1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Type 2 diabetes mellitus (T2DM) impairs hepatic clearance of atherogenic postprandial remnant lipoproteins. Our work and that of others have identified syndecan-1 heparan sulfate proteoglycans (HSPGs) as remnant lipoprotein receptors. Nevertheless, defects in the T2DM liver have not been molecularly characterized, and neither has the correction that occurs upon caloric restriction. We used microarrays to compare expression of proteoglycan-related genes in livers from control db/m mice; obese, T2DM db/db littermates fed ad libitum (AL); and db/db mice pair-fed to match the intake of db/m mice. Surprisingly, the arrays identified only one gene whose dysregulation by T2DM would disrupt HSPG structure: the heparan sulfate glucosamine-6-O-endosulfatase-2 (Sulf2). SULF2 degrades HSPGs by removing 6-O sulfate groups, but had no previously known role in diabetes or lipoprotein biology. Follow-up quantitative polymerase chain reaction assays revealed a striking 11-fold induction of Sulf2 messenger RNA in the livers of AL T2DM mice compared with controls. Immunoblots demonstrated induction of SULF2 in AL livers, with restoration toward normal in livers from pair-fed db/db mice. Knockdown of SULF2 in cultured hepatocytes doubled HSPG-mediated catabolism of model remnant lipoproteins. Notably, co-immunoprecipitations revealed a persistent physical association of SULF2 with syndecan-1. To identify mechanisms of SULF2 dysregulation in T2DM, we found that advanced glycosylation end products provoked a 10-fold induction in SULF2 expression by cultured hepatocytes and an approximately 50% impairment in their catabolism of remnants and very low-density lipoprotein, an effect that was entirely reversed by SULF2 knockdown. Adiponectin and insulin each suppressed SULF2 protein in cultured liver cells and in murine livers in vivo, consistent with a role in energy flux. Likewise, both hormones enhanced remnant lipoprotein catabolism in vitro. Conclusion: SULF2 is an unexpected suppressor of atherogenic lipoprotein clearance by hepatocytes and an attractive target for inhibition. (HEPATOLOGY 2010;.)

Accelerated atherosclerotic cardiovascular disease remains the major cause of death in patients with diabetes and related syndromes.1 Atherosclerosis arises from the retention of cholesterol-rich, apolipoprotein B (apoB)-containing lipoproteins within the vessel wall.2 Importantly, patients with diabetes suffer from a unique and typically neglected aspect of cardiovascular risk—namely, the striking persistence of postprandial apoB lipoproteins, called remnants, in their plasma after each meal. A major cause is a defect in hepatic clearance of these harmful particles.3-6

A conspicuous impediment in this area had been our ignorance regarding pathways for remnant uptake into liver. Over a quarter century ago, hepatic uptake of remnants was shown to occur independently of low-density lipoprotein (LDL) receptors.7 This realization launched a long, difficult search for the responsible molecules. Our work8 and the work of others9-12 have implicated heparan sulfate proteoglycans (HSPGs) as remnant lipoprotein receptors (reviewed by Williams,5 Williams and Chen,6 and Fuki et al.13). Each HSPG molecule consists of a protein strand onto which the cell assembles sugar polymers called heparan sulfate that we have shown could capture lipoproteins. Thus, the ligand-binding domains of these molecules are carbohydrate, not polypeptide. The liver contains several different species of HSPGs, from which our laboratory identified syndecan-1 as a strong candidate to participate in remnant lipoprotein clearance, based on its abundance along the sinusoidal surface of hepatic parenchymal cells and our finding that it directly mediates endocytosis of lipoproteins.13-15 These results were recently extended by the demonstration that syndecan-1 knockout mice exhibit substantially impaired remnant lipoprotein clearance.16 Thus, the syndecan-1 HSPG is a major hepatic receptor for remnant lipoproteins.13, 15, 16

In the current study, we focused on type 2 diabetes mellitus (T2DM), a disease of considerable and increasing clinical impact.1, 5, 6 In T2DM and other syndromes of insulin resistance, the proteoglycans from the liver17 and elsewhere18 exhibit subnormal charge density,17 decreased heparan sulfation,17, 18 and impaired binding to model remnant lipoproteins.17 Hyperphagic mice provide a suitable animal model, because they mimic humans who overeat and thereby develop the same complications, including obesity, insulin resistance, T2DM, and impaired remnant catabolism by the liver.19 To assess the effects of T2DM on the roughly 50 genes involved in HSPG assembly and disassembly, we used a highly annotated glycomic microarray20 to compare hepatic expression profiles in hyperphagic mice versus controls. Surprisingly, these arrays identified just one gene whose dysregulation would impair HSPG structure: the heparan sulfate glucosamine-6-O-endosulfatase-2 (Sulfatase 2, Sulf2). This gene encodes an enzyme that degrades cell surface HSPGs by removing 6-O sulfates. Thus far, the SULF2 enzyme has been studied only within the context of growth factor and morphogen signaling,21, 22 with no prior suspicion of a role in lipoprotein biology or diabetes.5, 6 Preliminary reports of this work were presented at the International Symposium on Atherosclerosis (June 2009, Boston, MA)23 and at the American Heart Association Scientific Sessions (November 2009, Orlando, FL).24

Materials and Methods

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References


Antibodies against SULF2 and the syndecan-1 core protein (catalogue #sc68436 and #sc-5632, Santa Cruz Biotechnology, Santa Cruz, CA), rat Sulf2 small interfering RNA (siRNA) (#M-093673-00-0010; Smartpool, Dharmacon, Lafayette, CO), control siRNA, enzyme-linked immunosorbent assay kits to quantify mouse adiponectin (#RD293023100R; BioVendor, Candler, NC) and insulin (#EZRMI-13K; Millipore, Billerica, MA), full-length adiponectin prepared in eukaryotic cells (Pepro Technology, Rocky Hill, NJ), and insulin (#I2643; Sigma Chemical Company, St. Louis, MO) were obtained commercially. Advanced glycosylation end products (AGEs) were made by way of a 6-week incubation of sterile, endotoxin-free bovine serum albumin (BSA) (100 mg/mL) with D-glucose (90 mg/mL). As a control, an aliquot of BSA was simultaneously incubated for 6 weeks in the same buffer but without D-glucose.

Primers and probes for quantitative real-time reverse-transcription polymerase chain reaction (qRT-PCR) were synthesized by the Gene Expression Facility at the University of North Carolina (Chapel Hill, NC; Dr. Hyung Suk Kim, director). Sequences for the mouse and rat Sulf2 reactions were: 5′-ggc tta gag acg gag gaa g-3′ (sense primer), 5′-ggt ctc ttc att tct ggc ca-3′ (antisense primer), and F-5′-tg aac aat aca ggc agt ttc agc ctg g-3′-Q (probe [F and Q denote the positions of the fluorophore and quencher]). Sequences for the mouse and rat Ppia (cyclophilin A) reactions were: 5′-atc tgc act gcc aag act ga-3′ (sense primer), 5′-cgc tcc atg gct tcc aca at-3′ (antisense primer), and F-5′-ct tcc caa aga cca cat gct tgc cat c-3′-Q (probe).

T2DM Mice and Microarrays.

We prepared three groups of male mice to study using microarrays: phenotypically lean db/m mice at age 14 weeks (controls); obese, T2DM db/db (Leprdb/db) mice fed ad libitum (AL) until 14 weeks; and db/db mice pair-fed (PF) from 8 to 14 weeks to match the intake of the db/m mice during that period. The background strain was C57BLKS (Jackson Laboratory, Bar Harbor, ME). Despite mild phenotypic changes, db/m mice do not develop T2DM under these conditions.25, 26 At age 14 weeks, mice were euthanized. Plasma samples were obtained for in-house assays of adiponectin and insulin concentrations, as well as lipid, lipoprotein, and hemoglobin A1c levels, by the Vanderbilt Mouse Metabolic Phenotyping Center (Nashville, TN). The livers of these mice were promptly flushed in situ with cold phosphate-buffered saline, and tissue samples were obtained for RNA purification by way of Qiagen column and protein extraction using Sigma tissue lysis buffer. To accommodate biologic variability, each RNA sample for array analysis was pooled from the livers of three littermates undergoing the same treatment.20 Unpooled RNA samples from each mouse liver were set aside for subsequent qRT-PCR to confirm differentially expressed targets found on the arrays. Quality of RNA samples was verified with an Agilent Bioanalyzer (Agilent Technologies, Palo Alto, CA). The MessageAmp II-Biotin Enhanced Amplification kit (Ambion, Inc., Austin, TX) was used to label each pooled RNA preparation before array hybridization. Arrays on three pooled RNA samples were run for each of the three treatment groups (n = 9 animals/group [27 mice total]).

We used the GLYCOv3 oligonucleotide array, a custom Affymetrix GeneChip designed for the Consortium for Functional Glycomics. A complete description is available at Our inspection of the list of probe set identifiers indicated that only heparanase was missing, and at our suggestion it has been added to newer versions of the array. Hybridization and scanning to the GLYCOv3 chip were performed using the Affymetrix GeneChip Scanner 3000 according to the manufacturer's protocol.27 Chips had a background of <50 intensity units and a glyceraldehyde 3-phosphate dehydrogenase (GAPDH) 3′/5′ ratio below 1.5.

Cell Culture and Analyses.

McArdle 7777 rat hepatoma cells were grown as described.28 To examine the regulatory effects of AGEs, adiponectin, and insulin, cells were switched to medium with 2% serum for 2 hours and then supplemented with these reagents. Media supplemented with AGEs or insulin contained 2% fetal bovine serum, whereas serum from adiponectin knock-out mice was used for adiponectin supplementation.29 For dose–response studies, cells were assessed after 24 hours of exposure. For time course studies, all cells were plated simultaneously and then harvested simultaneously; AGEs, adiponectin, and insulin were added at different times before the end of the experiment. At the end of these treatments, cells were assessed for protein expression by way of immunoblotting, messenger RNA (mRNA) quantification by qRT-PCR, and catabolism of model remnant lipoproteins following our published methods.13, 28, 30 Model remnant lipoproteins were prepared by adding lipoprotein lipase (LpL), a key protein on remnants recognized by HSPGs, to 125I-labeled human LDL that we had methylated to an extent that blocks its binding to LDL receptors (125I-mLDL), thereby mimicking apoB48.13, 30 In some experiments, we also examined cellular catabolism of 125I-labeled very low-density lipoprotein (VLDL),31, 32 which we prepared by ultracentrifugation of fasting human plasma33 followed by radioiodination.34 The concentration (25 μg VLDL protein/mL medium) was chosen to mimic the plasma triglyceride levels in our mice.

Regulatory Effects of Adiponectin and Insulin In Vivo.

To assess the regulatory effects of adiponectin in vivo, we peritoneally injected adiponectin knockout mice (C57BL6/J background) with full-length adiponectin three times a day for 4 days. Four hours after the last injection, 100 μL blood samples from each mouse were collected for adiponectin enzyme-linked immunosorbent assays, the mice were euthanized, and liver samples were obtained. To assess regulatory effects of insulin in vivo, we performed qRT-PCR and immunoblots on preexisting frozen liver samples from control rats and rats subjected to euglycemic-hyperinsulinemic clamps, as described.35 All animals received humane care according to the criteria outlined in the Guide for the Care and Use of Laboratory Animals prepared by the National Academy of Sciences and published by the National Institutes of Health (NIH publication 86-23 revised 1985).

Statistical Analyses.

Raw expression values from the microarrays were normalized using the Robust Multichip Average expression summary (http://rmaexpress. Differential expression was determined using the open-source, open-development Limma package37 in the R programming language for statistical computing, available under the terms of the Free Software Foundation's GNU General Public License. Comparisons between two or more experimental conditions were made by calculating the fold change in expression levels and the adjusted P value. The adjusted P value corrects for multiple testing using Benjamini and Hochberg's method to control the false discovery rate.38 The lists of differentially expressed genes were filtered for expression ratios >1.3.

Quantitative data from our other experiments were analyzed using SigmaStat version 3 (SPSS Inc., Chicago, IL). Normally distributed data are reported as the mean ± SEM. For comparisons between a single experimental group and a control, an unpaired, two-tailed t test was used. For comparisons involving several groups simultaneously, analysis of variance (ANOVA) was initially used, followed by pairwise comparisons using the Student-Newman-Keuls q statistic. Nonnormally distributed data are displayed as medians with individual measurements. Comparisons were performed using ANOVA on ranks followed by a Student-Newman-Keuls post hoc test. In some cases, we log-transformed nonnormally distributed data. Relationships between selected parameters were tested using the Spearman rank correlation coefficient. Differences and correlations were considered to be significant at P ≤ 0.01, which also holds for the adjusted P values from the microarrays.


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

T2DM Robustly Induces Hepatic Overexpression of SULF2 In Vivo, and Caloric Restriction Corrects Hepatic Levels of this Enzyme Toward Normal.

Control db/m, AL db/db, and PF db/db mice continuously gained weight throughout the study. By 14 weeks, body weights and plasma concentrations of hemoglobin A1c, triglycerides, and LDL cholesterol were significantly higher in AL versus control mice but improved toward normal or were completely normalized in the PF group (Table 1). Consistent with these metabolic data, plasma adiponectin values were significantly lower in the AL group than in control or PF mice. The pattern of hemoglobin A1c and insulin levels indicate insulin resistance in the AL group that improved in the PF group (Table 1). Thus, induction of T2DM in the AL group was successful, as was caloric restriction of the PF group, similar to previous results.19

Table 1. Characteristics of Control db/m, AL db/db, and PF db/db Mice at 14 Weeks of Age (n = 9 Animals/Group)
CharacteristicsControl db/mAL db/dbPF db/dbP Value
  1. All data are presented as the mean ± SEM (ANOVA). For each characteristic, numbers labeled with different lowercase letters (a, b, c) are statistically different (P < 0.01 [Student-Newman-Keuls test]). Normally distributed data were analyzed without transformations; nonnormally distributed data became normally distributed after log transformation.

Body weight, g28.5 ± 0.3a47.6 ± 1.2b42.0 ± 1.2c<0.001
Plasma levels
 Hemoglobin A1c, %4.16 ± 0.03a11.3 ± 0.64b7.70 ± 0.32c<0.001
  Ln, insulin (antilog of the mean), ng/mL0.128 ± 0.183a (1.14)1.01 ± 0.25b (2.75)1.10 ± 0.20b (3.00)<0.01
  Ln, triglycerides (antilog of the mean), mg/dL4.24 ± 0.06a (69.4)4.75 ± 0.10b (116)4.66 ± 0.06b (106)<0.001
 LDL cholesterol, mg/dL73.2 ± 1.8a111 ± 7.6b71.9 ± 5.4a<0.001
 Adiponectin, ng/mL876 ± 31a547 ± 30b742 ± 37c<0.001

Glycomic microarrays of hepatic RNA from these mice revealed three clusters of mRNA signals for proteins that directly affect HSPG structure (the entire dataset is publicly available under Core E request #928 at The first cluster of mRNAs from the arrays comprises participants in HSPG assembly that were not significantly affected by T2DM. Examples include the heparan sulfate Ndst2 isoform; the C5-epimerase (Glce); Ext2, one of two elongase isoforms that transfer glucuronyl and N-acetylglucosaminyl residues onto the growing HS chain; and the core proteins perlecan, agrin, and syndecan-2. Hepatic mRNA levels for each of these differed by less than 25% between AL db/db mice versus db/m controls. Of note, the arrays revealed no significant change in hepatic levels of Ndst1 mRNA, in contrast to our prior finding that it is suppressed in type 1 diabetes mellitus (T1DM) livers.28, 39 Thus, T1DM and T2DM induce different molecular derangements in hepatic HSPG assembly and disassembly.

The second cluster comprises transcripts from a limited subset of HSPG assembly genes that were paradoxically increased by T2DM. There were exactly three in this cluster: the heparan sulfate 6-O-sulfotransferase-3 (6ost3) (adjusted P = 0.0034); Ext1, which is the other HS elongase isoform (P = 0.0053); and the syndecan-1 core protein (Sdc1) (P = 0.00014). An increase in Sdc1 mRNA had been reported in livers from obese Zucker rats.17 Based on our work on syndecan-1 regulation,40 we speculate that the increase in syndecan-1 mRNA might arise from elevated FXR expression in db/db mouse livers.41

The third cluster comprises any gene whose dysregulation by T2DM would disrupt normal HSPG structure. Surprisingly, the arrays revealed only one member in this cluster: Sulf2, which showed a highly significant induction in T2DM livers (adjusted P = 0.0081). In contrast, our statistical analyses showed no significant difference in Sulf2 or any murine mRNA signal on the entire glycogene microarray between calorically restricted and AL-fed db/db mice (PF versus AL groups). We28 and others42 have reported that several key HSPG-related enzymes undergo substantial posttranscriptional regulation, which consequently became a major focus of our current study of T2DM and SULF2.

To confirm the array findings, our follow-up qRT-PCR demonstrated a remarkable 11-fold increase in hepatic Sulf2 mRNA levels in AL db/db mice versus db/m controls (Fig. 1A). Also confirmed was the absence of a statistically significant difference in hepatic Sulf2 mRNA levels between AL versus PF db/db mice, with the PF livers showing an unexpected trend toward an increase (Fig. 1A). To address posttranscriptional effects, we performed immunoblots to detect hepatic SULF2 protein in the three groups. Consistent with the induction of Sulf2 mRNA that we found in T2DM, our immunoblots revealed a large increase in SULF2 protein levels in livers from AL T2DM db/db mice (AL versus control group) (Fig. 1B). Importantly, pair-feeding substantially lowered SULF2 protein levels in db/db livers to nearly normal (PF group) (Fig. 1B), despite continued elevation of Sulf2 mRNA.

thumbnail image

Figure 1. T2DM robustly induces hepatic overexpression of SULF2 in association with hyperlipoproteinemia, and caloric restriction corrects hepatic levels of this enzyme toward normal. (A,B) RNA and protein were extracted from livers of 14-week-old phenotypically lean db/m mice (controls); their obese, AL T2DM db/db littermates (AL); and db/db mice that were pair-fed to match the intake of the db/m controls (PF). (A) Individual Sulf2 mRNA levels were assessed by way of qRT-PCR, normalized to Ppia mRNA levels (ΔCt), and then expressed relative to the median control value (2−ΔΔCt). Median values are indicated by short horizontal black lines, and individual values for all mice in each group are indicated by the symbols X, ▵, and ▿, respectively (P < 0.001 [ANOVA on ranks]). **P < 0.01 for pairwise comparisons (Student-Newman-Keuls test). n.s., not significant. (B) Immunoblots in triplicate for SULF2 and, as a loading control, GAPDH. Each lane represents a sample from a different animal. (C) Plot of plasma triglyceride concentrations versus hepatic Sulf2 mRNA levels for all control (X) and all T2DM AL (▵) mice. (D) Plot of plasma LDL concentrations versus hepatic Sulf2 mRNA levels from the same mice.

Download figure to PowerPoint

To assess possible relationships between plasma lipid/lipoprotein concentrations and hepatic Sulf2 mRNA levels, we examined statistical correlations in control db/m and AL db/db mice (PF db/db mice were omitted from this analysis, because their Sulf2 mRNA levels in liver failed to track protein expression). A graph of plasma triglyceride concentrations versus hepatic Sulf2 mRNA levels showed complete segregation of control mice, in the lower left hand corner, from AL mice in the upper right hand sector (Fig. 1C). A plot of plasma LDL concentrations versus hepatic Sulf2 mRNA levels showed the same pattern (Fig. 1D). In both cases, the calculated Spearman rank correlation coefficients were statistically significant.

Endogenous SULF2 Strongly Inhibits Catabolism of Model Remnant Lipoproteins by Hepatic Cells.

Removal of 6-O sulfates by the SULF2 enzyme inhibits growth factor signaling,21 but by contrast, prior studies of LpL binding and remnant clearance emphasized the role of 2-O sulfate groups.43, 44 To determine whether endogenous SULF2 could affect HSPG-mediated uptake of model remnant lipoproteins, we used siRNA to partially knock it down in cultured rat McArdle 7777 hepatocytes. Knockdown of SULF2 significantly increased HSPG-mediated cell surface binding, internalization, and degradation of model remnant lipoproteins by 70%-90% (Fig. 2A,B). Thus, endogenous SULF2 acts an inhibitor of remnant lipoprotein catabolism. Because the importance of heparan 6-O–linked sulfates in lipoprotein binding has been called into question,44 we sought an additional explanation for the clear inhibitory effect of SULF2 on the catabolism of remnants. Prior evidence suggests rapid substrate turnover by SULF enzymes,45 yet our co-immunoprecipitations of McArdle cell homogenates revealed an unexpectedly persistent physical association of SULF2 with the syndecan-1 HSPG (Fig. 2C), raising the possibility of nonenzymatic actions, such as ligand competition, independent of 6-O sulfation. Knockdown of SULF2 depleted the amount associated with syndecan-1 to nearly undetectable levels (data not shown).

thumbnail image

Figure 2. Endogenous SULF2 strongly inhibits the catabolism of model remnant lipoproteins by cultured liver cells. McArdle hepatoma cells were preincubated for 24 hours with 100 nM nontarget (control) or Sulf2 siRNA, as indicated, followed by an additional 44 hours at 37°C. Cells were then incubated for 4 hours at 37°C with model remnant lipoproteins that were prepared by combining 125I-labeled methylated LDL with LpL, a molecule that bridges between lipoproteins and HSPGs (background values were assessed in the absence of LpL). (A) Immunoblots of cellular homogenates, using anti-SULF2 antibodies or, as a loading control, anti–β-actin antibodies. (B) LpL-dependent cellular catabolism of model remnant lipoproteins—shown as surface binding (Surf), internalization (Inter), and degradation (Degr)—normalized to control values from cells treated with the nontarget siRNA (mean ± SEM, n = 3; nonnormalized control values were 194 ± 2.9, 605 ± 20.6, and 112 ± 4.3 ng/mg, respectively). **P < 0.01 (two-tailed Student t test). The data shown are from a representative experiment with a total of four independent knockdown studies. (C) Co-immunoprecipitation of SULF2 with the syndecan-1 HSPG. McArdle hepatocytes were extracted into NP-40 and subjected to immunoprecipitation with nonimmune immunoglobulin G (IP: Mock) or anti–syndecan-1 immunoglobulin G (IP: anti-SDC1), followed by electrophoretic separation. Shown are immunoblots that were performed to detect SULF2 (IB: anti-SULF2), and then the same blots were stripped and reprobed to detect syndecan-1 (IB: anti-SDC1).

Download figure to PowerPoint

Regulation of SULF2 Expression and Remnant Catabolism by Metabolic Factors in T2DM.

We investigated the effects of three key metabolic factors in T2DM—namely, AGEs, adiponectin, and insulin—on hepatocyte expression of SULF2 and on lipoprotein catabolism. Given our findings in T2DM in vivo (Fig. 1), we were particularly interested in factors affecting SULF2 protein levels that did or did not also alter Sulf2 mRNA levels. Exposure of McArdle cells to AGEs caused a 10-fold induction of SULF2 protein in a dose- and time-dependent manner (Fig. 3A). Likewise, AGEs caused a 20-fold induction of Sulf2 mRNA (Fig. 3B). Functionally, the exposure of McArdle cells to AGEs impaired catabolism of model remnant lipoproteins by ≈50% (P < 0.01), and this inhibitory effect was completely reversed by SULF2 knockdown (Fig. 3C). The same effects were seen with catabolism of 125I-labeled VLDL (Fig. 3D). Thus, AGEs substantially induce SULF2 protein and mRNA and impair catabolism of remnant lipoproteins and VLDL. Importantly, the entire ability of AGEs to alter catabolism of these lipoproteins by hepatocytes appears to be mediated through its induction of SULF2.

thumbnail image

Figure 3. AGEs augment expression of SULF2 by cultured liver cells, thereby inhibiting catabolism of model remnant lipoproteins and VLDL. (A) Dose–response (upper) and time course (lower) of SULF2 protein induction by AGEs. The dose–response involved a 24-hour incubation of McArdle hepatoma cells with the indicated concentrations of AGEs, plus an amount of BSA to bring the total supplemented protein in each well to 400 μg/mL. As described in Materials and Methods, cells in each time course of SULF2 regulation in vitro were harvested simultaneously. Here, 200 μg AGEs/mL was added at the indicated times before harvest. Immunoblotting of cellular homogenates was performed using anti-SULF2 or anti–β-actin antibodies. (B) Time course of Sulf2 mRNA induction by AGEs (200 μg/mL). Sulf2 mRNA levels were assayed by qRT-PCR, normalized to Ppia mRNA levels, and then expressed relative to the unexposed control at 0 hours (mean ± SEM, n = 3; P < 0.001 [ANOVA]). **P < 0.01 versus unexposed control (Student-Newman-Keuls test). (C,D) Effects of AGEs on catabolism of (C) remnant lipoproteins or (D) VLDL by cultured liver cells, with and without SULF2 knockdown. McArdle hepatoma cells were incubated for three consecutive 24-hour periods at 37°C. For the first 24-hour period, 100 nM nontarget siRNA (Sulf2 siRNA −) or Sulf2 siRNA (Sulf2 siRNA +) was used for incubation. Cells were rinsed to remove the siRNAs, then incubated in serum-containing medium for the second 24-hour period. During the third 24-hour period, 200 μg/mL of either BSA (AGEs −) or AGEs (AGEs +) was added. Lipoprotein catabolism was examined during the last 4 hours of this final 24-hour period. The upper images in these two panels show immunoblots of cellular homogenates. (C) LpL-dependent surface binding, internalization, and degradation of model remnant lipoproteins, normalized to control values from cells treated with nontarget siRNA and without AGEs (mean ± SEM, n = 3; nonnormalized control values were 242 ± 3.4, 626 ± 7.1, and 103 ± 4.6 ng/mg, respectively). (D) Normalized values for surface binding, internalization, and degradation of labeled VLDL (mean ± SEM, n = 3; nonnormalized control values were 18.58 ± 0.3, 78.86 ± 0.9, and 59.89 ± 0.35 ng/mg, respectively). P < 0.001 for each of the three groups (ANOVA); columns labeled with different lowercase letters (a, b, c) are statistically different (P < 0.01 [Student-Newman-Keuls test]). The data are representative of a total of four independent dose–response, time course, and lipoprotein catabolism experiments.

Download figure to PowerPoint

We next examined the regulatory effects of adiponectin. Suppression of plasma adiponectin levels is a prominent feature of obesity and T2DM46 (Table 1), but not of T1DM. Addition of physiologic concentrations of adiponectin to cultured McArdle hepatocytes suppressed SULF2 levels by ≈90%, with essentially the entire effect apparent by 2 hours and persisting for at least 36 hours (Fig. 4A). Physiologic concentrations of adiponectin also suppressed Sulf2 mRNA (Fig. 4B). To determine the role of this adipokine in vivo, we found that livers from adiponectin knockout mice29 exhibited substantially increased levels of SULF2 protein and mRNA, both of which were normalized by day 4 of intraperitoneal adiponectin injections (Fig. 4C,D). Similar to the effects of siRNA knockdown, addition of adiponectin to cultured McArdle hepatocytes suppressed SULF2 protein levels and enhanced surface binding, internalization, and lysosomal degradation of model remnant lipoproteins (Fig. 4E).

thumbnail image

Figure 4. Adiponectin suppresses SULF2 expression by cultured liver cells and in murine livers in vivo, and enhances remnant lipoprotein catabolism in vitro. (A) Dose–response (upper) and time course (lower) of SULF2 protein suppression by adiponectin. The dose–response involved a 24-hour incubation of McArdle hepatoma cells with the indicated concentrations of adiponectin, where zero indicates adiponectin-free medium. In the time course, 6 μg adiponectin/mL was added at the indicated times before harvest. Displayed are immunoblots of cellular homogenates. (B) Time course of Sulf2 mRNA suppression by a physiologic concentration of adiponectin (6 μg/mL). Sulf2 mRNA levels were normalized to Ppia and then expressed relative to the unexposed control at 0 hours (mean ± SEM, n = 3; P < 0.001 [ANOVA]). **P < 0.01 versus unexposed control (Student-Newman-Keuls test). (C) Regulation of hepatic SULF2 protein levels by adiponectin in vivo. Protein was extracted from livers of wild-type C57BL/6J mice (WT, n = 3) and adiponectin knockout mice that had been injected with either saline (KO + Saline, n = 4) or full-length adiponectin (KO + fAd, n = 3; 25 μg/mouse, three times/day for 4 days). Plasma adiponectin concentrations in these mice 4 hours after the last injection were 20.86 ± 1.23, 0.41 ± 0.013, and 6.16 ± 1.08 μg/mL, respectively. Immunoblots for SULF2 and GAPDH are shown. (D) Regulation of hepatic Sulf2 mRNA levels by adiponectin in vivo. RNA was extracted from the same livers as in Fig. 4C. Levels of Sulf2 mRNA were assessed by way of qRT-PCR, normalized to Ppia, and then expressed relative to wild-type (mean ± SEM, n = 3-4; P < 0.001 [ANOVA]). Columns labeled with different lowercase letters (a, b) are statistically different (P < 0.01 [Student-Newman-Keuls test]). (E) Effects of adiponectin on remnant lipoprotein catabolism by cultured liver cells. McArdle hepatoma cells were incubated for 24 hours in adiponectin-free medium (−) or medium supplemented with 6 μg adiponectin/mL (+). The upper images show immunoblots of cellular homogenates. The column graph displays LpL-dependent surface binding, internalization, and degradation of model remnant lipoproteins, normalized to control values from cells incubated without adiponectin (mean ± SEM, n = 3; the nonnormalized control values were 179 ± 3.5, 618 ± 43.4, and 37 ± 1.1 ng/mg, respectively). **P < 0.01 (two-tailed Student t test).

Download figure to PowerPoint

Insulin resistance is the hallmark of T2DM. Exposure of McArdle hepatoma cells to physiologic concentrations of insulin for 24 hours suppressed cellular SULF2 protein levels by ≈80% (3 and 10 nM insulin) (Fig. 5A). In contrast to the rapidity of insulin-induced signaling, which takes seconds to minutes, suppression of SULF2 by insulin took 4-8 hours to become apparent and lasted at least 32 hours (Fig. 5A). This delay in onset may reflect the need to degrade preexisting SULF2 protein to lower cellular levels. Unlike AGEs or adiponectin, but like caloric restriction, insulin altered SULF2 protein levels without significantly affecting Sulf2 mRNA levels (Fig. 5B). To examine regulation in vivo, we found that a 4-hour euglycemic, hyperinsulinemic clamp suppressed SULF2 protein in rat livers to nearly undetectable levels while producing a nonsignificant increase in hepatic Sulf2 mRNA (Fig. 5C,D). Consistent with a key role for SULF2 in the regulation of remnant lipoprotein catabolism, the addition of insulin to suppress SULF2 expression by cultured hepatoma cells significantly enhanced their catabolism of model remnant lipoproteins (Fig. 5E).

thumbnail image

Figure 5. Insulin suppresses SULF2 protein, but not mRNA, in cultured liver cells and in murine livers in vivo and enhances remnant lipoprotein catabolism in vitro. (A) Dose–response (upper) and time course (lower) of SULF2 protein suppression by insulin. The dose–response involved a 24-hour incubation of McArdle hepatoma cells with the indicated concentrations of insulin. In the time course, 3 nM insulin was added at the indicated times before harvest. Displayed are immunoblots of cellular homogenates. (B) Time course of Sulf2 mRNA levels during exposure to a physiologic concentration of insulin (3 nM). Sulf2 mRNA levels were normalized to Ppia and then expressed relative to the unexposed control at 0 h (mean ± SEM, n = 3; P value not significant [ANOVA]). (C) Regulation of hepatic SULF2 protein levels by insulin in vivo. Sprague-Dawley rats were subjected to euglycemic-hyperinsulinemic clamps (4.8 mU insulin/kg/minute) for 4 hours (Insulin injected); control rats received saline/glycerol. Protein was extracted from frozen liver samples. Immunoblots for SULF2 and GAPDH are shown. (D) Hepatic Sulf2 mRNA levels after insulin injection in vivo. RNA was extracted from the same livers as in Fig. 5C. Levels of Sulf2 mRNA were assessed by way of qRT-PCR, normalized to Ppia, and then expressed relative to control (mean ± SEM, n = 3-4; P value not significant [ANOVA]). (E) Effects of insulin on remnant lipoprotein catabolism by cultured liver cells. McArdle hepatoma cells were incubated for 24 hours without (−) or with (+) 10 nM insulin. The upper images show immunoblots of cellular homogenates. The column graph displays LpL-dependent surface binding, internalization, and degradation of model remnant lipoproteins, normalized to control values from cells incubated without insulin (mean ± SEM, n = 3; this experiment was performed simultaneously with the one displayed in Figure 4E). **P < 0.01 (two-tailed Student t test).

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

In the current study, we identified an unexpected participant in lipoprotein clearance—namely, SULF2, which exhibits a striking induction in the livers of obese T2DM mice and robustly inhibits the catabolism of model remnant lipoproteins and VLDL by cultured liver cells. Moreover, despite the existence of roughly 50 genes involved in hepatic HSPG assembly and disassembly, our results suggest dysregulation of only a few of them in diabetes. The Sulf2 mRNA was the sole transcript identified on the GLYCOv3 arrays that was affected by T2DM in a way that would impair HSPG function as remnant lipoprotein receptors.

Several groups have advocated an atherogenic role for postprandial particles, which carry substantial amounts of cholesterol.47-51 Like LDL, postprandial apoB lipoproteins penetrate into the arterial wall and are retained there.50, 52, 53 Importantly, postprandial dyslipoproteinemia has been linked to the development of arterial lesions49 and cardiovascular events.50, 54, 55 Regarding reversibility, moderate caloric restriction of obese, T2DM mice corrects their defect in remnant lipoprotein clearance,19 and similar effects occur in human subjects with diabetes after short-term weight loss.56 The improvements in hepatic clearance of remnant lipoproteins upon weight loss had remained completely uncharacterized on a molecular level.5, 6 Our current results indicate a role for induction of SULF2 by T2DM and then suppression of SULF2 by caloric restriction (Fig. 1B). Notably, regulation of SULF2 under these conditions occurs at both the level of mRNA (in the AL group) and posttranscriptionally (in the PF group), consistent with a role for insulin resistance as well as other factors.

The identification of SULF2 may simplify this therapeutic problem in T2DM considerably. It is generally easier to inhibit or suppress a protein than it is to boost it. Our results could convert the challenge of impaired remnant receptor function into the far more tractable pharmacologic problem of one overproduced protein. Thus, SULF2 should now become an attractive target for inhibition or suppression to correct postprandial dyslipoproteinemia. Moreover, if SULF2 is overexpressed in other tissues in T2DM (such as the skin), local inhibition of excess enzyme might improve growth factor function and hence wound healing.

Taken together, our work provides a compelling pathophysiologic model in which AGEs, hypoadiponectinemia, and insulin resistance in T2DM induce abnormal up-regulation of SULF2 in the liver, thereby impairing syndecan-1 HSPG function in remnant clearance. Hepatic overexpression of SULF2 provides a novel molecular mechanism that contributes to postprandial dyslipoproteinemia and hence arterial harm in T2DM and related disorders.


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

We thank Lawrence Chan (Baylor College of Medicine) for his generous gift of adiponectin knockout mice.


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  • 1
    Buse JB, Ginsberg HN, Bakris GL, Clark NG, Costa F, Eckel R, et al. Primary prevention of cardiovascular diseases in people with diabetes mellitus: a scientific statement from the American Heart Association and the American Diabetes Association. Diabetes Care 2007; 30: 162-172.
  • 2
    Williams KJ, Tabas I. The response-to-retention hypothesis of early atherogenesis. Arterioscler Thromb Vasc Biol 1995; 15: 551-561.
  • 3
    Dane-Stewart CA, Watts GF, Barrett PH, Stuckey BG, Mamo JC, Martins IJ, et al. Chylomicron remnant metabolism studied with a new breath test in postmenopausal women with and without type 2 diabetes mellitus. Clin Endocrinol (Oxf) 2003; 58: 415-420.
  • 4
    Hogue JC, Lamarche B, Tremblay AJ, Bergeron J, Gagne C, Couture P. Evidence of increased secretion of apolipoprotein B-48-containing lipoproteins in subjects with type 2 diabetes. J Lipid Res 2007; 48: 1336-1342.
  • 5
    Williams KJ. Molecular processes that handle—and mishandle—dietary lipids. J Clin Invest 2008; 118: 3247-3259.
  • 6
    Williams KJ, Chen K. Recent insights into factors affecting remnant lipoprotein uptake. Curr Opin Lipidol 2010; 21: 218-228.
  • 7
    Kita T, Goldstein JL, Brown MS, Watanabe Y, Hornick CA, Havel RJ. Hepatic uptake of chylomicron remnants in WHHL rabbits: a mechanism genetically distinct from the low density lipoprotein receptor. Proc Natl Acad Sci USA 1982; 79: 3623-3627.
  • 8
    Williams KJ, Fless GM, Petrie KA, Snyder ML, Brocia RW, Swenson TL. Mechanisms by which lipoprotein lipase alters cellular metabolism of lipoprotein(a), low density lipoprotein, and nascent lipoproteins. Roles for low density lipoprotein receptors and heparan sulfate proteoglycans. J Biol Chem 1992; 267: 13284-13292.
  • 9
    Ji ZS, Fazio S, Mahley RW. Variable heparan sulfate proteoglycan binding of apolipoprotein E variants may modulate the expression of type III hyperlipoproteinemia. J Biol Chem 1994; 269: 13421-13428.
  • 10
    Mann WA, Meyer N, Weber W, Meyer S, Greten H, Beisiegel U. Apolipoprotein E isoforms and rare mutations: parallel reduction in binding to cells and to heparin reflects severity of associated type III hyperlipoproteinemia. J Lipid Res 1995; 36: 517-525.
  • 11
    Ebara T, Conde K, Kako Y, Liu YZ, Xu Y, Ramakrishnan R, et al. Delayed catabolism of apoB-48 lipoproteins due to decreased heparan sulfate proteoglycan production in diabetic mice. J Clin Invest 2000; 105: 1807-1818.
  • 12
    MacArthur JM, Bishop JR, Stanford KI, Wang L, Bensadoun A, Witztum JL, et al. Liver heparan sulfate proteoglycans mediate clearance of triglyceride-rich lipoproteins independently of LDL receptor family members. J Clin Invest 2007; 117: 153-164.
  • 13
    Fuki IV, Kuhn KM, Lomazov IR, Rothman VL, Tuszynski GP, Iozzo RV, et al. The syndecan family of proteoglycans: novel receptors mediating internalization of atherogenic lipoproteins in vitro. J Clin Invest 1997; 100: 1611-1622.
  • 14
    Williams KJ, Fuki IV. Cell-surface heparan sulfate proteoglycans: dynamic molecules mediating ligand catabolism. Curr Opin Lipidol 1997; 8: 253-262.
  • 15
    Fuki IV, Meyer ME, Williams KJ. Transmembrane and cytoplasmic domains of syndecan mediate a multi-step endocytic pathway involving detergent-insoluble membrane rafts. Biochem J 2000; 351: 607-612.
  • 16
    Stanford KI, Bishop JR, Foley EM, Gonzales JC, Niesman IR, Witztum JL, et al. Syndecan-1 is the primary heparan sulfate proteoglycan mediating hepatic clearance of triglyceride-rich lipoproteins in mice. J Clin Invest 2009; 119: 3236-3245.
  • 17
    Olsson U, Egnell AC, Lee MR, Lunden GO, Lorentzon M, Salmivirta M, et al. Changes in matrix proteoglycans induced by insulin and fatty acids in hepatic cells may contribute to dyslipidemia of insulin resistance. Diabetes 2001; 50: 2126-2132.
  • 18
    Rohrbach DH, Hassell JR, Kleinman HK, Martin GR. Alterations in the basement membrane (heparan sulfate) proteoglycan in diabetic mice. Diabetes 1982; 31: 185-188.
  • 19
    Martins IJ, Tran JM, Redgrave TG. Food restriction normalizes chylomicron remnant metabolism in murine models of obesity as assessed by a novel stable isotope breath test. J Nutr 2002; 132: 176-181.
  • 20
    Comelli EM, Head SR, Gilmartin T, Whisenant T, Haslam SM, North SJ, et al. A focused microarray approach to functional glycomics: transcriptional regulation of the glycome. Glycobiology 2006; 16: 117-131.
  • 21
    Uchimura K, Morimoto-Tomita M, Rosen SD. Measuring the activities of the Sulfs: two novel heparin/heparan sulfate endosulfatases. Methods Enzymol 2006; 416: 243-253.
  • 22
    Lai JP, Sandhu DS, Yu C, Han T, Moser CD, Jackson KK, et al. Sulfatase 2 up-regulates glypican 3, promotes fibroblast growth factor signaling, and decreases survival in hepatocellular carcinoma. HEPATOLOGY 2008; 47: 1211-1222.
  • 23
    Chen K, Liu M-L, Schaffer L, Li M, Williams KJ. Type 2 diabetes strongly induces hepatic overexpression of SULF2, a novel factor that suppresses uptake of remnant lipoproteins [Abstract]. Atheroscler Suppl 2009; 10: e280.
  • 24
    Chen K, Liu M-L, Li M, Boden G, Wu X, Williams KJ. Metabolic factors in type 2 diabetes augment hepatocyte expression of SULF2, a novel suppressor of remnant lipoprotein uptake [Abstract]. Circulation 2009; 120(Suppl.):S1175.
  • 25
    Chung WK, Belfi K, Chua M, Wiley J, Mackintosh R, Nicolson M, et al. Heterozygosity for Lepob or Leprdb affects body composition and leptin homeostasis in adult mice. Am J Physiol 1998; 274: R985-R990.
  • 26
    Yamashita H, Shao J, Ishizuka T, Klepcyk PJ, Muhlenkamp P, Qiao L, et al. Leptin administration prevents spontaneous gestational diabetes in heterozygous Leprdb/+ mice: effects on placental leptin and fetal growth. Endocrinology 2001; 142: 2888-2897.
  • 27
    Lockhart DJ, Dong H, Byrne MC, Follettie MT, Gallo MV, Chee MS, et al. Expression monitoring by hybridization to high-density oligonucleotide arrays. Nat Biotechnol 1996; 14: 1675-1680.
  • 28
    Williams KJ, Liu M-L, Zhu Y, Xu X, Davidson WR, McCue P, et al. Loss of heparan N-sulfotransferase in diabetic liver: role of angiotensin II. Diabetes 2005; 54: 1116-1122.
  • 29
    Ma K, Cabrero A, Saha PK, Kojima H, Li L, Chang BH, et al. Increased beta-oxidation but no insulin resistance or glucose intolerance in mice lacking adiponectin. J Biol Chem 2002; 277: 34658-34661.
  • 30
    Williams KJ. Interactions of lipoproteins with proteoglycans. Methods Mol Biol 2001; 171: 457-477.
  • 31
    Bierman EL, Stein O, Stein Y. Lipoprotein uptake and metabolism by rat aortic smooth muscle cells in tissue culture. Circ Res 1974; 35: 136-150.
  • 32
    Goldstein JL, Basu SK, Brown MS. Receptor-mediated endocytosis of low-density lipoprotein in cultured cells. Methods Enzymol 1983; 98: 241-260.
  • 33
    Havel RJ, Eder HA, Bragdon JH. The distribution and chemical composition of ultracentrifually separated lipoproteins in human serum. J Clin Invest 1955; 34: 1345-1353.
  • 34
    Bilheimer DW, Eisenberg S, Levy RI. The metabolism of very low density lipoprotein proteins. I. Preliminary in vitro and in vivo observations. Biochim Biophys Acta 1972; 260: 212-221.
  • 35
    Boden G, Song W, Pashko L, Kresge K. In vivo effects of insulin and free fatty acids on matrix metalloproteinases in rat aorta. Diabetes 2008; 57: 476-483.
  • 36
    Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, Speed TP. Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res 2003; 31: e15.
  • 37
    Smyth GK. Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol 2004; 3: Article3.
  • 38
    Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Series B Stat Methodol 1995; 57: 289-300.
  • 39
    Goldberg IJ, Hu Y, Noh HL, Wei J, Huggins LA, Rackmill MG, et al. Decreased lipoprotein clearance is responsible for increased cholesterol in LDL receptor knockout mice with streptozotocin-induced diabetes. Diabetes 2008; 57: 1674-1682.
  • 40
    Anisfeld AM, Kast-Woelbern HR, Meyer ME, Jones SA, Zhang Y, Williams KJ, et al. Syndecan-1 expression is regulated in an isoform-specific manner by the farnesoid-X receptor. J Biol Chem 2003; 278: 20420-20428.
  • 41
    Zhang Y, Lee FY, Barrera G, Lee H, Vales C, Gonzalez FJ, et al. Activation of the nuclear receptor FXR improves hyperglycemia and hyperlipidemia in diabetic mice. Proc Natl Acad Sci USA 2006; 103: 1006-1011.
  • 42
    Grobe K, Esko JD. Regulated translation of heparan sulfate N-acetylglucosamine N-deacetylase/N-sulfotransferase isozymes by structured 5′-untranslated regions and internal ribosome entry sites. J Biol Chem 2002; 277: 30699-30706.
  • 43
    Parthasarathy N, Goldberg IJ, Sivaram P, Wagner WD. Isolation of heparin-derived oligosaccharides containing 2-O-sulfated hexuronic acids, by lipoprotein lipase affinity chromatography. J Biochem Biophys Methods 1996; 32: 27-32.
  • 44
    Stanford KI, Wang L, Castagnola J, Song D, Bishop JR, Brown JR, et al. Heparan sulfate 2-O-sulfotransferase is required for triglyceride-rich lipoprotein clearance. J Biol Chem 2010; 285: 286-294.
  • 45
    Frese MA, Milz F, Dick M, Lamanna WC, Dierks T. Characterization of the human sulfatase Sulf1 and its high affinity heparin/heparan sulfate interaction domain. J Biol Chem 2009; 284: 28033-28044.
  • 46
    Hotta K, Funahashi T, Arita Y, Takahashi M, Matsuda M, Okamoto Y, et al. Plasma concentrations of a novel, adipose-specific protein, adiponectin, in type 2 diabetic patients. Arterioscler Thromb Vasc Biol 2000; 20: 1595-1599.
  • 47
    Moreton JR. Atherosclerosis and alimentary hyperlipemia. Science 1947; 106: 190-191.
  • 48
    Zilversmit DB. Atherogenesis: a postprandial phenomenon. Circulation 1979; 60: 473-485.
  • 49
    Karpe F, Boquist S, Tang R, Bond GM, de Faire U, Hamsten A. Remnant lipoproteins are related to intima-media thickness of the carotid artery independently of LDL cholesterol and plasma triglycerides. J Lipid Res 2001; 42: 17-21.
  • 50
    Nordestgaard BG, Benn M, Schnohr P, Tybjaerg-Hansen A. Nonfasting triglycerides and risk of myocardial infarction, ischemic heart disease, and death in men and women. JAMA 2007; 298: 299-308.
  • 51
    Kannel WB, Vasan RS. Triglycerides as vascular risk factors: new epidemiologic insights. Curr Opin Cardiol 2009; 24: 345-350.
  • 52
    Proctor SD, Vine DF, Mamo JC. Arterial retention of apolipoprotein B48- and B100-containing lipoproteins in atherogenesis. Curr Opin Lipidol 2002; 13: 461-470.
  • 53
    Williams KJ, Tabas I. Lipoprotein retention–and clues for atheroma regression. Arterioscler Thromb Vasc Biol 2005; 25: 1536-1540.
  • 54
    Bansal S, Buring JE, Rifai N, Mora S, Sacks FM, Ridker PM. Fasting compared with nonfasting triglycerides and risk of cardiovascular events in women. JAMA 2007; 298: 309-316.
  • 55
    Freiberg JJ, Tybjaerg-Hansen A, Jensen JS, Nordestgaard BG. Nonfasting triglycerides and risk of ischemic stroke in the general population. JAMA 2008; 300: 2142-2152.
  • 56
    Ybarra J, James RW, Makoundou V, Bioletto S, Golay A. Effects of short-term modest weight loss on fasting and post-prandial lipoprotein sub-fractions in type 2 diabetes mellitus patients. Diabetes Metab 2001; 27: 701-708.