Transient Intermittent Hyperglycemia Accelerates Atherosclerosis by Promoting Myelopoiesis

Abstract

Rationale:

Treatment efficacy for diabetes is largely determined by assessment of HbA1c levels, which poorly reflects direct glucose variation. People with pre-diabetes and diabetes spend >50% of their time outside the optimal glucose range. These glucose variations, termed transient intermittent hyperglycemia (TIH) appear to be an independent risk-factor for cardiovascular disease (CVD) but the pathological basis for this association is unclear.

Objective:

To determine whether TIH per se promotes myelopoiesis to produce more monocytes and consequently adversely affects atherosclerosis.

Methods and Results:

To create a mouse model of TIH we administered 4 bolus doses of glucose at 2hr intervals intraperitoneally once to wild-type (WT) or once weekly to atherosclerotic prone mice. TIH accelerated atherogenesis without an increase in plasma cholesterol, seen in traditional models of diabetes. TIH promoted myelopoiesis in the bone marrow, resulting in increased circulating monocytes, particularly the inflammatory Ly6-C^hi subset, and neutrophils. Hematopoietic-restricted deletion of S100a9, S100a8 or its cognate receptor Rage, prevented monocytosis. Mechanistically, glucose uptake via GLUT-1 and enhanced glycolysis in neutrophils promoted the production of S100A8/A9. Myeloid-restricted deletion of Slc2a1 (GLUT-1) or pharmacological inhibition of S100A8/A9 reduced TIH-induced myelopoiesis and atherosclerosis.

Conclusions:

Together, these data provide a mechanism as to how TIH, prevalent in people with impaired glucose metabolism, contributes to CVD. These findings provide a rationale for continual glucose control in these patients and may also suggest that strategies aimed at targeting the S100A8/A9-RAGE axis could represent a viable approach to protect the vulnerable blood vessels in diabetes.

Graphical Abstract

INTRODUCTION

Atherosclerotic-cardiovascular disease (CVD) is accelerated in diabetes¹, which translates into premature mortality. Many different factors contribute to this risk; however, hyperglycemia is likely to be one cause of greater atherosclerosis. We have previously reported that chronic and persistent hyperglycemia promotes myelopoiesis, resulting in enhanced production and infiltration of these monocytes into plaques². This process was driven by systemic increase in the neutrophil-derived alarmins, S100A8/A9, which signals through the receptor for advanced glycation end-products (RAGE) to promote myelopoiesis in the bone marrow, resulting in monocytosis². Whether transient hyperglycemia has similar effects on myelopoiesis and atherosclerosis remains unknown.

Large trials in patients with diabetes have failed to demonstrate reductions in major cardiovascular events through glucose lowering alone³. One reason for this failure may be that while significant reductions in ambient hyperglycemia are achieved with conventional therapy, most patients with treated diabetes continue to experience abnormal fluctuations in glycemia⁴, including transient and intermittent hyperglycemia (TIH), especially post-prandially⁵. In non-diabetic individuals, post-meal plasma glucose seldom rises to >7.8 mmol/L (>140 mg/dL). In patients with diabetes this is often not the case, even for patients with presumed overall good glycemic control (HbA1c<7%; <53mmol/mol)⁴. Significant post-prandial hyperglycemia is also experienced by subjects with impaired glucose tolerance (IGT), who despite consistent mean glycemia (i.e. HbA1c) in the normal range, have increased incidence of CVD^{6, 7}, often comparable to those with ‘managed’ diabetes⁸. More recently, trials using continuous glucose monitoring (CGM) revealed that individuals with diabetes, as well as those with pre-diabetes (diagnosed under the current ADA Guidelines⁹), exhibit frequent glucose variability. Indeed, these individuals average a median of 50% of their time with hyperglycemia¹⁰. In addition, the mean amplitude of glycemic excursions obtained through CGM correlates with increased plaque vulnerability¹¹. However, the direct impact of TIH on the molecular pathways implicated in atherogenesis remains to be established.

As TIH occurs in individuals with dysglycemia, it is likely that glucose variability directly contributes to the atherogenic profile, potentially through these aforementioned S100A8/A9-driven effects on monocyte production². Given that neutrophils are almost exclusively glycolytic and contribute to the majority of circulating S100A8/A9 in diabetes^{2, 12}, we postulated that neutrophils act as the innate sensors of glycemic stress initiating this pathway. Thus, understanding the changes to the innate immune system following TIH could help us understand why people who are perceived to have good glucose control, based on HbA1c levels, still carry considerable cardiovascular risk. Therefore, we sought to determine if TIH stimulates myelopoiesis and atherogenesis, while exploring the underlying mechanism(s).

METHODS

Materials and methods are included in the data supplement.

All data and supporting materials are available within the article and Online Data Supplement.

Any further clarification of methods and analysis can be sought bt contacting the corresponding author.

RESULTS

Transient intermittent hyperglycemia promotes myelopoiesis and atherosclerosis in Apoe^−/− mice

We aimed to model TIH in mice using the rationale that humans, particularly those who are insulin resistant and obese eat more frequently^13–15 and often under report their snacking habits¹⁶ resulting in many postprandial glucose spikes. Thus, Apoe^−/− mice received four i.p. injections of glucose (2g/kg) or isovolumic saline (control), delivered two hours apart, achieving TIH with plasma glucose peaking at ~15–20mM over the 8-hours (Figure 1A). This procedure was repeated weekly for 10 weeks, modelling TIH, but not increasing HbA1c Figure IA in the data supplement). We found that Apoe^−/− mice subjected to weekly TIH accumulated significantly more plaques across their aortic arch and sinus (Fig. 1B–C). Moreover, TIH increased lipid and macrophage content within the plaques (Fig. 1D, E). Collagen content was also increased in the lesions of TIH mice over control (Fig. 1F). While reduced collagen content is usually a sign of unstable lesions, this is typically in the context of much larger and complex lesions than achieved in this study¹⁷. Thus, increased collagen content, along with more abundant lipids and macrophages seen in the plaques of the TIH mice suggests that these lesions had progressed at an accelerated rate compared to saline-treated mice. Importantly, while previous studies have demonstrated increased plasma cholesterol in diabetic mice¹⁸, TIH did not increase plasma cholesterol indicating that cholesterol was not responsible for the increase in atherosclerosis in these mice (Fig. 1G). Exploring circulating leukocytes, we noted that TIH increased both monocytes and neutrophils in Apoe^−/− mice (Fig. 1H). This was driven by an expansion and increased proliferation of the common-myeloid progenitor (CMPs) and granulocyte-macrophage progenitors (GMPs) in the bone marrow (BM) (Fig. 1I, J). This change in blood leukocytes is consistent with observations in mice with chronic diabetes². Apoe^−/− mice are also known to exhibit intrinsic defects in their HSPCs which promotes mobilisation of these cells to the spleen where they undergo extramedullary myelopoiesis to produce atherogenic Ly6-C^hi monocytes^19–21. Examining the spleen, we observed an increase in GMPs (Figure IB in the data supplement), suggesting that TIH may also promote extramedullary myelopoiesis likely contributing to monocytosis and accelerated atherosclerosis as observed in this model. As cholesterol levels were not increased in these mice, we postulate that the cause of greater atherosclerosis was due to enhanced myelopoiesis leading to greater numbers of circulating inflammatory white blood cells².

Figure 1: TIH induces atherosclerosis in Apoe^−/− mice.

Apoe^−/− mice were subjected to the TIH procedure (4 injections of 2g/kg glucose, 2 hours apart) or saline (control) once a week for 10 weeks. A) Blood glucose levels measured prior to as well as 15min and 60min following each injection (indicated by dotted lines). B) Atherosclerosis in the aortic arch quantified by ORO. C) H&E, D) ORO (lipid), E) CD68 (macrophage) and F) Picrosirius red (collagen) in the aortic sinus (n=9–10/group). Scale bars: 50μM. G) Total plasma cholesterol (n=9/group). H) Blood monocytes and neutrophils (including flow plots with mean percentages ± SEM), and I) populations and J) proliferation of BM HSPCs, CMPs and GMPs quantified by flow cytometry (n=6/group). Data are presented mean ± SEM and analyzed using Student’s t-test (B, F-J) or for plaque areas as median with 95% CI and analyzed using a Mann-Whitney U test (C-E).

A single day of TIH induces myelopoiesis and monocytosis.

To delineate the kinetics of myelopoiesis during TIH, Wild type (WT) mice were randomly assigned to receive a single day of TIH or saline (control) (Fig. 2A, B). One day after the treatment, mice exposed to TIH had increased abundance of CMPs and GMPs in the BM (Fig. 2C). This directly translated into increased numbers of circulating monocytes, notably the highly inflammatory Ly6-C^hi subset, and neutrophils seven days post-TIH (Fig. 2D). The expansion of BM progenitors preceded monocytosis as there were no acute changes in circulating numbers of monocytes or neutrophils on day 1 (Fig. 2D). Notably, the numbers of CMPs and GMPs normalized within 7 days following a single day of glucose variability (Fig. 2C) and the blood leukocytes by 14 days (not shown) suggesting that TIH stimulation of myelopoiesis is transient. Without the mobilisation of HSPCs induced by deletion of Apoe as in Figure IB in the data supplement, we did not observe extramedullary myelopoiesis within the spleen in response to TIH (data not shown).

Figure 2: TIH induces myelopoiesis.

A) WT mice were subjected TIH or saline injections and assessed one or seven days later (n=6/group). B) Plasma glucose measurements were recorded prior to, 15min and 60min following each injection (dotted lines). C) BM HSPCs, CMPs and GMPs and D) blood monocytes and neutrophils were quantified by flow cytometry (flow plots with mean percentages ± SEM). Data are presented mean ± SEM and analyzed using a One-way ANOVA followed by Tukey’s multiple comparisons test, all pairwise comparisons among 3 groups (3 times).

TIH-accelerated atherosclerosis is RAGE dependent.

Hyperglycaemia is known to promote signaling via the receptor for advanced glycation end-products (RAGE). Accordingly, we determined if the enhanced atherogenesis in response to TIH was mediated through RAGE by subjecting Apoe^−/− and Rage^−/−Apoe^−/− mice (littermates) to TIH or saline control as in Figure 1 (Fig. 3A). Consistent with the data in Figure 1, Apoe^−/− mice subjected to TIH had increased lesions in the aortic arch, however, we observed no increase in atherosclerotic plaques in the Rage^−/−Apoe^−/− mice after TIH or chronic hyperglycemia (Fig. IIIB and Figure II in the data supplement). RAGE is expressed on both endothelial and hematopoietic cells²² and activation of endothelial RAGE is known to increase adhesion molecule expression and increase leukocyte adhesion²³. Thus, using a dynamic shear-flow cell adhesion assay to assess endothelial inflammation, we labelled leukocytes from human whole blood and perfused ex vivo through aortas collected from Apoe^−/− and Rage^−/−Apoe^−/− mice 7 days after the final dosing of TIH or saline control. Importantly, TIH stimulated the same degree of leukocyte recruitment in aortas from Apoe^−/− and Rage^−/−Apoe^−/− mice (Fig. 3C), suggesting that RAGE expression on the endothelium may not be the major contributing factor to the atherosclerosis in the setting of TIH.

Figure 3: Deletion of RAGE protects against TIH-induced myelopoiesis and atherosclerosis.

A) Apoe^−/− (n=9) and Apoe^−/−/Rage^−/− (n=10) mice received weekly TIH or saline treatments for 10 weeks, B) Atherosclerosis in the aortic arch quantified by ORO. C) Leukocyte-endothelial adhesion was measured using whole blood perfused through aortas obtained from Apoe^−/− and Apoe^−/−/Rage^−/− mice 7 days following TIH or saline (n=4 for Apoe^−/−/Rage^−/− + TIH and n=5/group for all other groups). D-J) WT and Rage^−/− mice were subjected to D) TIH and 1 day later BM E) CMPs, F) GMPs, G) GMP proliferation and 7 days later blood H) monocytes, I) Ly6-C^hi monocytes and J) neutrophils quantified by flow cytometry; n=8 for Rage^−/− + TIH and n=6/group for all other groups. Data are presented mean ± SEM, analyzed using a Two-way ANOVA followed by Tukey’s multiple comparisons test, all pairwise comparisons among 4 groups (6 times) (B, E-J), or a repeated measures Two-way ANOVA (P_Time=3.97×10⁻⁶, P_TIH=3.25×10⁻⁴, P_TimexTIH=1.65×10⁻⁴, no significant variation observed from KO, Time × KO, KO × TIH, or KO × Time × TIH) and Tukey’s multiple comparison’s test, all pairwise comparisons among 4 groups over 3 timepoints (66 times) (C).

Next, to determine whether RAGE signaling is responsible for the myelopoiesis and monocytosis induced by TIH, we performed the TIH model in Rage^−/− mice (Fig. 3D). Indeed, these mice demonstrated protection from TIH-induced myelopoiesis, with Rage^−/− mice exhibiting no increase in the numbers of CMPs or GMPs and no increase in GMP proliferation as observed in the WT mice after 1 day, nor did we see any increase in total or Ly6-C^hi monocytes or neutrophils after 7 days as was observed in WT mice. (Fig. 3E–J). These data, along with the data in Figure 3C, indicates that deletion of RAGE within the hematopoietic system is the reason for reduced atherogenesis in the TIH mice.

Hematopoietic expressed RAGE is required for TIH-induced myelopoiesis.

To test the hypothesis that TIH-induced monocytosis was mediated via activation of RAGE specifically in haematopoietic cells in the BM, a BM transplant (BMT) was performed where lethally irradiated WT mice were engrafted with BM from Rage^−/− or WT mice subjected to a single day of TIH (Fig. 4A). As expected, mice receiving WT BM, displayed TIH-induced myelopoiesis, whereas the mice with Rage^−/− BM were protected (Fig. 4B–E). These data confirm the involvement of RAGE on hematopoietic cells in the BM in mediating TIH-induced myelopoiesis.

Figure 4: Disrupting the RAGE-S100A8/A9 axis protects against TIH-induced myelopoiesis.

A) Experimental outline of Rage^−/− BM transplant study: Following reconstitution recipient mice were subjected TIH and assessed 1 or 7 days later. Seven days post-TIH blood B) Monocytes, C) Ly6-C^hi monocytes were quantified by flow cytometry (n=6/saline group, n=7 for WT + TIH and n=9 for Rage^−/− + TIH). One day post-TIH BM D) progenitor abundance and E) proliferation quantified by flow cytometry (n=6/saline group, n=9–10/TIH group). F) S100a9 mRNA expression in the blood leukocytes in WT mice measured 1 and 7 days following TIH (n=4–6/group). G-J) WT mice transplanted with either G, H) S100a9^−/− or WT BM or I, J) S100a8^−/− or WT BM and subjected to the G, I) TIH procedure and H, J) circulating Ly6-C^hi monocytes quantified after 7 days by flow cytometry (n=3–7/group). K) S100A8/A9 concentrations from plasma collected one day following TIH procedure from mice transplanted with S100a8^−/− or WT BM; n.d.: not detected (n=3–6/group). Data are presented mean ± SEM and analyzed using a Two-way ANOVA, all pairwise comparisons among 4 groups (6 times) (B-E, H,J,K), or a One-way ANOVA followed by Tukey’s multiple comparisons test, all pairwise comparisons among 3 groups (3 times) (F).

S100A8/A9 is essential for TIH-induced myelopoiesis.

RAGE can be activated by an array of ligands including S100A8/A9 which is responsible for monocytosis in chronic diabetes². When we assessed the expression of S100a9 from circulating WBCs, we noted an increase in S100a9 mRNA 1 day following TIH coinciding with the changes in RAGE-mediated increase in CMPs and GMPs in the BM (Fig. 4F). Thus, to directly test whether S100A8 and S100A9 play a role in TIH-induced myelopoiesis, we performed in vivo TIH studies using mice deficient for S100A8 or S100A9 in the hematopoietic compartment. WT mice transplanted with S100a9^−/− BM were subjected to a single day of TIH and were protected from TIH-induced myelopoiesis compared to mice receiving WT BM. (Fig. 4G, H). Similar protection was also observed in mice deficient for S100a8 (Fig. 4I, J). Consistent with the hypothesis, we found increased plasma S100A8/A9 from WT mice one day following TIH (Fig. 4K). Importantly, genetic deletion of S100A8 (which also deletes S100A9) completely removed detectable circulating S100A8/A9 validating the specificity of the ELISA.

Enhanced glycolytic flux stimulates secretion of S100A8/A9 from neutrophils.

S100A8/A9 is primarily derived from neutrophils, and neutrophil depletion normalizes the levels of S100A8/A9 in the circulation of STZ-induced diabetic mice^{2, 12}. Likewise, in our TIH model we observed higher levels of S100A9 in these neutrophils than other cells known to express S100A8/A9, namely monocytes, and these levels in neutrophils were decreased following exposure to TIH (Figure III in the data supplement). Thus, we hypothesized that TIH acts directly on neutrophils to alter their metabolism and promote S100A8/A9 release. Using the Seahorse Extracellular Flux Analyzer to explore real-time glycolytic rate, we found that exposure of isolated human neutrophils to glucose levels observed in our TIH model significantly increased their glycolytic rate, which could be inhibited when the cells were pre-incubated with the pan hexokinase inhibitor, 2-deoxyglucose (2-DG) (Fig. 5A, B). Under these conditions, stimulation of human neutrophils with glucose upregulated S100a9 mRNA, which was also blocked by 2-DG (Fig. 5C). To determine whether enhanced glycolytic flux was linked to the loss (due to release) of S100A8/A9 from neutrophils, we assessed the content of S100A8/A9 in human neutrophils following acute glucose treatment by flow cytometry. High glucose exposure rapidly reduced the content of S100A8/A9 (<30 min), which was largely prevented by 2-DG (Fig. 5D). These neutrophils also exhibited a substantial increase in superoxide accumulation after 30 min of high glucose exposure, which was ameliorated by pre-treatment with 2-DG or the reactive oxygen species (ROS) scavenger N-acetylcysteine (NAC) (Fig. 5E). NAC also prevented the loss of S100A8/A9 from neutrophils challenged with glucose, confirming the importance of ROS in mediating this event (Fig. 5F). As ROS is produced in neutrophils through the action of NADPH oxidases (NOX), we pre-treated neutrophils with the pan-NOX inhibitor apocynin prior to exposure to hyperglycemia and observed protection from hyperglycemia induced loss of S100A8/A9 (Fig. 5G).

Figure 5: Human neutrophils release S100A8/A9 under hyperglycemic conditions in vitro.

Isolated human blood neutrophils were exposed to 25mmol/L glucose after being pre-treated for 1hr with or without 2-deoxy-D-glucose (2-DG, 5mmol/L) or N-acetylcysteine (NAC, 1mmol/L). A) Extracellular acidification rate (ECAR; glycolysis) in response 25mmol/L glucose (dotted line) represented as change in from baseline. B) ECAR represented as area under curve (AUC). C) Neutrophil S100A9 mRNA expression after 4hrs in 25mmol/L glucose. D) S100A8/A9 content measured by flow cytometry at 30min of 25mmol/L glucose exposure with 2-DG pre-treatment. E) Superoxide levels (DHE) measured by flow cytometry at 30min of 25mmol/L glucose exposure (2-DG or NAC pre-treatment). Neutrophil S100A8/A9 content measured by flow cytometry at 30min of 25mmol/L glucose exposure with F) NAC and G) apocynin pre-treatments. Data are presented mean ± SEM with n=3–7/group and analyzed using a One-way ANOVA followed by Dunnett’s multiple comparisons test, pairwise comparisons to the glucose treatment group (B-D, F-G: 2 times, E: 3 times).

Myeloid GLUT-1 promotes TIH-induced monocytosis and macrophage accumulation in atherosclerotic plaques.

GLUT-1 is the most enriched non-insulin dependent glucose transporter in neutrophils compared with all other cells^{24, 25}, which we confirmed using data from the Immgen consortium analyzed as a hierarchical differentiation tree using the online software BloodSpot (Figure IVA in the data supplement)²⁵. To further explore if acute glucose uptake in neutrophils is required for myelopoiesis, we deleted glucose transporter 1 (GLUT-1, Slc2a1) in mature myeloid cells by crossing Slc2a1^fl/fl with LysM Cre mice²⁶. The LysM-driven Cre recombinase is expressed in all mature myeloid cells with a similar profile and efficiency to the neutrophil touted S100a8 Cre²⁷. Following TIH, we observed significantly lower glucose uptake in neutrophils and Ly6-C^hi monocytes of mice transplanted with Slc2a1^fl/flLysM^cre/cre BM compared with mice transplanted with BM from Slc2a1^fl/fl littermates, but no significant difference was observed in glucose uptake in the overall leukocyte population (Figure IVB–D in the data supplement). Notably, we found that monocytes downregulate GLUT-1 in a hyperglycemic setting whereas neutrophils (which express more GLUT-1) do not, leaving the neutrophils potentially vulnerable to fluctuating glucose levels (Figure IVE in the data supplement). Following TIH, which was equal in both groups of mice, the Slc2a1^fl/flLysM^cre/cre BM mice were protected from TIH-induced monocytosis compared to the Slc2a1^fl/fl BM mice (Fig. 6A–C). Importantly, there were no significant difference was observed in circulating leukocytes between the two genotypes prior to inducing TIH (data not shown). We also formally explored the role of GLUT-6, another insulin-independent glucose transporter enriched in neutrophils at the message level, in mediating glucose uptake and myelopoiesis in response to TIH. This was achieved by performing BMTs from mice lacking GLUT-6 (Slc2a6^−/−) or WT littermate controls into C57Bl/6 mice and subjecting them to TIH. Overall the data revealed no role for GLUT-6 in mediating glucose uptake in neutrophils or TIH-induced myelopoiesis (Figure V in the data supplement).

Figure 6: Myeloid cell-specific deletion of GLUT-1 protects against TIH-induced myelopoiesis.

A-C) WT or D-I) Ldlr^−/− mice transplanted with BM from Slc2a1^fl/fl LysM^cre/cre Cre mice or their Slc2a1^fl/fl littermate controls were subjected to TIH. WT mice underwent one day of TIH while Ldlr^−/− underwent the TIH procedure twice weekly for 9wks. B) Blood glucose levels during the TIH procedure (n=5/group). C) Blood Ly6-C^hi monocytes 7 days following TIH by flow cytometry (n=5/group). D) Atherosclerosis experimental overview. E) Total plasma cholesterol (n=5 for Slc2a1^fl/fl and n=6 for Slc2a1^fl/fl LysM^cre/cre). F) Blood monocytes and neutrophils (n=4 for Slc2a1^fl/fl and n=6 for Slc2a1^fl/fl LysM^cre/cre), G) BM HSPCs, CMPs and GMPS measured by flow cytometry (n=5 for Slc2a1^fl/fl and n=6 for Slc2a1^fl/fl LysM^cre/cre). H) Atherosclerotic plaque size (H&E) in the aortic sinus and I) plaque macrophages (CD68⁺) (n=5 for Slc2a1^fl/fl and n=6 for Slc2a1^fl/fl LysM^cre/cre). Scale bars: 50μM. Data are presented mean ± SEM and analyzed using a Student’s t-test.

As GLUT-1 appears to be an important transporter of glucose in neutrophils, we explored the relevance of this pathway to atherogenesis. This was achieved by transplanting BM from the Slc2a1^fl/fl and Slc2a1^fl/flLysM^cre/cre mice into irradiated Ldlr^−/− mice (Fig. 6D). Six weeks later the mice were placed on a 10-week WTD to induce atherosclerosis and were subjected to bi-weekly challenges of TIH, as we expected there would be some masking effects of the TIH by the significantly higher cholesterol levels in the WTD-fed Ldlr^−/− mice compared to chow-fed Apoe^−/− mice. Importantly, we found no significant difference between the cholesterol levels of Slc2a1^fl/flLysM^cre/cre/Ldlr^−/− mice and their Slc2a1^fl/fl/Ldlr^−/− controls (Fig. 6E). The Ldlr^−/− mice transplanted with the Slc2a1^fl/flLysM^cre/cre BM had fewer circulating Ly6-C^hi monocytes and neutrophils, and fewer BM GMPs (Fig. 6F–G). Although we did not observe an overall reduction in plaque size with the Slc2a1^fl/flLysM^cre/cre/Ldlr^−/− model, the lesions exhibited fewer plaque macrophages, indicating reduced infiltration of monocytes due to protection from TIH induced-myelopoiesis (Fig. 6H–I). Thus, taken together, increased glucose uptake via GLUT-1 and enhanced glycolytic rate in neutrophils appears to contribute to enhanced myelopoiesis and atherosclerotic macrophage accumulation in response to TIH.

Pharmacological targeting of S100A8/A9 inhibits TIH-induced myelopoiesis and atherogenesis.

As maintaining continuous glucose control is a formidable challenge in people with varying degrees of metabolic dysfunction and diabetes, we aimed to assess whether the RAGE-S100A8/A9 pathway could be targeted therapeutically. We sought to test the utility of the S100A8/A9 small molecule inhibitor, ABR-215757 (Paquinimod)²⁸. Apoe^−/− mice were treated with ABR-215757 (10mg/kg/day in their drinking water) or vehicle, whilst undergoing weekly TIH for 10-weeks. Treatment with ABR-215757 reduced circulating monocytes and neutrophils by decreasing the abundance of myeloid progenitors, independent of cholesterol levels and hyperglycaemic response (Fig. 7A–D & Figure VI in the data supplement). This translated into a reduction in plaque formation, macrophage and lipid content (Fig. 7E–H). In addition, aortic plaques in the ABR-215757 treated mice were more immature and thus had reduced collagen content (Fig. 7I). These data suggest that preventing S100A8/A9-RAGE signaling may reduce cardiovascular risk in the setting of TIH.

Figure 7: Blocking S100A8/A9 with ABR-215757 decreases myelopoiesis and atherosclerosis in TIH mice.

Apoe^−/− mice were treated with or without ABR-215757 (10mg/kg/day in drinking water) and subjected to weekly TIH for 10 weeks. A) Blood monocytes (n=8 for vehicle, n=10 for ABR-215757), B) neutrophils and C) BM HSPCs, CMPs and GMPs were measured by flow cytometry (n=9 for vehicle, n=10 for ABR-215757). D) Total plasma cholesterol. E) Atherosclerosis in the aortic arch quantified by Oil Red O staining. F-I) In the aortic sinus plaques were assessed by F) H&E (size), G) CD68⁺ (macrophages), H) Oil Red O (lipid) and I) Picrosirius red (collagen). Scale bars: 50μM. Data are presented mean ± SEM and analyzed using a Student’s t-test (A, C-I) or as median with 95% CI and analyzed using a Mann-Whitney U test (B).

DISCUSSION

Hyperglycemia is causally linked to atherogenesis and CVD, as evidenced by the increased risk of major cardiovascular events in patients with diabetes. This risk is not dependent on chronic hyperglycemia alone, since patients with normal ambient glucose levels who experience intermittent post-prandial hyperglycemia also have increased cardiovascular risk after adjusting for other risk-factors⁶. We discovered that TIH alone, in the absence of diabetes or insulin resistance, is sufficient to activate pathogenic pathways in the BM that lead to myelopoiesis, monocytosis and accelerated atherogenesis.

We previously discovered that chronic hyperglycemia promotes monocytosis by increasing the number and proliferation of myeloid progenitors in the BM². We now show, that fluctuations in blood glucose, more closely mimicking clinical glucose variation, in an arguably more clinically relevant model, is sufficient to cause proliferation and differentiation of HSPCs in the BM, leading to a rapid expansion of CMPs and GMPs along with a subsequent increase in circulating leukocytes, especially the atherogenic Ly6-C^hi monocytes. The importance of this finding is that epidemiological studies demonstrate that peripheral blood leukocyte counts, particularly monocytes and neutrophils and their inflammatory status, are predictive of incident CVD^29–31. Monocyte recruitment to the nascent atherosclerotic lesion is regarded as an initiating step in plaque development^{19, 32–34}, and monocytosis causes accelerated atherogenesis and also impairs plaque resolution^{2, 35}. We have also shown that monocytosis in diabetes is a significant risk factor that is not addressed by lipid lowering therapy^{2, 35}. Other risk-factors seen in prediabetes with obesity, such as inflammation, dyslipidemia, oxidative stress and signaling from adipose tissue macrophages may also promote the expansion and differentiation of BM HSPCs and subsequent leukocytosis^36–38. The synergy between these cardiac risk-factors in the clinical setting may be partly explained by their effects on the common endpoint of hematopoiesis, albeit it by distinct mechanisms. Delineation of these pathways may provide novel targets for anti-atherogenic therapy.

In the setting of TIH, RAGE was found to be important in mediating the increase in atherosclerosis. While we found a dominant role for RAGE expressed within the haematopoietic compartment, particularly with respect to myelopoiesis, we should note that we have previously found that RAGE expression in non-haematopoietic cells also plays an important role in mediating atherosclerosis²². Upstream of RAGE, S100A8/A9 levels were found to be the initiating ligands to drive myelopoiesis. S100A8 and S100A9 are constitutively expressed in neutrophils where they make up to 40% of neutrophil cytosolic proteins¹². In response to oxidative stress, PKC activation and potassium ion flux, S100A8/A9 are actively secreted³⁹. We found that exposure of neutrophils to acute hyperglycemia rapidly triggers a reduction in S100A8/A9 content and increased S100A8/A9 in the plasma following TIH in mice. Increased glucose flux is a well-known stimulus for oxidative stress. Unlike other cells, neutrophils are almost exclusively glycolytic and do not downregulate GLUT1 in the setting of hyperglycemia, leaving them sensitive to fluctuations in circulating glucose levels. Thus, in this setting where glucose levels are inappropriately high, neutrophils rapidly reach their maximal rate of glycolysis. This means that glycolytic intermediates accumulate and divert into other metabolic pathways leading to the generation of ROS and activation of PKC⁴⁰ to trigger S100A8/A9 release. We demonstrate here that glucose induced S100A8/A9 loss from neutrophils is mediated via ROS signaling. Furthermore, myeloid-restricted GLUT-1 deletion reduced glucose uptake in neutrophils and protected against TIH-induced myelopoiesis and atherogenesis. This is consistent with previous findings from the Yvan-Charvet group, where Apoe^−/− mice transplanted with Apoe^−/−Slc2a1^+/− BM had reduced circulating monocytes and plaque macrophages compared to Apoe^−/− mice transplanted with Apoe^−/−Slc2a1^+/+ BM⁴¹. The important distinction between our study is the deletion of GLUT-1 being restricted to mature myeloid cells, and that Apoe^−/− HSPCs, unlike the Ldlr^−/− HSPCs, have a basal cell intrinsic myeloproliferative phenotype driven by cholesterol accumulation¹⁹, supporting the notion that glucose uptake in neutrophils following TIH is important in driving CVD.

Given the key role of the S100A8/A9-RAGE axis in hyperglycemia-induced myelopoiesis and subsequent monocytosis following TIH, this pathway represents an important target to abrogate hyperglycemia-induced atherosclerosis. We demonstrated the role of this axis using the S100A8/A9 inhibitor, ABR-215757²⁸, a drug currently being developed to treat systemic sclerosis, which blocks the binding of S100A8/A9 to its receptors including RAGE. ABR-215757 is also known to inhibit human S100A12 which is expressed in myeloid cells, is increased in the circulation in diabetes and is implicated in atherogenesis, but importantly is not expressed in mice^{42, 43}. We have also recently reported that ABR-215757 treatment in STZ-induced chronic diabetes normalizes circulating platelet levels and reduces atherogenesis⁴⁴. HSPCs from the BM can mobilize to the spleen where they undergo extramedullary myelopoiesis to produce atherogenic Ly6-C^hi monocytes, however, this mobilisation is impaired in chronic diabetes. In response to TIH we also observed no increase in the mobilisation of HSPCs or extramedullary myelopoiesis in WT mice. However, in the Apoe^−/− mice, which we have previously reported to have HSPC-intrinsic defects promoting HSPC mobilisation^{19, 20}, we observed that TIH exacerbated myelopoiesis within the spleen. This is likely to be driven by the amplification of the GMPs which have already differentiated from these mobilized HSPCs. Notably, ABR-215757 reduced TIH-induced myelopoiesis within the spleen, thereby likely contributing to its atheroprotective phenotype. Additionally, while ABR-215757 clearly antagonizes the effects of glucose-induced myelopoiesis on the BM, other actions on platelets⁴⁴ or the vasculature which contribute to its vasculoprotective effects cannot be excluded and are being investigated.

In summary, we show that TIH can accelerate preclinical atherosclerosis. This is consistent with the increased CV risk observed in patients with IGT, independent of known risk-factors, as well as the substantial residual CV risk in individuals with diabetes despite what is perceived to be adequate glycemic control^6–8. While beyond the scope of this study, it is important to note that hypoglycemic events have also been associated with an increased risk of cardiovascular events, particularly in those with more advanced disease^{45, 46}. Frequency of hypoglycemia has also been associated with increased intima-media thickness⁴⁷ indicating a potential role for hypoglycemia in accelerating atherosclerosis, however, these studies did not correct for TIH present in these patients. Hypoglycemia has been shown to result in acute neutrophil mobilisation and activation^{48, 49} and it would be interesting to explore whether such activation could also contribute to increased circulating levels of S100A8/A9 in patients with IGT. Importantly, however, no hypoglycemic events were observed in our TIH model. Moreover, ABR-215757 acts downstream of changes in glucose levels and is thus not at risk of inducing hypoglycemic events.

We propose that abnormal transient elevations in circulating glucose concentrations are sufficient to induce pathogenic pathways via the hematopoietic system, leading to plaque formation and progression, as well as interfering with the ability of plaques to regress. Significantly, TIH induces a similar pathway to that seen in chronic hyperglycemia, suggesting therapies targeting this pathway (i.e. S100A8/A9) could have significant reach across the array of glucose-associated metabolic diseases to lower cardiovascular events and mortality. In the absence of methods to restore physiological glucose control, an alternative strategy to modulate the pathogenic impacts of TIH should be a priority, given that TIH affects over a billion-people world-wide.

NOVELTY AND SIGNIFICANCE.

What Is Known?

• In mouse models of diabetes, chronic hyperglycemia induces signaling via the S100A8/A9-RAGE axis to promote myelopoiesis which consequently impairs the regression of atherosclerosis.

• People with diabetes and pre-diabetes experience frequent fluctuations in glycemia and greater hyperglycaemic responses post-prandially.

What New Information Does This Article Contain?

• Neutrophil sensing of hyperglycemia promotes the release of S100A8/A9, which is increased in the plasma following transient and intermittent hyperglycemia.

• Transient and intermittent hyperglycemia, independent of other complications of diabetes or chronic hyperglycemia, promotes S100A8/A9-RAGE-induced myelopoiesis and accelerates the development of atherosclerosis.

• Pharmacological inhibition of S100A8/A9 in mice exposed to transient and intermittent hyperglycemia reduces myelopoiesis and atherosclerosis.

Current standards for assessing hyperglycemia are based on HbA1C measures which detect chronic hyperglycemia, however increasing evidence suggests that people with diabetes and pre-diabetes experience frequent fluctuations in glycemia including post-prandial hyperglycemia. Furthermore, impaired glucose tolerance is associated with an increased incidence of CVD, despite presumably good glucose control based on HbA1c levels. Standard mouse models of diabetes exhibit chronic elevation of glucose not reflective of human disease. By modelling post-prandial hyperglycemia, this study demonstrates that, independent of effects of chronic hyperglycemia and cholesterol, transient and intermittent hyperglycemia promotes myelopoiesis and monocytosis and accelerates atherosclerosis. Mechanistically, we show that neutrophils act as a sensor of hyperglycaemia to release S100A8/A9 which signals via RAGE to promote myelopoiesis and atherosclerosis. From a clinical perspective, our findings highlight the importance integrating continual glucose monitoring into standard care and suggest that pharmacological inhibition of S100A8/A9 could prevent the inflammatory and cardiovascular consequences of hyperglycemia where methods of glucose control fail to prevent transient hyperglycemia.

Nonstandard Abbreviations and Acronyms:

TIH

Transient intermittent hyperglycemia

CVD

Cardiovascular disease

WT

Wild-type

RAGE

Receptor for advanced glycation end-products

IGT

Impaired glucose tolerance

CGM

Continuous glucose monitoring

WTD

Western-type diet

BM

Bone marrow

HSPC

Hematopoietic stem and progenitor cell

GMP

Granulocyte macrophage progenitor

CMP

Common myeloid progenitor

STZ

Streptozotocin

2-DG

2-Deoxyglucose

ROS

Reactive oxygen species

NAC

N-acetylcysteine

NOX

Nicotinamide adenine dinucleotide phosphate (NADPH) oxidase

GLUT

Glucose transporter