effect of mir-142-3p on the M2 Macrophage and therapeutic efficacy Against Murine Glioblastoma

The immune therapeutic potential of microRNAs (miRNAs) in the context of tumor-mediated immune suppression has not been previously described for monocyte-derived glioma-associated macrophages, which are the largest infiltrating immune cell population in glioblastomas and facilitate gliomagenesis. An miRNA microarray was used to compare expression profiles between human glioblastoma-infiltrating macrophages and matched peripheral monocytes. The effects of miR-142-3p on phenotype and function of proinflammatory M1 and immunosuppressive M2 macrophages were determined. The therapeutic effect of miR-142-3p was ascertained in immune-competent C57BL/6J mice harboring intracerebral GL261 gliomas and in genetically engineered Ntv-a mice bearing high-grade gliomas. Student t test was used to evaluate the differences between ex vivo datasets.

Profound immune suppression exists in the glioblastoma microenvironment that promotes glioma invasion and progression and systemically inhibits antitumor immunity. MicroRNAs (miRNAs) are noncoding molecules involved in post-transcriptional gene regulation that have been shown to modulate tumor cell proliferation and apoptosis and to act as oncogenes or tumor-suppressor genes (1,2). miR-142-3p has been shown to be a useful biomarker and potential therapeutic target in a variety of malignancies (3)(4)(5)(6). Although most studies on miR-142-3p focus on expression frequency or prognostic impact, emerging data indicate that miR-142-3p is involved in immunological reactivity and activation (7)(8)(9), including T-cell immune suppression (10)(11)(12). Moreover, miR142-3p may influence the differentiation state of immune cell populations (13,14).
Tumor-associated macrophages, the largest infiltrating immune cell population in glioblastomas (15), originate from the peripheral blood monocytes and are recruited by a variety of tumor-derived signals. Glioblastomas actively recruit circulating macrophages to the tumor site and induce them to adopt a tumor-supportive M2 phenotype capable of mediating immune suppression and promoting invasion (16). This shift away from the desired proinflammatory/antitumor M1 phenotype (17) by cytokines such as transforming growth factor beta (TGF-β) may be regulated by miR-142-3p (14,18,19). Although the immune suppressive properties of glioblastoma-infiltrating macrophages have been investigated, the relationship to miRNA dysregulation has not been evaluated to date. Therefore, we hypothesized that miR-142-3p, by interacting with the TGF-β pathway, regulates the tumor-supportive properties of glioma-associated macrophages and potent antitumor immune effects can be exerted by manipulating miR-142-3p levels in vivo.

Syngeneic Subcutaneous Model
The murine glioma GL261 cell line was obtained from the National Cancer Institute-Frederick Cancer Research Tumor Repository and cultured in a humidified atmosphere of 5% carbon dioxide and 95% air at 37°C in Dulbecco's modified Eagle medium (Life Technologies) supplemented with 10% fetal bovine serum (Sigma-Aldrich, St. Louis, MO) and 1% penicillin/streptomycin/L-glutamine (Mediatech, Manassas, VA). The cells were kept in −80°C and used within 15 passages after the thawing. To induce subcutaneous tumors, logarithmically growing GL261 cells were injected into the right hind flanks of 6-week-old C57BL/6J female mice or nude mice at a dose of 1 × 10 6 cells suspended in 100 µL of matrigel basement membrane matrix (BD Biosciences, San Jose, CA). When palpable tumors formed that were approximately 0.5 cm in diameter, the mice were treated by local tumor injection in blinding and randomization (n = 5 per group). Tumors were measured twice per week. Tumor volume was calculated with slide calipers using the following formula: V = (L x W x H)/2, where V is volume (mm 3 ), L is the long diameter, W is the short diameter, and H is the height.

Syngeneic Intracranial Clonotypic Glioma Model
To induce intracerebral tumors in C57BL/6J mice, GL261 cells were collected in logarithmic growth phase and mixed with an equal volume of 10% methyl cellulose in improved modified Eagle medium (zinc option; Life Technologies), and loaded into a 250-μL syringe (Hamilton, Reno, NV) with an attached 25-gauge needle. The mice were anesthetized with ketamine (100 mg/kg) and xylazine (10 mg/kg), and the needle was positioned 2 mm to the right of bregma and 4 mm below the surface of the skull at the coronal suture using a stereotactic frame (Kopf Instruments, Tujunga, CA), as previously described (21). The intracerebral tumorigenic dose for GL261 cells was 5 × 10 4 in a total volume of 5 μL. Mice were randomly assigned to scramble control or the miR-142-3p treatment group (n = 10 per group) and observed and weighed three times per week. When they showed signs of neurological symptoms, they were compassionately killed by prolonged carbon dioxide inhalation. Their brains were removed, placed in 4% paraformaldehyde, and embedded in paraffin.

Pharmacokinetic Study
Non-tumor-bearing C57BL/6J mice (n = 3 per time point) were administered miRNA + lipofectamine 2000 intravenously once and subsequently killed at 0, 15 minutes, 1, 4, 8 and 24 hours. The liver, peripheral blood mononuclear cells, and serum were subsequently analyzed for miRNA expression by quantitative PCR. Further details are provided in the Supplementary Methods (available online).

Genetically Engineered Murine Model of Glioma
The use of immune-competent Ntv-a mice with RCAS-plateletderived growth factor B + RCAS-B-cell lymphoma 2 induced high-grade gliomas for testing immune therapeutics has been previously described (22). To transfer genes by RCAS vectors, DF-1 producer cells transfected with a particular RCAS vector (1 × 10 5 DF-1 cells in 1-2 µL of PBS) were injected on postnatal day 1 or 2 into the frontal lobes of Ntv-a mice at the coronal suture. Twentyone days after introducing the glioma-inducing transgenes, the mice were randomly assigned to the treatment or control group (n = 9 per group). Mice were treated intravenously on Monday, Wednesday, and Friday for 3 weeks. The mice were killed 90 days after injection or sooner if they demonstrated morbidity related to tumor burden.

Statistical Analysis
All quantified data represent a mean of triplicate samples ± standard deviation (SD) or as indicated. Statistical significance was determined by Student t test or as indicated. P values less than .05 were considered statistically significant. Outliers were eliminated by the Grubbs test. All of the statistical analyses were conducted using GraphPad Prism 5 (GraphPad Software, La Jolla, CA), except for .95-fold decrease in level relative to matched peripheral blood monocytes, miR-142-3p emerged as a leading downregulated candidate. B) Total RNA was extracted from a validating set of glioma cancer stem cells (gCSCs; round dots; n = 5), glioma cell lines (cubes; n = 2), glioblastoma tumor tissues (triangles; n = 4), healthy donor peripheral blood CD14 + monocytes (inverted triangles; n = 3), glioblastoma patient peripheral blood CD14 + monocytes (diamonds; n = 6), and glioblastoma infiltrating CD11b + macrophages (empty circles; n = 3). Analysis by quantitative reverse-transcription polymerase chain reaction (qRT-PCR) demonstrated the various miR-142-3p expressions among different samples. Of note, miR-142-3p is downregulated within the glioblastoma-infiltrating macrophages relative to peripheral blood monocytes. An unpaired t test was used to calculate the two-sided P values. Central horizontal lines are means, and error bars represent standard deviations. *P < .001. the survival comparison with log-rank test performed by SAS version 9.3 (SAS Institute, Cary, NC) and TIBCO Spotfire S+ version 8.2 (Somerville, MA). For the subcutaneous tumor measurement, linear mixed models and F test were fit to assess tumor growth after adjusting for treatment effect and taking into account the associations among repeated measures within each subject. All statistical tests were two-sided.

miR-142-3p Expression in Monocyte-Derived Glioblastoma-Associated Macrophages
Using the Human miRNA OneArray Microarray v2 to assess miRNA expression profiling, the miRNA expression profile in glioblastoma-infiltrating macrophages was matched to that of monocytes from the peripheral blood. With a mean 4.95-fold decrease relative to the level in matched peripheral monocytes, miR-142-3p emerged as a leading candidate ( Figure 1A; Table 1). Meanwhile, miR-142-3p downregulation was not detected in glioblastoma tissue relative to normal brain tissue by total tissue miRNA microarray (Supplementary Table 1, available online). Subsequent real-time PCR analysis revealed that, although miR-142-3p expression was detected in total glioblastoma tumor tissues, glioblastoma cell lines or glioma cancer stem cells seldom express this miRNA. Rather it is the monocytes that are the main source of miR-142-3p. Intriguingly, miR-142-3p is downregulated within the glioblastoma-infiltrating macrophages relative to peripheral blood monocytes (mean relative expression of infiltrating CD11b+ macrophages = 2.03, SD = 1.17; mean relative expression of peripheral CD14+ monocytes = 9.45, SD =0.69; P < .001) ( Figure 1B). There was no statistically significant difference of miR-142-3p expression in peripheral monocytes from patients with glioblastoma compared with healthy donors.

miR-142-3p Interacts With the TGF-β Pathway
Using multiple bioinformatics tools (TargetScan 6.2, PicTar, miRanda, and miRDB), potential target genes of miR-142-3p were identified. A cluster of TGF-β pathway genes with conserved target sites in their 3'-untranslated regions (UTRs) were identified, including transforming growth factor beta receptor 1 (TGFBR1) and transforming growth factor beta 2 (TGFB2). The TGF-β pathway is well known for its potent role in immune suppression and tumorigenesis, including polarization of tumor-promoting M2 macrophages (26). Thus, we investigated the effect of miR-142-3p overexpression on the TGF-β pathway in the monocyte-derived M1 and M2 subsets (Supplementary Figure 1A, available online). In the miR-142-3p overexpressing cells, the TGFBR1 mRNA levels remained unchanged (Supplementary Figure 1B, available online), whereas the TGFBR1 protein levels were repressed in the M2 macrophages ( Figure 3, A and B), which indicates that miR-142-3p mediates its effect on TGFBR1 by post-transcriptional regulation rather than by targeting mRNA degradation. TGF-β2 cytokine secretion was not inhibited by miR-142-3p in either the M1 or M2 macrophages (Supplementary Figure 1C, available online). Anti-miR-142-3p treatment resulted in increased TGFBR1 protein levels only in the M2 macrophages ( Figure 3A), signifying that TGF-β pathway regulation by miR-142-3p is contextual and cell specific. Other predicted targets of miR-142-3p such as integrin beta 8 (ITGB8), integrin alpha V (ITGAV), SMAD4, NF-κB, and TGFB2 were not found to be inhibited in either the M1 or M2 macrophage populations ( Figure 3A). Furthermore, we detected the activation of the TGFBR1 downstream protein, SMAD2, after stimulation of M1 and M2 cells with TGF-β1. As expected, p-SMAD2 was inhibited by miR-142-3p only in M2 macrophages but not in the M1 macrophages ( Figure 3B). TGFBR1 SMADindependent targets such as p-AKT and RhoA were not appreciably altered by miR-142-3p. p-TAK1 was not expressed in either M1 or M2 populations (data not shown).

miR-142-3p Overexpression Induces M2 Macrophage Apoptosis
During ex vivo culture of miR-142-3p overexpressing macrophages, we observed a significant decrease in the number of M2 macrophages. Further evaluation revealed that miR-142-3p Similar results were observed in 3 replicates. The gray histogram denotes M1 macrophages, black denotes M2 macrophages, and the dotted line represents the isotype control. C) The phagocytic activities of M1 and M2 macrophages were measured by fluorescent uptake and summarized. Error bars represent standard deviations. A paired t test was used to calculate the two-sided P values. n = 6. ***P < .001. D) The miR-142-3p expression was determined by quantitative reverse-transcription polymerase chain reaction (qRT-PCR) analysis. Its expression was downregulated during macrophage differentiation but preferentially in the immunosuppressive M2 macrophages relative to the proinflammatory M1 macrophages. Error bars represent standard deviations. A paired t test was used to calculate the two-sided P values. n = 6. *P = .03.
induce selective apoptosis. As expected, more apoptosis was observed within the M2-committed macrophages treated with either SB431542 (mean fold relative to control of M2-comitted macrophages = 2.38, SD = 1.04; mean fold relative to control of M1-comitted macrophages = 1.23, SD = 0.26; P = .04; n = 6) or LY364947 (mean fold relative to control of M2-comitted macrophages = 2.29, SD = 0.35; mean fold relative to control of M1-comitted macrophages = 1.28, SD = 0.37; P = .001; n = 6) compared with the M1-committed macrophages at 48 hours ( Figure 4B). Identical results were obtained with specific TGFBR1 small interfering RNA (Supplementary Figure 3, A and B, available online), and this appeared to be cell-specific because miR-142-3p overexpression did not induce apoptosis in glioma cancer stem cells (Supplementary Figure 3, C and D, available online).

miR-142-3p Inhibits Glioma Growth
We next investigated whether miR-142-3p could exert an antiglioma effect in vivo ( Figure 5A). Aggressive subcutaneous GL261 glioma growth was observed in the control group of syngeneic C57BL/6J mice, whereas glioma growth was markedly inhibited during the 3-week course of miR-142-3p local treatment (P = .03; n = 5 per group) ( Figure 5B).To determine whether miR-142-3p treatment can exert a similar effect against intracerebral glioma, syngeneic C57BL/6J mice harboring GL261 were treated with miR-142-3p intravenously for 3 weeks ( Figure 5C). The median survival duration for the scramble control group was 23.5 days, which was extended to 31 days for mice treated with miR-142-3p (P = .03; n = 10 per group) ( Figure 5D). During the treatment course, no behavioral or neurological abnormalities were observed in the mice. Necropsies of the brains of miR-142-3p-treated mice demonstrated no evidence of demyelination, macrophage infiltration, or lymphocytic infiltration in the non-tumor-bearing areas that would indicate the induction of autoimmunity. Furthermore, administration of miR-142-3p did not induce a peripheral lymphocytosis or monocytosis (Supplementary Figure 4, available online).

Discussion
The objective of this study was to identify potential therapeutic immune modulatory miRNAs. As such, we cataloged downregulated miRNAs using a novel subtractive approach to identify the preferential expression profile of miRNAs within immune-suppressive (M2) glioblastoma-infiltrating macrophages relative to the monocyte precursor. This screening strategy was intended to identify those miRNAs that are downregulated by the tumor microenvironment that could have biological roles in the monocyteto-macrophage differentiation state, skewing to or maintenance of the M2 phenotype, or mediating M2 immune suppression. We then selected and prioritized potential candidates on the basis of binding to immunosuppressive pathways or mechanisms. Although several alternative candidates identified in the human miRNA microarray expression library may have roles in modulating regulatory T-cell induction pathways, miR-142-3p was predicted to interfere with immune suppressive TGF-β signaling, which is known to influence the M2 phenotype (31,32). Other candidates such as miR-378 have been shown to inhibit macrophage proliferation (33); however, preferential inhibition of only the M2 phenotype would be desirous from a cancer therapeutic perspective. As additional mechanisms and pathways of glioblastoma-mediated immune suppression become elucidated, additional miRNAs are likely to be categorized as potential therapeutic immune modulatory miRNAs, Figure 5. miR-142-3p inhibits in vivo glioma growth. The local treatment schema (A) and volume of subcutaneous (s.c.) GL261 tumors (B) in C57BL/6J mice treated with scramble control (ctrl) and miR-142-3p (miR) starting on day 5 (n = 5 per group). Empty circles are scramble controls, and triangles are miR-142-3p-treated mice. Standard deviations are shown. Linear mixed models were fit to assess tumor growth, and two-sided F test was used. *P = .03. The in vivo experiment was duplicated with similar results. The intravenous (i.v.) treatment schema (C) and graph of the Kaplan-Meier estimate of survival time in C57BL/6J mice implanted with intracerebral (i.c.) GL261 gliomas (D), which showed that miR-142-3p improved survival in the miR-142-3p-treated group (miR; round dots) relative to the scramble controls (ctrl; empty triangles). Log-rank test was used to compare overall survival between groups. n = 10 per group. *P = .03. and these could ultimately be used in a complementary or alternative fashion with miR-142-3p.
The biological role of miR-142-3p in the tumor-associated M2 macrophage has not been previously described. Overexpression of miR-142-3p induces selective apoptosis in the M2 macrophage population due to the inhibition of autocrine-dependent TGFBR1. The specificity of miR-142-3p for TGFBR1 was predicted by multiple binding algorithms and confirmed by luciferase reporting assays and mutational analyses. Furthermore, despite the off-target effect of LY364947 on vascular endothelial growth factor (34), which can induce macrophage to M2 skewing under selective conditions (35), the more specific TGFBR1 blockade data (ie, SB431542 and small interfering RNA) indicate that the induced M2 macrophage apoptosis is secondary to inhibition of its dependent TGFBR1 pathway. Besides the autocrine growth dependence on TGF-β reported in cancer cells (30), we now extend these observations to the immune-suppressive M2 macrophage. However, we have not examined the exact biochemical mechanisms for cell death induced by TGFBR1 blockade We have previously demonstrated that glioma cancer stem cells can induce M2 macrophages (16) and that the M2 population is a negative prognosticator in murine models of high-grade gliomas (22). The specific targeting of the M2 immune population while sparing the M1 population in vivo for therapeutic intent is desirable. Prior indirect macrophage targeting strategies have included inhibiting macrophage differentiation and cytokine production (36) or trafficking to the tumor microenvironment (37). We are now showing that targeting of TGFBR1 with miR-142-3p, likely in the circulating monocytes as they become differentiated to the M2, can exert a therapeutic effect against malignant gliomas in vivo.
The exploitation of the immune system to mediate the therapeutic effects of miRNA, such as miR-142-3p, can circumvent previous limitations of miRNA delivery, including getting past the blood-brain barrier. Furthermore, circulating immune cells are the first point of contact to administrated miRNAs, affording an opportunity to directly modulate their functional activity. Despite a therapeutic effect of miR-142-3p against intracerebral gliomas, therapeutic "cures" were not observed, especially in the heterogeneous, genetically engineered murine models. The heterogeneity of immune suppressive mechanisms and pathways within malignant gliomas is widely acknowledged and documented; however, patients that have glioblastomas enriched with M2 macrophages may be particularly responsive to treatment with miR-142-3p, especially because the therapeutic effect of miR-142-3p in vivo was associated with a decreased glioma macrophage infiltration. A greater understanding of the compartmentalization and the Log-rank test was used to compare overall survival between groups. n = 9 per group. *P = .03. C) Immunohistochemistry demonstrating staining with anti-F4/80 antibodies to identify glioma-infiltrating macrophages. Left panel: representative images of mice treated with scramble control (ctrl) and miR-142-3p (miR), respectively. Scale bar = 50 µm. Right panel: quantification of glioma-infiltrating macrophage and comparison between the two groups. Triangles are scramble controls, and round dots are miR-142-3p-treated mice. A paired t test was used to calculate the two-sided P values. n = 8 per group. *P = .009. D) The correlation between the percentage of glioma-infiltrating F4/80+ macrophages and the survival duration of miR-142-3p treated mice. n = 8. R 2 = .303. pharmacokinetics of in vivo miRNA administration is needed, especially in the context of more stable and efficient vectors before clinical trial implementation.
This study did have some limitations in that we have not examined the exact biochemical mechanisms for cell death induced by TGFBR1 blockade. Furthermore, our therapeutic approach is not scalable or appropriate for use in human subjects and will require further refinement of formulation, likely nanoparticles. However, miR-142-3p treatment was well tolerated, and there was no evidence of central nervous system toxicity or induced autoimmunity.
Although miR-142-3p induces preferential apoptosis in the M2 population, the peripheral monocyte counts of miR-142-3ptreated mice were not affected, probably secondary to their preexisting expression of miR-142-3p and their lack of dependency on the TGFBR1 pathway. Moreover, an oncogenic effect of miR-142-3p in human T-cell acute lymphoblastic leukemia has been reported (38); however, we did not observe any statistically significant differences in peripheral blood lymphocyte counts between the miR-142-3p-treated and control mice. Reconciliation of these results suggests that miR-142-3p plays differential roles (oncopromoter vs oncosuppressor) in different malignancies.