Severity of Diabetes Governs Vascular Lipoprotein Lipase by Affecting Enzyme Dimerization and Disassembly

The investigation conforms to the Guide for the Care and Use of Laboratory Animals published by National Institutes of Health and the University of British Columbia. Adult male Wistar rats (250–320 g) were used. To induce acute hyperglycemia, diazoxide (DZ), a selective, ATP-sensitive K⁺ channel opener was administered intraperitoneally at 100 mg/kg. DZ causes a rapid decrease of insulin secretion within 1 h, and the animals remain hyperglycemic for 4 h (28–30). Four hours after DZ, animals were killed and hearts were removed. For chronic diabetes, streptozotocin (STZ), a β-cell–specific toxin was administered as a single dose intravenously to more closely mimic type 1 diabetes (12,13). STZ was given at doses of 55 (D55) or 100 (D100) mg/kg. With D55, animals are insulin deficient but do not require insulin supplementation for survival (moderate diabetes). Unlike D55, D100 animals also develop hyperlipidemia and are not able to survive beyond 4 days without insulin supplementation (severe diabetes). STZ animals were kept for 4 days before hearts were removed.

Hearts were isolated and perfused retrogradely (14). To measure coronary LPL, perfusion solution was changed to buffer containing heparin (5 units/mL). Perfusion effluent was collected to measure LPL activity. Subsequent to LPL displacement, hearts were used for heparin–sepharose chromatography. To evaluate the effect of Angptl-4 on LPL at the vascular lumen, hearts from control and D55 animals were perfused with or without 1 ng/mL purified Angptl-4 for 1 h in a recirculating mode. Perfusates were collected and run on a heparin–sepharose column. After Angptl-4 perfusion, the hearts were perfused with heparin to release LPL activity remaining at the vascular lumen.

Ventricular, calcium-tolerant myocytes were prepared by a previously described procedure (31).

The glucosidase inhibitor castanospermine (Cs; 50 mmol/L, 2 h) was used to interrupt the association between LPL and calnexin. Colocalization of LPL and calnexin was observed using immunofluorescence. LPL activity released into the medium and remaining in the myocytes was also measured. Myocytes were also exposed to 25 mmol/L glucose (high glucose [HG]) and 1.0 mmol/L palmitic acid (PA) bound to 1% BSA (HG+PA) for 2 h. Cells were then lysed and run on a heparin–sepharose column. To evaluate the impact of HG+PA on processing of newly synthesized LPL, cells were preincubated with 50 μmol/L cyclohexamide (CHX; 1 h) to inhibit protein synthesis. Subsequently, cells were washed and treated with HG+PA for another hour, and LPL activity released into the media was determined. Calnexin and LPL colocalization was also visualized at the indicated times. To study the effect of FA on Angptl-4 expression, isolated myocytes were treated with 1.0 mmol/L PA for 4–24 h, and Angptl-4 mRNA was determined using real-time PCR. To study the direct effect of Angptl-4 on cardiomyocytes, isolated cardiomyocytes from control animals were incubated with Angptl-4 for the indicated times, and LPL activity in the media and on the cell surface (by incubating cardiomyocytes with media containing 8 units/mL heparin for 1 min) was determined. In a separate experiment, at the end of treatment, cell lysates were loaded onto a heparin–sepharose column, and dimeric LPL was determined.

Heparin–sepharose chromatography was carried out as described previously to separate dimeric LPL from monomers (20,27). Heart perfusates or equal amounts (total protein) of tissue homogenates or cell lysates were loaded onto a HiTrap HP column and sequentially eluted with 0.25, 0.75, 1.0, and 1.5 mol/L NaCl at 0.2 mL/min. Fractions (1 mL each) collected were used to detect LPL activity and protein by ELISA (USCNLIFE, Wuhan, China) or Western blot after trichloroacetic acid (TCA) precipitation. To determine the total amount of monomeric or dimeric LPL, fractions from 0.75 or 1.0 mol/L NaCl were combined before TCA precipitation. To determine the processing stage of dimeric LPL, aliquots from the 1.0 mol/L NaCl fractions were subjected to endoglycosidase H (endo H) digestion. Digested products of LPL were visualized by Western blotting after TCA precipitation.

After the indicated treatments, cells were double stained with Alexa635 and Alexa488 to colocalize LPL (red) and calnexin (green), respectively. 4,6-Diamidino-2-phenylindole (DAPI) was used to stain nuclei. Slides were visualized using a confocal microscope.

LPL from vascular lumen was obtained by perfusing hearts from D55 animals with heparin. Purified human Angptl-4 was incubated with perfusates containing peak LPL activity at a final concentration of 1 ng/mL, 0.1 μg/mL, and 1 μg/mL. The reaction mix was incubated at 37°C, and at the indicated times, 100 μL reaction solution was removed for measuring LPL activity.

LPL catalytic activity was determined by measuring the in vitro hydrolysis of a sonicated [³H]triolein substrate emulsion (15). The standard assay conditions were 0.6 mmol/L glycerol tri[9,10-³H]oleate (1 mCi/mmol; 1 Ci = 37 GBq), 25 mmol/L piperazine-N,N’-bis(2-ethanesulfonic acid) (pH 7.5), 0.05% (wt/vol) albumin, 50 mmol/L MgCl₂, and 2% (vol/vol) heat-inactivated chicken serum as a source of apo-CII. The reaction mix was incubated at 30°C for 30 min. Reaction product, sodium [³H]oleate, was extracted and determined by liquid scintillation counting. Results are expressed as nmol/h/mL (perfusate, elution fraction, or media) or nmol/h/mg protein (cell lysate).

LMF1 and Angptl-4 mRNA levels were determined using SYBR Green real-time quantitative PCR (Roche, Indianapolis, IN). Primer sequences were as follows: LMF1 forward, 5′-TGATCCTGCAGGGCACA-3′; LMF1 reverse, 5′-GTCCAGGCGGTAGTGGTA-3′ (32); Angptl-4 forward, 5′- CTCTGGGATCTCCACCATTT-3′; Angptl-4 reverse, 5′-TTGGGGATCTCCGAAGCCAT-3′ (33). All values obtained were normalized to 18S ribosomal RNA.

Blood samples were collected from the tail vein before termination and serum isolated. Concentrations of nonesterified fatty acid (NEFA) and TG were determined using diagnostic kits (Wako, Osaka, Japan and Stanbio, Boerne, TX). Angptl-4 was analyzed using an ELISA kit from Raybiotech (Burlington, CA).

DZ, STZ, PA, Cs, and CHX were obtained from Sigma (St. Louis, MO). [³H]triolein was purchased from Amersham Canada (Ontario, Canada). Purified human Angptl-4 protein (Genway Biotech, San Diego, CA) is a human recombinant and 100% homologous to the 26–229 amino acid sequence of human Angptl-4. Anti-calnexin antibody was from Stressgen (Victoria, British Columbia, Canada). Anti-LPL 5D2 antibody was a gift from J. Brunzell, University of Washington (Seattle, WA). It is a monoclonal antibody recognizing LPL from different species, but not related lipases like hepatic lipase (34). Anti-rabbit True Blot was from eBioscience (San Diego, CA). All other antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA).

Values are means ± SE. Wherever appropriate, one-way ANOVA followed by the Bonferroni test was used to determine differences between group mean values. The level of statistical significance was set at P < 0.05.

Injection of DZ resulted in stable hyperglycemia within 4 h. These animals also exhibited an increase in circulating levels of NEFA. Although administration of STZ at both doses produced hyperglycemia of equal intensity as that seen with DZ, only D100 animals showed a robust elevation in circulating FA and TG levels (Table 1).

LPL is an enzyme present in two forms based on its affinity to a heparin–sepharose column. Monomeric LPL has low affinity to the column, is predominantly eluted in 0.4–0.75 mol/L NaCl fractions, and has no catalytic activity. Active LPL is a homodimer eluted at higher concentrations of NaCl (1.0–1.5 mol/L) (20,27). We found that the majority of LPL in the heart is monomeric (eluted at 0.75 mol/L NaCl), with only ∼30% of the enzyme in the form of dimers (eluted in 1.0 mol/L NaCl fractions) accounting for all of the LPL activity (Supplementary Fig. 1). Consistent with our previous studies, total LPL protein remained unchanged after acute hyperglycemia induced by DZ (Fig. 1A). However, when equal amounts of homogenate protein were loaded onto a heparin–sepharose column, the 1.0 mol/L fractions from DZ hearts demonstrated higher LPL activity (Fig. 1B) and protein (Fig. 1C), suggesting the presence of more dimeric LPL in these hearts.

Dimeric LPL could be located either at the vascular lumen or within cardiomyocytes. To determine vascular LPL, hearts were perfused with heparin, and LPL activity was measured in the perfusate. DZ hearts possessed increased LPL activity at this site (Fig. 2A). After stripping of the dimeric enzyme from the vascular lumen with heparin, hearts with predominantly myocyte LPL were then subjected to a heparin–sepharose column. Interestingly, in addition to the vascular lumen, DZ hearts also harbored increased dimeric LPL in cardiomyocytes (Fig. 2B, lane 1 vs. 3). Maturation of LPL including dimerization occurs within the ER of cardiomyocytes. To further localize myocyte LPL dimers, we used endo H to digest dimeric LPL. Endo H cleaves the glycan chains of LPL based on their processing stages in different cellular organelles; the 57- and 55-kDa products represent LPL that has either passed or is present within Golgi apparatus, respectively, whereas the 52-kDa band refers to LPL undergoing processing inside the ER (19,20). Interestingly, DZ hearts had a higher proportion of dimeric LPL at a post-ER processing stage (Fig. 2B, lane 2 vs. 4), implicating augmented maturation of LPL after DZ. To examine whether this is due to acute exposure to a hyperglycemic and hyperlipidemic environment seen in DZ animals, isolated cardiomyocytes were incubated with HG+PA for 2 h. Using the same conditions, we have previously found an accelerated LPL trafficking from Golgi to the myocyte surface. HG+PA also increased dimerization of LPL (Fig. 2C), indicating that processing of LPL could be a prerequisite for trafficking.

LPL is a glycan protein whose processing into active dimers requires its association with ER-resident chaperones like calnexin/calreticulin to allow proper folding. Using immunofluorescence, LPL and calnexin were found to be intensively colocalized at the perinuclear region (Fig. 3A). LPL association with this chaperone was further confirmed by immunoprecipitation (Fig. 3A, inset). Cs is a glucosidase inhibitor that blocks trimming of LPL glycan chains, such that it cannot be recognized by calnexin (22). Using Cs, we were able to inhibit the association between LPL and calnexin; a complete absence of the perinuclear colocalization of these proteins was observed (Fig. 3B, inset). This hindered LPL secretion and processing; both secreted (Fig. 3B) and intracellular (Fig. 3C) LPL activity declined in the presence of Cs, likely due to reduced LPL dimerization (Fig. 3C, inset). To determine whether HG+PA augmented LPL processing by affecting the association between calnexin and newly synthesized LPL, cardiomyocytes were pretreated with the protein synthesis inhibitor CHX. After 1 h, as a result of no newly synthesized LPL entry into the ER, CHX abolished the association between LPL and calnexin (Fig. 3D, inset 2). Upon removal of CHX and renewal of protein synthesis, association between newly synthesized LPL and calnexin was visualized in the presence or absence of HG+PA. One hour after re-entry of LPL into the ER, newly synthesized LPL was observed colocalized with calnexin at the perinuclear region (Fig. 3D, inset 3). However, in contrast to Cs, even though HG+PA attenuated this colocalization (Fig. 3D, inset 4), more LPL was observed present at the cell surface, suggesting that more enzyme had completed processing under these conditions. Increased LPL processing was also reflected by more active LPL being released into the media (Fig. 3D).

To examine LPL processing subsequent to chronic diabetes, 55 mg/kg of STZ was administered to generate “moderate diabetes”. Similar to DZ, although chronic diabetes did not increase total LPL protein (data not shown), D55 hearts also demonstrated higher amounts of dimeric LPL (Fig. 4A). The importance of calnexin in LPL processing is also suggested in D55 hearts; augmented LPL dimerization in these hearts was accompanied by increased calnexin expression (Fig. 4B). However, we did not observe the same increase in calnexin when cardiomyocytes were acutely exposed to HG+PA (data not shown). Another crucial factor for LPL dimerization, LMF1, was found to be unchanged after D55 (Supplementary Fig. 2). Overall, our data suggest that after diabetes, more LPL is processed into its active dimeric form, which appears to involve the ER chaperone calnexin.

Increasing the dose of STZ to 100 mg/kg (D100) produced hyperglycemia similar to that observed in D55 animals. However, unlike D55, D100 animals developed severe hyperlipidemia (Table 1). Additionally, in contrast to D55, D100 animals showed a decline in dimeric LPL, both in the whole heart (Fig. 5A) and at the vascular lumen (Fig. 5B). This effect on LPL was not a result of decreased calnexin expression in D100 (Fig. 5C) but was accompanied by a robust increase in Angptl-4 serum concentrations (Fig. 5D). Furthermore, cardiac expression of Angptl-4 protein (Fig. 5E) and mRNA (Fig. 5E, inset) was also found to be higher in D100 animals.

To test the direct inhibitory effect of Angptl-4 on LPL, dimeric LPL collected from the vascular lumen of D55 animals was incubated with purified Angptl-4 in vitro. The inhibition of LPL activity by Angptl-4 was fast (within 15 min) and dose dependent, with 1 μg/mL Angptl-4 inhibiting ∼85% of the enzyme activity at 2 h (Fig. 6A). It should be noted that the serum concentration of Angptl-4 in D100 animals (1 ng/mL) was effective in inhibiting ∼20% of LPL activity in vitro (Fig. 6A). When this concentration of Angptl-4 was perfused through D55 hearts for 1 h, LPL activity remaining at the vascular lumen was reduced (Fig. 6C). Interestingly, this reduction in LPL activity correlated with an increased appearance of monomeric LPL in the perfusion buffer (Fig. 6B) and a decline in dimeric LPL remaining at the coronary lumen (Fig. 6C, inset), suggesting that the inhibitory action of Angptl-4 on LPL is through the conversion of dimeric LPL into monomers, which in turn reduces its affinity to the binding sites at the vascular lumen. This inhibitory effect of Angptl-4 was also observed in isolated hearts (Supplementary Fig. 3A) and cardiomyocytes (Supplementary Fig. 3B and C) from control animals. To duplicate the elevated serum NEFA level seen in D100 animals, isolated cardiomyocytes were treated with 1.0 mmol/L PA for 4–24 h. Within 4 h of exposure to PA, Angptl-4 mRNA did not change significantly. However, at 12 h, increase in Angptl-4 mRNA was observed, an effect that was even more dramatic after a 24-h exposure to PA (Fig. 6D).

The majority of FA provided to the heart comes from breakdown of circulating TG, a process catalyzed by LPL located at the vascular lumen (35). In mice with cardiac overexpression of LPL, excess lipid uptake is evident, with development of dilated cardiomyopathy (36). Paradoxically, chronic cardiac depletion of LPL is also associated with a decrease in cardiac ejection fraction (37). Thus, cardiac LPL is of crucial importance for regulating lipid metabolism in hearts, and disturbing its innate function is sufficient to cause cardiomyopathy. Diabetes is a unique metabolic disorder during which either an increase (moderate type 1 diabetes) or decrease (severe type 1 diabetes) in vascular LPL activity is observed. Data from this study suggest that after acute hyperglycemia and moderate diabetes, more LPL is processed into an active dimeric form, whereas severe diabetes is associated with increased conversion of LPL into inactive monomers (Fig. 7).

In type 1 diabetic patients, insulin supplementation can never mimic the exquisite control of glucose by pancreatic insulin seen in normal individuals. Multiple finger pricks and insulin injections (three to four per day) mean poor patient compliance, repeated exposure to bouts of inadequate glucose control (and a shift in cardiac metabolism), and cardiovascular disease in the long term. To imitate this poorly controlled type 1 diabetic patient exposed to acute hyperglycemia, we used DZ. Measurement of LPL protein expression in the heart revealed no change after acute hyperglycemia. However, the drawback with this measurement is that it does not distinguish active LPL from its inactive monomeric form. Using the heparin–sepharose column, our data for the first time show that acute hyperglycemia can indeed increase the amount of dimeric LPL in the heart. We are unaware of a similar increase in dimeric LPL in other physiology or pathology. In fact, the reverse is often seen with lipid metabolism disorders. In one case of human familial LPL deficiency, the Tyr262→His mutation of the LPL gene results in shifting of dimeric LPL to monomers (38). In addition, in cell lines derived from animals with a mutation (cld) that impairs posttranslational processing of LPL, a significant amount of inactive LPL is produced as aggregated monomers (21). The increased dimeric LPL in the whole heart after acute hyperglycemia correlated well with the elevated LPL activity at the vascular lumen. Our data suggest that after a rapid rise in circulating glucose, a unique augmentation in LPL dimerization contributes toward increased coronary enzyme, delivering more FA to the heart.

After stripping of dimeric LPL from the vascular lumen, more LPL dimers were still evident in the myocytes from acute diabetic animals. These dimeric LPL could be located inside the ER, Golgi, or secretion vesicles or at the myocyte surface. Interestingly, the majority of dimeric LPL displayed resistance to endo H digestion (20). This suggested that after DZ, LPL processing is very efficient, with more LPL completing its maturation in the ER before moving to the Golgi. Early processing of LPL into a conformation competent for dimerization is suggested to involve chaperones like calreticulin/calnexin. For example, in a transfected sf21 cell line, folding/dimerization of human LPL is only promoted when calreticulin is coexpressed (39). In addition, the impaired LPL processing seen in cld/cld mice is related to a decreased expression of calnexin (40). In our study, LPL in control cardiomyocytes was found associated with calnexin. As interruption of this association by Cs hindered LPL dimerization and secretion, our data imply that calnexin is important for LPL processing in the heart.

We have previously found that HG+PA increased LPL trafficking from the Golgi to myocyte surface (12,15). The current data suggest that HG and PA are also responsible for enhanced LPL dimerization. To test whether the effect of an acute diabetic environment on LPL processing is not a transient response, we turned to a more chronic diabetes model using D55 animals. A significant increase in the amount of LPL dimers correlated with an increased expression of calnexin in D55 hearts. As no change in LMF1 gene expression was observed, our data suggest that chronic diabetes may have a profound impact on the early processing events of LPL dimerization. It is possible that the increased calnexin expression after chronic diabetes is related to the unfolded protein response (UPR), an adaptive mechanism that augments protein processing efficiency after ER stress (41). One of the three well-defined arms of UPR includes up-regulation of ER chaperone expression through activation of inositol-requiring kinase 1 (IRE-1) on the ER membrane (42,43). It should be noted that acute exposure to HG+PA had no effect on calnexin expression. Nevertheless, this environment could still alter the processing efficiency of LPL by affecting chaperone activity and accessibility in the ER through calcium and extracellular signal–regulated kinase (ERK-1). Calcium is an important cofactor for protein folding in the ER, facilitating LPL dimerization, and can also impact the structural stability and substrate recognition of calnexin (44,45). ERK-1, activated in an IRE-dependent manner, can directly phosphorylate calnexin at Ser563, resulting in its recruitment to ribosomes, potentially rendering it more accessible to newly synthesized proteins (46,47). The role of calcium and ERK-1 on LPL maturation, and how diabetes could potentially alter LPL processing through these pathways, requires investigation.

In contrast to D55, D100 animals have a more profound loss of pancreatic β-cells and reduction in circulating insulin. Under these conditions, excessive lipolysis and hepatic lipoprotein secretion increases serum FA and TG, respectively. Thus, these doses of STZ represent a continuum of poor glycemic control, from moderate (D55) to severe (D100). To avoid lipid overload, LPL activity at the coronary lumen decreased in D100 hearts, and this could be due to enzyme inactivation by conversion from a dimeric to monomeric conformation (26). Angptl-4, identified to bring about this conversion, is expressed in various tissues, including white adipose tissue, liver, heart, and skeletal muscle (48). This secreted protein can function in circulation, breaking down vascular LPL (49). Our data, for the first time, report a robust increase of Angptl-4 protein, both in the serum and heart tissue from D100 animals, an effect that was not seen in D55. Given that the serum concentration of Angptl-4 in D100 was capable of inhibiting the activity of vascular LPL released from D55 hearts in vitro, we assumed that introduction of Angptl-4 to D55 hearts would give these hearts a D100 phenotype. Indeed, a drop of LPL activity was observed at the vascular lumen in D55 hearts perfused with Angptl-4, an effect that correlated well with the inhibitory efficiency in vitro. The reduction of coronary LPL activity could be due to the increased conversion of dimeric LPL into monomers, with subsequent detachment from the vascular binding sites. This effect of Angptl-4 was also evident in control hearts. Long chain FA, through activation of proliferator-activated receptor δ, has been reported to increase the transcription of Angptl-4 in skeletal muscle cells (50). In our study, as Angptl-4 expression in cardiomyocytes was also up-regulated after exposure to PA (more than 4 h), the decrease of LPL in D100 could be through an FA–Angptl-4 pathway. As this effect was absent in DZ that had short term hyperlipidemia (less than 4 h), our data suggest that “turning off” LPL requires prolonged exposure to FA. Whether the increased expression of Angptl-4 seen in D100 hearts could contribute toward inhibition of LPL at the coronary lumen is currently unknown. Nevertheless, purified Angptl-4 was capable of directly inhibiting LPL in or on cardiomyocytes. In addition, in mice overexpressing Angptl-4 specifically in the heart, a reduced cardiac LPL activity was found without affecting this enzyme in other LPL-expressing tissues, implying a tight link between tissue-specific Angptl-4 and local or proximal vascular LPL activity (48).

In summary, after diabetes, recruitment of LPL to the vascular lumen could represent an immediate compensatory response to guarantee FA supply. In moderate type 1 diabetes, an exaggerated LPL processing to dimeric, catalytically active enzyme increases vascular LPL, delivering more FA to the heart when glucose utilization is compromised. In severe diabetes, FA-induced expression of Angptl-4 leads to conversion of LPL to inactive monomers at the coronary lumen to impede TG hydrolysis and FA supply. We suggest that both compensatory processes that alter the innate function of LPL at the vascular lumen during diabetes could be the forerunner for cardiomyopathy seen with this disease.

When examining LPL in humans, heparin has been used to displace LPL, and either a decrease or no change in its activity has been reported in type 1 diabetic patients. As this approach releases LPL from other tissues, including skeletal muscle and adipose tissue, it is incapable of establishing cardiac LPL content. However, even if heart homogenates from patients with diabetes were used, it would still be inappropriate, as it would reflect total cardiac LPL and not the functional pool of enzyme at the coronary lumen. The animal models used in this study may not perfectly mimic the clinical condition. Nevertheless, they allowed us to distinguish functional LPL from total enzyme, and uncovered how inadequate glycemic control could lead to cardiovascular disease through manipulation of LPL-governed lipid metabolism.

This study was supported by an operating grant from the Canadian Diabetes Association. P.P., F.W., and M.S.K. are the recipients of Doctoral Student Research Awards from the Canadian Diabetes Association.

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

Y.W. conceived the idea, generated most data, and wrote the manuscript. P.P., F.W., M.S.K., and A.A. helped with obtaining some data. B.R. helped with writing the manuscript.