Hyperglycemia Induces Emt6 Cancer Growth
In three different mouse models of hyperglycemia, BC growth was accelerated by the high glucose levels (Fig. 1). In a model of Type 1 Diabetes, BALBc mice that were injected with streptozotocin (STZ) to induce hyperglycemia had 2.11-fold increase in tumor weight compared to normoglycemic (citrate buffer injected) controls. (Fig. 1a; 0.585 ± 0.2080 g vs 0.277 ± 0.1213 g, P < .001). In Leprdb/db mice, a genetically diabetic mouse that models Type 2 Diabetes, tumor weights were increased by 4.22-fold compared to their genetically normoglycemic Dock7m+/Leprdb heterozygote control group (LeprDock/db) (Fig. 1b; 0.373 ± 0.136 g vs 0.088 ± 0.037 g, P = .0079). In a diet-induced hyperglycemia model, WT C57BL6 mice fed a “Western diet” for 16 weeks was also used to more closely model chronic hyperglycemia in humans. There was a 2.95-fold increase in tumor weights Western diet-fed mice compared to chow-fed controls (Fig. 1c; 0.374 ± 0.157 g vs 0.127 ± 0.047 g, P < .001) and had significantly elevated glucose levels (129.4 ± 21.44 mg/dL vs 86.35 ± 9.172 mg/dL, P < .001).
Increased expression of pro-inflammatory markers and markers of macrophages in BC from hyperglycemic mice
Expression of macrophage and pro-inflammatory markers were assayed in EMT6 tumors from hyperglycemic mice (Fig. 2, 3). Expression of pro-inflammatory markers Il6, Ccl2, and Tnf were significantly increased in STZ-treated hyperglycemic mice compared to a normoglycemic, citrate buffer control group (Fig. 2a). To further understand the type of macrophage populations present in these tumors, a marker of total macrophages (Cd68), pro-inflammatory (Cd38), or pro-resolving (Egr2) macrophages were analyzed. There was a 124.6- and 4.37- fold increase in Cd68 expression in hyperglycemic BALBc and Leprdb/db mice, respectively, suggesting influx of macrophages into the tumor tissue under hyperglycemic conditions. Additionally, hyperglycemic mice had higher expression levels of both Cd38 and Egr2. Similarly, expression of pro-inflammatory markers was upregulated in tumors from hyperglycemic Leprdb/db mice (Fig. 2b). Levels of Tnf mRNA were significantly increased compared to a normoglycemic Dock7m+/Leprdb heterozygote control group (LeprDock/db). Macrophage pro-inflammatory marker Cd38 was increased, but not statistically significant, in Leprdb/db, which may explain why both Il6 and CCl2 expression were upregulated in hyperglycemic mice, but not statistically significant. In contrast, WT mice on a Western diet tended to decrease the expression of each pro-inflammatory marker (Fig. 2c). Expression of Cd38 was significantly decreased (P = .04), which may explain the overall decreased inflammatory expression in these tumors.
Inhibition of miR-467 prevents macrophage accumulation in BCs from hyperglycemic mice
To understand how miR-467 affects macrophage accumulation in BC, a miR-467 antagonist was used in BALB/c mice injected with STZ, and sections of EMT6 tumors were stained with antibodies against markers of macrophages, anti-CD68 and MOMA-2. Tumors from hyperglycemic mice had a 24.3% increase in miR-467 levels (Fig. 3a, P = .01) and the macrophage infiltration was also increased (Fig. 3b,c). In the group that received injections of the control oligonucleotide, we detected a 2.06-fold increase in CD68 macrophage staining in BC sections (Fig. 3b, P = .03) and a 2.4-fold increase in staining with the MOMA-2 antibody (Fig. 3c, P = .02) in hyperglycemic mice (STZ-treated). In normoglycemic mice, CD68-positive staining decreased 3.65-fold in response to injections of the miR-467 antagonist (Fig. 3b, P = .001), and MOMA-2 staining decreased 2.89-fold (Fig. 3c, P = .02). Inhibition of miR-467 with an antagonist, significantly blunted CD68 macrophage staining in hyperglycemic (STZ-injected) mice by 5.56-fold (Fig. 3b, P = .02) and also decreased MOMA-2 staining by 4.08-fold (Fig. 3c, P = .01).
Expression of miR-467 in human breast tissue positively correlates with glucose levels
In collaboration with the Cleveland Clinic biorepository, we obtained normal and malignant breast tissue from chronically hyperglycemic (HbA1c > 7) or normoglycemic (HbA1c < 6) patients. These specimens were assessed for miR-467 expression to determine whether the miR-467-dependent pathway is present in humans. There was a 3.7-fold increase in miR-467 expression in normal breast tissue, comparing hyperglycemic and normoglycemic groups (Fig. 4a, P = .02). In normoglycemic patients, there was no difference in miR-467 levels in malignant vs normal breast tissue. In hyperglycemic patients, miR-467 expression was dramatically increased 258-fold in malignant tissue when compared to normoglycemic patients (P < .001) and increased 56-fold compared to normal tissue (P = .008). Additionally, there was a positive correlation between increasing glucose levels and miR-467 expression in normal breast tissue (Fig. 4b, r2 = 0.6109).
Increased macrophage accumulation in malignant breast tissue from hyperglycemic patients
To understand how macrophage accumulation in BC tumors is changed by hyperglycemia, sections of normal and malignant breast tissue samples from patients were stained for a marker of macrophages, CD68 (Fig. 4c). Comparing breast tissue from hyperglycemic patients, macrophage staining increased in both the normal tissue (2.18-fold, P = .04) and malignant tissue (2.17-fold, P = .002). Macrophage staining was significantly increased in malignant tissue as compared to normal tissue in both normoglycemic (1.65-fold, P = .02) and hyperglycemic patients (1.64-fold, P = .04).
Increased angiogenesis and decreased TSP-1 in human diabetic BC samples
To determine whether the negative regulation of TSP-1 by miR-467 is present within the human specimens, breast tissue specimens from patients with HbA1c > 7 were assessed for TSP-1 protein levels (by an anti-TSP-1 antibody) and levels of the angiogenesis markers (anti-CD31 and anti-α-actin antibodies) (Fig. 4d-f). As expected, staining of TSP-1 tended to decrease along with an increase in CD31 staining, a marker of endothelial cells, in hyperglycemic patients compared to the normoglycemic group. α-actin, a marker of smooth muscle cells and maturation of the vessels, was either decreased in normal breast tissue or not changed in tumor tissue of hyperglycemic patients (Fig. 4f), suggesting growth of less mature blood vessels in hyperglycemic tissues. In normoglycemic patients, α-actin staining was lower in tumors as compared to the normal tissue, consistent with the less mature leaky blood vessels expected in cancer tissue.
Increased expression of pro-inflammatory markers and markers of macrophages in BC from hyperglycemic patients
Similarly to mouse tumors, expression of macrophage and pro-inflammatory markers were analyzed in tumors from hyperglycemic patients (Fig. 5). Expression of pro-inflammatory markers IL6 and CCL2 were increased in tumors compared to the normal breast tissue (Fig. 5a,b). IL6 levels were further increased 5-fold in malignant specimens from hyperglycemic patients compared to the normoglycemic group (Fig. 5a, P = .03). To further understand the type of the increased macrophage population present in these tumors, markers of pro-inflammatory (CD38, Fig. 5c) and pro-resolving (EGR2, Fig. 5d) macrophages were measured. Both populations were increased in tumors. However, there was a 7.5-fold increase in CD38 expression in specimens of tumors from hyperglycemic patients as compared to the specimens of tumors from normoglycemic patients (Fig. 5c, P = .04), but the levels of pro-resolving (EGR2) macrophages marker were decreased 2.4-fold in hyperglycemic tumors (Fig. 5d, P = .006), suggesting that infiltrating TAMs are more pro-inflammatory in human BC tumors of hyperglycemic patients compared to normoglycemic patients.
Level of circulating plasma miR-467 is higher in hyperglycemic patients
We found that miR-467 is a circulating miRNA: we obtained discarded blood plasma samples from both normoglycemic and hyperglycemic patients and quantified the levels of miR-467. Plasma miR-467 in hyperglycemic patients was increased by 80.5% (Fig. 6a, P = .04).
Level of circulating plasma miR-467 is higher in hyperglycemic mice
We measured the levels of miR-467 in blood plasma from hyperglycemic mouse models. We did not detect increased levels of miR-467 in STZ-treated BALB/c mice (Fig. 6b), but plasma from Leprdb/db mice had a 2.17-fold increase in plasma miR-467 as compared to normoglycemic LeprDock/db heterozygote controls (Fig. 6c, P = .03). In a mouse model of diet-induced obesity and chronic hyperglycemia, mice on a Western diet had a 15-fold increase in plasma miR-467 compared to chow-fed mice (Fig. 6d, P < .001).
Level of circulating plasma miR-467 is higher in mice with EMT6 BC tumors
Dramatically increased levels of miR-467 in tumor of hyperglycemic patients and the finding of circulating miR-467 suggested that the levels of circulating miR-467 may be elevated in the presence of a tumor and that miR-467 may clinically useful as a BC biomarker.
We measured the levels of miR-467 in blood plasma samples from hyperglycemic mouse models injected with EMT6 BC cells. In mice injected with STZ to induce hyperglycemia, mice with EMT6 tumors had an 85% increase in plasma miR-467 compared to mice without cancer (Fig. 6e, P = .047). We did not detect increased levels of plasma miR-467 in Leprdb/db and mice on Western diet (data not shown). Because mouse xenografts do not fully imitate the human BC growth, we further explored the changes in miR-467 levels in blood cells and bone marrow (BM) as described below, in order to detect any early signs of miR-467-dependent response in mouse models.
Additionally, the presence of EMT6 BC resulted in increased plasma miR-467 levels in all three hyperglycemia models in response to elevated blood glucose levels: e.g., there was a 4.5-fold increase in plasma miR-467 in STZ-injected hyperglycemic mice with tumors compared to the normoglycemic controls with tumors (Fig. 6f, P = .054). In Leprdb/db mice with tumors, there was a 46.0% increase in plasma miR-467 compared to normoglycemic mice with tumors, but it did not reach statistical significance (Fig. 6g). In WT mice with tumors on the Western diet, plasma miR-467 had a 6-fold increase compared to chow-fed mice with tumors, although it did not reach statistical significance (Fig. 6h).
miR-467 levels in whole blood, blood cells, and bone marrow of mice with BC tumors.
To further understand the source of circulating miR-467 in response to BC, we analyzed fractions of whole blood for miR-467 in WT mice on a chow or Western diet. This model of hyperglycemia did not have increased levels of miR-467 in plasma, but we expected to find early changes in response to EMT6 tumors. miR-467 levels in whole blood did not change (Fig. 7a). miR-467 levels in red blood cells (RBC) and white blood cells (WBC) fractions were also unchanged (Fig. 7b,c). Lack of changes in miR-467 levels in blood in response to tumor prompted us to look at the effects in bone marrow (BM).
Increased miR-467 expression in bone marrow (BM) from hyperglycemic mice
BM analysis demonstrated that, in hyperglycemic mice (STZ-injected), there was a 153% increase in miR-467 levels as compared to normoglycemic mice (Fig. 8a, P = .004). A similar effect was observed in WT mice on Western diet: miR-467 was increased by 74.3% in BM (Fig. 8b, P < .001). In a genetic model of diabetes, Lepr mice, there was no significant differences between the hyperglycemic (Leprdb/db) group compared to the normoglycemic heterozygous control (Dock/db) (data not shown). In mice with EMT6 tumors, we detected a significant increase in miR-467 levels in response to hyperglycemia (Figs. 8c-d). In BM from STZ-treated hyperglycemic mice with tumors, there was a 2.63-fold increase in miR-467 expression compared to BM from normoglycemic mice with tumors (Fig. 8c, P < .001). A similar effect was observed in Western diet-fed mice: in mice with tumors, miR-467 was increased by 40.0% in hyperglycemic mice as compared to normoglycemic mice on chow (Fig. 8d, P < .001).
Increased miR-467 expression in bone marrow from mice with EMT6 BC cells
The levels of miR-467 in BM were increased in presence of EMT6 tumor as compared to BM from mice without tumors: in hyperglycemic STZ-treated mice with tumors, levels were increased 2.5-fold as compared to BM from hyperglycemia mice without tumors (Fig. 8e, P < .001); in mice on Western diet with tumors, the levels of miR-467 were increased 2.7-fold as compared to mice on Western diet without tumors (Fig. 8f, P < .001). There was no difference in miR-467 BM expression in normoglycemic BALBc mice with or without tumors (Fig. 8g), but in mice fed a Western diet, presence of BC increased miR-467 expression in BM by 3.4-fold (Fig. 8h, P < .001).