or 24 h after Ang II stimulation (Fig. 3B), reversed PDE3A reduction, demonstrating that ICER is critical for maintenance of Ang II-induced PDE3A reduction.

Role of PDE3A Enzymatic Activity in the Regulation of ICER Expression. Next, we determined the specific role and mechanism of PDE3A in the regulation of ICER expression. As shown in Fig. 4A, we first examined the effect of decreased PDE3A expression on ICER expression using adenovirus-mediated expression of rat PDE3A antisense (PDE3A-AS) (12). Down-regulation of PDE3A expression by PDE3A-AS induced sustained ICER expression.

Because we reported that inhibition of PDE3 activity with PDE3 inhibitors induced ICER expression (12), it would be predicted that inhibition of PDE3 activity by PDE3 inhibitors leads to down-regulation of PDE3A expression via ICER-mediated repression of PDE3A gene expression. As predicted, we found that inhibition of PDE3 enzyme activity by milrinone caused a sustained down-regulation of PDE3A protein levels in a time- (Fig. 4B) and dose-dependent manner (Fig. 9), accompanied by sustained ICER up-regulation. The down-regulation of PDE3A protein by PDE3 inhibitors was caused by the down-regulation of PDE3A mRNA (Fig. 4C), consistent with ICER-mediated suppression of PDE3A transcription. Similarly, another structurally different PDE3 inhibitor cilostamide also caused PDE3A protein down-regulation (see Fig. 10).

To further confirm the importance of PDE3 enzymatic activity in the regulation of ICER expression, we examined the effects of wild-type PDE3A1 and a catalytically inactive mutant of PDE3A1 (with two histidine residues H836 and H840 in the PDE signature mutated to alanines). As shown in Fig. 4D, Ang II-induced ICER expression was blocked by overexpression of wild-type PDE3A1, but not the catalytically inactive PDE3A1.

To determine the molecular mechanism responsible for PDE3 inhibition-induced down-regulation of PDE3A expression, we focused on PKA, CREB, and ICER because a decrease in PDE3A activity causes elevation of cAMP (12, 17), which may stimulate PKA and subsequent CREB activation, then induce ICER expression (10). To determine the roles of PKA, CREB, and ICER, we used adenovirus-mediated expression of PKI (a 75-aa endogenous PKA inhibitor), DN-CREB (a dominant negative form of CREB, with phosphorylation site serine133 mutated to alanine), and ICER-AS (antisense ICER-IIγ). As shown in Fig. 4E, PKI, DN-CREB, and ICER-AS blocked ICER up-regulation and PDE3A down-regulation induced by PDE3 inhibitor, milrinone, indicating the critical roles of PKA, CREB, and ICER in PDE3 activity inhibition-induced down-regulation of PDE3A expression.

Mechanisms in the Initiation and Continuation of the PDE3A-ICER Feedback Loop. Thus far, our data demonstrated that PDE3A and ICER mutually regulate each other directly or indirectly. ICER functions as a repressor of PDE3A gene expression (Fig. 2) and PDE3A reduction induces ICER expression probably via activation of PKA and CREB (Fig. 3), which forms an autoregulatory positive feedback loop. Because AT1R activation by Ang II is involved only in the initiation but not the continuation of PDE3A down-regulation and ICER up-regulation (Fig. 1), it suggests that the feedback loop might be initiated and sustained through different signaling mechanisms. Therefore, we investigated the roles of classical PKC, PKA, and CREB in the initiation and continuation of Ang II-induced PDE3A-ICER feedback loop by inhibiting PKC, PKA, and CREB before (Fig. 5 A and C) or after (Fig. 5 B and D) Ang II stimulation. To determine the role of classical PKC, we used a myristolated PKCβ C2-4 inhibitor (Biomol) that inhibits the translocation and function of all classical PKC (cPKC) isoforms (referred as to cPKC inhibitor). PKA and CREB function was inhibited by adenovirus-mediated overexpression of PKI and DN-CREB, respectively. We found that pretreatment with cPKC inhibitor (Fig. 5A) or preexpressing PKI (Fig. 5A) or DN-CREB (Fig. 5C) before Ang II stimulation abolished Ang II-induced PDE3A down-regulation and ICER up-regulation. However, after 24 h of Ang II stimulation, the cPKC inhibitor (Fig. 5B) and DN-CREB (Fig. 5D) could not prevent the changes of PDE3A and ICER expression. In contrast, 24 h after Ang II treatment PKI expression inhibited Ang II-mediated PDE3A down-regulation and ICER up-regulation (Fig. 5B). These data show that cPKC and CREB activation are important for the initiation but have limited effects on the continuation of Ang II-induced PDE3A-ICER feedback loop. In contrast, PKA activation is important for both initiating and maintaining the PDE3A-ICER feedback loop.

Role of PKA in the Regulation of ICER Protein and mRNA Stability. The expression level of ICER is mainly regulated by CREB-dependent ICER gene transcription as well as proteasome-dependent ICER protein ubiquitination and degradation (18). Because PKA but not CREB is necessary for maintaining PDE3A-ICER feedback loop (Fig. 5) and cAMP has been shown to attenuate ICER protein ubiquitination and degradation (18), we speculated that PKA activation due to PDE3A down-regulation may lead to ICER protein elevation by stabilizing the ICER protein. To investigate this possibility, we examined the effect of PKA and CREB activation on Ang II-induced ICER protein stability. As shown in Fig. 6A, cardiomyocytes were stimulated with Ang II for 24 h to induce ICER expression and transduced with Ad-LacZ, Ad-DN-CREB, or Ad-PKI for 6 h (the minimal time allowing a sufficient expression of LacZ, DN-CREB, or PKI, respectively; data not shown). Cells were then treated with cycloheximide for 2, 6, or 12 h to inhibit global protein translation. As shown in Fig. 6B, in the presence of lacZ, Ang II-induced ICER protein levels declined slowly, and DN-CREB did not influence ICER protein degradation as expected. In contrast, PKI significantly decreased Ang II-induced ICER protein stability, indicating that PKA plays an important role in stabilizing the ICER protein.

Role of PDE3A-ICER Feedback Loop in Ang II-Induced Cardiomyocyte Apoptosis. Because Ang II stimulates the PDE3A-ICER feedback loop that leads to persistent ICER induction, and ICER induces cardiomyocyte apoptosis (12, 19), we believe that the PDE3A-ICER feedback loop is a critical mechanism of Ang II-induced cardiomyocyte apoptosis. As shown by TUNEL staining (Fig. 7), preinhibition of AT1R, cPKC, PKA, CREB, and ICER before Ang II stimulation ablated Ang II-induced apoptosis (Fig. 7A). However, 24 h after Ang II stimulation, only the inhibition of PKA and ICER but not AT1R, cPKC, and CREB was able to reverse Ang II-induced cardiomyocyte apoptosis (Fig. 7B). In addition, Ang II washout was no longer able to prevent cardiomyocyte apoptosis (Fig. 7B). We obtained similar results with an ELISA method that detects cytoplasmic accumulation of histone-associated DNA fragments (data not shown). These results reveal that Ang II induced cardiomyocyte apoptosis in a manner that parallels with the regulation of the PDE3A-ICER feed back loop, suggesting that the PDE3A-ICER feedback loop is an important mechanism in Ang II-induced cardiomyocyte apoptosis. To verify the role of PDE3A reduction and ICER induction in the regulation of adult cardiomyocytes apoptosis, we showed that overexpression of PDE3A1 or antisense ICER in rat adult cardiomyocytes blocked Ang II-induced ICER induction, Bcl-2 reduction, and apoptosis (Fig. 11), suggesting the similarity between neonatal and adult cardiomyocytes.

Discussion

In the present study, we define a PDE3A-ICER feedback loop that is critical for sustaining proapoptotic ICER expression and subsequent cardiomyocyte apoptosis. We found that ICER transcriptionally represses PDE3A gene expression and PDE3A down-regulation feeds back to activate PKA that leads to ICER elevation via PKA-dependent stabilization of ICER protein; this constitutes an autoregulatory PDE3A-ICER positive feedback loop (Fig. 8). Ang II induces this PDE3A-ICER feedback loop by the activation of classical PKC and CREB pathways, which are also the mechanisms by which Ang II induces cardiomyocyte apoptosis. The data support the role of heterologous signaling via AT1R, in that PDE3A-dependent cAMP signaling is important for Ang II-induced cardiomyocyte apoptosis. We have also found that β-AR agonist ISO induced a similar PDE3A-ICER feedback loop, which is important for ISO-induced cardiomyocyte apoptosis (Figs. 12-14). The difference from Ang II-induced PDE3A-ICER feedback loop was that the activation of cPKC was not involved (Figs. 12 and 14). In addition, we have reported the reciprocal regulation of PDE3A and ICER in human failing hearts with both dilated cardiomyopathy and ischemic cardiomyopathy (12), in a mouse model of pressure-overload (12), as well as in a rabbit postinfarction heart failure model (unpublished observation). Taken together, these results suggest that the PDE3A-ICER autoregulatory positive feedback loop may represent a common mechanism of cAMP signaling in cardiomyocyte apoptosis and the pathologic progression of heart failure of various etiologies. This finding may provide therapeutic targets to prevent cardiomyocyte apoptosis and the development and progression of heart failure.

Under normal physiological conditions, ICER induction is transient because of negative down-regulation of its expression by competing with CREB for CRE-binding sequences in the ICER promoter (10, 20). This negative feedback mechanism is responsible for returning stimulated cells to the steady-state level and allows cAMP-dependent signaling to be activated in a cyclic fashion. Sustained elevation of exogenous ICER levels have been shown to elicit pathological consequences in various cell types, such as death of neuronal cells (20) and cardiomyocytes (12, 19). In previous studies, we have found that endogenous ICER expression in cardiomyocytes was persistently elevated under pathological circumstances, such as in failing human hearts, in mouse hearts with chronic pressure overload, and in cardiomyocytes stimulated with the proapoptotic stimuli Ang II and ISO (12). The induction of ICER is important for cardiomyocyte apoptosis because cardiomyocytes with persistently high levels of ICER achieved by ectopic ICER expression induced cell apoptosis, whereas the opposite effect was obtained in cells expressing an antisense ICER (12). ICER-mediated apoptosis occurs in part through inhibition of CREB-mediated transcription and down-regulation of Bcl-2 (12, 19).

In the present study, we further demonstrated that the sustained induction of ICER is permitted by the autoregulatory PDE3A-ICER positive feedback loop where PDE3A is a key mediator (Fig. 8). Persistent but not transient ICER induction is critical for promoting cardiomyocyte apoptosis because cardiomyocyte apoptosis was prevented by blocking ICER expression using antisense ICER 24 h after ICER has been induced (Fig. 3B). The initiation of ICER induction depends on CREB activation, likely via CREB-dependent ICER gene transcription, whereas persistent ICER induction due to the positive feedback loop is no longer dependent on CREB (Fig. 5 C and D). However, chronic PKA activation due to PDE3A down-regulation causes persistent ICER elevation by stabilizing ICER protein, which is a critical mechanism for the continuation of the PDE3A-ICER feedback loop and the subsequent cardiomyocyte apoptosis. This finding is consistent with the observation that transgenic mice with cardiac specific overexpression of constitutively active PKA developed dilated cardiomyopathy (21). The stability of ICER protein is regulated by the process of proteasome-dependent degradation (22). The precise mechanism by which PKA-regulated ICER protein stability will require further investigation; however, cAMP has been previously shown to attenuate ICER protein ubiquitination and degradation probably via cAMP-dependent inhibition of mitogen-activated protein kinase ERK1/2 that phosphorylates and targets ICER to ubiquitin-mediated destruction (18). It is, of course, very likely that the effect of cAMP on ERK1/2 is mediated by PKA, although PKA was not examined in their study (18).

The putative important regulatory regions of human PDE3A gene were determined based on the conserved regions via cross-species comparison between human and mouse. Within the 10-kb 5′ flanking region of PDE3A gene, there are clusters of conserved sequences located within the 3-kb 5′ upstream flanking region (Fig. 2C). When the 3-kb 5′ upstream flanking sequence was tested by using several reporter constructs, including P-1kb, P-2kb, and P-3kb, the promoter activity was varied with P-1kb ≫ P-2kb ≈ P-3kb (Fig. 2D), suggesting the existence of silencer(s) in the region from -1 to -2 kb. However, the region responsible for ICER appears to be located within the first 1-kb 5′ flanking region because overexpression of ICER decreased promoter activity to a similar extent for all three reporter constructs (Fig. 2D). ICER serves as very powerful repressor of CRE-mediated transcription (10). Most of the genes regulated by ICER contain a CRE sequence in their promoters. However, ICER has also been shown to inhibit gene activity by binding to sequences that are recognized by other transcription factors such as NFAT/AP1 (10). Analysis of the 5′ flanking region of PDE3A gene using Transcription Factors Database (www.ifti.org/cgi-bin/ifti/Tfsitescan.pl) revealed one putative CRE sequence and several AP1 sites within the first 1-kb 5′ flanking region. The precise regulatory mechanism and the ICER responsible site in the PDE3A regulatory region need to be further investigated.

PDE3 inhibitors, such as amrinone, milrinone, and enoximone, have been shown to improve the cardiac index and lower pulmonary capillary pressure in patients with heart failure. However, in long-term clinical trials, the hemodynamic improvements seen early in therapy were typically not sustained (23, 24). This may be explained by PDE3 inhibitor-induced down-regulation of PDE3A expression observed in this study (Fig. 4). In addition, an increase in mortality after chronic PDE3 inhibitor treatment was also found, primarily because of arrhythmias and sudden death (23, 24). Our finding that PDE3 inhibitors induce a autoregulatory positive PDE3A-ICER feedback loop leading to cardiomyocyte apoptosis may provide a molecular mechanism at least for the part of adverse effects of chronic PDE3 inhibitor therapy. The development of therapeutic strategies that intervene in the PDE3A-ICER feedback loop may allow for promoting the beneficial effects while preventing the detrimental effects of PDE3 inhibitor therapy.

Supplementary Material