Potential conflict of interest: E.S.G.S. has received a consultancy fee for the apoB antisense program from ISIS. B.P.M. and S.G. are employees of ISIS Pharmaceuticals.
This work was supported by the Netherlands Organisation for Scientific Research (VENI grant 016.096.044; to M.N.) and by the National Institutes of Health (grants HL94277 and HL93321; to K.J.W.).
Type 2 diabetes mellitus (T2DM) impairs hepatic clearance of atherogenic postprandial triglyceride-rich lipoproteins (TRLs). We recently reported that livers from T2DM db/db mice markedly overexpress the heparan sulfate glucosamine-6-O-endosulfatase-2 (SULF2), an enzyme that removes 6-O sulfate groups from heparan sulfate proteoglycans (HSPGs) and suppresses uptake of TRLs by cultured hepatocytes. In the present study, we evaluated whether Sulf2 inhibition in T2DM mice in vivo could correct their postprandial dyslipidemia. Selective second-generation antisense oligonucleotides (ASOs) targeting Sulf2 were identified. Db/db mice were treated for 5 weeks with Sulf2 ASO (20 or 50 mg/kg per week), nontarget (NT) ASO, or phosphate-buffered saline (PBS). Administration of Sulf2 ASO to db/db mice suppressed hepatic Sulf2 messenger RNA expression by 70%-80% (i.e., down to levels in nondiabetic db/m mice) and increased the ratio of tri- to disulfated disaccharides in hepatic HSPGs (P < 0.05). Hepatocytes isolated from db/db mice on NT ASO exhibited a significant impairment in very-low-density lipoprotein (VLDL) binding that was entirely corrected in db/db mice on Sulf2 ASO. Sulf2 ASO lowered the random, nonfasting plasma triglyceride (TG) levels by 50%, achieving nondiabetic values. Most important, Sulf2 ASO treatment flattened the plasma TG excursions in db/db mice after corn-oil gavage (iAUC, 1,500 ± 470 mg/dL·h for NT ASO versus 160 ± 40 mg/dL·h for Sulf2 ASO\P < 0.01). Conclusions: Despite extensive metabolic derangements in T2DM mice, inhibition of a single dys-regulated molecule, SULF2, normalizes the VLDL-binding capacity of their hepatocytes and abolishes postprandial hypertriglyceridemia. These findings provide a key proof of concept in vivo to support Sulf2 inhibition as an attractive strategy to improve metabolic dyslipidemia. (HEPATOLOGY 2012;55:1746–1753)
The prevalence of type 2 diabetes mellitus (T2DM) and related syndromes is rising at an alarming pace worldwide, and the overwhelming majority of affected individuals die from accelerated atherosclerotic cardiovascular disease.1, 2 Atherosclerosis is exacerbated in these patients in large part from their characteristic dyslipidemia, which includes increased fasting levels of very-low-density lipoprotein (VLDL) and its major component, triglyceride (TG), as well as impaired clearance of postprandial triglyceride-rich lipoprotein (TRL) remnants.3-5 Atherosclerosis arises from the subendothelial retention of these lipoproteins, and increased plasma levels of VLDL and, particularly, postprandial TRL remnants have been linked to atherosclerotic cardiovascular events in human cohorts.6-9
Unfortunately, current therapeutic strategies have shown limited success in lowering fasting or postprandial TRL concentrations as a way to reduce cardiovascular morbidity or mortality. A major step forward in atherosclerotic cardiovascular risk reduction has been achieved in T2DM by the introduction of statins, a class of medicines that lower plasma levels of low-density lipoprotein cholesterol.10, 11 Nonetheless, T2DM subjects treated with optimal statin therapy exhibit considerable residual risk for cardiovascular disease, which may occur, in part, because statins lower TRL levels by only 10%-25%.12 Although fibrates are widely used in the treatment of hypertriglyceridemia, there is no definitive evidence that fenofibrate, when added to statin therapy, reduces the risk of coronary events in subjects with T2DM.13, 14 In addition, we lack therapeutic strategies that specifically restore postprandial remnant lipoprotein clearance to normal in T2DM.
Healthy metabolism of TRLs involves a series of steps that culminate in the uptake of TRL remnants by hepatocytes.8, 15, 16 During the past decades, we17 and others18, 19 have implicated hepatic heparan sulfate proteoglycans (HSPGs) in TRL removal, specifically, the syndecan-1 HSPG.20-22 The syndecan-1 HSPG comprises a single-span transmembrane core protein that has three extracellular covalent attachment sites for heparan sulfate (HS),23 which is an unbranched polysaccharide that captures lipoproteins. Roughly 50 genes are involved in HSPG assembly and disassembly, affecting core protein expression, HS side-chain length, epimerization of glucuronyl residues, and sites and extent of sulfation.24 To molecularly characterize HSPG defects in the T2DM liver, we recently used a highly annotated glycomic microarray to compare hepatic expression profiles in obese, T2DM db/db (Leprdb/db) mice versus lean, nondiabetic db/m controls.25 Despite the complexity of HSPG biology, just one gene was identified whose dys-regulation could impair syndecan-1 HSPG structure or function: the HS glucosamine-6-O-endosulfatase-2 (Sulf2).25 This gene encodes an enzyme, SULF2, that removes 6-O sulfate groups from HSPGs.26, 27 Livers of obese T2DM mice were found to markedly overexpress SULF2, and SULF2 was shown to inhibit the catabolism of TRLs by cultured liver cells.25 Moreover, hepatic Sulf2 messenger RNA (mRNA) expression was positively related to plasma TG levels.25 These experimental findings imply that SULF2-mediated disruption of hepatic HSPGs may contribute to impaired TRL clearance in T2DM.
In the present study, we evaluated whether inhibition of this single overexpressed target, Sulf2, could correct the characteristic postprandial dyslipidemia of T2DM mice in vivo. To address this question, second-generation antisense oligonucleotides (ASOs) were identified that selectively inhibit hepatic Sulf2 mRNA expression. We studied the effects of Sulf2 inhibition in vivo on hepatic HSPG sulfation, binding of TRLs to isolated primary hepatocytes, and, most important, plasma TG excursions after corn-oil gavage under diabetic conditions.
Materials and Methods
Antisense therapy relies on base-pair hybridization through which ASOs selectively bind to their complementary mRNA target.28 This binding typically results in selective, catalytic degradation of the target mRNA by RNase H29 and thereby reduces levels of the encoded protein. All ASOs used in these studies were 20 nucleotides in length and chemically modified with phosphorothioate in the backbone and 2'-O-methoxyethyl on the wings with a central deoxy gap (5-10-5 gapmer).28 Oligonucleotides were synthesized using an Applied Biosystems 380B automated DNA synthesizer (PerkinElmer-Applied Biosystems, Foster City, CA) and purified as previously described.29 To identify a potent Sulf2 ASO for experiments in mice, a series of ASOs was designed and tested in primary mouse hepatocytes for their relative abilities to suppress Sulf2 mRNA levels. From these experiments, the optimal Sulf2 ASO was selected and its efficacy was then verified by its ability to suppress hepatic Sulf2 mRNA levels in wild-type C57BL6 mice. An oligonucleotide that is not complementary to any known murine RNA sequence was used as nontarget (NT) ASO. In C57BL6 mice (Jackson Laboratory, Bar Harbor, ME), Sulf2 ASO treatment for 4 weeks (described below) resulted in an 80% ± 3% reduction of hepatic Sulf2 mRNA levels, compared to levels after administration of the NT ASO (two-sided, unpaired Student's t test; P < 0.0001; n = 4/group).
Animals and Oligonucleotide Dosing.
Seven-week-old male T2DM db/db (Leprdb/db) mice and lean, nondiabetic control db/m mice from the same colony on the C57BLKS background were used (Jackson Laboratory). Animals were injected intraperitoneally twice-weekly with Sulf2 ASO (10 or 25 mg/kg per dose, i.e., 20 or 50 mg/kg per week), NT ASO (50 mg/kg per week), or phosphate-buffered saline (PBS) for 5 weeks. Animals were housed in microisolator cages on a constant 12-hour light-dark cycle with controlled temperature and humidity and were given access to food and water ad libitum (Purina LabDiet #5008; PMI Nutrition International, Brentwood, MO). Two days after the final dose, mice were weighed, and plasma samples were taken for in-house assays of plasma glucose, insulin, and markers of liver function, as well as plasma lipids (Olympus Analyzer; Olympus America, Inc., Center Valley, PA). Plasma insulin levels were analyzed using a commercially available enzyme-linked immunosorbent assay (Catalog No.: 90080; Crystal Chem Inc., Downers Grove, IL). Homeostasis model assessment-estimated insulin resistance (HOMA-IR) was defined as fasting plasma insulin [μU/mL] × fasting plasma glucose [mmol/L] divided by 22.5. All animal procedures were approved by the institutional animal care and use committee.
Measurements of Hepatic mRNA Levels.
Mouse livers were homogenized in guanidine isothiocyanate solution (Invitrogen, Carlsbad, CA), supplemented with 8% 2-mercaptoethanol (Sigma-Aldrich, St. Louis, MO). Total RNA was prepared using RNeasy Mini Kits (QIAGEN Inc., Valencia, CA) and reversed transcribed with a complementary DNA synthesis kit (Bio-Rad Laboratories, Inc., Berkeley, CA). Quantitative real-time reverse-transcription polymerase chain reaction (qRT-PCR) assays for Sulfs were performed using an ABI Prism 7700 sequence detector (Applied Biosystems). Sequences of primers and the probe for mouse Sulf2 were: 5'-TGGACGGTGAGATATACCACGTA-3' (forward), 5'-CAGTGCGGCTTGCTAAGGTT-3' (reverse), and F-5'-CTTGGATACTGTGCCTCAGCCCCG-3'-Q (probe) (Integrated DNA Technologies, Coralville, IA). Primers for mouse Sulf1 were: 5'-TCATTCGTGGTC CAAGCATAGA-3' (forward), 5'-TGGTAGGAGCTA GGTCGATGTTC-3' (reverse), and F-5'-CCAGGGTC GATAGTCCCACAGATTGTTC-3' (probe). 18S RNA was used to normalize gene expression, with the following primers: 5'-GCAATTATTCCCCATGAACG-3' (forward), 5'GGGAC TTAATCAACGCAAGC-3' (reverse), and 5'-TTCCCA GT-3' (probe).
Purification and Analysis of Heparan Disaccharides From Liver.
HS disaccharides from murine liver tissue were prepared and measured as described previously.30 Briefly, 50 mg of liver tissue were homogenized in 300 μL of NH4Ac/Ca(Ac)2 (pH 7) and digested by a mixture of recombinant heparinases I, II, and III (5 IU each; kind gifts from Dr. Jian Liu, University of North Carolina, Chapel Hill, NC) for 2 hours at 37°C. Samples were heat inactivated and centrifuged (16,000×g for 5 minutes). The supernatant was transferred to an Amicon ultracentrifugal filter (Millipore, Bedford, MA) with a 5-kDa cutoff. The filtered samples containing heparan disaccharides were applied to liquid chromatography (LC)/dual mass spectrometry (Acquity UPLC; Waters Corporation, Milford, MA) and Quattro Premiere XE (Micromass; Waters) using multiple reactions monitoring in negative ion mode. Separation of HS disaccharides by LC was performed using a Hypercarb column (2.1 mm i.d. × 100 mm, 5 μm; Thermo Fisher Scientific, Waltham, MA) with a gradient elution (10 mM NH4HCO3, pH 10, to 100% acetonitrile). HS disaccharide standards were purchased from Iduron (Manchester, UK). SULF2 activity in vivo was analyzed as the ratio of trisulfated (D2S6) versus disulfated (D0S6 and D2S0) heparan disaccharides, the nomenclature of which has been described previously.31 The two disulfated disaccharides could not be separated.
Isolation and DyLight Labeling of TRL Fraction.
Human TRLs (d < 1.006 g/mL) were isolated by density-gradient ultracentrifugation (SW41 rotor; 19 hours, 36,000 rpm, 10°C) from serum obtained from fasting healthy volunteers. The TRL fraction was labeled with DyLight fluorophore (Amine-Reactive Fluors 488; Thermo Fisher Scientific), which allows a high dye-to-protein ratio. Labeling was performed according to the manufacturer's protocol.
Primary Hepatocyte Isolation and Lipoprotein Binding.
After the administration of ASO or PBS to db/db and db/m mice, primary hepatocytes were isolated using collagenase perfusion, as described previously.32 Isolated hepatocytes were plated on Primaria multiwell plates (Becton Dickinson Labware, Lincoln Park, NJ) using Williams' Medium E 1× (GIBCO-Invitrogen, Grand Island, NY). After 3 hours, the original culture medium was replaced by serum-free Williams' Medium E containing 1% bovine serum albumin (BSA), followed by 6 hours of incubation at 37°C. Several minutes before the binding experiments, cells were prechilled on ice, followed by a wash with Medium E/1% BSA at 4°C. Cells were incubated with a combination of DyLight-TRL (50 μg/mL) and bovine lipoprotein lipase (LPL; 5 ug/mL; L2254; Sigma-Aldrich) for 30 minutes at 4°C. Cells were rinsed once with cold PBS and lysed in 200 μL of radioimmunoprecipitation assay buffer, supplemented with protease inhibitors (Roche, Basel, Switzerland). Cell lysates were collected and transferred into a black 384-well plate, and fluorescence was measured using the Cytofluor Multi-Well Plate Reader 4000 (PerSeptive Biosystems, Framingham, MA).
Postprandial Fat-Tolerance Testing.
A stock preparation of 1 mL of corn oil (Catalog No.: C8267; Sigma-Aldrich) was supplemented with 27 μCi of [11.12-3H]retinol (44.4 Ci/mmol; PerkinElmer Life Sciences, Waltham, MA) in ethanol. Mice were fasted for 4 hours, after which each mouse received 10 μL of the corn-oil/[3H]retinol mixture per gram of body weight by gastric gavage. Blood was sampled at the indicated times by submandibular bleeding. TGs were measured on an Olympus Clinical Analyzer (Beckman Coulter, Inc., Brea, CA), and [3H] was quantified by scintillation counting.
Normally distributed data are presented as mean ± standard error of the mean (SEM), unless otherwise stated. For comparisons between a single treatment group and a control, the unpaired, two-tailed Student's t test was used. For comparisons among several groups, analysis of variance (ANOVA) was initially used, followed by pairwise comparisons using the Student-Newman-Keuls q statistics. P values less than 0.05 were considered significant. Data and graphics were analyzed and constructed by GraphPad Prism software (version 5 for Windows; GraphPad Software, Inc., La Jolla, CA).
ANOVA, analysis of variance; ASOs, antisense oligonucleotides; BSA, bovine serum albumin; HOMA-IR, homeostasis model assessment-estimated insulin resistance; HS, heparan sulfate; HSPGs, heparan sulfate proteoglycans; iAUC, incremental area under the curve; LC, liquid chromatography; LPL, lipoprotein lipase; mRNA, messenger RNA; NT, nontarget; PBS, phosphate-buffered saline; qRT-PCR, quantitative real-time reverse-transcription polymerase chain reaction; SEM, standard error of the mean; SULF2, glucosamine-6-O-endosulfatase-2; T2DM, type 2 diabetes mellitus; TG, triglyceride; TRLs, triglyceride-rich lipoproteins; VLDL, very-low-density lipoprotein.
Treatment of db/db Mice With Sulf2 ASO Specifically Restores Hepatic Expression of Sulf2 to Normal.
By the end of the 5-week treatment period, body weights, random nonfasting plasma glucose and insulin levels, and HOMA-IR values were significantly higher in PBS-treated db/db mice, compared to PBS-treated db/m mice (Table 1). These parameters were not corrected by the NT ASO or by either dose of the Sulf2 ASO in db/db mice. Markers of liver function were mildly elevated after NT and Sulf2 ASO (Table 1). Liver, kidney, and spleen histology did not show remarkable differences between saline- and oligo-treated animals (data not shown).
Table 1. Characteristics of db/db Mice After Treatment With Sulf2 ASO
Seven-week-old male db/db and db/m mice were given PBS, Sulf2 ASO, or NT ASO for 5 weeks at indicated weekly doses. Two days after the final dose, body weights and random plasma levels of glucose, insulin, ALT, and AST were measured. Displayed are means ± SEM, n = 5-8 animals/group.
P value by ANOVA; columns labeled with different lowercase letters (a, b, and c) are statistically different from each other by the Student-Newman-Keuls test (P < 0.05).
Body weights (g)
43.9 ± 1.0a
43.6 ± 1.0a
47.9 ± 1.0b
47.0 ± 0.8b
28.6 ± 0.5c
36.8 ± 2.4a
32.6 ± 1.8a
32.9 ± 2.0a
29.1 ± 2.4a
11.4 ± 0.3b
172 ± 46a,b
179 ± 14a,b
253 ± 63b
215 ± 30b
56 ± 5a
267 ± 59a
259 ± 28a
345 ± 63a
249 ± 39a
28 ± 3b
49 ± 4a
56 ± 2a
53 ± 2a
75 ± 9b
20 ± 1c
127 ± 12a
83 ± 4b
65 ± 3b
81 ± 5b
58 ± 4c
Hepatic Sulf2 mRNA expression was strongly induced in PBS-treated db/db mice to five times the levels in db/m mice (Fig. 1A; PBS-treated db/db versus PBS-treated db/m), consistent with our previous report.25 Importantly, administration of Sulf2 ASO to db/db mice suppressed hepatic Sulf2 mRNA levels by 70%-80%, and the higher dose restored hepatic Sulf2 mRNA to levels indistinguishable from db/m (Fig. 1A). Sulf2 ASO had no effect on hepatic Sulf1 mRNA levels (Fig. 1B), indicating specificity. These data show that Sulf2 ASO effectively and selectively normalizes hepatic Sulf2 mRNA expression in db/db mice.
Administration of Sulf2 ASO In Vivo to db/db Mice Increases Trisulfated Heparan Disaccharides in Liver and Completely Restores the Ability of Primary Hepatocytes to Bind TG-Rich Lipoproteins.
To assess the effects of Sulf2 inhibition in vivo, we began by analyzing heparan sulfation in liver homogenates from db/db mice treated with Sulf2 versus NT ASO. Consistent with the pattern of hepatic Sulf2 expression shown in Fig. 1, the administration of low- and high-dose Sulf2 ASO to db/db mice significantly raised the ratio of tri- to disulfated heparan disaccharides in their livers, whereas the NT ASO had no effect over PBS (0.73 ± 0.04 and 0.72 ± 0.06 versus 0.56 ± 0.08, respectively; P < 0.05; Fig. 2A). Next, we analyzed the binding of DyLight-labeled VLDL to primary hepatocytes isolated from db/db mice after treatment with Sulf2 or NT ASOs and from db/m mice after treatment with PBS. Compared to db/m hepatocytes, hepatocytes isolated from db/db mice after administration of the NT ASO exhibited a significant impairment in VLDL binding that was completely corrected in hepatocytes from db/db mice after treatment with Sulf2 ASO (Fig. 2B). These data collectively show that Sulf2 inhibition in vivo in db/db mice increases hepatic HSPG sulfation and restores hepatocyte binding of TRLs.
Treatment of db/db Mice With Sulf2 ASO Corrects Their Random Nonfasting Hypertriglyceridemia.
Consistent with previous reports,25 PBS-treated db/db mice exhibited a significant nonfasting hypertriglyceridemia (Fig. 3A; PBS-treated db/db versus PBS-treated db/m; P < 0.05). Administration of the NT ASO to db/db mice had no detectable effect on their nonfasting TG levels. In contrast, the Sulf2 ASO caused a dose-dependent improvement in nonfasting hypertriglyceridemia, reaching a 50% reduction in nonfasting TG levels at the higher dose (Fig. 3A; 102 ± 8 mg/dL in db/db Sulf2 ASO 50 mg/kg versus 171 ± 23 mg/dL in db/db NT ASO and 212 ± 18 mg/dL in db/db PBS; P < 0.05), thereby restoring this parameter to a level indistinguishable from PBS-treated db/m mice (125 ± 7 mg/dL). Fasting plasma TG levels (not shown) and nonfasting plasma total cholesterol concentrations (Fig. 3B) were significantly higher in PBS-treated db/db mice, compared to the db/m mice, and were not corrected by either dose of Sulf2 ASO.
Treatment of db/db Mice With Sulf2 ASO Completely Abolishes Their Postprandial Dyslipidemia.
After 5 weeks of treatment, db/db animals were fasted for 4 hours, then given a gavage of corn oil enriched with [3H]retinol. Sulf2 ASO administration to db/db mice flattened their postprandial TG excursions (Fig. 4A,B). The iAUC was 1,500 ± 470 mg/dL·h in db/db mice given the NT ASO, which fell to just 160 ± 40 mg/dL·h in mice treated with the higher dose of Sulf2 ASO (Fig. 4A,B). Likewise, Sulf2 ASO lowered plasma [3H]retinol excursions by >50%, indicating a profound improvement in the clearance of chylomicron remnant particles (Fig. 4C,D).
In the present study, we show that Sulf2 inhibition in T2DM db/db mice increases heparan sulfation, normalizes the ability of their hepatocytes to bind TRLs, substantially decreases nonfasting plasma TG levels, and abolishes postprandial hypertriglyceridemia. Thus, despite extensive, persistent metabolic derangements in these animals (Table 1), inhibition of a single overexpressed molecule, Sulf2, down to control levels normalizes the hepatic metabolism of atherogenic remnant lipoproteins. These findings provide the first proof of concept in vivo to support Sulf2 inhibition as an attractive strategy to improve metabolic dyslipoproteinemia. Moreover, our current results bolster the concept that diabetes dys-regulates a surprisingly small number of key molecules involved in the function of hepatic syndecan-1 as a receptor for TRL remnants.15, 25, 33
After Sulf2 ASO administration, nonfasting plasma TG levels were decreased by 50%. Nonfasting TG levels closely reflect persistent postprandial TRL particles.8 Likewise, by examining plasma TG excursions after corn-oil gavage, we found robust improvement after Sulf2 ASO administration to db/db mice (>90% reduction in iAUC). The magnitude of this improvement vastly exceeds the effects on postprandial TG excursions of conventional lipid-lowering interventions, such as statins (10%-15% reduction in iAUC) and fibrates (10%-20% reduction in iAUC).34-38 Unlike Sulf2 ASO, these conventional interventions fail to specifically target the key molecular derangement in T2DM liver. Although NT ASOs also produce a mild, nonspecific effect, the effects of Sulf2 ASOs are most consistent, greater in magnitude, and hence clearly directly related to Sulf2 inhibition.
Residual atherosclerotic cardiovascular risk in T2DM patients remains substantial, even during maximal conventional treatment with currently available therapies. Recent work has implicated nonfasting TG levels, a marker of persistent remnant lipoproteins, as an independent risk factor for atherosclerotic cardiovascular disease,7, 9 but there have been no therapeutic strategies that selectively target persistent postprandial remnants. Our present findings demonstrate that hepatic Sulf2 inhibition in vivo corrects postprandial dyslipidemia in T2DM mice. Translation of these findings to the clinic will benefit from the relative maturity of ASO technology. In other circumstances, ASO administration has been selective and effective against hepatic targets.39, 40 Importantly, ASOs have been reported to be safe and effective during short-term administration to humans.41 In our system, the Sulf2 ASOs lowered abnormally high levels of Sulf2 mRNA in T2DM mouse livers to normal, but not below normal, which is highly desirable from the standpoint of safety. In conclusion, our work provides a key proof of concept in vivo for a novel therapeutic approach to improve metabolic dyslipidemia through restoration of hepatic HSPG function in diabetes.
The authors greatly acknowledge J.A. Sierts for excellent laboratory assistance as well as H. van Lenthe and L. IJlst for technical assistance and expertise in the disaccharide analysis.