The emerging role of heparanase in tumor initiation, growth, metastasis, and chemoresistance is well recognized, encouraging the development of heparanase inhibitors as anticancer drugs. Unlike the function of heparanase in cancer cells, little attention has been given to heparanase contributed by cells composing the tumor microenvironment. Here, we focused on the cross-talk between macrophages, chemotherapy, and heparanase and the combined effect on tumor progression. Macrophages were markedly activated by chemotherapeutics paclitaxel and cisplatin, evidenced by increased expression of proinflammatory cytokines, supporting recent studies indicating that chemotherapy may promote rather than suppress tumor regrowth and spread. Strikingly, cytokine induction by chemotherapy was not observed in macrophages isolated from heparanase-knockout mice, suggesting macrophage activation by chemotherapy is heparanase dependent. paclitaxel-treated macrophages enhanced the growth of Lewis lung carcinoma tumors that was attenuated by a CXCR2 inhibitor. Mechanistically, paclitaxel and cisplatin activated methylation of histone H3 on lysine 4 (H3K4) in wild-type but not in heparanase-knockout macrophages. Furthermore, the H3K4 presenter WDR5 functioned as a molecular determinant that mediated cytokine induction by paclitaxel. This epigenetic, heparanase-dependent host-response mechanism adds a new perspective to the tumor-promoting functions of chemotherapy, and offers new treatment modalities to optimize chemotherapeutics.
Chemotherapy-treated macrophages are activated to produce proinflammatory cytokines, which are blunted in the absence of heparanase.
Heparanase is an endo-β-D-glucuronidase capable of cleaving heparan sulfate (HS) side chains at a limited number of sites (1, 2). Heparanase activity is highly implicated in the metastatic potential of tumor-derived cells, a consequence of remodeling of the extracellular matrix (ECM) underlying epithelial and endothelial cells (2–4). Similar considerations tie heparanase activity with neovascularization, inflammation, and autoimmunity, facilitating the motility of vascular endothelial cells and activated cells of the immune system (5, 6). Compelling evidence gathered in the last two decades revealed that heparanase expression is upregulated in an increasing number of human carcinomas, sarcomas, and hematologic malignancies. Most often, heparanase induction correlated with increased tumor metastasis and shorter survival of cancer patients (6–9), thus providing strong clinical support for the protumorigenic function of the enzyme and encouraging the development of heparanase inhibitors as anticancer drugs (10, 11).
Although novel therapeutics modalities are developed and implemented successfully (i.e., immune-checkpoint inhibitors), chemotherapy is still the leading and most powerful treatment for cancer patients. Recent studies, nonetheless, suggest that chemotherapy, in addition to its cytotoxic effects on tumor cells, can support tumor regrowth and spread (12). For example, mice that had been pretreated with paclitaxel or cisplatin before intravenous injection of tumor cells succumbed to metastatic disease earlier than control mice (12–14). Similarly, fibrosarcoma cells intravenously injected into C57Bl/6 mice that had been previously treated with bleomycin developed increased pulmonary metastases (15). Importantly, this paradoxical effect of chemotherapy that emerged from preclinical research is likely to have human relevance (16).
Given the established role of heparanase in tumor metastasis, we examined whether increased metastasis by chemotherapy is due to induction of heparanase. This notion emerged from previous reports showing that heparanase expression is increased substantially in myeloma patients and cells treated with chemotherapy (17, 18).
Materials and Methods
Cells and cell culture
Human U87 glioma, MSTO-211H, and NCI-2052H mesothelioma and SGC7901 gastric carcinoma cells have been described previously (19–21). The cell lines were authenticated in June 2018 by the short tandem repeat profile of 15 loci plus amelogenin for sex determination (X or XY) method according to the manufacturer's (Promega) instructions, as described previously (19, 20). Lewis lung carcinoma (LLC) and J774 murine macrophage cells have been described previously (22). Mouse peritoneal monocytes/macrophages (MPM) were harvested from the peritoneal fluid of WT or Hpa-KO C57BL/6 mice 3 days after intraperitoneal injection of thioglycolate (3 mL; 40 mg/mL), essentially as described (22). Peritoneal exudate cells (5 × 106) were plated in 60-mm dish for 24 hours and cultured in Dulbecco's Modified Eagle's Medium supplemented with glutamine, pyruvate, antibiotics, and 10% fetal calf serum in a humidified atmosphere containing 5% CO2 at 37°C. Nonadherent cells were removed after 24 hours by washing, and the cells remaining attached were considered as macrophages (22). Macrophages were then grown for additional 24 hours in 2% FCS followed by treatment with paclitaxel, cisplatin, or doxorubicin for the times and concentrations indicated. Cells were free of Mycoplasma contamination.
Cell lysates and protein blotting
Preparation of cell lysates and protein blotting were carried out essentially as described (22).
Heparanase enzymatic activity
Preparation of ECM-coated 35-mm dishes and determination of heparanase activity were performed as described in detail elsewhere (23). Briefly, to evaluate heparanase activity in live cells, cultures of U87 glioma cells were left untreated or were treated with the indicated concentration of H1001 for 18 hours. Cells were then washed and lysed by three freeze/thaw cycles. The resulting cell extracts were incubated (18 hours, 37°C, pH 5.8) with 35S-labeled ECM, in a 35-mm dish. The incubation medium was then collected and subjected to gel filtration on a Sepharose CL-6B column. Fractions of 0.2 mL are eluted with PBS and measured for their radioactivity in a β-scintillation counter. Degradation fragments of HS side chains are eluted at 0.5 < Kav <0.8 (fractions 15–30). H1001 was similarly incubated with recombinant active heparanase (200 ng) for 5 hours at 37°C and heparanase activity was evaluated as above (22, 24).
Real-time PCR analyses
Total RNA was extracted with TRIzol (Sigma), and RNA (1 μg) was amplified using the one-step PCR amplification kit, according to the manufacturer's (ABgene) instructions. The PCR primer sets utilized in this study are listed in Supplementary Table S1. Cytokines expression was normalized to actin. Data are expressed as the mean level of expression normalized to actin, and data represent the mean ± SEM of triplicate samples; results are representative of three independent experiments (22).
Antibodies and reagents
Rat anti-mouse F4/80 antibody was purchased from Serotec; rat anti-mouse CD31 was purchased from Dianova. Anti-actin and anti-smooth muscle actin (SMA) monoclonal antibodies were purchased from Sigma. Antibodies directed against phopsho-AMPK alpha, phospho-p38 and phospho-JNK, histone H3 and H3K4-di (H3K4me2) and -tri (H3K4Me3) methylation, H3K27 trimethylation and acetylated H2BK5 were purchased from Cell Signaling Technology. Anti-WDR5, anti-BiP, and anti-Ly6g antibodies were purchased from Abcam. The selective inhibitors of LSD1 (GSK-2879552), WDR5 (OICR-9429), MKL1 (CCG-203971), CXCR2 (SB 225002), p38 (SB208530), and JNK (sp600125) were purchased from ApexBio and were dissolved in DMSO as stock solutions. DMSO (Sigma D2438) was added to the cell culture medium as control. The heparanase inhibitors H1001 and PG545 were kindly provided by HepaRx Ltd and Zucero Therapeutics, respectively. Mouse CXCL2/MIP-2 Quantikine ELISA kit was purchased from R&D Systems. The MyD88 peptide inhibitory set was purchased from Novus Biologicals. Latent heparanase was purified from medium conditioned by CHO cells overexpressing heparanase essentially as described (25) and was added to cell cultures at 1 μg/mL.
Migration assay was performed using modified Boyden chambers with polycarbonate Nucleopore membrane (Corning). Filters (6.5 mm in diameter, 8-μm pore size) were coated with fibronectin (30 μL; 10 μg/mL). WT macrophages (2 × 105) in 100 μL of serum-free medium were seeded in triplicate on the upper part of each chamber, and the lower compartment was filled with 600 μL medium conditioned by WT or KO macrophages that were not treated or treated with paclitaxel for 24 hours. After incubation for 24 hours at 37°C in a 5% CO2 incubator, noninvading cells on the upper surface of the filter were wiped with a cotton swab, and migrated cells on the lower surface of the filter were fixed, stained with 0.5% crystal violet (Sigma) and counted by examination of at least five microscopic fields, as described (26).
Tumorigenicity and IHC
Coinjection of LLC and macrophages was carried out essentially as described (22). Briefly, LLC cells were detached with trypsin/EDTA, washed with PBS, and brought to a concentration of 4 × 106 cells/mL. Control (untreated) and paclitaxel-treated macrophages were mixed with LLC cells at a ratio of 1:1, and cell suspension (8 × 105/0.1 mL) was inoculated subcutaneously at the right flank of 6- to 8-week-old WT (Envigo RMS Ltd) and Hpa-KO (in-house bred) C57Bl/6 mice. Xenograft size was determined by externally measuring tumors in two dimensions using a caliper. At the end of the experiment, mice were sacrificed, and tumors were removed and weighed. RNA was extracted from a small portion of the tumor, and the remaining portion was fixed in formalin. Paraffin-embedded 5-μm sections were subjected to immunostaining applying the indicated antibodies using the Envision kit according to the manufacturer's (Dako) instructions, as described (22). Pictures were captured with a Nikon Digital Sight camera attached to Nikon Eclipse microscope. Immunofluorescent staining was performed on methanol-fixed macrophages essentially as described (27). For Matrigel experiments, control- and paclitaxel-treated macrophages were prepared similarly, detached, and resuspended in ice-cold Matrigel (3 × 106/mL). Matrigel-cell suspension (0.5 mL) was implanted subcutaneously in WT mice (n = 6), and Matrigel plugs were excised 2 weeks later. Matrigel plugs were fixed in formalin, embedded in paraffin, and 5-μm sections were subjected to histologic evaluation and immunostaining. Matrigel was also implanted in mice without cells as control. All experiments were performed in accordance with the Technion's Institutional Animal Care and Use Committee (IL-049-03-2017; OPRR-A5026-01).
Control and paclitaxel/cisplatin-treated macrophages were subjected to flow cytometry essentially as described previously (22, 28). The antibodies utilized for flow cytometry analyses are listed in Supplementary Table S2.
Results are shown as means ± SE. GraphPad Instat software was used for statistical analysis. The differences between the control and treatment groups were determined by Student t test/one-way ANOVA, and posttest analyses were done using Dunnett/Bonferroni multiple comparison test. A value of P ≤ 0.05 was considered statistically significant. All experiments were repeated at least three times with similar results.
Heparanase is required for macrophage activation by chemotherapy
Previous reports have shown that heparanase expression is increased substantially in multiple myeloma patients and cells exposed to chemotherapy (17, 18, 29). Treatment of mesothelioma (MSTO, 2052) and gastric carcinoma (SGC7901) cells with cisplatin, doxorubicin, or paclitaxel resulted, nonetheless, in only modest 2- to 3-fold increase in heparanase expression (Supplementary Fig. S1A). Given the role of heparanase in cells of the tumor microenvironment (24), and more specifically macrophages (22), we examined heparanase expression by J774 macrophages exposed to chemotherapies and found a similar magnitude of heparanase induction (Fig. 1A, top). Unlike heparanase, MIP2 (=CXCL2, GROβ) expression was markedly induced (over 60-fold) in J774 cells exposed to cisplatin or paclitaxel (Fig. 1A, bottom), suggesting that chemotherapy activates macrophages. In order to ascertain this finding, we exposed primary peritoneal macrophages to paclitaxel and examined the expression of selected cytokines. We found that the expression of TNFα (Fig. 1B, top), MIP2 (Fig. 1B, middle), and IL10 (Fig. 1B, bottom) was increased noticeably (15-, 60-, and 10-fold, respectively) in macrophages isolated from C57BL/6 mice exposed to paclitaxel (Fig. 1B, WT). In striking contrast, paclitaxel failed to stimulate cytokine expression in peritoneal macrophages isolated from heparanase-knockout (Hpa-KO) mice (Fig. 1B, KO). However, once Hpa-KO macrophages were supplemented exogenously with recombinant heparanase and then treated with paclitaxel, cytokine induction was regained (Fig. 1C). We further examined the induction of MIP2 at the protein level by ELISA. We found that treatment of peritoneal macrophages and J774 cells with paclitaxel and cisplatin (but not doxorubicin) resulted in a marked increase in MIP2 levels (Fig. 2A, middle and bottom), comparable to MIP2 induction quantified by qPCR (Fig. 2A, top). Moreover, induction of MIP2 and TNFα by paclitaxel was attenuated by the heparanase inhibitor PG545 (Fig. 2B, PG) and even a more prominent inhibition was obtained by the small-molecule heparanase inhibitor H1001 (Fig. 2B). The latter compound (Supplementary Fig. S1B) appears unique in its ability to inhibit intracellular (Supplementary Fig. S1C) as well as extracellular (Supplementary Fig. S1D) heparanase, altogether resulting in decreased cell invasion (Supplementary Fig. S1E). In addition, H1001 appears to inhibit heparanase processing (Supplementary Fig. S1F), suggesting that this compound affects heparanase activity at different levels.
Paclitaxel-treated macrophages enhance chemoattraction
To reveal the biological consequences of macrophage activation by chemotherapeutics, we first examined the migration capacity of macrophages in a Boyden chamber apparatus. To this end, WT macrophages were plated in the upper compartment, and conditioned medium (CM) collected from control (untreated) and paclitaxel-treated WT and KO macrophages was added as a chemoattractant to the lower compartment. Clearly, medium conditioned by paclitaxel-treated macrophages was far more efficient in eliciting cell migration than medium conditioned by control, untreated, macrophages (Supplementary Fig. S2A, WT; P < 0.01). Moreover, CM isolated from Hpa-KO macrophages was less efficient in driving macrophage migration (Supplementary Fig. S2A, KO; P < 0.01). To examine this aspect in an in vivo setting, control (untreated) and paclitaxel-treated macrophages were suspended in Matrigel and implanted subcutaneously in WT C57BL/6 mice. Hematoxylin and eosin staining of Matrigel plugs collected 12 days later showed that plugs containing paclitaxel-treated macrophages attract far more cells (Fig. 2C, top), associating with increased blood vessels density (CD31; Fig. 2C, second plot). Moreover, paclitaxel-treated macrophages induced accumulation of SMA-positive cells, most likely fibroblasts, at the plug periphery (SMA; Fig. 2C, third plots) and far more macrophages evident by immunostaining and real-time PCR (F4/80; Fig. 2C, bottom; Supplementary Fig. S2B). This suggests that cytokines induced by paclitaxel attract more immune (i.e., macrophages) and nonimmune (i.e., fibroblasts) cells to the plug, and promote angiogenesis. Implantation of Matrigel devoid of macrophages failed to attract host cells (Supplementary Fig. S3A).
We next examined the polarization of untreated (control) and paclitaxel/cisplatin-treated macrophages by FACS analyses utilizing cell-surface markers typical of M1 (CD206−, CD11c+) and M2 (CD206+, CD11c−) macrophages. Untreated WT macrophages were mostly comprised of M1 cells with marginal (less than 1%) polarized M2 cells (Con; Fig. 2D, WT). Paclitaxel and cisplatin treatment resulted in a marked, 10-fold increase in the number of M2 macrophages and macrophages that showed both M1 and M2 markers, accompanied by decreased M1 type cells (WT; Fig. 2D, paclitaxel, Cis). In striking contrast, paclitaxel and cisplatin failed to polarize macrophages isolated from Hpa-KO mice (Fig. 2D, KO; Supplementary Fig. S3B), further signifying the role of heparanase in macrophages' responses to chemotherapeutic drugs.
Paclitaxel-treated macrophages enhance tumor growth
To elucidate the role of macrophage polarization by paclitaxel in the context of tumor growth, we implanted LLC cells without or with an equal number of untreated (+Con) or paclitaxel-treated (+PCT) macrophages subcutaneously, and tumor growth was inspected. This model system was preferred because it was used successfully in a previous study (22). Once implanted in WT C57BL/6 mice, paclitaxel-treated macrophages modestly promoted the growth of LLC tumors (+PCT; Supplementary Fig. S3C), yet this increase in tumor weight was statistically insignificant (P = 0.1). However, when implanted in Hpa-KO mice, the inclusion of paclitaxel-treated macrophages together with LLC cells resulted in a noticeable increase in tumor weight (Fig. 3A). Thus, although LLC alone or LLC together with control (untreated) macrophages yielded tumors with an average weight of 192 (LLC) and 218 mg (+Con; Fig. 3A), respectively, paclitaxel-treated macrophages resulted in a 4-fold increase in tumor weight (+PCT; 821 ± 79 mg, Fig. 3A), differences that were statistically highly significant (P = 0.004 for LLC vs. LLC +PCT). This increase in tumor weight was associated with a 7-fold increase in tumor VEGFA expression levels (Fig. 3B) and tumor vascularity (Fig. 3C). Furthermore, tumors produced by LLC cells and paclitaxel-treated macrophages showed increased MIP2 expression (Fig. 3D, left, +PCT), in agreement with our in vitro results (Fig. 1). Also, we found that LLC + PCT tumors show a substantial increase in the expression of Ly6g, a typical marker of neutrophils (+PCT; Fig. 3D, right). Immunostaining of tumor sections revealed that in LLC tumors, neutrophils are mainly detected at the tumor periphery (Fig. 3E, left). In contrast, the inclusion of paclitaxel-treated macrophages together with LLC cells resulted in massive recruitment of neutrophils to the tumor periphery and the center of the tumor (Fig. 3E, right).
Consistent and robust induction of MIP2 expression by paclitaxel and cisplatin in vitro and in vivo (Figs. 1B and C; 2A and B; and 3D), and its critical role in tumorigenesis (30) led us to examine the role of this cytokine in paclitaxel-enhanced tumor growth. To this end, Hpa-KO mice were implanted with LLC +paclitaxel-treated macrophages and were left untreated (+PCT) or were treated with an inhibitor of CXCR2 (SB 225002; +PCT + CXCR2 Inh; ref. 31), the high-affinity receptor of MIP2 (32). Growth of tumors produced by LLC cells inoculated together with paclitaxel-treated macrophages was increased 4-folds versus tumors produced by LLC alone or LLC+Con macrophages (Fig. 4A), in perfect agreement with the previous experiment (Fig. 3A). Notably, the growth of tumors produced by LLC +paclitaxel-treated macrophages was attenuated markedly by the CXCR2 inhibitor (Fig. 4A, +PCT + CXCR2 Inh). Recruitment of macrophages to tumors produced by LLC +PCT-treated macrophages was increased 6-fold versus tumors produced by LLC or LLC + Con macrophages (Fig. 4B), and this recruitment, as well as the recruitment of fibroblast activation protein (FAP)/SMA-positive cells, was also reduced by the CXCR2 inhibitor (Fig. 4B, top; Fig. 4C). Moreover, FACS analyses of tumors' single-cell suspensions revealed a decrease in M1 macrophages in tumors produced by LLC +PCT-treated macrophages versus LLC alone or LLC + Con macrophages (Fig. 4B, second plot), accompanied by a parallel increase of M2 macrophages (Fig. 4B, bottom). Decreased M1 and increased M2 macrophages in tumors produced by LLC +PCT-treated macrophages was reversed by the CXCR2 inhibitor (+PCT + CXCR2 Inh; Fig. 4B, second and bottom plots), altogether signifying the critical role of CXCR2 in paclitaxel-mediated macrophages attraction, polarization, and tumor growth.
Activation of macrophages by chemotherapeutics involves heparanase-mediated histone methylation
In order to reveal the molecular mechanism underlying heparanase-dependent macrophage activation by chemotherapeutics, we examined the methylation status of histones postulated previously to involve heparanase (33). We found that paclitaxel stimulates dimethylation and trimethylation of lysine 4 of histone 3 (H3K4Me2, Me3) in macrophages isolated from WT mice (WT; Fig. 5A, top and second plots). In striking contrast, no such increase in H3K4 methylation was noted in macrophages isolated from Hpa-KO mice (KO; Fig. 5A, top and second plots). Increased H3K4 methylation in WT versus Hpa-KO macrophages appeared unique because methylation of histone 3 on lysine 27 was induced by paclitaxel to comparable magnitude in macrophages isolated from WT and KO mice (H3K27Me3; Fig. 5A, third plot). Acetylation of histone H2B on lysine 5 was higher in Hpa-KO macrophages at time 0 and was modestly increased following paclitaxel treatment (Fig. 5A, fourth plots), pointing to H3K4 methylation as the histone alteration most relevant to the differential induction of cytokines in WT vs KO macrophages (Fig. 1). We further found that cisplatin elicits a comparable increase in di- and tri-H3K4 methylation in WT but not Hpa-KO macrophages (Fig. 5B) and, moreover, that increased H3K4 methylation by paclitaxel is markedly reduced by the small-molecule heparanase inhibitor H1001 (Fig. 5C). Increased H3K4 methylation by paclitaxel and cisplatin in WT but not KO macrophages was further evident by immunofluorescent staining (Fig. 5D and E). To further tie H3K4 methylation with cytokine expression, we treated cells with an inhibitor of LSD1 (GSK-2879552), an H3K4 demethylase. Indeed, H3K4 methylation was increased substantially (over 7-fold) in WT macrophages treated with GSK-2879552 (Fig. 5F), associating with marked induction of MIP2, IL6, and TNFα (Fig. 5G, 1st, 2nd and 3rd plots, respectively) expression. These results strongly suggest that induced cytokine expression by paclitaxel/cisplatin involves H3K4 methylation.
To reveal the molecular mechanism underlying the increase in heparanase-dependent H3K4 methylation, we first examined the possible involvement of Toll-like receptors (TLR). This was anticipated based on previous reports showing that paclitaxel activates macrophages via TLR (34, 35). We found that the expression of TLR2 and TLR4 is lower in macrophages isolated from Hpa-KO versus WT mice (Supplementary Fig. S4A, top and second plots, 0), in agreement with our previous report (22). We also found that TLR2 and TLR4 expression is increased by paclitaxel, but this modest increase was noted in macrophages isolated from WT and KO mice (Supplementary Fig. S4A, top and second plots). Moreover, MIP2 induction by paclitaxel was not affected in macrophages treated with inhibitor of Myd88, an adapter protein used by almost all TLRs (except TLR3) to activate the transcription factor NF-κB (Supplementary Fig. S4A, bottom), suggesting that the differential response of WT and KO macrophages to paclitaxel in terms of cytokine induction is mediated by mechanism(s) other than TLRs. H3K4 methylation is catalyzed by the highly evolutionarily conserved multiprotein complex of methyltransferases known as Set1/COMPASS or MLL/COMPASS-like complexes (36). We, therefore, examined the expression levels of Set1 and MLLs in WT and Hpa-KO macrophages treated with paclitaxel. We found that Set1A and MLL1 expression is profoundly lower in KO versus WT macrophages, yet their expression was not induced by paclitaxel (Supplementary Fig. S4B, top and second plots). Moreover, we could not detect a noticeable increase of MLL2, MLL3, MLL4, or RBBP5 in response to paclitaxel nor a differential response of WT versus KO macrophages (Supplementary Fig. S4B, third–fifth plots; Supplementary Fig. S4C). In striking contrast, the expression of WDR5, an essential component of H3K4 methyltransferase complexes (37), was highly induced by paclitaxel in WT, but not in KO macrophages (Fig. 6A, left). WDR5 expression was highly induced, nonetheless, in KO macrophages supplemented exogenously with heparanase and then treated with paclitaxel (Fig. 6A, middle and right), closely resembling the induction of cytokines in this experimental setting (Fig. 1C). We further confirmed WDR5 induction in WT but not Hpa-KO macrophages by immunoblotting (Fig. 6B). WDR5 induction by WT macrophages was also evidenced by immunofluorescent staining (Fig. 6C). Notably, induction of MIP2, TNFα, and IL6 by paclitaxel was reduced prominently in cells treated with a WDR5 inhibitor (OICR-9429; Fig. 6D; Supplementary Fig. S5A). These results point to WDR5 as the molecular determinant that mediates macrophage activation by paclitaxel and reveal WDR5 as a novel gene under heparanase regulation.
Evidence accumulating in the last two decades has critically staged heparanase at the heart of tumor progression and metastasis (6–8). This led basic researchers and biotechnology companies to develop heparanase inhibitors, some of which (i.e., PG545 = pixatimod) are being evaluated in advanced clinical trials alone, and in combination with other drugs (38). Less attention was, nonetheless, directed toward deciphering the role of heparanase in cells that constitute the tumor microenvironment and thought to play an instrumental role in tumorigenesis (39, 40).
We and others have reported previously that heparanase is expressed by macrophages and is intimately involved in cytokine gene regulation (22, 25, 41). Here, we show that activation and polarization of macrophages by chemotherapy is also heparanase-dependent, thus extending the repertoire of heparanase function in macrophages.
In agreement with earlier reports (42), we found that macrophages are stimulated by paclitaxel and cisplatin to express much higher levels of cytokines. Notably, and unlike WT macrophages, peritoneal macrophages isolated from Hpa-KO mice failed to increase cytokine expression following paclitaxel and cisplatin treatment (Fig. 1B); these macrophages, nevertheless, retained the capacity to respond to paclitaxel once heparanase is provided (Fig. 1C), or upon treatment with medium conditioned by WT macrophages exposed to paclitaxel (Supplementary Fig. S5B). Moreover, cytokines induction by paclitaxel was attenuated substantially by the heparanase inhibitor PG545 (HS mimetic; ref. 43) and even more so by the small-molecule heparanase inhibitor H1001 (Fig. 2B). This compound appears unique in its ability to inhibit endogenous heparanase. This emerged from reduced heparanase activity in cell extracts following the addition of H1001 to the cell culture medium (Supplementary Fig. S1C), thus possibly inhibiting heparanase functions that take place inside the cell (44). Even more dramatic was the ability of chemotherapy to polarize WT, but not Hpa-KO, macrophages toward M2-like phenotype (Fig. 2D), clearly indicating that heparanase is critical for macrophages polarization by chemotherapy. Consequently, paclitaxel-treated WT macrophages promoted the growth of LLC tumors. This trend was evident in WT mice (Supplementary Fig. S3C) and became highly significant in Hpa-KO mice (Figs. 3A and 4A), associating with a marked increase in VEGFA expression and tumor vascularity (Fig. 3B and C). The reason for the more extreme phenotype in the growth of tumors implanted in Hpa-KO versus WT mice is not entirely clear, but likely involves the inability of KO macrophages to populate the tumor. We have shown previously that LLC cells develop smaller tumors in Hpa-KO versus WT mice (22). Decreased tumor growth was associated with a lower number of macrophages being recruited to the tumors and, moreover, their localization. Thus, although macrophages populated the entire tumor mass in WT mice, they were arrested at the periphery of tumors developed in Hpa-KO mice (22), implying that heparanase is required for macrophage penetration into the tumors. This may suggest that host macrophages attracted to the tumor compromise the effect of paclitaxel-treated macrophages implanted together with LLC cells. In Hpa-KO mice, on the other hand, host macrophages do not populate the tumors, and the effect of paclitaxel on the implanted macrophages is not compromised. Unlike macrophages, neutrophils seem not to be dependent on heparanase for their migration into inflamed tissues (45) and massively populate the tumors produced by coinjection of LLC cells and paclitaxel-treated macrophages (Fig. 3E, right), likely promoting tumor growth (46). Increased recruitment of neutrophils was also confirmed by qPCR for Ly6g (Fig. 3D, right) and was associated with a comparable increase of MIP2 (Fig. 3D, left) that functions as a chemoattractant for neutrophils and macrophages (47). Indeed, blocking CXCR2 blunted tumor growth evoked by paclitaxel (Fig. 4A), and prevented the recruitment and polarization of macrophages toward M2 while increasing the M1 type (Fig. 4B). This points to CXCR2 as an important player in the adverse effects of chemotherapy and justifies its targeting along with paclitaxel (48). It should be noted that CXCR2 functions as a receptor to cytokines other than MIP2 so that the observed phenotypes cannot be attributed solely to MIP2. However, we did not find changes in the expression of CXCL1 or CXCL5 (Supplementary Fig. S5C and S5D) in this experimental setting, and their relevance to these results is questionable.
The molecular mechanism underlying the activation of WT, but not Hpa-KO, macrophages by paclitaxel/cisplatin appears to involve histone methylation. Clearly, dimethylation and trimethylation of H3K4 were prominently increased in WT macrophages by paclitaxel and cisplatin (Fig. 5A, B, and D; WT). In striking contrast, no such increase was found in paclitaxel/cisplatin-treated Hpa-KO macrophages (Fig. 5A, B, and E; KO). Likewise, H3K4 methylation by paclitaxel was attenuated markedly by the heparanase inhibitor H1001 (Fig. 5C), correlating with decreased MIP2 and TNFα expression (Fig. 2B). Moreover, inhibition of LSD1 that functions as H3K4 demethylase (49, 50), resulted in increased H3K4 methylation, as expected, and most importantly increased cytokine expression (Fig. 5F and G). This result evidently links H3K4 methylation and cytokine expression, in agreement with the notion that H3K4 methylation marks active transcription (49). Importantly, meta-analysis revealed that cancer patients exhibiting a lower level of H3K4 trimethylation are expected to have longer overall survival (51). This, and the finding that increased H3K4 methylation by paclitaxel/cisplatin is heparanase dependent (Fig. 5A–E), may provide another explanation for the shorter overall survival of cancer patients exhibiting high levels of heparanase (6–9), yet this possibility awaits further confirmation and clinical validation. Finally, we discovered that cytokine induction by paclitaxel involves WDR5. Unlike other components of the COMPASS complex (i.e., SET1A, MLL1-4, RBBP5, Ash2L, LEDGF; Supplementary Figs. S4B, S4C, and S5E), expression of WDR5 was induced over 10-fold by paclitaxel in WT but not in Hpa-KO macrophages. Furthermore, increased WDR5 expression was evident already 2 hours after paclitaxel addition and persisted for further 24 hours (Fig. 6A and B), likely leading to cytokine induction at later time points (Fig. 1B). Moreover, the expression of megakaryocytic leukemia 1 (MKL1) that potentiates the binding of WDR5 to specific promoter region of target genes (37) exhibited a similar expression pattern (Supplementary Fig. S6A), further supporting the role of WDR5 in cytokine induction by paclitaxel. Most importantly, inhibition of WDR5 or MKL1 practically prevents the induction of cytokines by paclitaxel (Fig. 6D; Supplementary Figs. S5A and S6B), thus supporting the critical role of WDR5 in macrophage induction by chemotherapy. In addition, overexpression of WDR5 enhanced, whereas WDR5 gene silencing reduced MIP2 expression (Supplementary Fig. S6C and S6D), thus critically linking the COMPASS complex with cytokine gene expression. The observation that SET1A, MLL1, and MKL1 expression is significantly lower in KO macrophages (Supplementary Fig. S4B and S6A) strongly implies that the COMPASS complex is impaired in the absence of heparanase, resulting in decreased H3K4 methylation (Figs. 5A and B; 7).
The mode by which heparanase regulates the expression of COMPASS genes is not entirely clear but may involve the interaction of heparanase with the promoter and transcribed regions of transcriptionally active genes as was found in Jurkat T cells (33). Alternatively, gene regulation by heparanase may result from cleavage of nuclear HS, altering the chromatin structure (52), or signal transduction potentiated by heparanase. For example, phosphorylation of JNK was increased markedly in WT macrophages treated with paclitaxel. In striking contrast, only minimal increase in JNK phosphorylation was elicited in KO macrophages (Supplementary Fig. S7A, top; Supplementary Fig. S7B). Importantly, increased H3K4 methylation and cytokine induction by paclitaxel was prevented by JNK and p38 inhibitors (Supplementary Fig. S7C and D), thus connecting paclitaxel, stress signals (JNK, p38), histone methylation, and cytokine expression. Noteworthy, the expression of BiP, indicative of ER stress, was higher in Hpa-KO versus WT macrophages (Supplementary Fig. S7A, right bottom), suggesting that paclitaxel effect is specific for the stress arm of the MAPK pathway.
Taken together, we show for the first time that activation of macrophages by chemotherapy is heparanase dependent. We further delineate the molecular mechanism underlying this novel function of heparanase to involve WDR5 induction and H3K4 methylation. This is highly significant given the key roles of WDR5 in the progression of a variety of cancers involving transcriptional activation of oncogenes, EMT-related genes, and genes involved in lymph-angiogenesis and tumor metastasis, among other protumorigenic properties (37, 53, 54). This unique mechanism provides a new perspective for the unfortunate procancer function of chemotherapy (12, 55), yet offers new treatment modalities (i.e., WDR5 inhibitors, H1001) to optimize chemotherapeutics. This approach deems promising, but it should be kept in mind that heparanase also plays a role in the recruitment of cells with antitumor activity such as NK cells (56). Thus, heparanase inhibitors should be applied in a personalized manner, where NK cells are less abundant.
Disclosure of Potential Conflicts of Interest
Y. Shaked is a consultant for OncoHost. No potential conflicts of interest were disclosed by the other authors.
Conception and design: N. Ilan, I. Vlodavsky
Development of methodology: U. Bhattacharya, N. Ilan
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): U. Bhattacharya, L. Gutter-Kapon, T. Kan, I. Boyango, U. Barash, S.-M. Yang, J. Liu, M. Gross-Cohen, R.D. Sanderson, N. Ilan, I. Vlodavsky
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): U. Bhattacharya, L. Gutter-Kapon, T. Kan, I. Boyango, U. Barash, S.-M. Yang, M. Gross-Cohen, R.D. Sanderson, Y. Shaked, N. Ilan, I. Vlodavsky
Writing, review, and/or revision of the manuscript: Y. Shaked, N. Ilan, I. Vlodavsky
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): U. Barash
Study supervision: I. Vlodavsky
We are grateful to Dr. Nili Schutz (HepaRx Ltd., Ness-Ziona, Israel) for kindly providing the heparanase inhibitor H1001. This study was generously supported by research grants awarded to R.D. Sanderson and I. Vlodavsky by the NIH (CA211752) and the United States–Israel Binational Science Foundation. It was also supported by grants from the Israel Science Foundation (grant 601/14), the ISF-NSFC joint research program (grant No. 2572/16 awarded to I. Vlodavsky and S.-m. Yang), and the Israel Cancer Research Fund (ICRF) awarded to I. Vlodavsky, and by an ERC grant (number 771112) awarded to Y. Shaked. U. Bhattacharya was supported by a postdoctoral fellowship awarded by the Israeli Council for Higher Education. I. Vlodavsky is a research professor at the ICRF.
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.