Macrophage ABHD5 Suppresses NFκB-Dependent Matrix Metalloproteinase Expression and Cancer Metastasis

Metastasis, the dissemination of cancer cells from the primary tumor to a distant organ, is a key characteristic of malignancies and a leading cause of cancer-associated death. In breast cancer, the vast majority of deaths are attributed to metastasis (1). Pathologically, cancer metastasis consists of multiple discrete steps. The genetic aberrations seen as the primary drivers of cancer initiation and metastasis account for only a subset of the necessary steps required for progression (2). The tumor microenvironment, established by tumor cells and, more importantly, by resident stromal cells (immune cells, fibroblasts, etc.), is thought to be a major contributor to most metastasis (2, 3).

In the tumor microenvironment, macrophages are causally linked to cancer metastasis (4–6). Modulation of macrophage activity can regulate the migration and invasion of cancer cells. It has been reported that macrophages can regulate cancer metastasis via direct cell–cell interactions or the secretion of various factors (6–8), events in which matrix metalloproteinases (MMP) are intensively investigated (9). For example, macrophages express a variety of MMPs, including MMP-2, 3, 8, 9, 11, 12, 13, and 19 as well as Adamts-1, 4, and 7 (10). MMP-9 plays an important role in tumorigenesis and metastasis formation by regulating tumor growth, angiogenesis, cell growth, and invasion (11–13). A mechanistic study indicated that MMP-9 is a collagenase and promotes cancer metastasis via degradation of cell adhesion molecules and removal of extracellular matrix (ECM). Adamts-1 is also reported to be an active contributor to cancer metastasis (14). Interestingly, the abovementioned MMPs are commonly regulated by transcription factors such as NFκB and AP-1 (15).

Metabolic reprogramming is the basis of macrophage activity (16–20). Transcriptional reprogramming of mitochondrial metabolism regulates the macrophage inflammatory response (16). Alpha-ketoglutarate orchestrates macrophage activation through metabolic and epigenetic reprogramming (17). In addition, triglyceride metabolism is closely correlated with macrophage activation (19). The anabolism of triglycerides relies on monoacylglycerol O-acyltransferases (MGAT) and diacylglycerol O-acyltransferases (DGAT), and the catabolism of triglycerides involves factors such as adipose triglyceride lipase (ATGL), abhydrolase domain-containing 5 (ABHD5), hormone-sensitive lipase (HSL), and monoglyceride lipase (MGLL; ref. 21). In particular, ABHD5 functions as a coactivator of ATGL, catalyzing the hydrolysis of cellular triglycerides (22). It is well established that saturated free fatty acids stimulate proinflammatory activation of macrophages through Toll-like receptor 4 (TLR4) and downstream NFκB as well as JNK pathways (18, 20). We reveal that ABHD5 deficiency in obesity-associated macrophages results in lipid accumulation, potentiates inflammasome pathway activity, and aggravates chronic inflammation (19). However, ectopic expression of ABHD5 in tumor-associated macrophages (TAM) enhances the growth of colorectal cancer via suppression of polyamine synthesis (23). Recently, we identified MGLL as a regulator of lipid accumulation in TAMs (24). MGLL deficiency greatly potentiates CB2/TLR4-dependent alternative activation of TAMs and suppression of tumor immunity (24). These studies highlighted the roles of triglyceride metabolism in macrophage activation and cancer progression.

However, the relationship between macrophage triglyceride metabolism and cancer metastasis is still obscure. In this study, we aimed to screen the cancer metastasis- and triglyceride metabolism–associated enzymes in macrophages and further decipher their roles in cancer metastasis.

MC-38 cells were purchased from JENNIO Biological Technology and maintained in our laboratory. The Raw264.7 (RAW cells), CT-26 and B-16 cell lines (B-16F10) were provided by ATCC. All cells had been authenticated and tested for Mycoplasma. The primary mouse macrophages or cell lines were maintained in regular medium (DMEM or RPMI supplemented with 10% FBS) at 37°C in a humidified 5% CO₂ atmosphere.

ABHD5-KD RAW cells were previously established with stable knockdown of murine ABHD5 expression by shRNA, and the control cells were stably transfected with PLKO vector (19, 23). The establishment of stable cell lines was verified by immunoblotting assays of target proteins.

The protocol was described previously (24). Briefly, each mouse was injected intraperitoneally with 3 mL of 3% thioglycollate (#T9032, Sigma) on day 1 and sacrificed with isoflurane on day 3. After intraperitoneal injection of 5 mL of DMEM cell culture medium supplemented with 10% FBS as well as penicillin and streptomycin, the peritoneal cells were collected in cell culture dishes. Two hours later, the floating cells were removed by washing the cells with PBS. The attached cells were taken as peritoneal macrophages (PM; purity: ∼90%) and were used in further studies.

The protocol was described previously (19). Bone marrow medium (BMM) for BM-derived macrophages (BMDM) culture was made by mixing 30 mL of conditioned medium from L929 murine fibroblasts with 70 mL of DMEM containing 10% FBS and penicillin/streptomycin. On day 1, mice were sacrificed by CO₂ inhalation and the femur and tibia were separated. Each end of the bones was cut off, and the BM was washed out with a 25-gauge needle attached to a 10 mL syringe. The BM-derived cells were centrifuged (500 × g), resuspended in BMM and plated in 6-well cell culture plates containing BMM at 4 mL/well. Fresh BMM (1 mL/well each time) was added to the cultured cells on days 3 and 5. The medium was replaced with DMEM containing 10% FBS and penicillin/streptomycin on day 6, and cells attached to the bottom of the culture dishes were taken as BMDMs.

PMs or RAW cells with different gene modifications were cultured in 250 mL flasks in regular medium. At 80% confluence, 10 mL of DMEM supplemented with 1% FBS was added to each flask and recollected 48 hours later to obtain macrophage-primed medium. CM was obtained by mixing the macrophage-primed medium with regular medium (v/v = 1:1). These conditioned media (CM) were used to treat the cultured cancer cells in vitro.

At 80% confluence, 10 mL of DMEM supplemented with 1% FBS and inhibitors of components of the inflammatory pathway, including AG490 (#S1509, Beyotime), SP600125 (#S1876, Beyotime), NAC (#S0077, Beyotime), and BAY 11-7082 (#S1523, Beyotime), were added to each flask. After incubating for 4 hours, the supernatant was replaced with 10 mL of DMEM supplemented with 1% FBS and was recollected 48 hours later. This medium was mixed with regular medium (v/v = 1:1) to obtain inhibitor-primed CM.

At 80% confluence, primary macrophages were infected with control lentivirus (GV358) or specific lentivirus overexpressing murine p65 (GV358-p65) for 6 hours and were then incubated in regular medium for 36 hours and treated with DMEM supplemented with 1% FBS for 48 hours. The supernatant was mixed with regular medium (v/v = 1:1) to generate virus-primed CM. The lentivirus (#GOSL0141471) was provided by GeneChem.

To investigate whether ABHD5-regulated cancer cell migration was MMP-dependent, we blocked MMP expression in mRNA levels with RNA interference and MMP activity in protein levels with antibodies. For RNA knockdown, the Ctrl or ABHD5-KD RAW cells were transfected with a scramble shRNA as control or a mixture of shRNAs targeting MMPs including MMP-2, 3, 8, 9, 11, 12, 13, and 19 as well as Adamts-1, 4, and 7. The shRNA sequences were provided by Sigma-Aldrich (MISSIONshRNA) and were listed in Supplementary Table S1. For protein blockade, the CM from the Ctrl or ABHD5-KD RAW cells was added with IgG (#MAB002) or a mixture of antibodies including MMP-2 (#NB200-114), MMP-3 (#MAB905), MMP-8 (#MAB908), MMP-9 (#NBP2-22181), MMP-11 (#MAB3657), MMP-12 (#MAB919), MMP-13 (#NBP2-45887), MMP-19 (#NBP2-17311), Adamts-1 (#MAB2197), Adamts-4 (#MAB4307), and Adamts-7 (#NBP2-38539). The working concentration for each antibody was 100 ng/mL. All the above antibodies were provided by Novus Biologicals. SB-3CT (#s7430, Selleckchem), the inhibitor of MMP-2 and MMP-9, was used for in vivo experiment.

To investigate whether ABHD5-regulated MMP production and cancer cell migration was IL1β-dependent, we silenced IL1β expression in mRNA levels with RNA interference. The shRNA sequences were listed in Supplementary Table S1. We blocked IL1β activity with anti-IL1β antibody (#12242, Cell Signaling Technology, 4 mg/kg) in vivo.

These mouse studies were approved by the Institutional Animal Care and Use Committee of Third Military Medical University (Chongqing, China) and were performed in accordance with relevant guidelines. All mice were housed in a specific pathogen-free environment. All the mice were housed in a pathogen-free facility with a 12-hour light, 12-hour dark cycle in TMMU. All the mice were provided with food and purified water ad libitum. Each cage contained no more than 5 mice. The transgenic mice with myeloid overexpression of ABHD5 (tg^ABHD5) were constructed previously by our group (23). The FVB/N-Tg(MMTV-PyVT)/Nju mice (#N000228), which spontaneously develop primary mammary tumors with lung metastases (25, 26), were provided by National Model Animal Resource Information Platform (Nanjing University, China). To observe the regulatory role of macrophage ABHD5 in breast cancer, the MMTV-PyVT mice were crossed with tg^ABHD5 mice.

This model was used to investigate the peritoneal metastasis of cancer cells and was intensively introduced in our previous study (27). Briefly, MC-38 or B-16 cells (5.0 × 10⁶ cells in 0.1 mL of PBS) were intraperitoneally injected into the abdominal cavity of 6-week-old male C57BL/6 mice. Then, the tumor-inoculated mice were sacrificed on days 14 or 18. The peritoneal tumor nodes were collected and weighed. This study was approved by the Institutional Animal Care and Use Committee of the Third Military Medical University (Chongqing, China) and was performed in accordance with the relevant guidelines.

The 6-week-old C57BL/6 WT or transgenic mice were intravenously injected with MC-38 or B-16 cells (5.0 × 10⁶ cells per mouse) via the tail vein. The mice were sacrificed 14 or 18 days after tumor inoculation. The spleens were observed and weighed. The lungs were dissected for pathologic observation of tumor lesions with hematoxylin and eosin (H&E) staining. This study was approved and performed in accordance with relevant guidelines of the Third Military Medical University (Chongqing, China).

At 80% confluence, uniform wounds were created in the cell culture system by using a pipette to make a wound 500-μm wide. The resulting gaps (or wounds) were washed with serum-free medium and each well was refilled with 1 mL of CM as described in Materials and Methods. Images of the scratches were captured using a fluorescence inverted microscope (Nikon 2000 E, Nikon GmbH Duesseldorf). The wound closure distance was calculated using ImageJ software. The migration rate was calculated as the ratio of the wound closure distance to the original wound width. The relative migration rate was calculated as the ratio of the migration rate of the experimental group to that of the control group (the rate of the control group was set as 1).

The migration ability of cells was assessed using Transwell chambers with polycarbonate membrane filters with 24-well inserts (6.5-mm diameter and 8-mm pore size; Corning Life Sciences). Serum-starved cancer cells (1.0 × 10⁵ cells/150 μL) were added to the top chambers in serum-free medium and allowed to migrate through to the bottom chambers, which contained different kinds of CM. After 13 hours (B-16 cells and CT-26 cells) or 18 hours (MC-38 cells), cells that migrated through the insert were stained with crystal violet and imaged (10× objective). The relative migration rate was calculated as the ratio of the number of migrated cells in the experimental group to that in the control group (the number in the control group was set as 1).

Migration-active macrophages or cancer cell migration-associated macrophages were collected via the Transwell system. As shown in the schematic of the Transwell system (Fig. 1A), three sets of culture conditions were used in this system. First, murine PMs were cultured alone in the bottom chambers of the Transwell units as the control group (Ctrl). Second, PMs were cocultured with CT-26 cells in the bottom chambers of the Transwell units as Group T (T), which represented all TAMs. Third, PMs were cocultured with CT-26 cells in the top chambers of the Transwell units. The macrophages in the bottom chamber that migrated from the top chamber were taken as group M (M). The macrophages remaining in the top chamber were taken as group (P), which contained nonmigratory TAMs. All sets of PMs were cultured in regular medium for 24 hours and were then isolated by FACS (F4/80⁺ cells) and subjected to gene microarray analysis.

Total RNAs were extracted using a kit provided by Thermo Fisher Scientific (#10296010). RNAs were transcribed into cDNAs using PrimeScript (DRR047A, Takara). qPCR was performed using a 7900HT Fast Real-Time PCR system or Applied Biosystems 7500 Real-Time PCR system (Applied Biosystems). Expression levels were normalized to β-actin. Reactions were performed in duplicate using Tli RNaseH Plus and Universal PCR Master Mix (#RR820A, TakaRa). The relative expression was calculated by the 2(^−ΔΔC_t) method (24). The primers were displayed in Supplementary Table S2.

Cell proteins were extracted using RIPA Lysis Buffer (#P0013, Beyotime) and quantified using a BCA kit (#P0009, Beyotime). Fifty micrograms of each protein sample was separated by 10% SDS-PAGE and transferred to a polyvinylidene difluoride membrane. The membranes were blocked with 5% BSA and incubated with primary antibodies for 10 hours at 4°C. The membranes were rinsed five times with PBS containing 0.1% Tween 20 and incubated for 1 hour with the appropriate horseradish peroxidase–conjugated secondary antibody at 37°C. Membranes were extensively washed with PBS containing 0.1% Tween 20 three times. The signals were stimulated with Enhanced Chemiluminescence Substrate (#NEL105001 EA, PerkinElmer) for 1 minute and detected with a Bio-Rad ChemiDoc MP System (170-8280). The primary antibodies included anti-ABHD5 (#H00051099-M01, Novus Biologicals; the dilution ratio was 1:1,000), anti-Phospho-NFκB p65 (Ser536; P-p65; #3033S, Cell Signaling Technology, the dilution ratio was 1:2,000), anti-p65 (#GTX107678, GeneTex, the dilution ratio was 1:1,000), anti-GAPDH (#2118, Cell Signaling Technology; the dilution ratio was 1:2,000), and anti-β-actin (#3700, Cell Signaling Technology; the dilution ratio was 1:2,000; ref. 24).

Primary TAMs (CD45⁺F4/80⁺ cells) from the control or tg^ABHD5 MMTV-PyVT mice were isolated and plated on the coverslips for 2 hours. Then, the coverslips were fixed in 4% ice-cold paraformaldehyde in PBS for 20 minutes, washed with PBS 3 times (5 minutes each) and incubated for 30 minutes at room temperature in a protein-blocking solution. The coverslips were incubated with the primary antibodies (anti-ABHD5, #H00051099-M01, Novus Biologicals; the dilution ratio was 1:1,000; anti-p65, #GTX107678, GeneTex, the dilution ratio was 1:1,000) at 37 °C for 1 hour and then at 4°C overnight. After washed, the coverslips were incubated at 37°C for 1 hour with fluorescent secondary antibodies (#715-606-150 or #111-545-003, Jackson ImmunoResearch). The cells were counterstained with 4′, 6-diamidino-2-phenylindole (DAPI) to reveal cell nuclei. The specificity of the primary antibody was verified by omitting that antibody in the reaction.

H&E staining was used to observe the lung metastasis in intravenous tumor models or MMTV-PyVT transgenic mice. Lung tissues were fixed in 4% buffered paraformaldehyde. The fixed tissues were embedded in paraffin, and sections were stained with after H&E and photographed (4× objective). The area of the tumor lesions was calculated using ImageJ software and the data was incorporated into Excel program for further analysis. The relative tumor lesions of each group were the ratio of the area of tumor lesions to the total area of the tissue (the number in the control group was normalized as 1).

The human tumor-associated macrophages were isolated and enriched from the patients in PLA 324 Hospital and Southwest Hospital (Chongqing, China). All the participants gave written informed consent. All tumors were primary and untreated before surgery and the specimens were anonymized. Tumor tissues were collected in compliance with the regulations approved by the Scientific Investigation Board of the hospitals. Fifty-three cases of colorectal cancer tissues were collected. Isolation of tissue macrophages were intensively described in our previous study (24). The macrophages in the human tumor tissues were assessed as CD45⁺CD68⁺ and were isolated by flowcytometry (Accuri C6, BD Biosciences). The purity of the isolated macrophages was high reaching up to 95%. The patient information was included in Supplementary Table S3. The investigator performing PCR assays was blinded to the patient information.

Statistical analyses were performed using GraphPad Prism (GraphPad Software, Inc.). All data were expressed as the means ± SEM and were analyzed using one-way ANOVA, two-tailed unpaired Student t test or Gehan–Breslow–Wilcoxon test. For each parameter of all data presented, * indicates P < 0.05, ** indicates P < 0.01, and *** indicates P < 0.005.

TAMs are important contributors to cancer metastasis. Mechanistically, TAMs can migrate ahead of premetastatic cancer cells or even migrate together with cancer cells (4, 6). However, TAMs are heterogeneous in gene expression and metabolic activity (6, 28). To crudely screen the macrophage subpopulation facilitating cancer cell migration, we established a Transwell system to collect migration-active macrophages or cancer cell migration–associated macrophages (Fig. 1A). Subsequently, migration-active macrophages were subjected to gene microarray analysis and triglyceride metabolism-associated enzymes were fully investigated (Fig. 1B). We found that in colorectal cancer cell–primed macrophages, the expression of MOGAT1, DGAT1, and ABHD5 was increased, while the mRNA levels of MOGAT2 and MGLL were decreased according to gene microarray analysis (Fig. 1C), consistent with our previous reports (23, 24). Surprisingly, the mRNA expression of ABHD5, although generally induced by cancer cells (23), was strikingly decreased in migration-active macrophages (Fig. 1B and C). These results indicated that a subgroup of macrophages with low ABHD5 expression (ABHD5^low macrophages) might be enriched in the bottom chambers of the Transwell units, because ABHD5 was indeed heterogeneously expressed in TAMs according to the single-cell sequencing results (Fig. 1D). As expected, compared with the nonmigratory macrophages in the top chambers of the Transwell units, the migratory macrophages in the bottom chambers had a notable decrease in ABHD5 mRNA levels (Fig. 1E) and an increase in cellular triglyceride levels (Fig. 1F). Notably, the migratory macrophages were resistant to induction of ABHD5 expression by cancer cells (Fig. 1G) and had higher migration activity than nonmigratory cells (Fig. 1H). Furthermore, we demonstrated that TAMs in the lung metastases had lower levels of ABHD5 mRNA (Fig. 1I) and higher levels of triglycerides (Fig. 1J) than those in primary tumor tissues in MMTV-PyVT mice. Collectively, the aforementioned findings indicated a potential correlation between ABHD5^low macrophages and cancer metastasis.

To validate the above assumption, we modified ABHD5 expression in macrophage-like RAW cells and primary PMs (Supplementary Fig. S1A and S1B). Then, we prepared CMs derived from these macrophages for further assays of cancer cell growth and migration. Consistent with the results of our previous study (29), macrophage ABHD5 potentiated the growth of MC-38 colorectal cancer cells in vitro and in vivo (Supplementary Fig. S1C and S1D). Surprisingly, we found that CM from ABHD5-KD RAW cells greatly potentiated the migration of MC-38 cells and that CM from ABHD5-transgenic mouse PMs obviously suppressed MC-38 migration in the wound-healing assays (Fig. 2A and B). The inhibitory effect of macrophage ABHD5 on cancer cell migration was also confirmed in another colorectal cancer cell line, CT-26 (Supplementary Fig. S1E), and in a melanoma cell line, B-16 (Supplementary Fig. S1F).

To further verify the inhibitory role of macrophage ABHD5 in cancer cell migration, Transwell assays were performed. The results showed that CM from ABHD5-KD macrophages markedly promoted the migration of MC-38, CT-26, and B-16 cells (Fig. 2C; Supplementary Fig. S1G and S1H). Identical conclusions were obtained when those cancer cells were treated with CM from either WT or ABHD5-transgenic PMs (Fig. 2D; Supplementary Fig. S1G and S1H).

To confirm the function of macrophage ABHD5 in cancer metastasis, peritoneal and intravenous metastasis models were employed. In the MC-38–based peritoneal carcinomatosis model (Fig. 2E), MC-38 cells were intraperitoneally injected into WT or myeloid ABHD5-transgenic (tg^ABHD5) mice. We found that colorectal peritoneal carcinomatosis was notably inhibited by myeloid transgene expression of ABHD5 (Fig. 2F). Similar results were obtained when the peritoneal carcinomatosis model was established with B-16 cells (Supplementary Fig. S2A and S2B).

In the MC-38- and B-16–based intravenous tumor metastasis models, cancer cells were intravenously inoculated into the mice via the tail vein (Fig. 2G). We demonstrated that tumor-bearing tg^ABHD5 mice had smaller spleens than WT mice (Fig. 2H; Supplementary Fig. S2C). Consistent with this result, the size of pulmonary tumor lesions was significantly decreased in tg^ABHD5 mice relative to that in WT mice (Fig. 2I; Supplementary Fig. S2D). To further validate the inhibitory effect of macrophage ABHD5 on cancer metastasis, BMDMs were collected for adoptive cell therapy. We demonstrated that silencing ABHD5 in BMDMs greatly potentiated lung metastasis in the MC-38–based intravenous metastasis model (Supplementary Fig. S2E and S2F). Consistent with this finding, CM from BMDMs from tg^ABHD5 mice obviously suppressed the migration activity of MC-38 and B-16 cells (Supplementary Fig. S2G).

To determine which factors from ABHD5-modified macrophages affect cancer cell migration, we separated the macrophage-derived CMs into the metabolite fractions (<3 kDa) and the protein fractions (>3 kDa). Via migration tests, we demonstrated that macrophage ABHD5 deficiency–induced MC-38 migration mainly depended on protein factors, thus excluding the direct effects of triglycerides or other metabolites on cancer cells (Supplementary Fig. S3A). We next investigated the interaction between colorectal cancer cells and ABHD5-modified macrophages by performing gene microarray analysis, pathway enrichment analysis, and functional verification (Fig. 3A). Gene microarray analysis of ABHD5-KD and control macrophages was carried out and revealed enrichment of downstream pathways, including the JAK–STAT pathway, the Toll-like receptor pathway (which stimulates the NFκB and JNK pathways), and migration-associated signaling (Fig. 3B and C). In addition, the gene expression profiles of colorectal cancer cells cocultured with ABHD5-KD or control macrophages were investigated. The inflammatory pathways and migration-associated signaling were significantly altered in CT-26 cells (Fig. 3D and E). Simultaneously, we demonstrated that a variety of MMPs (including MMP-2, MMP-3, MMP-8, MMP-9, MMP-11, MMP-12, MMP-13, MMP-19, Adamts-1, Adamts-4, and Adamts-7) were intensively regulated by ABHD5 silencing in RAW cells, according to the microarray assay results (Fig. 3F). We next confirmed the inhibitory effect of ABHD5 on the mRNA of those MMPs in ABHD5-KD or ABHD5-transgenic macrophages by real-time PCR (Fig. 3G and H). In addition, the inhibitory effects of ABHD5 on MMP-9 and MMP-19 proteins (as representatives of MMPs) in RAW cells, PMs and BMDMs were verified by ELISA (Supplementary Fig. S3B–S3D). We also demonstrated that TAMs in the lung metastases had lower levels of ABHD5 and higher levels of MMP-9 and MMP-19 as well as IL1β than TAMs in the primary tumors. However, there were no obvious differences in the mRNA levels of TNFα, IL6, and IL10 between those two groups (Supplementary Fig. S3E).

Finally, we simultaneously silenced the aforementioned MMPs with pooled shRNA (Supplementary Fig. S3F) and found that ABHD5-KD–stimulated colorectal cancer cell migration was prevented by MMP silencing (Fig. 3I). Similar results were obtained when those MMP proteins were simultaneously blocked with their specific antibodies (Fig. 3J). Consistent with this result, blockade of MMP-2 and MMP-9 with the inhibitor SB-3CT attenuated macrophage ABHD5 silencing–induced peritoneal metastasis of MC-38 tumors to some extent (Supplementary Fig. S3G). These results indicated that the antimetastatic effect of macrophage ABHD5 was mediated by multiple MMPs, not merely by MMP-2 and/or MMP-9.

The aforementioned findings indicated that macrophage ABHD5 inhibited MMP-dependent cancer cell migration. However, the mechanism linking ABHD5 to MMP expression was still obscure. As shown in Fig. 3C, ABHD5 inhibited a variety of inflammatory pathways that are the upstream regulators of MMPs (10, 15). We therefore blocked those pathways with specific inhibitors and further observed whether ABHD5-inhibited MMP expression could be rescued. Notably, the NFκB inhibitor largely prevented ABHD5 deficiency–induced p65 phosphorylation and MMP expression in RAW cells (Fig. 4A and B), while the inhibitors of JAK–STAT or JNK exhibited only modest effects (Supplementary Fig. S4A and S4B). We next validated that ABHD5 transgene–suppressed MMP expression was largely rescued by p65 overexpression in PMs (Fig. 4C and D).

We previously reported that macrophage ABHD5 deficiency potentiated inflammasome activity, IL1β production, and subsequent NFκB activation (19, 30). Consistent with those findings, we found here that shRNA-mediated silencing of IL1β largely attenuated the ABHD5-KD–stimulated mRNA levels of MMPs (Supplementary Fig. S4C). Functionally, blockade of IL1β with neutralizing antibodies fully prevented peritoneal metastasis of MC-38 tumors induced by deficiency of macrophage ABHD5 (Supplementary Fig. S4D).

To observe whether NFκB p65 is involved in macrophage ABHD5-regulated cancer cell migration, we modulated NFκB activity or p65 expression in ABHD5-modified macrophages before CM preparation. We demonstrated that macrophage ABHD5 deficiency–stimulated MC-38 and B-16 cell migration was prevented by inactivation of NFκB in wound-healing assays and Transwell assays (Fig. 4E and F; Supplementary Fig. S5A and S5B). Consistent with these findings, macrophage ABHD5 transgene–inhibited MC-38 and B-16 cell migration was rescued by p65 overexpression in in vitro cell migration assays (Fig. 4G and H; Supplementary Fig. S5C and S5D).

To confirm the role of the macrophage ABHD5–NFκB axis in cancer metastasis, PMs with modified ABHD5 and p65 expression were mixed with cancer cells and used to establish the in vivo metastasis models in WT mice (Fig. 5A; Supplementary Fig. S6A). In the peritoneal carcinomatosis model, PMs from WT or tg^ABHD5 mice were collected and infected with p65-overexpressing lentivirus before inoculation. We revealed that tg^ABHD5 PMs largely inhibited the development of peritoneal carcinomatosis of MC-38 and B-16 cells, and this effect was abolished by restoring p65 expression in PMs (Fig. 5B; Supplementary Fig. S6B). In MC-38 cell–based intravenous tumor metastasis model, inoculation with ABHD5-transgenic PMs decreased the spleen weight and alleviated lung metastasis, and restoring p65 expression in tg^ABHD5 PMs abolished this effect (Fig. 5C and D). Similar results were obtained in the B-16–based intravenous tumor metastasis model (Supplementary Fig. S6C and S6D). Furthermore, we collected migration-active TAMs for adoptive therapy and demonstrated that these macrophages stimulated MC-38 cell migration in an NFκB signaling–dependent manner (Supplementary Fig. S7A and S7B).

The inhibitory effect of macrophage ABHD5 on cancer metastasis was validated in cell migration models and metastatic xenograft models. We next used a genetically engineered cancer model, the MMTV-PyVT mouse (25, 26), which spontaneously develops primary mammary tumors with lung metastases, to further confirm the role of macrophage ABHD5 in cancer metastasis. We crossed tg^ABHD5 mice with MMTV-PyVT mice and found that ABHD5 transgene expression in macrophages largely suppressed pulmonary metastasis and modestly promoted the growth of the primary mammary tumors (Fig. 6A and B). In comparison with the littermate controls, tg^ABHD5 mice had an obvious decrease in MMP mRNA levels in the primary mammary cancer tissues (Fig. 6C) and TAMs (Fig. 6D). Consistent with this result, nuclear translocation of p65 (active p65) in TAMs in the primary cancer tissues was also largely suppressed in the tg^ABHD5 mice relative to that in the control group (Fig. 6E). These results indicated that ABHD5 might inactivate p65 and suppress MMP expression in TAMs, ultimately inhibiting pulmonary metastasis from primary mammary tumors in the MMTV-PyVT cancer model.

To correlate the aforementioned findings with physiopathology in the clinic, we measured the expression of ABHD5 and MMP-9 (as a representative MMP) in TAMs from primary carcinoma tissues in patients with colorectal cancer (Supplementary Table S3). We verified that the ABHD5 mRNA level in TAMs gradually declined with the development of metastasis (Fig. 7A). In contrast, the mRNA level of TAM MMP-9 dynamically increased with metastasis development (Fig. 7B). We also revealed a negative correlation between ABHD5 and MMP-9 expression in TAMs from those patients (Fig. 7C). Interestingly, higher levels of ABHD5 or lower levels of MMP-9 in TAMs of patients with colorectal cancer could predict better survival (Fig. 7D and E). Especially for patients with stage T2–4N0M0 disease, ABHD5 and MMP-9 expression in TAMs might also be valuable in predicting the disease outcome, although our analysis could not detect statistical significance due to the limited sample size (Fig. 7F and G).

The role of macrophage triglyceride metabolism in cancer metastasis has yet to be elucidated. Here, via gene microarray analysis, we demonstrated that deficiency of ABHD5, an activator of triglyceride hydrolysis in macrophages, was associated with cancer cell migration. We further verified that macrophage ABHD5 suppressed cancer cell migration in in vitro cell assays, in genetically engineered xenograft models and in genetic cancer models. Mechanistic studies have indicated that macrophage ABHD5 regulates NFκB-dependent MMP expression and cancer metastasis. This mechanism was also validated by the correlation among ABHD5, MMP-9, and clinical parameters.

This study provided a more comprehensive understanding of the role of ABHD5 in cancer progression. We were the first to investigate the role of ABHD5 in tumor biology (23, 31, 32). We found that ABHD5 was deficient in cancer tissues and identified ABHD5 as a tumor suppressor in colorectal cancer (31). Mechanistically, ABHD5 deficiency enhanced glycolysis and promoted EMT in colon cancer (31). ABHD5 also physically interacted with beclin-1 to sustain appropriate autophagy and genomic stability, thus preventing tumorigenesis (32). In addition, we found that ABHD5 was highly expressed in TAMs in the tumor microenvironment (23). We revealed that macrophage ABHD5 enhanced the growth of colorectal cancer cells by suppressing spermidine synthesis (23). Unexpectedly, we found here that ABHD5 expression was heterogeneous in TAMs and that ABHD5^low macrophages potentially facilitate tumor cell migration and cancer metastasis. It is interesting that ABHD5^low or ABHD5^high macrophages promote cancer metastasis or growth, respectively. Perhaps the ratio of ABHD5^low/ABHD5^high (or ABHD5⁻/ABHD5⁺) macrophages is low in nonmetastatic cancers and high in metastatic cancers. These findings indicated that ABHD5 in TAMs should not be used as a universal target in cancer therapy. Given the heterogeneous functions of ABHD5 in cancer cells and TAMs (23, 31), ABHD5 could be targeted in specific cells of a certain type of cancer at a certain stage. In addition, the downstream pathways of ABHD5 that we identified here could be modified for precise cancer therapy.

It should be noted that the mechanism of heterogeneous ABHD5 expression in TAMs is still obscure. Our findings revealed a macrophage subpopulation with decreased expression of ABHD5 and increased migration activity. Interestingly, expression of ABHD5 in these ABHD5^low macrophages could not be induced by colorectal cancer cells. Deciphering the ontogeny of this macrophage subpopulation might be important for the management of cancer metastasis. Loss of copy number and heterozygosity of the abhd5 gene are seen in many human cancers, including colorectal cancers (COSMIC database; ref. 31). However, the following questions remain: how is ABHD5 expression downregulated in metastasis-associated TAMs? Does downregulation occur at the epigenetic, transcriptional, and/or posttranscriptional level? Is TAM ABHD5 genetically mutated? Extensive genetic and epigenetic screening of abhd5 gene sequences at the single-cell level is warranted in future studies.

The mechanism linking ABHD5 deficiency to NFκB activation was thoroughly investigated in our previous studies. We found that ABHD5 deficiency and triglyceride accumulation stimulated ROS-dependent NLRP3 inflammasome activation and IL1β secretion in macrophages. This macrophage-derived IL1β stimulated p65 signaling in an autocrine or paracrine manner (19, 30). Consistent with these results, we found here that ABHD5-regulated MMP expression and cancer metastasis are IL1β- and p65-dependent. Given the important role of IL1β in MMP expression, we assumed that clearance of ROS or blockade of inflammasome signaling in TAMs might also be effective in preventing ABHD5 deficiency–induced MMP production and subsequent cancer metastasis.

In summary, we demonstrated that macrophage ABHD5 inhibits cancer metastasis via suppression of NFκB-dependent MMPs. Our findings shed light on preventive strategies for such metastatic malignancies.

No potential conflicts of interest were disclosed.

Conception and design: X. Ji, H. Miao

Development of methodology: S. Shang, X. Ji, R. Shi, W. Xiang, H. Miao

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Shang, X. Ji, L. Zhang, J. Chen, C. Li, Y. Chen, H. Miao

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Shang, X. Ji, L. Zhang, P. Chen, Y. Li, H. Miao

Writing, review, and/or revision of the manuscript: S. Shang. R. Dai, H. Miao

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Kang, D. Zhang, F. Yang, R. Dai, S. Chen, Y. Chen, Y. Li, H. Miao

Study supervision: Y. Li, H. Miao

This work was supported in part by award numbers 81672693, 81872028 (to H. Miao) from the National Natural Science Foundation of China (NSFC) and cstc2017jcyjBX0071 (to H. Miao) from the Foundation and Frontier Research Project of Chongqing. This work was also, in part, supported by Southwest Medical University (Sichuan, China). We thank the CapitalBio Technology (Beijing, China) for the gene microarray analysis and the relevant microarray data have been deposited in GEO under the accession codes GSE135576 and GSE135673. We also thank Springer Nature for language editing (Certificate 8F43-805A-67DE-9E80-13F0).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.