Resolvin D1 and D2 inhibit tumour growth and inflammation via modulating macrophage polarization

Abstract Plastic polarization of macrophage is involved in tumorigenesis. M1‐polarized macrophage mediates rapid inflammation, entity clearance and may also cause inflammation‐induced mutagenesis. M2‐polarized macrophage inhibits rapid inflammation but can promote tumour aggravation. ω‐3 long‐chain polyunsaturated fatty acid (PUFA)‐derived metabolites show a strong anti‐inflammatory effect because they can skew macrophage polarization from M1 to M2. However, their role in tumour promotive M2 macrophage is still unknown. Resolvin D1 and D2 (RvD1 and RvD2) are docosahexaenoic acid (DHA)‐derived docosanoids converted by 15‐lipoxygenase then 5‐lipoxygenase successively. We found that although dietary DHA can inhibit prostate cancer in vivo, neither DHA (10 μmol/L) nor RvD (100 nmol/L) can directly inhibit the proliferation of prostate cancer cells in vitro. Unexpectedly, in a cancer cell‐macrophage co‐culture system, both DHA and RvD significantly inhibited cancer cell proliferation. RvD1 and RvD2 inhibited tumour‐associated macrophage (TAM or M2d) polarization. Meanwhile, RvD1 and RvD2 also exhibited anti‐inflammatory effects by inhibiting LPS‐interferon (IFN)‐γ‐induced M1 polarization as well as promoting interleukin‐4 (IL‐4)‐mediated M2a polarization. These differential polarization processes were mediated, at least in part, by protein kinase A. These results suggest that regulation of macrophage polarization using RvDs may be a potential therapeutic approach in the management of prostate cancer.


| INTRODUC TI ON
Macrophage possesses multiple functions in immunomodulation and tissue repair. Plastic polarization bestows the exact role of macrophage in diverse biological process. According to the Th1-Th2 classification, macrophage polarization can be roughly divided into M1 and M2 types. 1 Classically activated M1 macrophage can be found in tissues suffering acute inflammation or be experimentally induced by lipopolysaccharide (LPS) and interferon gamma (IFN-γ).
M1 macrophage plays a pro-inflammatory role by increasing the concentrations of superoxide anions, oxygen radicals and nitrogen radicals as well as secreting inflammatory factors including interleukin 1β (IL-1β), IL-6 and tumour necrosis factor α (TNF-α). 2,3 On the one hand, M1 macrophage can execute pathogen clearance and promote tumour cell killing via activating CD8 + T cell and NK cell [4][5][6][7][8] ; on the other hand, M1 macrophage also causes inflammation and involves various pathological process such as insulin resistance and inflammation-associated mutagenesis. 9,10 Activated M2 macrophage can be divided into various subsets including M2a, M2b, M2c and M2d (aka. tumour-associated macrophage, TAM). 4 Although these M2 subsets share some markers (eg CD206 and CD163) and immunosuppressive functions, different subsets are induced by different mechanisms and have diverse physiological functions. 11,12 M2a, induced by IL-4 and/or IL-13, is the mostly studied M2 subset. Immunosuppressive M2a macrophage expresses scavenger receptors and secretes IL-10, transforming growth factor β (TGF-β) and C-C motif chemokine ligand 17 (CCL17). 4,13 The immunomodulatory functions of M2b and M2c are similar to those of M2a but have distinct inducers (immune complex for M2b and glucocorticoids for M2c). 4,14 In general, M2a, M2b and M2c tend to attenuate inflammation and are considered as physiologic inhibitors to M1 macrophage. 15 Different from above-mentioned three M2 subsets, TAM is a highly heterogeneous collection with diverse activation modes and markers among different tumour tissues. The exact inducer of TAM is still not clear and might include the combinations among vascular endothelial growth factors (VEGFs), TGF-β, IL-4, CCL2, colony-stimulating factors (CSFs) and some extracellular matrix components. Therefore, common methods to acquire certain TAM are to isolate them from target tumour tissues in vivo or from the cancer cell-macrophage co-culture system in vitro. Despite the heterogeneity, different TAMs play a similar tumour promotive role in the microenvironment via secreting immunosuppressive factors (eg IL-10), angiogenic factors (eg VEGFs) and growth factors (eg EGF). 7,16 Docosahexaenoic acid (DHA) which is a dietary ω-3 polyunsaturated fatty acid (PUFA) exhibits remarkable anti-inflammatory effect, and this effect largely attributes to its oxidation products including resolvin, maresin and protectin. 17 Resolvin D (RvD) comprises a series of lipoxygenase metabolites from DHA and in which RvD1 and RvD2 get the most attentions.  or aspirin-modified cyclooxygenase-2 (COX-2) converts DHA to 17-hydroxy DHA (17-HDHA) and then 5-lipoxygenase (ALOX5) metabolized 17-HDHA to RvD1, RvD2, RvD3 and RvD4. 18 In general, RvDs exhibit anti-inflammatory effects via modulating the activation of monocytes, macrophages, T lymphocyte and epithelial cell. 19-21 A switch of M1 to M2 polarization and consequent decrease of pro-inflammatory mediators were reported as the critical mechanism of RvDs' anti-inflammatory function. 17,19,22,23 RvD1 can induce higher   levels of reparative macrophages expressing typical M2-like marker, CD206, in heart failure mice. 24 RvDs inhibit murine abdominal aortic aneurysm formation and increase M2 macrophage polarization 17,19,22,23 and may also improve cognition in mild cognitive impairment patients. 19,23 As tumour environment tends to induce a M2-like TAM, blocking TAM polarization could be a potential anti-tumour therapeutic strategy. 25 RvD was reported to suppress tumour growth by enhancing clearance of debris via macrophage phagocytosis 26 and to reduce the number of cancer mediator-induced CD11b + Ly6G − myeloid cells. 27 In addition, some studies have revealed the anti-tumour effect of ω-3 PUFAs is ALOX5-or ALOX15-dependent. [28][29][30] In light of these literature, we wonder whether RvD can affect TAM polarization. In this study, we found that RvDs had the opposite effect on M2a and TAM (M2d), namely RvD1 and RvD2 promoted M2a but inhibited TAM polarization. This process depended on RvD induced up-regulation of PKA pathway.

| Pten-knock out mice and diets
Prostate-specific Pten knockout (Pten −/− ) mice were generated as described previously. 31 All procedures were approved by the ethics committee of Jiangnan University. The prostate tissue was weighed, embedded into paraffin, cut into 5-micron-thick sections and stained with haematoxylin and eosin (H&E). Immunohistochemical staining was performed to detect the expression of total macrophages marker. The operation was according to previous studies of our laboratory. 31,32 Diet formulas were from the custom animal diet laboratory of the Animal Resources Program at Wake Forest University (US).
According to our previous researches, 31,32 the ω-6 diet was based on a typical American diet consisting of an ω-6 to ω-3 ratio of 40:1, 397 kcal/100 g with 30% of energy from fat, 50% from carbohydrates and 20% from proteins. And the isocaloric ω-3 diet had an ω-6 to ω-3 ratio of 1:1. Scale bar: 100 μm. B, Prostate weights of two-and six-month-old Pten −/− mice. Data were shown as mean ± SEM (n = 5). Student's t test was performed. *, P < .05. C and E, PC3 and 22RV1 cell were cultured alone or co-cultured with M0 THP-1 cell and received treatments with fatty acid and RvDs for 48 h. Cell proliferation was determined by MTT assays. Results are expressed as percentages relative to the control and correspond to the means ± SD of three independent experiments. For experiments where necessary, ANOVA (Tukey's test) was performed, and P < .05 was considered significant. Statistical differences were found among groups marked with different letters. D, Diagram of the co-culture system. The pore size of inserts was 0.4 μm. F, Immunohistochemical staining of F4/80 was performed to determine macrophage infiltration. AP, DL and VP: anterior, dorsolateral, and ventral prostate. Scale bar: 100 μm. F4/80 positive area (%) was calculated with ImageJ software. Data were shown as mean ± SEM (n = 50) ***P < .001. G, Detect expression of total macrophages and M2-like TAM markers in Pten −/− prostate with western blot supplemented with 5% (v/v) foetal bovine serum (FBS, 10099141; Gibco). THP-1 was grown in RPMI 1640 medium supplemented with 5% (v/v) FBS, 200 μmol/L glutamine (Thermo Fisher Scientific) and 0.2 μmol/L β-mercaptoethanol (Sigma). L929 cells were maintained in Dulbecco's modification of Eagle's medium (DMEM) supple-

| Cell culture
Bone marrow-derived macrophage (BMDM) was isolated from mouse tibia according to Spring Harbor Protocols. 33 In brief, femur and tibia bones are collected from 6 to 8 weeks male C57BL6/J mice and bone marrow cells were flushed out using Hank's buffer.

| Macrophage polarization
Differentiation of THP-1 cells into M0, M1 or M2a macrophages was performed as previously described. 34 Briefly, THP-1 cells were differ- were purchased from Cayman chemical. For co-culture assay, transwell inserts with 0.4 μm pore size were chosen to avoid cell shuttle.

| Western blot
Proteins were separated by SDS-PAGE, transferred onto an NC membrane (45-004-001, GE Healthcare) and then probed with relevant antibodies (Table S2) at 4°C overnight. HRP-labelled goat anti-mouse or anti-rabbit IgG secondary antibodies were used for ECL detection (WBKLS0500; Sigma).

| RNA extraction and quantitative real-time PCR (qPCR)
Total RNA was isolated using TRIZOL reagent (15596018, Thermo Fisher Scientific) according to the manufacturer's protocol. Then, reverse transcription was performed using Prime Script ® RT reagent Kit with gDNA Eraser (PR047A; Takara). Real-time PCR was performed on CFX96 Real-Time System (BIO-RAD) using SYBR Green PCR master mix (4367659; Thermo Fisher Scientific). Primers used for qPCR were listed in Table S1. All experiments were performed in triplicate, and the mRNA level of GAPDH was chosen as the internal reference.
Fold changes of target genes were calculated using the 2 (-ΔΔCT) .  Table S2. Isotype-matched controls were included. Stained cells were analysed on an Attune NxT flow cytometer (Thermo Fisher Scientific).

| MTT assay
To determine cell proliferation, MTT (Thiazolyl Blue Tetrazolium Bromide) assay was performed. In brief, medium was replaced by 0.5 mg/mL MTT (M5655; Sigma) in serum-free medium. MTT solution was discarded after 4 hours incubation under normal culture conditions, and insoluble formazan was dissolved by DMSO. Values of OD570 and OD630 were recorded to evaluate cell count.

F I G U R E 2 Suppression of TAM polarization by RvD1 and RvD2
. A, Flow chart shows THP-1 priming, TAM polarization and effect stages. B, PC3 and 22RV1 were treated with TAM-conditioned medium (CM), RvD treated CM (RvDCM) and RvD plus CM for 48 h. The results are expressed as percentages relative to the control and correspond to the means ± SD of three independent experiments. C, Cell growth curves were displayed according to the results of MTT assay. PC3 and 22RV1 were treated with CM, respectively. Four groups of cells (NC, CM, RvD1CM and RvD2CM) were used for MTT assay at different time points. D, Primed THP-1 cells were treated, stained for CD206 and analysed by flow cytometry. Mean fluorescence intensities were shown. Three independent experiments were integrated into a bar chart. ANOVA (Tukey's test) was performed and P < .05 was considered as significant. Statistical differences were found among groups marked with different letters. E, Messenger RNA level of indicated genes was measured by qPCR. Results are shown as means ± SD of three independent experiments. ANOVA (Tukey's test) was performed, and P < .05 was considered as significant. Statistical differences were found among groups marked with different letters. F, ELISA was performed to determine the concentration of VEGFA and EGF in TAM supernatant. Results are shown as means ± SD of three independent experiments. ANOVA (Tukey's test) was performed and P < .05 was considered as significant. Statistical differences were found among groups marked with different letters

| ELISA
Sandwich ELISA was performed to detect cytokines in TAM supernatant. The supernatant was directly used for ELISA in a 96-well plate without dilution or freeze-thaw. All operations were performed in accordance with manufacturer's protocol. After adding stop solution, plates were read immediately using a microplate reader set to 450 and 540 nm (reference wavelength). The concentration of cytokines was calculated according to the standard curve. Three independent experiments were performed. The information of kits can be found in Table S2.

| Statistics
All data were analysed using GraphPad Prism 6 software. The results from animal experiments were shown as mean ± SEM, and the results from cultured cell were shown as mean ± SD (n ≥ 3). Two-tailed Student's t test was used for the statistical comparison of two groups. One-way analysis of variance (ANOVA); then, Tukey's test was used for multiple comparisons. P < .05 was considered significant. When using letters to show statistical differences, there was NO significant difference between groups marked with the same letter and there were significant differences between any two groups marked with different letters.

| RvD1 and RvD2 reduce the ability of macrophage to stimulate cancer cell proliferation
Prostate-specific knockout of Pten (Pten −/− ) induces prostatic intraepithelial neoplasia (PIN) and prostate cancer in mice. 35 To verify the effect of dietary DHA on mouse prostate cancer, we generated two diets with high ω-6 PUFAs (mainly provided by Safflower seed oil) and high ω-3 PUFAs (mainly provided by fish oil with 70% DHA).
Compared with ω-6 PUFA diet which induced significant extension of malignant cells, ω-3 PUFA diet attenuated the pathogenesis of prostate cancer in Pten −/− mice ( Figure 1A). And, the weight of the prostate was also reduced in mice with ω-3 PUFA diet ( Figure 1B).
Our previous researches also indicated that ω-3 PUFA diet reduced  Figure 1C). Thus, we reasoned that the metabolites of ω-3 PUFA and the involvement of tumour microenvironment may mediate ω-3 PUFA induced tumour inhibition.
ALOX5 and ALOX15 which convert DHA to resolvin Ds (RvDs) were highly expressed in prostate cancer. 36 And they were also expressed in prostate cancer from mice fed with both ω-6 and ω-3 PUFA diet ( Figure S1A). RvDs showed their anti-tumour effects against some cancer cells. 27,37,38 Unexpectedly, neither RvD1 nor RvD2 inhibited proliferation of prostate cancer cells directly ( Figure 1E).

F I G U R E 3
Promotive effect of RvD1 and RvD2 on M2a polarization. A, Primed THP-1 was treated with IL-4 and RvDs, stained for CD163, and analysed by flow cytometry. Mean fluorescence intensities were shown. Three independent experiments were integrated into a bar chart. ANOVA (Tukey's test) was performed and P < .05 was considered as significant. Statistical differences were found among groups marked with different letters. B, Messenger RNA level of indicated genes were quantified by qPCR and results are shown as means ± SD of three independent experiments. C, BMDM cells were treated with IL-4 and RvDs, stained for CD11b and CD206, and analysed by flow cytometry. D, Messenger RNA level of indicated genes were determined with and results are shown as means ± SD of three independent experiments. For experiments where necessary, ANOVA (Tukey's test) was performed and P < .05 was considered as significant. Statistical differences were found among groups marked with different letters F I G U R E 4 Inhibitory effect of RvD1 and RvD2 on M1 polarization. A, Primed THP-1 was treated with LPS + IFN-γ in the presence or absence of RvDs, stained for CCR7 and iNOS, and analysed by flow cytometry. B, Messenger RNA level of indicated genes was measured by qPCR, and results are shown as means ± SD of three independent experiments. C, BMDM was treated with LPS + IFN-γ in the presence or absence of RvDs, stained for CD11b and iNOS, and analysed by flow cytometry. D, Messenger RNA level of indicated genes was measured by qPCR, and results are shown as means ± SD of three independent experiments. For experiments where necessary, ANOVA (Tukey's test) was performed and P < .05 was considered as significant. Statistical differences were found among groups marked with different letters | 8053

SHAN et Al.
Macrophage which has a M2-like phenomenon is an important tumour promotive component in tumour microenvironment. We found that ω-3 PUFA diet strongly decreased the infiltration of total macrophage ( Figure 1F). Moreover, referred to CD68 (a total macrophage marker), CD163 expression (an M2-like macrophage marker) was also inhibited by ω-3 PUFA diet ( Figure 1G). These suggested that tumour-associated macrophage (TAM) may involve with Pten −/− tumour development. As expected, proliferation of cancer cells was significantly promoted by the co-cultured THP-1 macrophage. Interestingly, DHA and RvDs effectively reverse this process although they cannot directly inhibit tumour growth ( Figure 1C-E). These results suggest that macrophage co-cultured with cancer cell can promote the proliferation of cancer cells in which RvD1 and RvD2 can dampen the oncogenic crosstalk between cancer cells and macrophages.

| RvD1 and RvD2 suppress TAM polarization
Cancer cells generate a tumour-promoting microenvironment in which TAM is both the 'victim' (polarization stage) and the 'perpetrator' (effect stage). 39 In order to determine the mechanism of RvDs' effect, we separated these two stages. Interestingly, we also found that COX-2, a hallmark of M1 macrophage, was down-regulated in TAM and could also be partially reversed by RvD ( Figure 2E). We also noted that the supernatant of M0 THP-1 had no effect on the growth of tumour cells, and the effect of RvDs on M0 THP-1 was trifling ( Figure S1B-D). These results suggest that RvD1 and RvD2 can inhibit TAM polarization, and this provides a potential mechanism for their anti-tumour effect.

| RvD1 and RvD2 modulate macrophage polarization via PKA pathway
The results above raised a question how RvDs regulate different macrophage subsets at the same time. There are three known RvD receptors, GPR32 and ALX/FPR2 for RvD1 and GPR18 for RvD2. 50,51 These receptors are expressed on macrophage ( Figure S1E,F) and regulate cell function via cAMP/PKA F I G U R E 5 RvD1 and RvD2 modulate macrophage polarization via PKA pathway. A, Western blot was performed to check the expression and phosphorylation of AKT1, PKA-C and PKC-β in TAM, M2a and M1 macrophages. The grey values of each band were extracted by ImageJ software. Results were shown as fold changes (mean ± SD), and ANOVA was performed. P < .05 was considered as significant. Statistical differences were found among groups marked with different letters. B, PKA inhibitor H-89 was added to examine the role of PKA in polarization of TAM, M2a and M1 macrophages. Cells were stained for CD206 and iNOS then analysed by flow cytometry. Mean fluorescence intensities from three independent experiments were integrated into bar charts. ANOVA (Tukey's test) was performed, and P < .05 was considered as significant. Statistical differences were found among groups marked with different letters. C, Messenger RNA level of indicated genes was measured by qPCR, and results are shown as means ± SD of three independent experiments. ANOVA (Tukey's test) was performed, and P < .05 was considered as significant. Statistical differences were found among groups marked with different letters pathway (ALX/FPR2, GPR32 and GPR18), 52-54 PI3K/AKT pathway (ALX/FPR2 and GPR18) [55][56][57] and PKC pathway (ALX/FPR2 and GPR32). 58

| DICUSS ION
Previous studies have reported that RvDs induce a switch of M1 polarization to M2 polarization, 17,19,22,23 while their role on TAM polarization is still unclear. Our study found that RvD1 and RvD2 can regulate the polarization of M1, M2a and prostate cancer-associated TAM. Besides the co-culture system, we adopted a medium transfer model and identified that RvD1 and D2 affected macrophage directly rather than tumour cells. This finding suggests that RvD1 and D2 can exert their anti-tumour effect on multiple stages during tumorigenesis. The inhibition of M1 and promotion of M2a macrophage attenuate the mutagenic inflammation before tumour initiation and the modulation of TAM reduces the support for tumour cell proliferation.
It is generally believed that the signal pathway activated by RvDs originate from their receptors. RvD1 interacts with both GPR32 and ALX/FPR2 while RvD2 binds to GPR18, 50,51 even though, ligands of these receptors are not limited to resolvins. However, signalling pathway activated by RvD may be different from those activated by other ligands. For example, GPR32 and GPR18 are both considered as the Gi type G protein-coupled receptor, 53,54 but RvD1 and RvD2 have been repeatedly proven to up-regulate cAMP. 53,60 Here, we also found that RvD1 and RvD2 could increase both phosphorylated and total PKA in M1 and M2a macrophages. This suggests that RvDs may activate their receptors through recruiting Gs type rather than generally reported Gi type effectors.
Cyclic AMP-PKA pathway is a critical determinant in M1-M2a polarization, potentially regulating the direction of macrophage polarization. 61 In this study, we showed that both total PKA and activation (activation degree, activation site and even subcellular localization) of PKA promotes the polarization process. However, our study observed a slight but stable increase of PKA in TAM and we also found that PKA inhibitor, H-89, can reduce TAM polarization. These suggest that activation of PKA pathway is necessary, but not sufficient for TAM polarization and RvD's effect can be multi-targeted. The activity of PKA depends on not only its activation but also its subcellular localization. A-kinase anchoring proteins (AKAPs) is the controller of PKA localization. Microtubule-binding protein MAP2 which is an AKAP can be significantly up-regulated by ω-3 PUFA rather than ω-6 PUFA. 67,68 This provides another potential mechanism by which RvDs regulate MAP2 (or other AKAPs) and alter the subcellular localization of PKA so that influence the signal transduction downstream of PKA without affecting the activation of PKA itself.

CO N FLI C T O F I NTE R E S T
The authors declared that they have no conflicts of interest to this work.

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