miR‐467 regulates inflammation and blood insulin and glucose

Abstract Obesity is associated with inflammation and insulin resistance (IR), but the regulation of insulin sensitivity (IS) and connections between IS and inflammation remain unclear. We investigated the role of miR‐467a‐5p, a miRNA induced by hyperglycaemia, in regulating inflammation and blood glucose handling. We previously demonstrated that miR‐467a‐5p is induced by hyperglycaemia and inhibits the production of thrombospondin‐1 (TSP‐1), a protein implicated in regulating inflammation. To investigate the role of miR‐467 in blood glucose handling and tissue inflammation, WT C57BL/6 mice were fed chow or Western diet from 5 to 32 weeks of age and injected weekly with miR‐467a‐5p antagonist. Inhibiting miR‐467a‐5p resulted in 47% increase in macrophage infiltration and increased Il6 levels in adipose tissue, higher plasma insulin levels (98 ng/mL vs 63 ng/mL), and 17% decrease in glucose clearance without increase in weight or HDL/LDL. The antagonist effect was lost in mice on Western diet. Mice lacking TSP‐1 lost some but not all of the miR‐467 effects, suggesting Thbs1 (and other unknown transcripts) are targeted by miR‐467 to regulate inflammation. miR‐467a‐5p provides a physiological feedback when blood glucose is elevated to avoid inflammation and increased blood glucose and insulin levels, which may prevent IR.


| INTRODUC TI ON
Genome-wide analyses have uncovered important roles of microR-NAs in the pathogenesis of diabetes mellitus, 1 including evidence to suggest a tight regulation between microRNAs, glucose metabolism and inflammation. [2][3][4] The expression of microRNAs can be further altered by a variety of stressors, for example changes in blood glucose or pro-inflammatory cytokines 5 ; regulation of miRNAs adds another layer of complexity in regulating targets. Diet-induced obesity and increased blood glucose levels correlate with chronic inflammation and development of IR. [6][7][8][9][10][11][12][13][14] However, the sequence and causality of pathological changes leading to IR, including naturally occurring feedback mechanisms preventing the transition to IR in response to elevated blood glucose, are poorly understood.
Macrophage infiltration in adipose tissue is thought to be a main contributor in promoting chronic inflammation and development of IR. Islet inflammation promotes impaired β-cell function and subsequent failure, which occurs before the onset of type 2 diabetes (T2D). [15][16][17] Thrombospondin-1 is an extracellular matrix protein involved in regulation of tissue remodelling and inflammation. Studies in Thbs1 −/− mice suggest that a lack of TSP-1 may alleviate macrophage accumulation and the pro-inflammatory phenotype observed in insulin-resistant metabolic organs, thus protecting the animals from diet-induced inflammation and IR. [18][19][20] We recently reported that miR-467a-5p is rapidly up-regulated by high glucose in vitro and in vivo and regulates angiogenesis by targeting Thbs1 mRNA. [21][22][23][24] Others report this miRNA prevents vascular inflammation by targeting lipoprotein lipase in macrophages. 21,22,[25][26][27][28] Yet, the physiological function of miR-467a-5p and the physiological significance of its rapid up-regulation by hyperglycaemia remained unknown.
In this work, the effects of a miR-467 antagonist on blood glucose and insulin levels and inflammation in adipose tissue and pancreas were examined in wild-type (WT) and Thbs1 −/− mice to understand the role of miR-467a-5p and its target, TSP-1, in regulating inflammation in tissues and in blood glucose handling.

| MATERIAL S AND ME THODS
Detailed description of methods is provided in the Online Supplement.

| Experimental animals
Animal procedures were approved by the Cleveland Clinic IACUC. Male WT C57BL6 (n = 10/group) or Thbs1 −/− (n = 7/group) mice were fed a chow or Western diet (TD.88137, 40%-45% kcal from fat, 34% sucrose by weight, Envigo) starting at 4 weeks of age and injected weekly with a miR-467a-5p antagonist (2.5 mg/kg bodyweight) or a control oligonucleotide with no predicted targets in mouse or human genomes 22,29 intraperitoneally, starting at 5 weeks of age until the end of the experiment.

| miR-467a-5p mimic and the miR-467a-5p antagonist
The miR-467a-5p mimic and the control oligonucleotide were purchased from Dharmacon. The custom LNA-modified miR-467a-5p antagonist and a control oligonucleotide were from Qiagen.

| Induction of diabetes in mice
Male mice were injected intraperitoneally with streptozotocin (STZ, 50 mg/kg, Sigma) for 5 consecutive days. Mice with blood glucose >250 mg/dL were selected for experiments.

| Blood cell counts, HDL/LDL cholesterol and cytokines in blood
Blood was collected by cardiac puncture, and circulating blood cell counts were analysed using an ADVIA 120 Hematology System (Siemens). Plasma insulin was measured using Insulin Mouse ELISA kit (Thermo).
HDL and LDL cholesterol were measured using the HDL and LDL/VLDL quantification kit (BioVision) at end of the experiment.

| Immunohistochemical staining
Visceral (omental) adipose tissue and pancreas were fixed in 4% formaldehyde (Electron Microscopy Sciences) for 24 hours and stained using VECTASTAIN ABC-HRP Kit (Vector Labs) with corresponding primary antibodies. Slides were scanned using Leica SCN400 or Aperio AT2 at 20X magnification. Quantification of positive staining was performed with Photoshop CS2 (Adobe) or Image Pro Plus (7.0).

| Isolation of bone marrow-derived macrophages
Isolation of bone marrow-derived macrophages (BMDM) was per-

| Oil Red O Staining
Differentiated 3T3-L1 cells were stained in the Oil Red O solution for 10' at RT.

| Statistical analysis
Data are expressed as the mean value ± SEM. Statistical analysis was performed with GraphPad Prism 5 Software. Student's t test and ANOVA were used to determine the significance of parametric data, and Mann-Whitney test was used for non-parametric data.

| Injections of miR-467a-5p antagonist increase macrophage accumulation in adipose tissue and pancreas
Inflammation and macrophage infiltration in tissues are associated with, and often precede, IR. [32][33][34][35][36][37] Macrophage accumulation in adipose tissue and pancreas was assessed by immunohistochemistry using an Male WT mice on chow diet were injected weekly with a miR-467a-5p antagonist (2.5 mg/kg) for 32 weeks, starting at 5 weeks of age. Injections of the miR-467a-5p antagonist increased macrophage accumulation in AT of chow-fed mice by 47% ( Figure 1A). Baseline AT macrophage infiltration was increased in mice on Western diet (65.4%, Figure S2A) without further increase in response to antagonist ( Figure 1B).
In pancreas, macrophage infiltration tended to increase in antagonist-injected mice on either diet (62.4% increase, chow-fed and 43.9% increase, Western diet ( Figure 1C,D) and was significantly increased by the Western diet (73.3%, Figure S2B).
Changes in tissue macrophage infiltration in response to the antagonist were not explained by the number of monocytes in blood F I G U R E 1 miR-467 antagonist increases macrophage accumulation in adipose tissue and pancreas from WT chow-fed mice. Macrophage accumulation in adipose tissue from WT mice on chow diet (A) and Western diet (B) was determined by MOMA-2 staining. Positive staining was normalized to mean adipocyte area for adipose tissue as adipocyte sizes were changed between groups. Macrophage accumulation in pancreas from WT mice on chow diet (C) and Western diet (D) was determined by anti-CD68 staining. Data are relative to control oligo. n = 10 mice/group. *P < 0.05

| miR-467a-5p antagonist has differential effects on the expression of inflammatory markers in adipose tissue
Adipose tissue expression of Il6, Tnf, Ccl2, Ccl4 or Il1b was assessed by RT-qPCR in chow or Western diet-fed WT mice (Figure 2A,B, respectively) injected with miR-467a-5p antagonist or control oligonucleotide.
Out of five cytokines, only Il6 expression was statistically significantly increased by the antagonist in chow-fed mice ( Figure 2A). Notably, all cytokine expression tended to be reduced in Western diet-fed mice in response to the antagonist ( Figure 2B).
Western diet increased baseline expression of inflammatory markers in AT of WT mice injected with control oligonucleotide ( Figure S4).

| TSP-1 knockout does not prevent macrophages infiltration in mice injected with the miR-467a-5p antagonist
We reported that TSP-1 is a direct target of miR-467 21

| miR-467a-5p antagonist has no effect on macrophage infiltration and levels of inflammatory markers in liver
Hepatic inflammation and macrophage infiltration in liver are as-

| Systemic injections of miR-467a-5p antagonist increase fasting insulin levels and decrease insulin sensitivity in chow-fed WT mice
In chow-fed mice, fasting blood glucose and insulin levels were measured at the end of the experiment. Inhibiting miR-467, using systemic injections of the antagonist, had no effect on fasting blood glucose levels in chow-fed mice but significantly increased fasting insulin levels (0.53 1.01 mg/dL vs 1.01 mg/dL) ( Figure 4A,B).
In glucose tolerance tests (GTT), no changes were observed ( Figure S7A). However, elevated glucose levels in antagonist-injected mice were observed during the ITT for all time-points ( Figure 4C).
Analysing the rate of glucose disappearance from plasma (based on the K itt analysis at 0-60 minutes, when the decrease in glucose levels was linear) revealed a significant decrease in glucose clearance in response to the antagonist ( Figure 4D).

| Systemic injections of miR-467a-5p antagonist do not affect mouse weight or blood lipid profile in chow-fed mice
Weight, HDL and LDL cholesterol, total cholesterol, and free cholesterol were measured. Antagonist injections had no effect on LDL ( Figure S7D), but decreased HDL, total and free cholesterol ( Figure S7C,E,F).
Similar to WT mice, weight was not affected by the antagonist in Figure S7H). The levels of HDL, total and free cholesterol were unchanged in response to the antagonist ( Figure S7I,K,L), but the LDL cholesterol levels were increased in mice injected with the antagonist in the absence of TSP-1 ( Figure S7J).

| TSP-1 knockout eliminates the effects of miR-467a-5p antagonist on insulin sensitivity in chowfed mice
As with WT mice, fasting blood glucose and insulin levels were measured in Thbs1 −/− mice at the end of the experiment. Thbs1 −/− mice were used in an identical experimental design as described above ( Figure 4). Without TSP-1, there was no effect by the antagonist on blood insulin levels ( Figure 4F). No differences were observed in GTT ( Figure S7G). In Thbs1 −/− mice, unlike in WT mice, blood glucose was not increased in the ITT in response to the antagonist at any time-point ( Figure 4G). Surprisingly, antagonist injections tended to improve IS, suggesting that other targets of miR-467 may become important in the absence of TSP-1. Loss of TSP-1 normalized, and even slightly increased, the plasma glucose disappearance rate in antagonist-injected mice ( Figure 4H).

F I G U R E 3
Macrophage accumulation in adipose tissue and pancreas from Thbs1 −/− mice on chow or Western diet. Macrophage accumulation in adipose tissue from Thbs1 −/− mice on chow diet (A) and Western diet (B) was determined by MOMA-2 staining. Positive staining was normalized to mean adipocyte area for adipose tissue as adipocyte sizes were changed between groups. Macrophage accumulation in pancreas from Thbs1 −/− mice on chow diet (C) and Western diet (D) was determined by anti-CD68 staining.
Data are relative to control oligo. n = 7 mice/group. *P < 0.05

| Systemic injections of miR-467a-5p antagonist increase blood glucose levels and decrease fasting insulin levels in WT mice on the Western diet
In Western diet-fed mice, similar experiments revealed a different response to miR-467 inhibition and a loss of its protective function. As expected, mice on a Western diet developed diet-induced IR: fasting blood glucose levels were increased compared with chow-fed mice (102.7 ± 2.88 vs 117.9 ± 2.42, P < 0.001 Figure 5A vs Figure 4A); insulin levels were twice higher in mice on Western diet (0.5301 ± 0.0820 vs 0.9327 ± 0.0660, P = 0.0015 Figure 5B vs Figure 4B). Antagonist injections further increased blood glucose levels by 20.85% (142.4 mg/dL vs 117.9 mg/dL) in mice on the Western diet ( Figure 5A) but significantly decreased blood insulin levels by 25.81% (0.69 mg/dL vs 0.93 ng/mL, Figure 5B).
Glucose clearance was delayed in the GTT ( Figure 5C). The antagonist did not increase glucose levels in ITT, but a higher baseline blood glucose levels was detected compared with mice on chow in Figure 4A ( Figure 5D). Analysis of the rate of glucose disappearance from plasma (K itt analysis) revealed a significant increase in glucose clearance in antagonist-injected mice, which was opposite in chow-fed mice ( Figure 5E vs Figure 4D). These data suggest that, in mice on Western diet, the protective effect of miR-467 (decreased insulin levels and accelerated clearance of glucose from blood) is lost and even further counteracted by new, pro-IR effects of miR-467.

| Systemic injections of miR-467a-5p antagonist do not affect mouse weight or blood lipid profile in WT mice on Western Diet
Similar to mice on the chow diet, the antagonist did not affect   Figure S7 vs Figure S8).

| Effects of the miR-467 antagonist on the expression of glucose transporters (GLUT1, GLUT2, GLUT4)
Expression of Slc2a1, a ubiquitous insulin-independent glucose transporter GLUT1, was measured in mouse AT, pancreas and liver in mice on chow and Western diet injected with the antagonist or control oligonucleotide ( Figure 6A-D).
No change in pancreas Slc2a1 expression was detected in response to the miR-467 antagonist ( Figure 6A).
Slc2a1 expression in AT from Western-fed mice was decreased by the antagonist (Figure 6B), but the effect was lost in Thbs1 −/− mice ( Figure 6C), suggesting TSP-1 as a target in regulation of GLUT1. In liver, Slc2a1 expression was unaffected by the antagonist, but was decreased by the Western diet ( Figure 6D).
In both pancreas and AT, there seemed to be a cumulative effect of the antagonist and Western diet: the expression was significantly decreased in antagonist-injected mice on Western diet compared with antagonist-injected mice on chow, without decreased expression in control oligonucleotide-injected mice ( Figure 6A,B).
Expression of the major glucose transporters was also measured: Slc2a2 (GLUT2) in pancreas and liver and Slc2a4 (GLUT4) in AT ( Figure S9). No changes were detected in these transporters in response to the antagonist. Western diet affected the expression in an organ-and transporter-specific manner: in pancreas, Slc2a1 expression was lower in mice on Western diet ( Figure 6A), while Slc2a2 expression was increased ( Figure S9A). In AT, Slc2a1 was decreased Western diet-fed mice ( Figure 6B), which was even more pronounced in Thbs1 −/− mice ( Figure 6C). AT Slc2a4 expression was unchanged in either genotype, although in Thbs1 −/− mice, expression tended to be lower in mice on the Western diet ( Figure S9C,D).

| miR-467a-5p in adipose tissue and the effects of the antagonist injections
To determine additional changes induced in AT by the antagonist, we evaluated the levels of miR-467 expression and TSP-1 protein, size of the adipocytes, and quantified ECM proteins ( Figure S10).  Mean areas and perimeters of adipocytes were increased by the Western diet and tended to increase in response to miR-467a-5p antagonist injections ( Figure S10E,F).

Fibrosis and ECM deposits between cells in AT affect remodel-
ling, growth and function of adipocytes. 32,39,41 To evaluate changes of ECM amounts in AT, sections were stained with Masson's Trichrome to assess ECM levels. There was no difference in staining between the mouse groups ( Figure S10G,H).

| miR-467a-5p in pancreas and the effects of the antagonist injections
To evaluate other effects of the antagonist in the pancreas, we examined miR-467 and TSP-1 levels, islet area and vascularization ( Figure S11).
Sections of pancreas were stained with the anti-insulin antibody and counterstained with haematoxylin ( Figure S11D). The total islet area in the pancreas was unchanged in any of the mouse groups.
We have previously reported that miR-467a-5p promotes angiogenesis as a result of regulation of production of its target, thrombospondin-1 (TSP1). 21 We also evaluated TSP-1, a target of miR-467, in the pancreas. At the end of the experiment, TSP-1 levels were not affected by miR-467 antagonist injections or by the diet (Figure S11G).

| Expression of gluconeogenesis gene in liver of mice injected with miR-467 antagonist
To evaluate whether changes in blood glucose were mediated due to regulation of gluconeogenesis, we examined expression of key gluconeogenesis enzymes in liver ( Figure S12). G6pc encodes glucose-6-phosphatase, a regulator of conversion of glucose 6-phosphate to glucose. Liver G6pc expression was decreased by the antagonist (significant in mice on the Western diet) ( Figure S12A). Thus, G6pc could not be responsible for the higher levels and slower clearance of blood glucose from Figure 4. G6pc expression was significantly decreased by the Western diet. Fbp1 is an enzyme catalysing the hydrolysis of fructose 1,6-bisphosphate to fructose 6-phosphate and acting as a rate-limiting enzyme in gluconeogenesis. Liver Fbp1 was unaffected by antagonist injections but, like G6pc, was decreased by the Western diet ( Figure S12B).
Increased miR-467a-5p levels were associated with a decrease in TSP-1 production by the cells (Figure 7B,D, P < 0.05). Consistent with the mechanisms of miR-467a-5p up-regulation by HG previously described by us, 21,31 both D-glucose and the biologically inactive L-glucose had similar effects: HG up-regulated miR-467a-5p expression and decreased secreted TSP-1, which indicated the osmolarity change as a stimulus for miR-467a-5p up-regulation in macrophages.

F I G U R E 7
High glucose up-regulates miR-467 in macrophages. A and C, Expression of miR-467 after 3 h of glucose stimulation in cultured mouse macrophages (RAW 264.7) and differentiated human monocyte (THP-1) cell lines was measured by RT-qPCR and normalized to β-actin, n = 3-8 independent replicates. B and D, TSP-1 secretion was assessed in cell supernatants after 24 h of glucose stimulation by Western blot. Quantification of densitometry is shown and relative to the control, n = 3 independent replicates. *P < 0.05 vs control. E, miR-467 expression in cultured WT bone marrow-derived macrophages (BMDM) 6 h post-glucose stimulation. Data are relative to low glucose (LG) control samples, n = 3 independent replicates. *P < 0.05. F, miR-467 expression in whole bone marrow (BM) from WT mice on chow or Western diet for 32 wk, n = 10 mice/group. Data are relative to chow mice *P < 0.05. G, miR-467 expression in BM monocytes or non-monocytes from male BALB/c mice injected with STZ to induce diabetes, or a citrate buffer control. Data are relative to citrate buffer control. n = 10 mice/group. *P < 0.05 vs citrate buffer control

| miR-467a-5p mimic and antagonist regulate production of inflammatory signals by the cultured macrophages
Cultured BMDMs from WT mice were transfected with a miR-467a-5p mimic or 467-antagonist as described in Methods, and expression of Tnf, Il6, Ccl2 and Ccl4 was measured. Expression of all four cytokines was increased by HG ( Figure S13A,B), but miR-467a-5p mimic had no additional effect ( Figure S13A).
Inhibiting miR-467a-5p with the antagonist in BMDMs from WT mice prevented the up-regulation of Tnf and Ccl4 in response to HG ( Figure S13B), suggesting that these two cytokines are regulated by HG through the miR-467-dependent mechanism, while others are not.
When cultured BMDMs isolated from Thbs1 −/− mice were transfected with the miR-467a-5p mimic or antagonist, up-regulation by high glucose was similar to BMDMs from WT mice ( Figure S13C  There was no effect of the antagonist or TSP-1 deletion on CXCL1 or VEGF-A levels ( Figure S13G,H), suggesting that these two markers are not regulated by miR-467 or TSP-1.

| D ISCUSS I ON
The sequence and causality of events in development of IR are still poorly understood, 42  Inflammation and the expression of glucose transporters and key enzymes of gluconeogenesis were examined in livers, but no changes were detected in response to the antagonist injections, confirming that the effect of the antagonist is associated with the regulation of glucose clearance rather than glucose production.
We previously reported that TSP-1 transcript is a target of miR-467 and mediates miR-467 effect on angiogenesis. Interestingly, all effects of the antagonist on the blood glucose and insulin levels were lost in Thbs1 −/− mice, suggesting that TSP-1 is the main target of miR-467, and the differential regulation in chow-fed and Western diet-fed mice is downstream of TSP-1. TSP-1 is a known regulator of insulin sensitivity and metabolic disorder. [18][19][20]48,49 In adipose tissue, pro-inflammatory molecules are released by adipocytes and activated macrophages to promote insulin resistance. [50][51][52][53] The inhibition of miR-467a-5p increased infiltration of macrophages in the adipose tissue and in the pancreas, suggesting that miR-467a-5p prevents inflammation. Additionally, our results and reports from others stress the importance of ECM, and TSP-1 (a target of miR-467a-5p) specifically, and other TSPs, in the recruitment of inflammatory cells into tissues. [18][19][20][54][55][56][57][58] The increase in macrophage infiltration in adipose tissue was associated with the increased Il6 levels, which was lost in Thbs1 −/− mice. However, the levels of Tnf, Ccl2, Ccl4 and Il1b were not changed by the antagonist injections.
The reduction of inflammation in the obese AT in response to the miR-467 antagonist may be due to a significant decrease in GLUT1 (Slc2a1) expression. Bone marrow-derived macrophages isolated from mice with a myeloid-specific knockout of GLUT1 (Slc2a1) were 'metabolically reprogrammed' such that they were unable to uptake glucose properly and had a decreased inflammatory phenotype. 59 In tissues from Western diet-fed mice, there may be additional miR- their up-regulation was prevented by the miR-467 antagonist. Only Ccl4 appears to be regulated through TSP-1 pathway: up-regulation, and the effect of miR-467 antagonist, was lost in BMDMs from Thbs1 −/− mice. These results suggested that inflammation is regulated by miR-467a-5p through multiple targets and in a cell-specific manner in various cell types.
Our results unveil the physiological role of miR-467a-5p: when this miRNA is up-regulated by high blood glucose, 21,22,31 it protects against the development of IR and inflammation in response to high glucose.
Interestingly, this protection is lost under a long-term Western diet, underscoring the negative effects of this chronic stressor.

ACK N OWLED G EM ENTS
This work was supported by R01CA177771 and R01HL117216, to OS-A and 17PRE33660475 from the American Heart Association (PI: JG, Sponsor: OS-A).

CO N FLI C T O F I NTE R E S T
The authors report no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.