Dahai Zhang,1 Fulong Wang,1 Nathaniel Lal,1 Amy Pei-Ling Chiu,1 Andrea Wan,1

Jocelyn Jia,1 Denise Bierende,1 Stephane Flibotte,1 Sunita Sinha,1 Ali Asadi,2

Xiaoke Hu,2 Farnaz Taghizadeh,2 Thomas Pulinilkunnil,3 Corey Nislow,1

Israel Vlodavsky,4 James D. Johnson,2 Timothy J. Kieffer,2 Bahira Hussein,1 andBrian Rodrigues1

Heparanase Overexpression InducesGlucagon Resistance and ProtectsAnimals From Chemically InducedDiabetesDiabetes 2017;66:45–57 | DOI: 10.2337/db16-0761

Heparanase, a protein with enzymatic and nonenzymaticproperties, contributes toward disease progression andprevention. In the current study, a fortuitous observation intransgenic mice globally overexpressing heparanase (hep-tg) was the discovery of improved glucose homeostasis.We examined the mechanisms that contribute toward thisimproved glucose metabolism. Heparanase overexpres-sion was associated with enhanced glucose-stimulatedinsulin secretion and hyperglucagonemia, in addition tochanges in islet composition and structure. Strikingly, thepancreatic islet transcriptome was greatly altered inhep-tg mice, with >2,000 genes differentially expressedversus control. The upregulated genes were enriched fordiverse functions including cell death regulation, extracel-lular matrix component synthesis, and pancreatic hormoneproduction. The downregulated genes were tightly linkedto regulation of the cell cycle. In response to multiple low-dose streptozotocin (STZ), hep-tg animals developed lesssevere hyperglycemia compared with wild-type, an effectlikely related to their b-cells being more functionally effi-cient. In animals given a single high dose of STZ causingsevere and rapid development of hyperglycemia relatedto the catastrophic loss of insulin, hep-tg mice continuedto have significantly lower blood glucose. In these mice,protective pathways were uncovered for managinghyperglycemia and include augmentation of fibroblastgrowth factor 21 and glucagon-like peptide 1. This study

uncovers the opportunity to use properties of heparanasein management of diabetes.

Heparan sulfate proteoglycans (HSPGs), located mainly on thecell surface and in the extracellular matrix, are composed of acore protein to which one or more heparan sulfate (HS) sidechains are attached (1). HSPGs function not only as structuralproteins, but also as anchors for bioactive molecules, as HS isnegatively charged. Highly expressed in pancreatic islets, HSbinds and guides the signaling and distribution of fibroblastgrowth factor (FGF) family members, which regulate pancre-atic endocrine cell differentiation, clustering, and development(2). It has been suggested that the presence of HSPG in thenucleus has a suppressive effect on histone acetyltransferaseactivity and may therefore modulate gene expression (3).

Heparanase is an endo-b-D-glucuronidase that is ubiq-uitously expressed in many organs, with blood and endo-thelial cells having the highest expression. Heparanase isencoded as a 65-kDa latent precursor (HepL) that requiresproteolytic cleavage to form an active enzyme (HepA) (4).Functionally, HepA cleaves HS at D-glucuronic acid resi-dues, an action associated with extracellular matrix dis-ruption (5) and release of cell surface–bound moleculessuch as FGF (2). Aside from the function of HepA incleaving HS, HepL can also activate numerous signaling

1Faculty of Pharmaceutical Sciences, The University of British Columbia, Vancou-ver, British Columbia, Canada2Department of Cellular & Physiological Sciences, Life Sciences Institute, TheUniversity of British Columbia, Vancouver, British Columbia, Canada3Faculty of Medicine, Department of Biochemistry and Molecular Biology,Dalhousie University, Saint John, New Brunswick, Canada4Rappaport Faculty of Medicine, Cancer and Vascular Biology ResearchCenter, Technion, Haifa, Israel

Corresponding author: Brian Rodrigues, rodrigue@mail.ubc.ca.

Received 22 June 2016 and accepted 1 October 2016.

This article contains Supplementary Data online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db16-0761/-/DC1.

© 2017 by the American Diabetes Association. Readers may use this article aslong as the work is properly cited, the use is educational and not for profit, and thework is not altered. More information is available at http://www.diabetesjournals.org/content/license.

Diabetes Volume 66, January 2017 45

METABOLISM

elements, including extracellular signal–regulated kinase1/2, phosphoinositide 3-kinase–AKT, RhoA, and Src (6–8).These signaling events, in addition to the entry of HepA

into the nucleus to regulate histone acetylation by cleavingHS, have been suggested as mechanisms for modulatinggene expression (3).

Heparanase has physiological functions in wound heal-ing and hair growth (9). However, intensive research hasalso hinted toward an additional role of heparanase inboth disease progression and prevention. Thus, in cancer,degradation of HS chains by the increased expression ofheparanase is associated with extracellular matrix andbasement membrane (BM) disruption; loss of this physi-cal barrier facilitates tumor cell invasion (10). In the pan-creas, islet-enriched HS blocks immune cell infiltration,reduces T cell–mediated inflammation (11), and supportsb-cell survival and function (12). Thus, removal of HS byheparanase is critical for the initiation and progression ofautoimmune (type 1) diabetes (T1D) (11). Additionally,heparanase expression is induced in many organs duringdiabetes. In the heart, such inappropriate expression mayamplify fatty acid delivery and use. These events, if sus-tained, can lead to lipotoxicity and cardiovascular disease(13). In contrast to these observations, heparanase hasalso been shown to have beneficial effects in some dis-eases. For instance, overexpression of this protein is pro-tective against adriamycin-induced kidney injury (14). Inthe brain, fragmentation of HS because of heparanasecleavage reduced amyloid deposition in Alzheimer’s dis-ease (15). Similarly, amyloid formation in the islets, ahallmark of type 2 diabetes, could be ameliorated by hep-aranase-induced matrix metalloproteinase-9 expression(16,17). Matrix metalloproteinase-9 constitutes an endog-enous islet protease that limits islet amyloid depositionand its concomitant toxic effects via degradation of theamyloid polypeptide (18). Thus, precise temporal and spa-tial control of heparanase may be essential for normal cellphysiology.

In the current study, when monitoring the metabolismof transgenic mice that globally overexpress heparanase(hep-tg), we made the fortuitous observation that glucosehomeostasis was improved in these animals. Comparedwith wild-type (WT) controls, these mice had lower plasmainsulin and blood glucose levels; however, they exhibitedhigher glucagon concentrations, suggesting the presenceof glucagon resistance. Given that glucagon receptorknockout (GRKO) mice exhibited increased insulin sensi-tivity and resistance to diabetes (19,20), we followed upthese preliminary observations by examining whether hep-aranase overexpression could contribute toward improv-ing glucose metabolism as well as resistance to chemicallyinduced diabetes. Our data indicate that heparanase over-expression is associated with dramatic shifts in hormonessecreted from the pancreas, reorganization of islet com-position and structure, significant changes in islet geneexpression, and protection from streptozotocin (STZ)–induced diabetes.

RESEARCH DESIGN AND METHODS

Experimental AnimalsWT C57BL/6J mice aged 10–12 weeks were purchasedfrom Charles River Laboratories. Hep-tg mice, in whicha constitutive b-actin promoter drives the expression ofhuman heparanase gene in a C57BL/6J genetic back-ground, were a gift from I.V. All animals were housed inpathogen-free conditions on a 12-h light/dark cycle.Hep-tg mice were previously crossed for 10 generationswith C57BL/6J mice to produce a stable homozygousbackground. Male homozygous hep-tg mice aged 12–15 weeks were used for all experiments. For confirmationof genotype, genomic DNA was prepared from 21-day-oldanimal ear-punched tissue and analyzed by PCR (Fig. 1A)as previously detailed (9). All experiments were approvedby the University of British Columbia Animal Care Com-mittee and performed in accordance with the CanadianCouncil on Animal Care Guidelines.

Metabolic Assessments and Treatment of AnimalsFor measurement of basal blood glucose and plasmahormones, animals were fasted for 6 or 16 h and bloodsampled from the tail vein. Blood glucose was measuredusing an Accu-Chek glucose monitor (Roche, Basel,Switzerland). Blood was also collected using a heparinizedmicrotube (Thermo Fisher Scientific, Waltham, MA) andcentrifuged immediately for separation of plasma. Circu-lating hormones in the plasma were measured using thefollowing kits: mouse insulin (ALPCO, Salem, NH),glucagon and GLP-1 (Meso Scale Discovery, Rockville,MD), and FGF21 (R&D Systems, Minneapolis, MN). Forthe oral glucose tolerance test (OGTT) or intraperitoneal(i.p.) glucose tolerance test (IPGTT), animals were fastedovernight (16 h), 2 g/kg glucose was administered orallyor i.p., and blood glucose was measured at the indicatedtimes. A different cohort of animals was treated in thesame manner for the assessment of insulin following theOGTT. To perform the glucagon challenge, glucagon (1 mg/kgbody weight) was administered i.p. into 6- or 16-h–fastedmice and blood glucose measured at the indicated times.The insulin (0.75 units/kg body weight) and L-arginine(2 g/kg, i.p.) challenges were performed the same way in6-h–fasted mice. In addition to glucose measurement fol-lowing injection of L-arginine, plasma insulin and glucagonwere also measured at the indicated times. For evaluationof skeletal muscle sensitivity to insulin, mice were fastedfor 6 h prior to an i.p. injection of 2 mg/kg insulin. At10 min post–insulin injection, skeletal muscle (soleus andgastrocnemius) was isolated for Western blot determina-tion of phospho- and total Akt (Cell Signaling Technology,Danvers, MA), with b-actin (Santa Cruz Biotechnology,Dallas, TX) used as an internal control.

Diabetes InductionSTZ is a b-cell–selective toxin. Using STZ, we used twodifferent strategies to induce diabetes. With the first pro-tocol, we injected multiple low doses of STZ (MLD-STZ;50 mg/kg i.p.) for 5 consecutive days. This procedure

46 Heparanase Overexpression and Pancreatic Islet Diabetes Volume 66, January 2017

Figure 1—Heparanase overexpression improves glucose homeostasis. A: PCR amplification of DNA extracted from WT and hep-tg (TG)mice ear tissue. L19 is used as an internal control. B and C: Following a 6-h fast, plasma insulin and blood glucose levels were measuredin WT and TG mice. n = 9 to 10. *P < 0.05 vs. WT. D: Intraperitoneal administration of insulin (2 mg/kg body weight) to 6-h–fasted mice,followed by skeletal muscle isolation for Western blot. n = 3. *P < 0.05 vs. control; #P < 0.05 vs. WT insulin. E and F: Following anovernight fast (16 h), WT and TG mice were administered an oral glucose (2 g/kg) gavage. At the indicated times, blood glucose andinsulin were measured. n = 4 to 5. *P < 0.05 vs. WT. G: Isolated islets (10) from the different groups were used to extract insulin withacid-ethanol, and the insulin content was measured with ELISA (WT, 2.38 6 0.23 ng/mL; hep-tg, 3.08 6 1.04 ng/mL; P = 0.218).Different islets from the same animals were exposed to 3 and 16.7 mmol/L glucose, respectively, and insulin secretion determined. n = 3.*P < 0.05.

diabetes.diabetesjournals.org Zhang and Associates 47

induces gradual b-cell death, followed by an immune sys-tem response (21). Body weight and plasma glucose levelswere monitored daily after the first STZ injection. Meta-bolic assessment and organ isolation were performed1 week after the last STZ injection, following confirmationof hyperglycemia. We also used a single high dose of STZ(SHD-STZ; 200 mg/kg i.p.), which is directly cytotoxic tob-cells, resulting in robust hyperglycemia within 24–48 h(22).

Staining and QuantificationThe pancreas from normal or MLD-STZ mice was per-fused with PBS, harvested, fixed in 4% paraformaldehyde,and stored in 70% ethanol before paraffin embedding. Allparaffin sections (5-mm thickness) were processed by Wax-itHistology laboratories (Vancouver, British Columbia, Can-ada). Immunofluorescence staining and quantification of in-sulin (Cell Signaling Technology), glucagon (Sigma-Aldrich,St. Louis, MO), GLUT2 (Chemicon International, Temecula,CA), and synaptophysin (Novus Biologicals, Littleton, CO)were performed as previously described (23). Alcian blue(0.65 mol/L MgCl2 [pH 5.8]; Sigma-Aldrich) staining andimaging was used to visualize HS.

Western BlotWestern blot was done as described previously (24). Hep-aranase (Abcam, Cambridge, U.K.) and HS (Santa Cruz Bio-technology) expression in the pancreas was determinedusing the appropriate antibodies.

Islet IsolationPancreatic islets were isolated using collagenase as de-scribed previously (12,25). Briefly, the pancreas was perfusedwith collagenase through the pancreatic duct. Additionalcollagenase digestion was performed in a water bath. Fol-lowing digestion, the tube containing the tissue sample wasshaken vigorously. Individual islets were handpicked undera bright-field microscope. Islets were cultured for 3 h (37°C,5% CO2) in RPMI 1640 medium (Invitrogen, Carlsbad, CA)with 5 mmol/L glucose (Sigma-Aldrich), 100 units/mL pen-icillin, 100 mg/mL streptomycin (Invitrogen), and 10% FBS(Invitrogen). For RNA sequencing, 100 islets were selected,rinsed with PBS, and snap frozen for later isolation ofRNA. For glucose-induced insulin secretion, 10 islets wererinsed three times with PBS to remove glucose and insulinand then transferred to RPMI 1640 medium with either 3 or16.7 mmol/L glucose for 15 min, followed by medium col-lection and insulin determination using ELISA. The totalinsulin content in islets was determined following acid-ethanol extraction.

RNA Sequencing and AnalysisRNA from five hep-tg and five WT mice was isolated usingan RNeasy purification kit (Qiagen, Hilden, Germany).Sequencing libraries were prepared from 400 ng totalRNA using the TruSeq Stranded mRNA Sample Preparationkit according to the manufacturer’s instructions (Illumina,San Diego, CA). Samples were checked for quality usinga Bioanalyzer (Agilent Technologies) and quantified using

a Qubit fluorometer (Thermo Fisher Scientific). Librarieswere then multiplexed and sequenced over one rapid runlane on the HiSeq2500 (Illumina), collecting 89 million100-bp paired-end reads. Kallisto version 0.42.4 was firstused to build an index file for the mouse reference tran-scriptome GRCm38 as downloaded from the Ensembl website (http://www.ensembl.org). The sequence reads foreach sample were then quantified with the quant functionof Kallisto. In-house Perl scripts were used to sum theread counts at the transcript level for each gene and cre-ate a matrix comprising the read counts for all of thegenes for all of the samples. Differential expressionanalysis was then performed on the data from that matrixusing the R package DESeq2 (26). Each sample wasassessed using the quality-control software RSeQC version2.6.3 (27) and the PtR script from the trinity suite (28).One potential outlier was detected when clustering thesamples and therefore removed for the differential ex-pression analysis. RNA sequencing transcriptomic datawere analyzed using Panther and Database for Annota-tion, Visualization and Integrated Discovery (DAVID).Network analysis was conducted using STRING.

Statistical AnalysisStatistics were performed using Sigma Plot (Systat Soft-ware Inc., Chicago, IL). For all analyses, the Student t testor one-way ANOVA was used to determine differences amonggroup mean values. Values are presented as means 6 SEMwith individual data points. A P value ,0.05 was consid-ered statistically significant.

RESULTS

Hep-tg Mice Have Improved Glucose HomeostasisAt 3 months of age, hep-tg mice had similar body weightscompared with WT (hep-tg, 22.03 6 1.11 g; WT, 22.11 61.08 g). However, the determination of circulating plasmainsulin in hep-tg animals after a 6-h fast revealed lowerlevels compared with WT (Fig. 1B). Interestingly, thetransgenic animals also demonstrated reduced basal bloodglucose (Fig. 1C), suggesting higher insulin efficiency inthese mice. This was tested by administering a bolus doseof insulin and measuring Akt activation in skeletal muscle.Samples of skeletal muscle from hep-tg mice showed anenhanced insulin response (Fig. 1D). To pursue this ob-servation, we performed an OGTT after an overnight fast.Unlike in the case of an acute fast, prolonged fasting elim-inated the difference in blood glucose between the twogenotypes (Fig. 1E). In addition, although hep-tg micetended to clear glucose faster, this improvement was notstatistically significant (Fig. 1E), suggesting that the im-provement in insulin sensitivity observed under basalconditions is not readily apparent during an OGTT. Fol-lowing an insulin tolerance test, we were also unable todetect any difference in blood glucose lowering betweenthe two groups up to 30 min after insulin injection. Inter-estingly, after 30 min of insulin, we start to see a separation(recovery of glucose levels) between the WT and hep-tg mice

48 Heparanase Overexpression and Pancreatic Islet Diabetes Volume 66, January 2017

(Supplementary Fig. 1). Unexpectedly, in response to theglucose challenge, insulin secretion in hep-tg animals wasalmost twofold higher than in WT (Fig. 1F). In addition,when islets from WT and hep-tg mice were isolated andexposed to high glucose, hep-tg islets were capable of re-leasing greater amounts of insulin compared with WT (Fig.1G). Although hep-tg islets appeared to contain slightlymore insulin compared with WT, this difference was notstatistically significant (Supplementary Fig. 2). Similarly, asneither Ins 1 nor Ins 2 expression demonstrated any sig-nificant difference between the two strains (data notshown), our data suggest that heparanase overexpressioninfluences the secretory process of insulin rather than itssynthesis.

Heparanase Overexpression Induces GlucagonResistance and Changes in Pancreatic Islet StructureIntriguingly, hep-tg mice showed an augmentation in circu-lating basal glucagon levels (Fig. 2A and SupplementaryFig. 3). However, these mice demonstrated resistance whengiven a glucagon challenge after 6 (Supplementary Fig. 3)or 16 h (Fig. 2B) of fasting. The glucagon resistance inhep-tg mice was evident in the absence of any change inthe glucagon receptor in the liver (Supplementary Fig. 4).In response to L-arginine, which drives islet hormonesecretion independent of glucose sensing, hep-tg miceshowed higher glucagon and lower insulin secretion, ef-fects that were reflected in a higher glucose excursion inthese animals (Fig. 2C). To determine the basis for thisobserved hyperglucagonemia, we quantified pancreatica-cells, and their ratio to b-cells, using fluorescent stain-ing. Expressed as glucagon-positive (WT, 0.000465 60.00026; hep-tg, 0.000764 6 0.000318; P = 0.433) andinsulin-positive (WT, 0.004056 6 0.000991; hep-tg,0.0046086 0.001674; P = 0.056) areas individually (signal-positive area versus total area of the section to repre-sent the a- and b-cell number in the pancreas), therewas no statistical difference between WT and hep-tganimals in the number of insulin- or glucagon-positivecells. However, when the data were expressed as a ratioof a/b-cells, the hep-tg islets demonstrated a-cells thatwere more abundant. Additionally, unlike WT islets, inwhich a-cells are mostly found in the periphery andb-cells in the core, hep-tg islets contained a-cells thatwere randomly distributed (Fig. 2D). Islets in hep-tg miceexhibit similar average size and size distribution (Fig. 2E)and no difference in their number compared with WT(Supplementary Fig. 5). Overall, heparanase overexpres-sion was associated with dramatic shifts in hormonessecreted from the pancreas, in addition to changes in isletcomposition and structure.

Heparanase Overexpression Greatly Alters thePancreatic Islet TranscriptomeEntry of heparanase into the nucleus to regulate histoneacetylation has been put forth as a mechanism to modulatetranscription (3). Given the impact of heparanase overex-pression on insulin and glucagon, we sought to characterize

the hep-tg islet transcriptome using RNA-sequencing (Sup-plementary Table 1). As expected, heparanase was the mostprofoundly altered mRNA in the islet transcriptome, exhib-iting a.100-fold change (Fig. 3B). This was mirrored by anequally robust increase in heparanase protein expression(Fig. 3B, inset). Strikingly, after correction for multipletesting, .2,000 genes were significantly differentiallyexpressed with P , 0.05. Of these, 1,176 were upregu-lated, and 985 were downregulated (Fig. 3A, inset). Fromthe mRNAs that increased above a cutoff (more thantwofold; P , 0.001), we identified 350 genes enrichedin multiple cellular processes, including in development(e.g., Crlf1 and Ifrd1) (29), metabolism (e.g., Acacb andPpargc1a) (30), and cell death regulation (e.g., Npas4,Igf1, and P2rx1) (Fig. 3A) (31,32). Of note, glucagon(Gcg) expression increased (Fig. 3B) and correspondedto the high plasma glucagon observed in hep-tg mice.Among the 112 genes that decreased (more than twofold;P , 0.001), the most compelling effect was observed forgenes related to the cell cycle (e.g., Cdc20 and Ccnb) (Fig.3A). In addition, glucagon receptor (Gcgr) expression wasalso decreased (P = 0.0004) (Fig. 3B). Combining these datawith a protein–protein interaction network model, weidentified two functional networks that were upregulatedrelevant to the current study (Fig. 4, circles): one associ-ated with hormone secretion (e.g., Ppy and Pyy) and theother with extracellular structure (e.g., Sdc1 [that en-codes for syndecan 1] and Hs3st3b1 [that encodes forHS–glucosamine 3-sulfotransferase 3B1]) (Figs. 3B and4). The amplification of pancreatic HS that was observedin hep-tg mice (Fig. 3C) could be attributed, at least inpart, to the increase in Sdc1 and Hs3st3b1. Among thedownregulated genes, a highly connected network ofgenes related to the cell cycle function was recognized(Fig. 4, box). Altogether, these data suggest that its abil-ity to modulate gene expression is consistent with therole of heparanase as a potent regulator of islet structureand function.

Hep-tg Mice Exhibit Resistance to MLD-STZ–InducedHyperglycemiaAs hep-tg islets are enriched in HS, which is important forb-cell survival (12), we attempted to chemically inducediabetes in these mice. Interestingly, although WT ani-mals showed robust hyperglycemia within 1 week ofSTZ injection, hep-tg mice failed to present with plasmaglucose levels comparable to WT (Fig. 5A). Measurementof plasma insulin provided one explanation for this ob-servation. STZ caused a precipitous drop of plasma insulinin WT animals, an effect that was absent in hep-tg mice(Fig. 5B). Prolonging the duration of the study did notchange the results: hep-tg mice remained resistant toSTZ-induced hyperglycemia for up to 8 weeks followinginjection (Supplementary Fig. 6). This resistance to STZwas likely not a result of any difference in expression ofpancreatic GLUT2 (Supplementary Fig. 7), the transporterrequired for STZ uptake. To establish whether the failure

diabetes.diabetesjournals.org Zhang and Associates 49

Figure 2—Glucagon resistance and altered islet cytoarchitecture in heparanase-overexpressing mice. A: Plasma glucagon levels in WT andhep-tg (TG) mice were determined after a 16-h fast. n = 4 to 5. *P < 0.05 vs. WT. B: For the glucagon challenge, 16-h–fasted mice wereadministered glucagon (1 mg/kg), and blood glucose was measured at the indicated times. n = 5. *P< 0.05 vs. WT. C: In 6-h–fasted mice, 2 g/kgL-arginine was administered i.p. and plasma glucagon, insulin, and blood glucose (fold change in glucose was calculated by normalizing to thecorresponding value at t = 0) was measured at the indicated times. n = 5. *P < 0.05 vs. WT. D: Higher magnification of insulin- and glucagon-positive cells in pancreas isolated from the two groups of animals using immunofluorescence staining, together with quantification of the a/b-cellratio. n = 3. *P < 0.05 vs. WT. E: Whole pancreas sections were stained for insulin (red) and glucagon (green) and scanned, and islets werecounted and quantified by size.

50 Heparanase Overexpression and Pancreatic Islet Diabetes Volume 66, January 2017

Figure 3—Islet transcriptome changes dramatically in hep-tg mice. A: Islet RNA from WT and hep-tg (TG) mice was sequenced, and the(P < 0.001) upregulated (red) and downregulated (green) genes were categorized based on the functions. Bar represents the enrichment ofthe genes in particular category over the total genes of input. B: Representative genes in extracellular matrix, signaling, and cell cycle areshown. Heparanase gene was used as positive control of the entire RNA sequencing, together with islet protein detection as inset. C:Pancreas staining for HS using Alcian blue (pH 5.8, 0.65 mol/L MgCl2) and its quantification using Western blot. n = 3. *P < 0.05 vs. WT.

diabetes.diabetesjournals.org Zhang and Associates 51

to develop STZ-induced hyperglycemia was a consequenceof b-cell preservation, we quantified b-cell mass. Surpris-ingly, upon injection of STZ, both the WT and hep-tg micedisplayed an identical loss of insulin-positive cells (Fig.5C). This implies that the resistance to experimental di-abetes in hep-tg mice is not a consequence of b-cell sur-vival, consistent with observations seen in GRKO mice,which do not develop T1D when injected with STZ (19).Given our observation that hep-tg mice secrete more in-sulin in response to an OGTT (Fig. 1F), it is possible thatthe islets, which survived STZ toxicity in hep-tg diabeticmice, have a greater insulin secretory capacity. In thisregard, hep-tg mice injected with STZ were still, albeitin limited capacity, able to secrete insulin in response toan IPGTT (Fig. 5D). Taken together, our data suggest thatthe protective effects of heparanase against STZ-induceddiabetes is unlikely to be a consequence of HS-mediatedb-cell survival, but could be related to the ability of anyresidual b-cells in these animals to continue their secre-tion of insulin.

Severity of Diabetes Remains Disparate Between WTand Hep-tg Mice Injected With SHD-STZMLD-STZ induces b-cell death by activating immunemechanisms (21). Hence, we also tested the response ofhep-tg mice to SHD-STZ, which produces severe diabetesby direct b-cell toxicity via DNA alkylation (22). In con-trast to our observation with MLD-STZ, both WT andhep-tg mice injected with 200 mg/kg STZ (SHD) exhibiteda significant reduction in plasma insulin levels (Fig. 6B).In agreement with the loss of insulin, WT diabetic micedemonstrated sustained hyperglycemia (Fig. 6A). Signifi-cantly, the values from day 3 onwards for these mice werebeyond the upper limit of detection (33.3 mmol/L) of theglucometer, suggesting that the actual blood glucose con-centrations were likely higher. Although high-dose STZwas competent to lower insulin and induce diabetes inhep-tg mice, the magnitude of hyperglycemia was lowerthan that seen in WT animals (Fig. 6A), suggesting thecontribution of an insulin mimetic factor. FGF21 has sig-nificant beneficial effects on glucose homeostasis (33).

Figure 4—Association network of genes that were significantly different between WT and hep-tg mice. Analysis of a protein–proteininteraction network assembled from RNA sequencing data. The blue square depicts the entire downregulated network that is related tocell cycle. The red circles encompass both the hormone-associated network and the extracellular elements network. Lines representassociations based on differential expression evidence.

52 Heparanase Overexpression and Pancreatic Islet Diabetes Volume 66, January 2017

Figure 5—Hep-tg mice are resistant to MLD-STZ–induced diabetes. A and B: To induce T1D, WT and hep-tg (TG) mice were administered50 mg/kg STZ for 5 consecutive days. Seven days after the last STZ injection and following a 6-h fast, tail vein blood was used for glucoseand plasma insulin determination. n = 5–10. *P < 0.05. C: Pancreas from STZ-injected mice was isolated and immune stained for insulin(red) and glucagon (green), and insulin-positive cells were quantified in the different groups. n = 3. *P < 0.05. vs. control (Con). D: Followingan IPGTT, plasma insulin in STZ-injected WT and hep-tg mice was measured over the indicated times. The data are expressed as a foldchange normalized to the corresponding values at t = 0 min for each animal within the two groups. n = 3–5. *P < 0.05 vs. WT.

diabetes.diabetesjournals.org Zhang and Associates 53

Measurement of FGF21 plasma concentration indicatedthat hep-tg mice had higher concentrations comparedwith WT (Fig. 6C). Moreover, although STZ-induced di-abetes caused a significant drop in FGF21 levels in WTmice, its levels in hep-tg mice were strikingly amplified bythreefold (Fig. 6C) and likely related to its increased he-patic production (Supplementary Fig. 8). Another hormonethat has glucose-lowering effects independent of insulin isGLP-1 (34). In response to high-dose STZ, GLP-1 trendedin the opposite direction between these two strains (Fig.6D). Thus, in hep-tg mice, driving circulating insulin con-centrations down to very low levels uncovers novel pro-tective pathways for the management of hyperglycemia.

DISCUSSION

Homozygous transgenic mice overexpressing human hep-aranase globally are fertile and have a normal life span (9).In addition, physiological functions associated with hepar-anase in these mice include roles in embryonic implantation,food consumption, tissue remodeling, and vascularization(9). In the current study, an unexpected feature of these

animals was our discovery of improved glucose homeosta-sis, but resistance to glucagon. Strikingly, compared withcontrols, hep-tg mice manifested significant changes in is-let gene expression of.2,000 genes. These mice also dem-onstrated resistance to chemically induced diabetes. Ourdata suggest a novel role for heparanase in mechanismsthat serve to correct hyperglycemia.

In hep-tg mice, both basal insulin and glucose levelswere lower than in WT following a 6-h fast, suggestingsuperior insulin sensitivity in these animals. This idea wasreinforced by an enhanced insulin response to activationof skeletal muscle Akt in this study. Supporting theseobservations is that, in myeloma cells, insulin receptor isthe predominant receptor tyrosine kinase activated byheparanase (6). The lower glucose concentration was ev-ident in the presence of high glucagon levels, suggesting adeficit in glucagon action. Interestingly, an insulin toler-ance test used to evaluate insulin efficiency was unableto detect any difference in blood glucose lowering be-tween the two groups. Moreover, the traditional OGTTwas also unable to confirm a higher rate of glucose disposal

Figure 6—SHD-STZ in hep-tg mice uncovers protective responses to manage hyperglycemia. A: SHD-STZ (200 mg/kg) was administeredi.p. and glucose levels monitored daily over 5 days. n = 4–8. *P < 0.05. B–D: One week after injection of SHD-STZ, insulin, FGF21, andGLP-1 levels from WT and hep-tg (TG) mice were determined. n = 4–8. *P < 0.05.

54 Heparanase Overexpression and Pancreatic Islet Diabetes Volume 66, January 2017

in hep-tg mice. We reasoned that the overnight fast priorto performing the OGTT could explain this inconsistency.Following fasting, hypersensitivity of the liver to glucagonaccounts for a multifold increase in hepatic glucose pro-duction (34). Should this happen, even marginally, inhep-tg mice that have higher basal glucagon levels, theadvantage of augmented insulin sensitivity in these ani-mals would be dampened, resulting in a similar glucoseclearance compared with WT. Remarkably, in hep-tgmice, the OGTT elicited higher insulin release, whereasisolated islets from these animals had a more robustsecretion of insulin in response to glucose, even with isletinsulin content and gene expression remaining un-changed. Our data suggest that although heparanase ap-pears to enhance insulin signaling, whole-body tests todetermine insulin sensitivity were unable to confirmthis, likely because of the overriding influence of an insulincounterregulatory hormone such as glucagon in the hep-tgmice.

In GRKO mice, a prolonged deficiency in glucagonreceptor signaling activates a-cell proliferation and leadsto a compensatory elevation in glucagon production (35).We were surprised to observe that these characteristics inGRKO mice were strikingly similar to features seen inhep-tg animals, which did not respond to a glucagon chal-lenge as well as WT mice. More importantly, hep-tg isletspresented an architecture that is conducive to hypersecre-tion of glucagon, under basal conditions, and also subse-quent to injection of arginine. In these mice, a-cells losttheir peripheral mantle-like localization and were distrib-uted randomly throughout the islets. In addition, thea/b-cell ratio was higher in hep-tg mice. These changesin a-cells were evident even though islet size and sizedistribution remained unaffected. Currently, it is unclearwhether these alterations in a-cell morphology are conse-quences of glucagon resistance or of disrupted pancreaticislet development through the action of heparanase onHS.

Measurement of HS in the pancreas produced anunanticipated result; rather than depleting HS, the hep-tgpancreas displayed the opposing phenotype, with in-creased expression of HS, and the gene (Hs3st3b1) thatencodes for its biosynthetic enzyme, 3-O-sulfotransferase,being dominant features. A similar observation has alsobeen made in the hep-tg liver, in which the authors de-scribed the accumulation of “heparin-like” HS degradationproducts, both within and between cells, resulting in anaccelerated biosynthesis of HS (36). It follows that as inthe liver, pancreatic heparanase processes HS into smallerheparin-like oligosaccharides. Regardless of whether a re-sult of its enzymatic activities or its nontraditional sig-naling activities, overexpression of heparanase produced adramatic alteration in the islet transcriptome. Among theclusters of genes that were increased, of particular in-terest were genes that regulate cell death, synthesizeextracellular matrix components, and produce pancreatichormones. Related to cell death, this was not unusual, as

cancer cells use the properties of heparanase to inducegene expression and cell survival (37). The impact on theextracellular matrix is especially meaningful, as compo-nents like collagen and laminin have roles in islet celldifferentiation, proliferation, and hormone secretion(38). Finally, peptide hormones like somatostatin andpancreatic polypeptide hormone are known to influencepancreatic islet function in a paracrine manner (39,40).Unlike the upregulated genes that were enriched for di-verse functions, the downregulated genes formed a coher-ent functional group, including those tightly linked to thecell cycle. Whether these changes in expression create anenergy-sparing environment that would be conducive to ahigher level of hormone production, as recently reported(41), is certainly consistent with our observations anddeserving of further investigation. Collectively, our dataprovide support for the notion that heparanase in thepancreas is particularly beneficial, reinforcing it againstthe effects of exogenous stress.

MLD-STZ mimics T1D by stimulating b-cell apoptosisthrough the recruitment of immune cells (21). Given therole that HS plays in modulating the innate immune re-sponse by effecting immune cell adhesion and cytokineand chemokine binding and providing a physical barrieragainst leukocyte infiltration (42), an increase in HS asseen in hep-tg mice could be potentially beneficial. This,when added to the genomic signature in these mice that isprotective against cell death, would anticipate a resistanceto MLD-STZ b-cell cytotoxicity. Indeed, our results, forthe first time, demonstrate that hep-tg animals developedless severe hyperglycemia compared with WT and thatthis was associated with an almost unchanged circulatinginsulin concentration. Inexplicably, the preserved plasmainsulin in these animals was not a consequence of higherb-cell mass, as quantification of these cells using immu-nostaining indicated that hep-tg b-cells were destroyed toan equal extent compared with WT. It is possible that, ofthe b-cells that survived STZ toxicity in both groups, anenhanced insulin secretory capacity in hep-tg animalscould explain this anomaly. In support, hep-tg diabeticanimals can still increase insulin secretion in responseto a glucose challenge, an effect that is lost in WT diabeticmice. Our data suggest that heparanase produces b-cellsthat are more functionally efficient, an effect likely relatedto an elevation of 3-O-sulfotransferases and accumulationof HS in these islets.

STZ in high doses induces diabetes by a mechanismdifferent from MLD-STZ; it directly destroys b-cells byalkylating DNA (22). Additionally, as there is a compara-tively more severe and rapid development of hyperglyce-mia related to a catastrophic loss of insulin, we testedwhether heparanase would continue to correct hypergly-cemia in this model. Despite the identical low levels ofresidual circulating insulin in both diabetic groups, hep-tgmice continued to have significantly lower blood glucose.In GRKO mice, FGF21 and GLP-1 have been suggested tobe the hormones responsible for glucose clearance in the

diabetes.diabetesjournals.org Zhang and Associates 55

absence of insulin (43). This was also true in the hep-tgdiabetic mouse. In these animals, regardless of their basalconcentrations, SHD-STZ caused an augmentation ofboth FGF21 and GLP-1, effects that can contribute totheir lower glucose levels. Altogether, the results indicatethat, with the onset of hyperglycemia, additional glucose-lowering mechanisms are triggered in hep-tg mice.

It has been proposed that HS in pancreatic islet BMfunctions as an obstruction against leukocyte infiltration,in addition to protecting the b-cell against reactive oxy-gen species–induced cell death (44). Thus, in the NODmouse, augmented production of active heparanase byimmune cells permits destruction of islet BM HS, entryof leukocytes, and degradation of intracellular HS. As theheparanase inhibitor PI-88 preserved intraislet HS andprotected NOD mice from T1D, the authors concludedthat heparanase inhibition is useful against T1D progres-sion (12). Contrasting results were observed in NOD miceinjected with exogenous heparanase, which amelioratedthe occurrence of diabetes (45). Unlike the NOD miceor metastatic cancer cells (46), in which the overproduc-tion of active heparanase is responsible for disease devel-opment, our model is one in which, in a disease-freebackground, latent heparanase is overexpressed, with anassociated amplification of islet HS. Thus, although activeheparanase has been considered a pathogenic marker, ourstudy has discovered multiple novel properties of latentheparanase related to the control of glycaemia. Overexpres-sion of this protein caused glucose lowering, potentiationof insulin secretion, HS induction, prodigious changes inislet gene expression, and protection against chemicallyinduced diabetes. This study unlocks the possibility ofusing these properties of heparanase in the managementof diabetes.

Funding. This work was supported by an operating grant from the CanadianInstitutes of Health Research (CIHR-MOP-133547 to B.R.). D.Z. and A.P.-L.C. arethe recipients of Doctoral Student Research Awards from the Canadian DiabetesAssociation.Duality of Interest. No potential conflicts of interest relevant to this articlewere reported.Author Contributions. D.Z. performed most of the experiments andwith B.R. generated the hypothesis, designed the study, performed the dataanalysis, and wrote the manuscript. F.W., N.L., A.P.-L.C., A.W., J.J., D.B., A.A.,X.H., F.T., and B.H. contributed in part to some of the experiments performed.S.F., S.S., and C.N. performed RNA sequencing and its analysis. T.P., I.V., J.D.J.,and T.J.K. contributed to the discussion and manuscript editing. B.R. is theguarantor of this work and, as such, had full access to all the data in the studyand takes responsibility for the integrity of the data and the accuracy of the dataanalysis.

References1. Sarrazin S, Lamanna WC, Esko JD. Heparan sulfate proteoglycans. ColdSpring Harb Perspect Biol 2011;3:a0049522. Gittes GK. Developmental biology of the pancreas: a comprehensive review.Dev Biol 2009;326:4–353. Purushothaman A, Hurst DR, Pisano C, Mizumoto S, Sugahara K, SandersonRD. Heparanase-mediated loss of nuclear syndecan-1 enhances histone

acetyltransferase (HAT) activity to promote expression of genes that drive anaggressive tumor phenotype. J Biol Chem 2011;286:30377–303834. Wu L, Viola CM, Brzozowski AM, Davies GJ. Structural characterization ofhuman heparanase reveals insights into substrate recognition. Nat Struct Mol Biol2015;22:1016–10225. Vlodavsky I, Ilan N, Naggi A, Casu B. Heparanase: structure, biologicalfunctions, and inhibition by heparin-derived mimetics of heparan sulfate. CurrPharm Des 2007;13:2057–20736. Purushothaman A, Babitz SK, Sanderson RD. Heparanase enhances theinsulin receptor signaling pathway to activate extracellular signal-regulated ki-nase in multiple myeloma. J Biol Chem 2012;287:41288–412967. Riaz A, Ilan N, Vlodavsky I, Li JP, Johansson S. Characterization of hep-aranase-induced phosphatidylinositol 3-kinase-AKT activation and its integrindependence. J Biol Chem 2013;288:12366–123758. Zetser A, Bashenko Y, Edovitsky E, Levy-Adam F, Vlodavsky I, Ilan N.Heparanase induces vascular endothelial growth factor expression: correlationwith p38 phosphorylation levels and Src activation. Cancer Res 2006;66:1455–14639. Zcharia E, Metzger S, Chajek-Shaul T, et al. Transgenic expression ofmammalian heparanase uncovers physiological functions of heparan sulfate intissue morphogenesis, vascularization, and feeding behavior. FASEB J 2004;18:252–26310. Arvatz G, Weissmann M, Ilan N, Vlodavsky I. Heparanase and cancer pro-gression: new directions, new promises. Hum Vaccin Immunother 2016;12:2253–225611. Bogdani M, Korpos E, Simeonovic CJ, Parish CR, Sorokin L, Wight TN.Extracellular matrix components in the pathogenesis of type 1 diabetes. Curr DiabRep 2014;14:55212. Ziolkowski AF, Popp SK, Freeman C, Parish CR, Simeonovic CJ. Heparansulfate and heparanase play key roles in mouse b cell survival and autoimmunediabetes. J Clin Invest 2012;122:132–14113. Wan A, Rodrigues B. Endothelial cell-cardiomyocyte crosstalk in diabeticcardiomyopathy. Cardiovasc Res 2016;111:172–18314. Assady S, Alter J, Axelman E, et al. Nephroprotective effect of heparanase inexperimental nephrotic syndrome. PLoS One 2015;10:e011961015. Jendresen CB, Cui H, Zhang X, Vlodavsky I, Nilsson LN, Li JP. Over-expression of heparanase lowers the amyloid burden in amyloid-b precursorprotein transgenic mice. J Biol Chem 2015;290:5053–506416. Purushothaman A, Chen L, Yang Y, Sanderson RD. Heparanase stimulationof protease expression implicates it as a master regulator of the aggressive tumorphenotype in myeloma. J Biol Chem 2008;283:32628–3263617. Aston-Mourney K, Zraika S, Udayasankar J, et al. Matrix metalloproteinase-9 reduces islet amyloid formation by degrading islet amyloid polypeptide. J BiolChem 2013;288:3553–355918. Meier DT, Tu LH, Zraika S, et al. Matrix Metalloproteinase-9 Protects Isletsfrom Amyloid-induced Toxicity. J Biol Chem 2015;290:30475–3048519. Sørensen H, Winzell MS, Brand CL, et al. Glucagon receptor knockout micedisplay increased insulin sensitivity and impaired beta-cell function. Diabetes2006;55:3463–346920. Conarello SL, Jiang G, Mu J, et al. Glucagon receptor knockout mice areresistant to diet-induced obesity and streptozotocin-mediated beta cell loss andhyperglycaemia. Diabetologia 2007;50:142–15021. McEvoy RC, Andersson J, Sandler S, Hellerström C. Multiple low-dosestreptozotocin-induced diabetes in the mouse. Evidence for stimulation of acytotoxic cellular immune response against an insulin-producing beta cell line.J Clin Invest 1984;74:715–72222. Deeds MC, Anderson JM, Armstrong AS, et al. Single dose streptozotocin-induced diabetes: considerations for study design in islet transplantation models.Lab Anim 2011;45:131–14023. Asadi A, Bruin JE, Kieffer TJ. Characterization of antibodies to products ofproinsulin processing using immunofluorescence staining of pancreas in multiplespecies. J Histochem Cytochem 2015;63:646–662

56 Heparanase Overexpression and Pancreatic Islet Diabetes Volume 66, January 2017

24. An D, Pulinilkunnil T, Qi D, Ghosh S, Abrahani A, Rodrigues B. The metabolic“switch” AMPK regulates cardiac heparin-releasable lipoprotein lipase. Am JPhysiol Endocrinol Metab 2005;288:E246–E25325. Alejandro EU, Kalynyak TB, Taghizadeh F, et al. Acute insulin signaling inpancreatic beta-cells is mediated by multiple Raf-1 dependent pathways. En-docrinology 2010;151:502–51226. Love MI, Huber W, Anders S. Moderated estimation of fold change anddispersion for RNA-seq data with DESeq2. Genome Biol 2014;15:55027. Wang L, Wang S, Li W. RSeQC: quality control of RNA-seq experiments.Bioinformatics 2012;28:2184–218528. Grabherr MG, Haas BJ, Yassour M, et al. Full-length transcriptome as-sembly from RNA-Seq data without a reference genome. Nat Biotechnol 2011;29:644–65229. Zimmer Y, Milo-Landesman D, Svetlanov A, Efrat S. Genes induced bygrowth arrest in a pancreatic beta cell line: identification by analysis of cDNAarrays. FEBS Lett 1999;457:65–7030. Oropeza D, Jouvet N, Bouyakdan K, et al. PGC-1 coactivators in b-cellsregulate lipid metabolism and are essential for insulin secretion coupled to fattyacids. Mol Metab 2015;4:811–82231. Sabatini PV, Krentz NAJ, Zarrouki B, et al. Npas4 is a novel activity-regulated cytoprotective factor in pancreatic b-cells. Diabetes 2013;62:2808–282032. Lingohr MK, Buettner R, Rhodes CJ. Pancreatic beta-cell growth and sur-vival–a role in obesity-linked type 2 diabetes? Trends Mol Med 2002;8:375–38433. Kwon MM, O’Dwyer SM, Baker RK, Covey SD, Kieffer TJ. FGF21-MediatedImprovements in Glucose Clearance Require Uncoupling Protein 1. Cell Reports2015;13:1521–152734. Dardevet D, Moore MC, Neal D, DiCostanzo CA, Snead W, Cherrington AD.Insulin-independent effects of GLP-1 on canine liver glucose metabolism: du-ration of infusion and involvement of hepatoportal region. Am J Physiol EndocrinolMetab 2004;287:E75–E81

35. Gelling RW, Du XQ, Dichmann DS, et al. Lower blood glucose, hyper-glucagonemia, and pancreatic alpha cell hyperplasia in glucagon receptorknockout mice. Proc Natl Acad Sci U S A 2003;100:1438–144336. Escobar Galvis ML, Jia J, Zhang X, et al. Transgenic or tumor-inducedexpression of heparanase upregulates sulfation of heparan sulfate. Nat Chem Biol2007;3:773–77837. Nadir Y, Brenner B. Heparanase procoagulant activity in cancer progression.Thromb Res 2016;140(Suppl. 1):S44–S4838. Meda P. Protein-mediated interactions of pancreatic islet cells. Scientifica(Cairo) 2013;2013:62124939. Hauge-Evans AC, King AJ, Carmignac D, et al. Somatostatin secreted byislet delta-cells fulfills multiple roles as a paracrine regulator of islet function.Diabetes 2009;58:403–41140. Kahleova H, Mari A, Nofrate V, et al. Improvement in b-cell function afterdiet-induced weight loss is associated with decrease in pancreatic polypeptide insubjects with type 2 diabetes. J Diabetes Complications 2012;26:442–44941. Szabat M, Page MM, Panzhinskiy E, et al. Reduced insulin production re-lieves endoplasmic reticulum stress and induces b cell proliferation. Cell Metab2016;23:179–19342. Simon Davis DA, Parish CR. Heparan sulfate: a ubiquitous glycosamino-glycan with multiple roles in immunity. Front Immunol 2013;4:47043. Omar BA, Andersen B, Hald J, Raun K, Nishimura E, Ahrén B. Fibroblastgrowth factor 21 (FGF21) and glucagon-like peptide 1 contribute to diabetesresistance in glucagon receptor-deficient mice. Diabetes 2014;63:101–11044. Simeonovic CJ, Ziolkowski AF, Wu Z, Choong FJ, Freeman C, Parish CR.Heparanase and autoimmune diabetes. Front Immunol 2013;4:47145. Bitan M, Weiss L, Zeira M, et al. Heparanase prevents the development oftype 1 diabetes in non-obese diabetic mice by regulating T-cell activation andcytokines production. Diabetes Metab Res Rev 2008;24:413–42146. Vlodavsky I, Friedmann Y. Molecular properties and involvement of hepar-anase in cancer metastasis and angiogenesis. J Clin Invest 2001;108:341–347

diabetes.diabetesjournals.org Zhang and Associates 57