Cleavage of Protein Kinase D After Acute Hypoinsulinemia Prevents Excessive Lipoprotein Lipase–Mediated Cardiac Triglyceride Accumulation

An expanded version of the research design and methods has been described in the online appendix that is available at http://diabetes.diabetesjournals.org/cgi/content/full/db09-0681/DC1. In this supplement, we have explained our animal model of hypoinsulinemia, perfusion of isolated hearts, isolation of cardiomyocytes, assay for LPL activity, adenoviral transfection of cardiac cells, Western blotting, immunoprecipitation, immunofluorescence, measurement of cardiac triglyceride, heart fractionation and estimation of lipid droplet protein, electron microscopy, plasma membrane CD36 determination, palmitate oxidation, and separation of golgi.

By exploiting two different doses of STZ, we induced moderate (55 mg/kg) and severe (100 mg/kg) hypoinsulinemia. Although both groups were equally hyperglycemic (Fig. 1,A, inset), d-100 rats showed a more profound loss of circulating insulin (Fig. 1,A). During hypoinsulinemia, excessive lipolysis and loss of adipose tissue mass results in an increase in both plasma fatty acids and triglycerides (24). Indeed, both plasma fatty acids (Fig. 1,B) and triglycerides (Fig. 1,B, inset) levels were elevated in d-100 animals. Interestingly, measurement of these parameters in d-55 showed no significant differences when compared to control subjects (Fig. 1,B). During hypoinsulinemia with its attendant loss of glucose utilization, LPL activity increases at the vascular lumen (13). Compared to control hearts, there was substantial increase in heparin-releasable LPL activity in d-55 animals (Fig. 1,C), with most of the enzyme located at the vascular lumen as shown using immunofluorescence (Fig. 1,E). Contrary to the results obtained with d-55 rats, heparin-releasable LPL activity (Fig. 1,C) and LPL immunofluorescence (Fig. 1,E) in severely hypoinsulinemic d-100 animals were unchanged when compared to control subjects. Both hypoinsulinemic groups showed no significant change in cardiac LPL gene expression (Fig. 1 D).

Recently, we have reported that phosphorylation of PKD is a key element to increase cardiac LPL after hyperglycemia (d-55) by regulating fission of vesicles from golgi membranes (18). In the present study, measurement of PKD revealed that only d-100 hearts showed a decrease in full-length PKD (Fig. 2,A, inset), with a corresponding robust increase in the cleaved form (Fig. 2,A), that increases its activity (26) but limits its golgi-binding capability (27). In U-937 leukemic monocyte lymphoma cells, caspase-3 has been suggested to control PKD by cleaving its cysteine-rich domain (26), which is necessary for binding to trans-golgi (27). Measurement of caspase-3 in d-100 rat hearts demonstrated an increase in its activity (Fig. 2,B). This increase was modest; PAC-1, a procaspase-3 activator, increased caspase-3 activity twofold higher than d-100. Transferase-mediated dUTP nick-end labeling (TUNEL) and DAPI nuclear staining were used to assess cell apoptosis. Hearts from d-100 animals exhibited a higher number of apoptotic cells. However, the overall percentage of these cells was very low (∼0.01%) (data not shown). Given the effect of caspase-3 to cleave PKD, purified glutathione S-transferase (GST)-PKD was incubated with active caspase-3 enzyme. As predicted, GST-PKD was rapidly cleaved by caspase-3 (Fig. 2,C). Replicating this experiment using PKD isolated from control hearts also showed cleavage of PKD by caspase-3. However, caspase-3 proteolysis of GST-PKD was faster than its cleavage of isolated PKD from control hearts (Fig. 2,D). Interestingly, although d-55 hearts also showed an increase in caspase-3 activity (Fig. 2,B), this was not associated with cleaved PKD (Fig. 2,A), suggesting that in d-55 hearts some protein is preventing PKD cleavage. Proteins 14–3-3 are binding proteins known to interact with a number of enzymes protecting them against proteolytic degradation (28), and diabetes is known to decrease the gene and protein expression of 14–3-3ζ (29). Measurement of 14–3-3ζ showed a decrease in its mRNA (Fig. 2,E) and protein (Fig. 2,E, inset), only in d-100 hearts. In addition, severe hypoinsulinemia also resulted in a decreased attachment of PKD to 14–3-3ζ (Fig. 2,F). Silencing of 14–3-3ζ in cardiomyocytes had no influence on PKD expression (Fig. 2,G, inset). Nevertheless, in these cells, exposure to caspase-3 accelerated PKD cleavage (Fig. 2 G, boxed inset). Our data suggest that both activation of caspase-3 together with loss of 14-3-3ζ are required for cleavage of PKD after severe hypoinsulinemia.

In cardiomyocytes, phosphomimetic (PKD-S744E/S748E) and kinase-dead mutants of PKD were overexpressed by adenoviral vectors to establish the relationship between PKD and LPL. First, we validated successful transduction using Western blotting (Fig. 3,A). Both PKD-S744E/S748E and kinase-dead mutants increased total PKD expression (Fig. 3,A, inset); only the phosphomimetic mutant demonstrated higher Ser⁹¹⁶ phosphorylation, which is related to its catalytic activity (30). The S744/S748 phosphorylation site has been shown to have a critical role in recruiting PKD to the trans-golgi (27). Simply increasing the expression of the phosphomimetic and kinase-dead mutants augmented their association with the trans-golgi (supplemental Fig. IA). However, PKD-S744E/S748E also increased the colocalization of LPL to the trans-golgi (Fig. 3,C), with a striking increase in heparin-releasable LPL activity (Fig. 3 B).

Using a different strategy, cardiomyocytes were transfected with PKD-D378A point mutated to prevent caspase-3–mediated proteolysis (26). Successful transfection of PKD wild type and PKD-D378A is demonstrated in supplemental Fig. 1B. Although PKD wild type and PKD-D378A themselves elevated PKD phosphorylation at Ser⁹¹⁶, phosphorylation of Ser^744/748 (supplemental Fig. IB) and heparin-releasable LPL activity (supplemental Fig. IC) remained unchanged compared to enhanced green fluorescent protein (EGFP). Bryostatin-1 is a potent activator of PKD (31). Control cardiomyocytes treated with bryostatin-1 increased phosphorylation (both S744/S748 and Ser⁹¹⁶) of PKD and heparin-releasable LPL activity (Fig. 3,D, inset). In wild-type and D378A cardiomyocytes, bryostatin-1 was not only capable of further increasing Ser⁹¹⁶ phosphorylation but also Ser^744/748 phosphorylation (supplemental Fig. IB). In these myocytes, bryostatin-1 also elicited a robust increase in heparin-releasable LPL activity (supplemental Fig. IC). Pretreating EGFP and wild-type cells with PAC-1 followed by bryostatin-1 induced significant cleavage of PKD, an effect that was not apparent in PKD-D378A cardiomyocytes (Fig. 3,D). More importantly, caspase-3 activation in D378A did not decrease LPL activity as seen in EGFP and wild type (Fig. 3 E). It should be noted that in this experiment, using nondiabetic myocytes 14-3-3ζ was unable to protect PKD from cleavage by PAC-1, which elicited a more robust increase in caspase-3 compared to d-100. Overall, these data suggest that LPL transport to the cell surface is dependent on PKD phosphorylation at S744/S748 rather than PKD's intrinsic activity and that cleavage of PKD by caspase-3 is sufficient to prevent PKD-mediated increase in LPL activity.

We used siRNA to silence caspase-3 in isolated cardiomyocytes. We first validated successful caspase-3 inhibition using Western blotting (Fig. 4,A, inset). In myocytes in which caspase-3 was silenced, PAC-1 was unable to increase caspase-3 activity (Fig. 4,A) or to increase PKD cleavage (Fig. 4,B). Furthermore, in myocytes in which caspase-3 was silenced, the PAC-1–induced decrease in LPL activity was prevented (Fig. 4 C).

PKD, by associating with vesicle-associated membrane protein (VAMP, one of v-SNAREs), is known to trigger vesicular movement (32). To further elucidate the interactions between caspase-3, PKD, and LPL, purified golgi were incubated in the presence or absence of GST-PKD and caspase-3 and vesicle formation determined using antibodies against VAMP. As predicted, VAMP-associated vesicle formation significantly increased on incubation with GST-PKD; the GST-PKD–induced vesicles contained high levels of LPL (Fig. 5,A and B). These effects of PKD on VAMP-associated vesicle formation and LPL content were impeded by caspase-3 (Fig. 5,A and B). We applied a similar approach in vivo, using Z-DEVD to inhibit caspase-3 in d-100 hearts. Interestingly, when the cytosolic fraction from Z-DEVD–treated d-100 hearts was incubated with isolated golgi, VAMP-associated vesicular formation together with LPL vesicle content increased significantly (Fig. 5 C and D).

Given the strong relationship between caspase-3 and LPL, d-100 rats were treated with the caspase-3 inhibitor Z-DEVD and LPL activity determined. Z-DEVD was very effective in preventing caspase-3 cleavage and its subsequent activation (Fig. 6,A). More importantly, this caspase-3 inhibitor prevented the cleavage of PKD that is seen in d-100 hearts (Fig. 6,B). This effect on PKD in Z-DEVD–treated d-100 hearts increased the appearance of LPL at the vascular lumen (Fig. 6,D), in addition to augmenting heparin-releasable LPL activity (Fig. 6,C) to levels observed in d-55 hearts. Z-DEVD had no influence on total LPL expression, which remained ∼20% lower than that in control subjects (Fig. 6 E). In addition, Z-DEVD had no further influence in increasing vascular LPL in d-55 hearts (data not shown). In summary, activation of caspase-3 in d-100 hearts restrains the transfer of LPL to the coronary lumen by affecting proteolysis of PKD.

Interestingly, the augmented amount of LPL in Z-DEVD–treated d-100 rats corresponded to an increased clearance of circulating triglyceride (Fig. 7,A), without affecting plasma fatty acids (Fig. 7,B). In these animals, unlike the drop in plasma triglyceride, there was a striking accumulation of cardiac triglyceride, which was almost twofold higher than that in the untreated d-100 animals (Fig. 7,C and D). Sucrose fractionation and subjection of the different fractions to Western blotting for OXPAT or determination of triglyceride revealed that in d-100, the fractions (top) that had highest amount of triglycerides were associated with OXPAT (supplemental Fig. 2); caspase inhibition in d-100 induced a further increase of OXPAT in these fractions (supplemental Fig. 2). Hearts from d-100 animals demonstrated an approximately twofold increase in CD36 protein (Fig. 7,E). Treatment of these animals with Z-DEVD did not initiate a further increase in plasma membrane CD36 (Fig. 7,E). Cardiac palmitate oxidation in d-100 rats treated and untreated with Z-DEVD remained high (Fig. 7 F).

During diabetes, when cardiac glucose uptake, glycolysis, and pyruvate oxidation are impaired, the heart rapidly adapts to using a greater amount of fatty acids for ATP generation, allowing it to maintain its function (33). This change occurs not only as a consequence of increased exogenous fatty acids (because of release of fatty acids from adipose tissue and hydrolysis of triglyceride-rich lipoproteins by LPL) and endogenous fatty acids (hormone-sensitive lipase hydrolysis of intracellular stored triglycerides) supply but also through an intrinsic adaptation or maladaptation to elevated fatty acids that includes an augmented fatty acid oxidation and expression of genes that control utilization of this substrate (33). To avoid oversupply of this potentially lethal substrate, the heart can exquisitely regulate its LPL content at the vascular lumen. Thus, following moderate acute STZ diabetes (that resembles a poorly controlled type 1 diabetes patient), when circulating fatty acids and triglycerides have yet to increase, significantly elevated luminal LPL protein and activity is observed (14). Given that vascular endothelial LPL is acquired from the cardiomyocyte, we previously examined the mechanisms by which diabetes increases cardiomyocyte cell surface LPL, a reservoir that can rapidly augment coronary luminal LPL when the need for fatty acids is intensified. In the myocyte, we reported that recruitment of LPL to the cell surface was controlled by PKD (that regulated LPL vesicle fission) (18), whereas stress kinases like AMPK and p38 MAPK allowed for actin cytoskeleton polymerization, providing the cardiomyocyte with an infrastructure to facilitate LPL movement (20). In the present study, our data demonstrate that after severe hypoinsulinemia (with its attendant elevation of plasma fatty acids and triglycerides), caspase-3 activation, by cleaving PKD, restricts amplification of LPL, thereby limiting fatty acid provision and overwhelming cardiac triglyceride accumulation.

The cytosolic serine-threonine kinase, PKC-μ, also known as PKD, regulates vesicular fission from the trans-golgi network, allowing cargo proteins to be delivered to the plasma membrane (27). Within PKD, a cysteine-rich domain has been shown to have a critical role in recruiting and binding PKD to the trans-golgi (27). We have recently reported that PKD is a key modulator of LPL vesicular trafficking within cardiomyocytes and is related not only to binding of PKD to the golgi (18) but also to phosphorylation of Ser^744/748 in its catalytic domain. In d-55 animals, PKD phosphorylation was strongly correlated to an increase in LPL on the luminal surface of the capillary endothelium. Given the observation that in hearts from d-100 animals LPL activity remained unchanged after 4 days or decreased after 1 week (14,25), we hypothesized that a signal deficiency for LPL translocation exists within these hearts. Interestingly, measurement of PKD revealed that only d-100 hearts showed a decrease in full-length PKD, with a corresponding robust increase in the cleaved form. Our data suggest that following severe hypoinsulinemia, proteolysis of PKD may be an important event to limit an increase in LPL activity at the vascular lumen.

In U-937 cells, caspase-3 has been suggested to control PKD by cleaving its cysteine-rich domain (26). Given the relationship between caspase-3 and PKD, we measured caspase-3 and determined that both d-55 and d-100 hearts demonstrated an increase in its activity, likely mediated by reactive oxygen species (34). It should be noted that the observed increase of caspase-3 in these hearts did not reach the upper limit seen with PAC-1, the caspase-3 activator that induces apoptosis (programmed cell death). Thus, although caspase-3 plays a key role in apoptosis (35) in the acute hypoinsulinemic heart, caspase-3 may also have a nonapoptotic function. Other studies have also suggested that caspase-3 has a regulatory function in cell proliferation, differentiation, and shaping (36). This secondary function of caspase-3 in d-100 hearts included cleavage of PKD, an observation that was absent after moderate hypoinsulinemia, even though these hearts showed an increase in caspase-3 activity. Given that 1) measurement of 14-3-3ζ showed a decrease in its mRNA and protein only in d-100 hearts, 2) caspase-3's proteolysis of cardiac PKD was slower than its cleavage of GST-PKD, and 3) silencing of 14-3-3ζ in cardiomyocyte accelerated PKD cleavage on exposure of these cells to caspase-3, our data suggest that this 14-3-3ζ binding protein protects PKD against proteolytic degradation by caspase-3 in d-55. Using control myocytes, we confirmed the relationship between PKD, LPL, and caspase-3. Thus, the phosphomimetic PKD colocalized with LPL in the trans-golgi, facilitating LPL secretion to cardiomyocyte cell surface. In addition, PKD mutated to prevent its cleavage by caspase-3 or silencing of caspase-3 was able to increase LPL activity. Overall, our data suggest that caspase-3, by cleaving PKD, limits LPL vesicle formation and eventual increase of LPL at the vascular lumen.

Z-DEVD-fmk is a potent, irreversible, and cell-permeable inhibitor of caspase-3. Using this compound in d-100 animals, we effectively lowered caspase-3 activity, prevented PKD cleavage, and increased LPL vesicle formation and translocation to the vascular lumen. This increase in LPL at the vascular lumen occurred even though the total LPL protein expression remained ∼20% lower than that in control subjects. It should be noted that electron microscopy using immunogold labeling established that in the heart, 78% of total LPL is present in cardiac myocytes, 3–6% in the interstitial space, and 18% at the coronary endothelium (6). Thus, LPL at the vascular lumen is likely more dependent on the signaling that promotes transfer of this enzyme from the cardiomyocyte to the coronary lumen rather than its absolute cardiomyocyte content.

Cardiac LPL is a major determinant of plasma triglyceride (37). Thus, the increase in cardiac luminal LPL in Z-DEVD–treated d-100 animals could explain the decline in circulating triglyceride. Whether other tissue LPL contributes toward this observed fall in plasma triglyceride is unknown because heparin perfusion to release vascular-bound LPL is difficult to accomplish in other tissues. As no apparent change was noted in plasma fatty acid levels in these animals, our data suggest that following LPL-mediated triglyceride hydrolysis, fatty acids can be taken rapidly and directly into tissues. In support of this suggestion, cardiac- and skeletal muscle–specific overexpression of LPL decreased plasma triglyceride and elevated triglyceride storage in muscle tissue but was without effect when plasma fatty acid was measured (38). In the event of an increased availability of fatty acid, its disposal can occur either via oxidation or via its conversion into triglyceride, which serves as the primary storage form of fatty acids. Caspase-3 inhibition did not alter the high fatty acid uptake and oxidation in d-100 rats. However, measurement and visualization of cardiac triglyceride revealed a striking accumulation in Z-DEVD–treated d-100 hearts, which was almost twofold higher than that in the untreated d-100 animals. This triglyceride was richly associated with the lipid-storage droplet protein, OXPAT (39). Overall, these data suggest that in d-100 hearts treated with Z-DEVD, fatty acid uptake and oxidation are likely operating at their maximum capacity. The additional triglyceride accumulation in this group is most likely a consequence of an increase in LPL and vascular lipolysis, with the released fatty acids rapidly taken up and stored as triglycerides with the assistance of OXPAT. Other studies have also reported excessive triglyceride overload when LPL is specifically overexpressed in the heart (40). It should be recognized that PKD has also been linked to GLUT4-mediated glucose uptake in contracting myocytes (41). However, as STZ diabetes rapidly decreases the myocardial content of total membrane GLUT4 (t_1/2 of 3.5 days) (42), the impact of PKD in regulating glucose uptake is likely minimal under conditions of hypoinsulinemia and hyperglycemia.

In summary, severe hypoinsulinemia with its associated increase in circulating fatty acids and triglycerides lowers 14-3-3ζ expression and activates caspase-3, which can cleave PKD, reduce LPL vesicle fission, and restrict LPL translocation to the vascular lumen (Fig. 7,G). When caspase-3 is inhibited, this compensatory response is lost, leading to accumulation of triglycerides (Fig. 7 G). It should be noted that in transgenic rabbits that have global overexpression of LPL, attenuation of hypertriglyceridemia was observed, an effect suggested to contribute toward amelioration of insulin resistance and obesity (43). Contrary to systemic overexpression, tissue-specific overexpression of LPL in skeletal muscle and heart is associated with lipid oversupply, insulin resistance, and severe myopathy, characterized by muscle fiber degeneration, extensive proliferation of mitochondria and peroxisomes, excessive dilatation, and impaired left ventricular systolic function (18,38,40,44,45). Interestingly, drugs such as pioglitazone that are known to decrease caspase-3 (46) have been reported to reduce circulating triglycerides (47,48) by increasing LPL in humans (49) and to increase the myocardial content of lipids in rats fed a high-fat diet (50). Whether this effect can explain the risk of heart failure when thiazolidenediones are used in patients with type 2 diabetes has yet to be determined (51,52). Thus, although caspase inhibition has been considered to have therapeutic effects in many human diseases, including heart attack, neurodegenerative diseases, and autoimmune disorders, its use in diabetes to prevent cardiac cell death could lead to significant cardiac triglyceride accumulation and warrants further investigation.

This study was supported by an operating grant from the Canadian Diabetes Association. M.S.K. is the recipient of Doctoral Student Research Awards from both the Canadian Diabetes Association and the Heart and Stroke Foundation of Canada.

No potential conflicts of interest relevant to this article were reported.

Parts of this study were presented in abstract form at the 69th Scientific Sessions of the American Diabetes Association, New Orleans, Louisiana, 5–9 June 2009.