Heparan sulfate proteoglycans in beta cells provide a critical link between endoplasmic reticulum stress, oxidative stress and type 2 diabetes

Heparan sulfate proteoglycans (HSPGs) consist of a core protein with side chains of the glycosaminoglycan heparan sulfate (HS). We have previously identified (i) the HSPGs syndecan-1 (SDC1), and collagen type XVIII (COL18) inside mouse and human islet beta cells, and (ii) a critical role for HS in beta cell survival and protection from reactive oxygen species (ROS). The objective of this study was to investigate whether endoplasmic reticulum (ER) stress contributes to oxidative stress and type 2 diabetes (T2D) by depleting beta cell HSPGs/HS. A rapid loss of intra-islet/beta cell HSPGs, HS and heparanase (HPSE, an HS-degrading enzyme) accompanied upregulation of islet ER stress gene expression in both young T2D-prone db/db and Akita Ins2WT/C96Y mice. In MIN6 beta cells, HSPGs, HS and HPSE were reduced following treatment with pharmacological inducers of ER stress (thapsigargin or tunicamycin). Treatment of young db/db mice with Tauroursodeoxycholic acid (TUDCA), a chemical protein folding chaperone that relieves ER stress, improved glycemic control and increased intra-islet HSPG/HS. In vitro, HS replacement with heparin (a highly sulfated HS analogue) significantly increased the survival of wild-type and db/db beta cells and restored their resistance to hydrogen peroxide-induced death. We conclude that ER stress inhibits the synthesis/maturation of HSPG core proteins which are essential for HS assembly, thereby exacerbating oxidative stress and promoting beta cell failure. Diminished intracellular HSPGs/HS represent a previously unrecognized critical link bridging ER stress, oxidative stress and beta cell failure in T2D.

Introduction Type 2 diabetes (T2D) is a metabolic syndrome disease in which obesity, insulin resistance, hyperglycemia, dyslipidemia and chronic low-grade inflammation of pancreatic islets contribute directly or indirectly to beta cell failure [1,2]. Under normal conditions, pancreatic beta cells are highly metabolically active, due to their specialized role in synthesizing proinsulin/ insulin. The ER is a subcellular compartment that functions to chemically modify and fold newly synthesized proteins, including proinsulin, into their mature state. Metabolic dysfunction in T2D and insulin (INS) gene mutations disrupt ER homeostasis, resulting in the accumulation of misfolded proteins in the ER lumen and ER stress [3][4][5]. ER stress is exacerbated by the reduced availability of protein-folding chaperones, inhibition of protein glycosylation, depletion of calcium ions from the ER and local oxidative stress [6,7]. The unfolded protein response (UPR) is an adaptive response for relieving ER stress by decreasing "global" protein synthesis (at the level of translation) and increasing the transcription of protective genes e.g., ER chaperones [6,7]. The capacity for the UPR to prolong beta function is determined by a delicate balance between these two processes.
Oxidative stress is also a key disorder outside the ER in T2D beta cells and is a consequence of increased metabolic demand [5,[8][9][10]. Under normoglycemic conditions, beta cells produce ROS at multiple sites by (i) oxidative phosphorylation in mitochondria, (ii) disulfide bond formation during proinsulin protein folding in the ER [11] and (iii) NADPH oxidase, localized at the cell membrane and in insulin granules [8,12]. Neutralization of intracellular ROS is an essential process for preventing oxidative damage and preserving beta cell integrity. However, beta cells have low basal levels of antioxidant enzymes, relative to other tissues [13,14], suggesting that a compensatory antioxidant mechanism exists. Recently we discovered that normal mouse and human beta cells have an unusually high intracellular level of highly sulfated heparan sulfate (HS), which is critical for their survival [15,16]. Mechanistically, beta cell HS provides a constitutive mechanism for quenching intracellular ROS generated during normal metabolism [15,16]. In T2D, oxidative stress and ER stress contribute to beta cell failure [4][5][6][17][18][19], with elevated ROS representing a mediator or by-product, respectively, in different sub-cellular compartments [5]. This dysregulation of ROS suggests that the antioxidant capacity of HS is impaired in beta cells in T2D.
Beta cell HS deteriorates acutely during the isolation of normal islets in vitro and chronically during the development of type 1 diabetes (T1D), due to excessive oxidative damage and HS degradation by leukocyte-derived HPSE, respectively [15,16]. In vitro, replacement of lost HS using HS mimetics preserves the survival of isolated beta cells and provides protection against hydrogen peroxide (ROS)-induced cell death [15,16]. Supporting a mechanistic role for HPSE in T1D, in vivo treatment of NOD mice with PI-88 (a heparanase inhibitor/HS replacer), reduced the incidence of diabetes, improved islet pathology and preserved beta cell HS [15]. We also found that beta cell HSPG core proteins were decreased in human pancreases from donors with T1D [16], potentially due to protease(s) produced by invading leukocytes and/or ER stress [33][34][35]. We reason that the substantial loss of beta cell HS following islet/ beta cell isolation in vitro and in T1D increases the sensitivity of beta cells to oxidative (ROSmediated) damage [15,16,28]. Against this background we have explored whether beta cell HSPGs and HS are altered in T2D and impact cell viability.
In this study we investigate the relationship between spontaneous ER stress and beta cell HSPGs/HS in the db/db mouse model of obesity-induced T2D and in Akita (Ins2 WT/C96Y ) mice. In addition, we experimentally induce ER stress in MIN6 beta cells in vitro and monitor intracellular HSPGs/HS. Intervention studies further interrogate the relationship between (i) ER stress and beta cell HSPGs/HS in db/db mice treated with TUDCA to relieve ER stress and (ii) intracellular HS, oxidative stress and the viability of db/db beta cells in vitro using HS replacement and acute exposure to hydrogen peroxide. Our findings reveal a previously unrecognized inter-relationship between ER stress, HSPGs, HS and oxidative stress in beta cells.

Animals
C57BL/KsJ db/db, heterozygous (db/+) and wild-type (wt; +/+) mice, heterozygous Akita (Ins2 WT/C96Y ) and wild-type Ins2 WT/WT mice were obtained from the Australian Phenomics Facility, The Australian National University, Canberra, ACT, Australia. Db/db mice carry a homozygous point mutation in the leptin receptor gene and develop obesity and hyperglycemia [36]; heterozygous Akita mice have a point mutation in the insulin 2 (Ins2) gene, resulting in misfolding of proinsulin, ER stress in beta cells and rapid onset of diabetes [37]. The mice were housed under specific pathogen free conditions in individually ventilated cages (up to 5 mice/cage). All mice were maintained with irradiated standard rodent chow and water ad libitum with environmental conditions set at~19˚C and an automated 12 hr light/dark cycle. Groups of male db/db mice at 4 weeks of age were randomly selected and treated daily (a.m.) with TUDCA (Tauroursodeoxycholic acid, 150 mg/kg; Calbiochem, EMD Millipore, Darmstadt, Germany) i.p. for 4 weeks [38,39]; control mice were treated with 8 μl/g saline (diluent) i.p. Non-fasting blood glucose levels and body weight were monitored up to 3x/week (a.m.). After termination of treatment, HbA1c levels were measured using a HemoCue HbA1c 501 analyzer (Angelholm, Sweden). Mice were euthanized by cervical dislocation at the completion of treatment and/or for pancreas harvest. All procedures and experiments were approved by The Australian National University Animal Experimentation Ethics Committee.

Real-time RT-PCR
Two-step qRT-PCR was used to measure transcripts of ER stress markers (Bip, p58, Chop and Atf3) in isolated islets and MIN6 cells. Total RNA was extracted from isolated islets (100-743 islets) using the RNeasy kit (Qiagen, Hilden, Germany) and cDNA was synthesized using Superscript RNase reverse transcriptase III (Invitrogen, Carlsbad, CA, USA). Real-time PCR was performed using TaqMan gene expression assays with FAM-labelled TaqMan 1 MGB probe (ThermoFisher, Waltham, MA, USA; S1 Table) and a AB17900HT real-time PCR machine with 7900HT Sequence Detection system (v2.2.2) software (Applied Biosystems, Foster City, CA, USA). Gene expression was quantified using the comparative C T method [15,27,45] with values normalized to the house-keeping gene Gapdh (for islets) or Ube2d1 (for MIN6 cells) and expressed as a fold change compared to control samples (control MIN6 cells, Ins2 WT/WT islets, wt or db/+ islets or mouse kidney (irrelevant tissue) samples).

Statistical analysis
Data are presented as mean ± SEM. Statistical analyses included one-way ANOVA with Fischer's unprotected least significant difference (LSD), unpaired t-test and Mann-Whitney test.

Age-dependent decline in intracellular HSPG core proteins and HS in db/ db islets
Firstly we investigated whether like T1D [15,16], T2D development in db/db mice also correlated with a deterioration in beta cell HS. We extended the analysis to HSPG core proteins, based on (i) the impact of ER stress on global translation/protein synthesis and (ii) the diminished levels of intra-islet HSPGs in T1D human pancreases [16]. Strong immunohistochemical staining for HSPG core proteins (COL18, SDC1 and CD44) and HS was observed in the islets of lean control (wt and db/+) males at 6 weeks of age; in contrast, the islet area stained for HSPGs and HS was significantly decreased in male db/db islets by as early as 3-4 weeks ( Fig  1A-1E and S1 File). Thereafter, the area of COL18, SDC1 and CD44 staining in db/db islets declined to 38.8-62.4% of controls by 5 weeks and 18.7-47.2% at 6-9 weeks (Fig 1B-1D and S1 File). Colocalization studies using immunofluorescence staining confirmed that COL18 in wt pancreas (Fig 2A-2D) and 4 wk db/db pancreas (Fig 2K-2N) was restricted to islet beta cells and not alpha cells. A striking loss of COL18 in db/db beta cells was observed in 9 wk db/ db islets with scattered insulin-positive beta cells that were COL18-negative (Fig 2G and 2H). The % HS+ve islet area in db/db islets was significantly decreased to 48.1% and 25.7% of lean control islets (wt) at 4 weeks and 9 weeks of age, respectively ( Fig 1E and S1 File). Unlike the early decline in % HSPG+ve and HS+ve islet area observed in young db/db islets, the decline observed in lean control islets at 16 weeks and/or 20 weeks may be due to an intrinsic slow core protein turnover [54] or possibly to age-related effects [55]. Compared to lean (wt and db/+) controls, male db/db mice were obese (1.3-fold heavier than corresponding lean controls) and hyperglycemic (db/db, 19.63 ± 2.54 mmol/l vs lean controls, 8.87 ± 0.49 mmol/l) by 6 weeks (S2 Fig). Overall, compared to lean controls, HSPG core proteins and HS were dramatically reduced in the islets of db/db mice, with this perturbation occurring before the onset of hyperglycemia. Similar changes were observed in female db/db islets in situ ( S3 Fig and S2 File). These findings raised the possibility that the intra-islet levels of HSPG core proteins, and indirectly HS, in db/db beta cells could be impacted by ER stress.

UPR gene expression is upregulated in db/db islets
Investigating the relationship between ER stress and levels of intracellular HSPGs and HS in db/db and lean (wt or db/+) beta cells, we found that db/db islets from young donors (4.0-4.4 weeks) showed a significant 15.9-fold upregulation in the early adaptive ER stress-associated gene Bip (p<0.05). We also observed trends for increased UPR transcripts for p58 (12.9-fold) as well as Chop (3.8-fold) and Atf (3.5-fold), i.e., deleterious UPR genes (Fig 3A and S3 File). The mRNA levels for Bip, p58, Chop and Atf3 were significantly enhanced 125-fold, 113-fold, 23-fold and 13.4-fold, respectively, in islets from older hyperglycemic db/db mice (5-8 weeks; Fig 3B and S3 File). For both age groups we found variable expression of each UPR transcript which we attribute to biological variability, including possible differences in the onset of ER stress between individual young donor mice, variable levels of ER stress in older donors and potentially variability in ER stress gene expression amongst islets from the same donor. The expression of adaptive UPR genes in db/db islets, particularly from older donors, remained selectively increased compared to pro-apoptotic UPR genes. These data suggest a possible mechanism responsible for the loss of intra-islet HSPG core proteins in db/db islets.

Loss of intra-islet HSPG core proteins and HS in Ins2 WT/C96Y mice
We further explored the relationship between ER stress and the loss of HSPG core proteins/HS in beta cells using Ins2 WT/C96Y (Akita) mice, an established model of ER stress-mediated diabetes [56]. TaqMan qRT-PCR showed a 287-383 fold increase in adaptive UPR mRNAs (Bip (p<0.05), p58 (p<0.05); Fig 4A and S4 File) in isolated heterozygous Ins2 WT/C96Y islets and a 33 fold increase in the expression of Chop, a deleterious UPR gene (p<0.05; Fig 4B and S4 File), compared to control Ins2 WT/WT islets. We also observed a trend for a 6-fold increase in Atf3 transcripts but this increase did not reach statistical significance. Immunohistochemistry showed that the COL18+ve islet area in Ins2 WT/C96Y islets was reduced to 24.5% of control Ins2 WT/WT islets by 5 weeks (p<0.0001; Fig 4C, S4 Fig and S4 File). The intra-islet SDC1 and CD44 core protein levels were also significantly reduced at 4-9 weeks (p<0.0001; Fig 4D and  4E, S4 Fig and S4 File). In parallel, the HS+ve islet area in Ins2 WT/C96Y islets was significantly decreased to 58.4% of controls at 6 weeks (p<0.001; Fig 4F, S4 Fig and S4 File). Ins2 WT/C96Y mice therefore demonstrated a significant loss of intra-islet HSPG core proteins and HS, correlating with acute ER stress and early diabetes development.

Intracellular HSPG core proteins, HS and HPSE in MIN6 cells ± ER stress
We next used MIN6 beta cells [44] to experimentally induce ER stress and to monitor beta cell HSPGs and HS levels. Firstly, we showed by flow cytometry that freshly harvested MIN6 cells resembled primary beta cells [15,27]  and Atf3 (p<0.01) and treatment with tunicamycin significantly increased the expression of Bip, p58, Chop and Atf3 mRNAs (p<0.05), confirming the induction of ER stress (Fig 5A and  S5 File). Extending the duration of in vitro treatment to 3 days, thapsigargin significantly decreased intracellular HPSG core proteins (COL18, p<0.01; SDC1, p<0.05) and HS (p<0.05) to 57.5-70.9% and 79.8% of controls (Fig 5B, 5C and 5E and S5 File); in parallel, HPSE levels declined to 42.2% of controls (p<0.001; Fig 5F and S5 File). Similarly, tunicamycin treatment also lowered intracellular SDC1 (p<0.05) and HPSE (p<0.001) levels to 52.0% and 33.6% of control levels (Fig 5C and 5F and S5 File). Neither treatment reduced intracellular CD44 core protein (Fig 5D and S5 File). However, the ER stress response induced pharmacologically in MIN6 cells was much less severe than that observed endogenously in native beta cells in vivo (Figs 3A, 3B, 4A and 4B). Accordingly, the reduced levels of a subset of intracellular HPSG core proteins and HS were less pronounced than those observed in mouse models of ER stress (Figs 3 and 4). Importantly, thapsigargin-treated MIN6 cells demonstrated a loss of both intracellular HPSE and HS, providing evidence that ER stress-induced HS loss was not due to HS degradation.

Intracellular HSPGs, HS and HPSE are reduced in isolated db/db beta cells
Using flow cytometry, we found that HSPG core proteins, HS and HPSE are 3.2-34.2 fold, 8.0 fold and 12.6 fold more highly expressed intracellularly than at the cell surface in freshly isolated wt beta cells, respectively (Fig 6, S6 Fig, S4 Table and S6 File). Furthermore, in agreement with immunohistochemical studies of host pancreas (Fig 1), we demonstrated significantly reduced intracellular expression of HSPG core proteins (COL18, SDC1) in db/db beta cells to 36-47% of wt cells (Fig 6A and 6D and S6 File). In addition, a rapid deterioration of COL18, HS and HPSE to 53%, 59% and 61% of day 0 levels, respectively, was observed inside db/db beta cells but not wt beta cells during culture for 2 days (Fig 6B and 6E, S4 Table and S6 File). Like ER-stressed MIN6 cells (Fig 5F), intracellular HPSE was significantly reduced in db/db beta cells to 26% of wt controls (Fig 6C and 6F, S4 Table and S6 File), indicating that the decline in db/db beta cell HS during culture was HPSE-independent. We next investigated whether the major loss of HS in cultured db/db beta cells contributes to their failure.

HS replacement improves db/db beta cell viability and protects against ROS-induced cell death
We tested whether db/db beta cells, like normal mouse and human beta cells [15,16], can be rescued by HS replacement during culture for 2 days with heparin (a highly sulfated HS analogue; 50 μg/ml; S7 Fig). Initially, the survival of freshly isolated wt beta cells (Cal + PI -; 41.49 ± 1.92%) was lower than for db/db beta cells from normoglycemic donors (58.60 ± 0.78%). After culture, the viability of wt beta cells and db/db beta cells (Cal + PI -) declined to 50.5% and 39.6% of Day 0 levels, respectively; similarly, the viability of beta cells from other db/db donors (with blood glucose levels of 10-15 mmol/l and > 15 mmol/l) was reduced to 42.2%-43.2% of Day 0 levels (S7 File). Culture of wt beta cells with heparin (which replaces highly sulfated HS that is selectively localized in beta cells [31] and lost during the isolation process [15]) significantly increased the live (Cal + PI -) population of beta cells by 3.5-fold and reduced damaged Cal + PI + beta cells and Cal -PI + dead cells to 10.5-19.6% (p<0.001) of untreated wt controls (Fig 7A and S7 File). Similarly, heparin significantly increased the survival of beta cells from male normoglycemic db/db mice (bg<10 mmol/l) by 2.8-fold (p<0.001), and from mildly hyperglycemic db/db donors (bg = 10-15 mmol/l) and hyperglycemic donors (bg>15 mmol/l) by 1.9-fold (p<0.05; Fig 7C, 7E and 7G and S7 File). PI-88, another highly sulfated HS mimetic and HS replacer, preserved the survival of wt beta cells as previously reported [15] but was less effective than heparin in protecting the viability of db/db beta cells (S8 Fig). HS replacement using heparin therefore vastly improved the survival of db/ db beta cells.
In parallel, db/db beta cells were analyzed for susceptibility to H 2 O 2 -induced damage by Sytox Green staining (S9 Fig). Freshly isolated (Day 0) wt beta cells (Fig 7B and S7 File) and beta cells from normoglycemic ( Fig 7D and S7 File), mildly hyperglycemic ( Fig 7F and S7 File) and severely hyperglycemic (Fig 7H and S7 File) db/db mice demonstrated a significant 2.7 (p<0.0001), 2.9 (p<0.05), 4.8 (p<0.0001) and 6.4-fold (p<0.05) increase in cell death following acute H 2 O 2 treatment. Hyperglycemic db/db beta cells were therefore more highly sensitive to acute oxidative damage than normoglycemic db/db beta cells or wt controls. This vulnerability was noticeably mitigated after 2 days of culture without heparin in all groups, probably reflecting protection of beta cell viability due to the induced expression of conventional oxidative stress-related genes e.g., ROS-scavenging enzymes [3,57]. In the presence of heparin, wt ( Fig  7B and S7 File) and normoglycemic db/db (Fig 7D and S7 File) beta cells significantly decreased Sytox Green+ve beta cell damage/death to 21.1% (p<0.001) and 23.4% (p<0.01) of untreated controls, resembling viability assessment using Calcein/PI staining (Fig 7A and 7C and S7 File). Strikingly, heparin treatment also protected wt and db/db beta cells from H 2 O 2induced death with only~8% Sytox Green+ve damaged/dead cells (versus 54% without heparin; Fig 7B; p<0.001; S7 File) and~12-30% Sytox Green+ve cells (versus 39-48% without heparin; Fig 7D, 7F and 7H; p<0.05; S7 File), respectively. These data illustrate that HS replacement in vitro using heparin (i) renders db/db beta cells resistant to acute oxidative stress and (ii) represents a more efficient mechanism than other endogenous pathways induced/enhanced during culture in response to oxidative stress.

TUDCA treatment of db/db mice improves glycemia and beta cell HSPG/ HS levels
Our studies have revealed that ER-stress in db/db islets correlates with a rapid decline in islet/ beta cell HSPGs and HS. Similarly, the pharmacological induction of ER stress in MIN6 cells results in loss of intracellular HSPGs and HS. Against this background we interrogated the relationship between ER stress, T2D and loss of beta cell HSPG/HS in vivo by treating db/db males from 4 weeks of age with the ER chemical chaperone TUDCA. TUDCA-treated male db/db mice showed impaired progression of T2D (Fig 8A and S8 File), decreased body weight (Fig 8B and S8 File) and improved control of glycemia (Fig 8C and S8 File). Also at termination, isolated db/db islets showed a significant 3.1-3.5 fold (p<0.01) increase in the expression of adaptive UPR genes (Bip and p58) and a 2.3-2.6 fold (p<0.01) increase in the expression of pro-apoptotic UPR genes (Chop and Atf3), compared to saline controls (Fig 8D and S8 File). These findings suggest that TUDCA elevated the UPR to enhance beta cell compensation, maintain beta cell function and thus improve glycemia.
We next assessed whether relieving ER stress influenced the intra-islet expression of HSPG core protein COL18 and HS. Immunohistochemical staining of host pancreas showed that the COL18+ve islet area was significantly increased by 1.7-fold in the islets of TUDCA-treated db/ db mice (23.79 ± 2.12%), compared to controls (13.65 ± 1.17%, p<0.001; Fig 8E and S8 File). Compared to saline-treated controls which demonstrated loss of intra-islet COL18 and only scattered insulin-positive beta cells (Fig 9A and 9B), TUDCA treatment resulted in widespread intra-islet COL18 staining ( Fig 9F) and uniform staining of intra-islet insulin (Fig 9G) which colocalized with COL18 ( Fig 9H). The HS+ve islet area also increased 1.6-fold (24.22 ± 3.95% versus 14.80 ± 1.72% in saline controls; Fig 8F and S8 File) and was accompanied by a 1.5-fold increase in the insulin+ve islet area (p<0.0001; Fig 8G and S8 File). The striking improvement in db/db beta cell HS therefore correlated with a significant increase in the expression of HSPG core proteins and an enhanced adaptive UPR to help compensate for ER stress and counter pro-apoptotic UPR effects.

Discussion
We report here that beta cell HSPG core proteins and highly sulfated HS are both significantly depleted in prediabetic 3-5 week db/db mice. These events correlated with the preferential increase in Bip and p58 (Dnajc3) mRNAs, representing markers of the UPR, and preceded both overt hyperglycemia and significant weight gain from 6 weeks of age. Similarly, we demonstrated by quantitative immunohistochemical analyses, a dramatic decline in beta cell HSPGs and HS in the islets of young 4-5 week Ins2 WT/C96Y mice, a transgenic model of ERstressed beta cells [56]. Flow cytometry studies revealed that, in vitro, acute pharmacological induction of ER stress in MIN6 cells by thapsigargin and tunicamycin, significantly reduced intracellular levels of the HSPG core proteins COL18 and SDC1. This outcome is attributed to a general decrease in protein expression resulting from the activation of UPR genes. Furthermore, the impact of ER stress is likely to be influenced by the rate of protein turnover, which may vary for different HSPGs. Ultimately a decline in HSPG core proteins would have the downstream effect of decreasing intracellular HS. We also observed a parallel decline in intracellular HPSE (HS-degrading endoglycosidase) in both ER stressed MIN6 cells and db/db beta cells, which precluded HS degradation as the mechanism responsible for HS loss.
We have previously reported that beta cell HS is depleted during the isolation of normal mouse and human islets in vitro, and renders isolated beta cells especially susceptible to ROSinduced death [15,16,27]. Furthermore, HS replacement using HS mimetics reverses this heightened vulnerability and preserves beta cell viability [15,16]. From our current study, we propose that the significant decrease in intracellular HSPG core proteins (COL18 and SDC1) and HPSE in freshly isolated db/db beta cells, compared to wt beta cells, is due to ER stress in the donor db/db mice. While there was a trend for a greater reduction in intracellular HS in db/db beta cells in vitro (Fig 6), we reason that any pre-existing impact of ER stress was masked by the acute loss of HS during the isolation of wt beta cells used for comparison. This in vitro manipulation-induced loss of HS is most likely due to the depolymerization of HS by excessive ROS generated during the islet isolation procedure [15,27,58,59]. Paradoxically, freshly isolated db/db beta cells initially demonstrated better viability than wt beta cells. We reason that in contrast to wt islets, islets derived from db/db donors have higher endogenous levels of antioxidant enzymes [3,57,60], induced in vivo in response to endogenous oxidative stress associated with ER stress-induced loss of HSPG core proteins/HS. We propose that this induced antioxidant response compensates for loss of HS (Fig 10) in the initial post-isolation period.
We observed a continued decline in HSPGs (COL18 and SDC1), HS and HPSE during the culture of db/db beta cells, a finding not observed in wt beta cells and consistent with the perpetuation of endogenous ER stress in vitro. Interestingly, treatment of both wt and db/db beta cells with the HS replacer heparin significantly improved the survival of beta cells (Cal + PIcells) from normoglycemic, mildly hyperglycemic and severely hyperglycemic db/db donors and protected them from hydrogen peroxide-induced death. On this basis, we propose that the loss of HS from beta cells of prediabetic db/db mice in vivo exacerbates oxidative stress. Surprisingly, another highly sulfated HS mimetic, PI-88, acted as a HS replacer and protected wt beta cells but not db/db beta cells (S8 Fig). It is possible that different heparin-binding proteins [25] may be used in the uptake and/or intracellular distribution of heparin and PI-88 in db/db beta cells and could lead to a difference in susceptibility to ER stress effects.
Consistent with previous reports of genetically obese mice treated with chemical chaperones [38,61], treatment of db/db mice with the ER chaperone TUDCA significantly reduced non-fasting blood glucose levels, body weight and HbA1c levels. In parallel, we observed a significant increase in the intra-islet distribution of COL18+ve which was accompanied by a striking increase in highly sulfated HS. In addition, we observed a significantly higher insulinpositive islet area, a finding consistent with improved islet insulin content after TUDCA treatment in a T1D mouse model [62]. Our findings therefore revealed a positive impact of TUDCA on beta cells and glycemic control. Paradoxically, TUDCA treatment of db/db mice, unlike a previous report of TUDCA-treated db/db mice [61], failed to mitigate the expression of apoptotic UPR mRNAs. The difference between the studies could be due to differences in the age of the mice at commencement of treatment, duration of treatment, tissues analyzed and to the lower dosage regimen of TUDCA in our study. Additionally, the activation of proapoptotic pathways could be due to ER stress-independent activation of c-Jun N-terminal kinase (JNK) by oxidative stress or lipotoxicity [63][64][65]. Increased Chop expression has previously been observed during beta cell adaptation in other obese rodent models (Zucker fatty rats, HFD mice), suggesting that a balance between Chop expression and adaptive UPR is important for maintaining beta cell function and survival [18,66]. Since TUDCA treatment of db/db mice improved glycemic control, we propose that the accompanying preferential increase in the expression of adaptive UPR genes compared to deleterious pro-apoptotic genes (Chop, Atf3) relieved ER stress, prolonged beta cell compensation and promoted recovery of beta cell HSPGs/HS. Together our in vitro and in vivo intervention studies highlight a previously unrecognized inter-relationship between HS/HSPGs, oxidative stress and ER stress. While ER stress and oxidative stress have long been recognized as accomplices in beta cell failure in T2D, mechanistic links have remained elusive. Based on our studies we propose a novel pathway which bridges ER stress and oxidative stress in T2D. Here, we postulate that ER stress, prior to the onset of In pre-T2D pancreatic beta cells, ER stress initiates the unfolded protein response (UPR) which acts to alleviate the stress by increasing molecular chaperones for protein folding, increasing ERAD and reducing the translation of mRNAs to a range of "general" proteins. As a consequence, HSPG core protein synthesis is impaired, which in turn severely reduces the synthesis of HS, a constitutive non-enzymatic antioxidant in beta cells. Depletion of beta cell HS increases intracellular ROS and elevates oxidative stress. Oxidative stress induces the expression of secondary antioxidant enzymes to help maintain beta cell homeostasis [57]. Failure to compensate for this stress results in UPR failure, the preferential expression of apoptotic genes and beta cell apoptosis/ death. ERAD, ER-associated degradation; T2D, Type 2 diabetes; HS, heparan sulfate; HSPG, heparan sulfate proteoglycan; ROS, reactive oxygen species.
https://doi.org/10.1371/journal.pone.0252607.g010 T2D, results in a UPR-induced decline in the synthesis of HSPG core proteins within the ER. Consequently, HSPG core proteins fail to be transferred to the Golgi compartment, impairing or blocking HS synthesis because HSPG core proteins are an essential scaffold for HS assembly [20,21,23]. HS deficiency in the beta cells results in increased intracellular ROS and thus, debilitating oxidative stress (Fig 10). This new understanding provides a rational basis for the previously observed upregulation of antioxidant enzyme transcripts [60] and the therapeutic effect of chemical and physiological antioxidants on beta cell mass and/or function in db/db mice [60,67,68], a strategy that would compensate for the depletion of endogenous beta cell HS. It is possible that glucotoxicity-associated ER stress in the db/db (genetic) model of T2D, resulting in the depletion of ER Ca 2+ , contributes to the decline in beta cell HSPG core proteins [69]. Future studies could investigate whether obesity, lipotoxicity and ER stress induced by a long-term high-fat diet (for 16 weeks) in normal mice [70] also impacts beta cell HSPGs and HS.
T2D and T1D have distinct metabolic and autoimmune aetiologies, respectively, and different mechanisms of beta cell death [65,71]. We demonstrate here that prediabetic db/db mice, like pre-diabetic NOD mice, display a loss of HS from beta cells. However, we reveal in this study that the HS-depleting pathways are vastly different. In T1D, the primary mechanism is due to HPSE-dependent degradation of HS [15,16]. In contrast, our data strongly suggest that HS depletion in T2D beta cells is HPSE-independent; instead, our findings are consistent with the inhibition of HS synthesis, a secondary effect of ER stress on the synthesis/maturation of HSPG core proteins. However, since loss of HSPGs [16] and ER stress have been reported in human T1D beta cells [6,7,34,35,65], we cannot exclude the possibility that ER stressinduced deprivation of HSPG core proteins may function as an ancillary mechanism for HS loss in T1D. Altogether, loss of beta cell HS is a fundamental defect underpinning pre-diabetes and culminates in insurmountable oxidative damage.
In summary, our studies highlight that intracellular HS acts as a critical link between ER stress and oxidative stress in T2D beta cells i.e., ER stress indirectly reduces beta cell HS, which in turn heightens the susceptibility of beta cells to oxidative stress. Based on the reported isletassociated ER stress in obesity-driven human T2D [17], we speculate that loss of beta cell HSPGs/HS is likely to precede beta cell failure in human T2D. We suggest that treatment with a heparin-like HS replacer, possibly combined with a chemical chaperone, could represent a novel therapeutic strategy for rescuing beta cell survival, prolonging beta cell compensation and alleviating oxidative stress in T2D.