Heparanase is required for activation and function of macrophages

Lilach Gutter-Kapona, Dror Alishekevitzb, Yuval Shakedb, Jin-Ping Lic, Ami Aronheimd, Neta Ilana, and Israel Vlodavskya,1

aCancer and Vascular Biology Research Center, Bruce Rappaport Faculty of Medicine, Technion, Haifa 31096, Israel; bDepartment of Cell Biology and Cancer Science, Bruce Rappaport Faculty of Medicine, Technion, Haifa 31096, Israel; cDepartment of Medical Biochemistry and Microbiology, University of Uppsala, SE-751 05 Uppsala, Sweden; and dDepartment of Molecular Genetics, the Bruce Rappaport Faculty of Medicine, Technion, Haifa 31096, Israel

Edited by Joseph Schlessinger, Yale University School of Medicine, New Haven, CT, and approved October 17, 2016 (received for review July 13, 2016) The emerging role of heparanase in tumor initiation, growth, The carcinoma microenvironment includes nontransformed metastasis, and chemoresistance is well recognized and is encouraging epithelial cells, fibroblasts, endothelial cells, and infiltrated immune the development of heparanase inhibitors as anticancer drugs. Unlike cells. Endothelial cells lining blood and lymph vessels are major the function of heparanase in cancer cells, very little attention has component of the tumor microenvironment, and antiangiogenesis been given to heparanase contributed by cells composing the tumor therapy, targeting vascular endothelial growth factor (VEGF) or microenvironment. Here we used a genetic approach and examined its receptor (VEGFR2), is implemented clinically (15). In addi- the behavior and function of macrophages isolated from wild-type tion, recent research has revealed the critical roles of inflam- (WT) and heparanase-knockout (Hpa-KO) mice. Hpa-KO macrophages matory responses in different stages of tumor development and express lower levels of cytokines (e.g., TNFα,IL1-β) and exhibit lower metastasis (16). motility and phagocytic capacities. Intriguingly, inoculation of control The most plentiful immune cells within the tumor microenviron- monocytes together with Lewis lung carcinoma (LLC) cells into Hpa-KO ment are tumor-associated macrophages (TAMs) (16, 17). Func- mice resulted in nearly complete inhibition of tumor growth. In tionally, two distinct states have been described for macrophages: striking contrast, inoculating LLC cells together with monocytes M1 (or classically activated) and M2 (or type II, alternatively acti- isolated from Hpa-KO mice did not affect tumor growth, indicating vated). The M1 phenotype is proinflammatory and characterized by that heparanase is critically required for activation and function of the release of inflammatory cytokines (e.g., IL-1β,TNFα), reactive macrophages. Mechanistically, we describe a linear cascade by nitrogen and oxygen intermediates, and microbicidal/tumoricidal ac- which heparanase activates Erk, p38, and JNK signaling in macro- tivity. In contrast, M2 macrophages are polarized by anti-inflammatory phages, leading to increased c-Fos levels and induction of cytokine molecules (e.g., IL-4, IL-13) and support angiogenesis, tissue expression in a manner that apparently does not require heparanase remodeling, and repair (18, 19). Thus, macrophages are thought enzymatic activity. These results identify heparanase as a key medi- to play a dual role in tumor growth, initiating an immune re- ator of macrophage activation and function in tumorigenesis and sponse against transformed cells on the one hand and promoting cross-talk with the tumor microenvironment. tumor growth and angiogenesis on the other hand (20–23). Here we used a genetic approach to examine the behavior heparanase | macrophage | tumor growth | cytokine expression | and function of macrophages isolated from wild-type (WT) and knockout mice heparanase-knockout (Hpa-KO) mice (24). Hpa-KO macro- phages express lower levels of cytokines (e.g,, TNFα,IL-1β) eparanase is an endo-β-glucuronidase that cleaves heparan previously shown to be induced by the addition of heparanase or1its overexpression (25), and appear less motile. Inoculating control Hsulfate (HS) side chains presumably at sites of low sulfation. + Traditionally, heparanase activity was correlated with the meta- monocytes (CD11b ) together with Lewis lung carcinoma (LLC) static potential of tumor-derivedcells,attributedtoenhanced cells into Hpa-KO mice resulted in nearly complete inhibition of cell dissemination as a consequence of HS cleavage and remodeling of the extracellular matrix (ECM) barrier (1, 2). Significance Intensive research efforts over the last decade have revealed that heparanase expression is up-regulated in an increasing The tumor microenvironment is now considered to play a major number of human carcinomas and hematologic malignancies. role in cancer growth and metastasis. Heparanase is the only In many cases, heparanase induction correlates with increased enzyme in mammals capable of cleaving heparan sulfate, an ac- tumor metastasis, vascular density, and shorter postoperative tivity that is highly implicated in tumor growth, metastasis, and survival of cancer patients (3–6), providing strong clinical inflammation. Here we provide evidence that heparanase is crit- support for the protumorigenic function of the enzyme and ically required for the activation and function of macrophages, an encouraging the development of heparanase inhibitors as anti- important constituent of the tumor microenvironment. Mecha- cancer drugs (3, 7, 8). More recent studies have provided com- nistically, we describe a linear cascade by which heparanase ac- pelling evidence associating heparanase level with all stages of tivates Erk, p38, and JNK signalinginmacrophages,leadingto tumor formation, including tumor initiation, growth, metastasis, increased c-Fos levels and induction of cytokine expression in a and chemoresistance (9–14). manner that apparently does not require heparanase enzymatic Although heparanase up-regulation by tumor cells is well activity. These results identify heparanase as a key mediator of documented, the protumorigenic contribution of heparanase pro- macrophage activation and function in tumorigenesis and cross- vided by cells composing the tumor microenvironment has not talk with the tumor microenvironment. been sufficiently explored. We recently reported that heparanase- neutralizing polyclonal and monoclonal antibodies attenuated the Author contributions: N.I. and I.V. designed research; L.G.-K., D.A., Y.S., and N.I. per- formed research; J.-P.L. and A.A. contributed new reagents/analytic tools; L.G.-K., D.A., growth of myeloma and lymphoma cells within bones (14). No- Y.S., J.-P.L., A.A., N.I., and I.V. analyzed data; and N.I. and I.V. wrote the paper. tably, the neutralizing antibodies also attenuated the growth of The authors declare no conflict of interest. Raji lymphoma cells, which do not express heparanase owing to This article is a PNAS Direct Submission. methylation of the gene, implying that neutralization of heparanase 1To whom correspondence should be addressed. Email: [email protected]. contributed by the tumor microenvironment is sufficient to restrain This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. tumor growth (14). 1073/pnas.1611380113/-/DCSupplemental.

E7808–E7817 | PNAS | Published online November 14, 2016 www.pnas.org/cgi/doi/10.1073/pnas.1611380113 Downloaded by guest on September 28, 2021 tumor growth. In striking contrast, inoculating LLC cells to- Results PNAS PLUS gether with monocytes isolated from Hpa-KO mice did not affect Reduced Cytokine Expression by Hpa-KO Macrophages. We have tumor growth, suggesting that heparanase is required for the previously shown that the exogenous addition or overexpression proper activation and function of macrophages. of heparanase activates macrophages and stimulates cytokine

Fig. 1. Reduced cytokine expression and motility of heparanase-deficient macrophages. (A) Heparanase activity. Cell exudates were collected from the peritoneum of control (WT) and Hpa-KO mice at 3 d after thioglycolate administration. (Left) After washing, cells (2 × 106) were lysed by three freeze/thaw cycles, and cell lysates were applied on sulfate-labeled ECM dishes. Determination of heparanase activity was carried out as described in Materials and Methods.(Right) Heparanase activity was evaluated similarly in freshly isolated WT macrophages (Mac) and an equal number (2 × 106) of LLC cells. (B) Cytokine expression. Cell exudates were collected from the peritoneum of control (WT) and Hpa-KO mice at 3 d after thioglycolate administration. Cells were plated on tissue culture dishes, and nonadherent cells were removed after 24 h. Total RNA was extracted from the adherent macrophages, and corresponding cDNAs were subjected to quantitative real-time PCR analyses using a set of primers specific for the indicated cytokines. Cytokine expression in Hpa-KO macrophages is shown graphically in relation to the level in control macrophages set arbitrarily to a value of 1. Note that the expression level of most cytokines is reduced in Hpa-KO macrophages. *P < 0.04. (C) WT C57BL/6 mice were administrated with the indicated anti–heparanase-neutralizing antibody (250 μg/mouse) or control rabbit IgG (Control) 30 min before the administration of thioglycolate. Cell exudate was collected 3 d later, and cytokine expression was evaluated as above. Note the reduced cytokine expression by peritoneal macrophages following treatment with the heparanase- neutralizing antibodies. *P < 0.02. (D) Heparanase small-molecule inhibitor. Cell exudate collected from WT mice was plated on tissue culture dishes for 24 h. The dishes were then washed, and macrophages were incubated under serum-free conditions without (0) or with latent heparanase (1 μg/mL) in the absence (Hepa) or presence of OGT2115 small-molecule heparanase inhibitor (10 μg/mL). Total RNA was extracted after 6 h, and expression of the indicated cytokines was evaluated by real-time PCR. Note that cytokine induction by heparanase is not significantly affected by OGT2115. (E and F) Cell motility. Cell exudate collected from the peritoneum of thioglycolate- treated control (WT) and Hpa-KO mice was plated on inserts coated with fibronectin (cell migration, 3 h; E) or Matrigel (cell invasion, 48 h; F). After washing, macrophages were maintained in serum-free medium, and chemoattraction was initiatedbyaddingmediumsupplementedwith10%FCStothelowercompartment.(Upper) Quantification of cell migration (E) and invasion (F). (Lower) Representative images of migrating (E) and invading (F) cells. (G) Quantification of cells collected from the peritoneum. Thioglycolate was administrated to control (WT; n = 15) and Hpa-KO (n = 14) mice, and cell exudate was collected from the peritoneum 3 d later. Red blood cells were removed, and remaining cells were counted. The number of Hpa-KO cells is presented graphically as the percentage of cells collected from control mice. MEDICAL SCIENCES

Gutter-Kapon et al. PNAS | Published online November 14, 2016 | E7809 Downloaded by guest on September 28, 2021 expression (25). Here we examined the expression profile of phages isolated from mice treated with heparanase-neutralizing selected cytokines in macrophages isolated from WT mice and antibodies (#1453, 1023; Fig. 1C) (14). Notably, Hpa-KO mac- Hpa-KO mice. We first established that WT macrophages rophages respond to the exogenous addition of heparanase, and exhibited typically high levels of heparanase activity (Fig. 1A, cytokine expression is increased to levels comparable to those in Left), comparable with or even greater than the activity in tumor- WT macrophages stimulated with heparanase (Table S1). derived cells (Fig. 1A, Right), whereas Hpa-KO macrophages Cytokine induction does not seem to require heparanase en- lacked such activity (Fig. 1A, Left). We also found that the ex- zymatic activity, because cytokines were induced to comparable pression of most cytokines examined was reduced significantly in levels also in the presence of a small-molecule heparanase in- Hpa-KO macrophages (Fig. 1B). Using a cytokine antibody ar- hibitor, OGT2115 (Fig. 1D). Moreover, the migration (Fig. 1E), ray, we found corresponding reduced cytokine levels (i.e., MIP-2, invasion (Fig. 1F), and phagocytic (Fig. S1B) capacities of Hpa- TNFα, CXCL1, BLC) in medium conditioned by Hpa-KO vs. KO macrophages were decreased compared with WT macrophages. control macrophages (Fig. S1A), in agreement with the PCR Indeed, fewer monocytes were collected from the peritoneum of analyses. Similarly, cytokine expression was reduced in macro- Hpa-KO mice after treatment with thioglycolate (Fig. 1G). These

Fig. 2. Host heparanase affects tumor growth. (A–E) Tumor growth in WT vs. Hpa-KO mice. LLC cells (0.5 × 106) were implanted s.c. in control (WT) or Hpa- KO mice (n = 7). At termination, tumors were excised and weighed (A), and single-cell suspension were prepared and subjected to FACS analyses. Graphical representations of macrophages, T cells, and neutrophils in WT and Hpa-KO tumors are shown in B, C, and D, respectively. For BM transplantation (E), WT mice were lethally irradiated, and the BM was replaced with BM cells (5 × 106) collected from WT (WT-WT; n = 6) or Hpa-KO (KO-WT; n = 6) tibias. LLC cells were implanted s.c. 6 wk later, and tumor weight was examined at termination. Note the modest yet statistically significant decrease in tumor weight inthe absence of heparanase in BM-derived cells. (F–H) MIP2 overexpression. LLC cells were transfected with MIP2 or an empty vector (Vo), and stably transfected cells were implanted s.c. in WT (n = 6) and Hpa-KO (KO; n = 6) mice. At termination, tumors were excised, weighed (F, Upper), photographed (F, Lower), and fixed in formalin for histological evaluation. Portions of each tumor were taken for RNA extraction and for the preparation of single-cell suspensions and FACS analyses. The number of tumor-resident F4/80-positive macrophages detected by FACS and quantitative real-time PCR analyses for F4/80 and lysozyme 1 are shown graphically in G, Upper and Lower, respectively). (H) Sections of tumor xenografts produced by LLC cells overexpressing MIP2 and implanted in WT or Hpa-KO (KO) mice were subjected to immunostaining applying rat anti-mouse F4/80 antibody. Shown are representative photomicrographs taken at low (4×) magnification (Upper) and high (10×) magnification at the tumor periphery (Middle) and center (Lower). Note that MIP2 overexpression restrains tumor growth in WT mice but not in Hpa-KO mice, associated with a robust increase in macrophage activation, as evidenced by lysozyme 1 expression in WT mice vs. lower induction in Hpa-KO mice. In Hpa-KO mice, macrophages reside mainly in the tumor periphery and fail to populate the tumor mass.

E7810 | www.pnas.org/cgi/doi/10.1073/pnas.1611380113 Gutter-Kapon et al. Downloaded by guest on September 28, 2021 PNAS PLUS

Fig. 3. Tumor elimination by WT, but not heparanase-deficient, monocytes. (A) FACS analyses. Cell exudate, collected from the peritoneum of WT (Upper) and Hpa-KO (Lower) mice 3 d after thioglycolate injection, was subjected to FACS analyses using antibodies directed against CD11b (monocytes), F4/80 + + (macrophages), and CD8-a (CD8 T cells). Note that the cell exudate comprised primarily monocytes (CD11b ), but no CD8 T cells. (B–D) Coinjection of LLC and monocytes. LLC cells (3.5 × 105) were implanted in Hpa-KO mice without (LLC; n = 9) or with an equal number of monocytes collected from thioglycolate- treated WT (+Con; n = 6) or Hpa-KO mice (+KO; n = 7). At termination, tumors were excised, weighed (B, Upper), and photographed (B, Lower). Tumor portions were taken for RNA extraction, and the remaining tumors were fixed in paraffin for histological evaluation. (C) Quantitative real-time PCR analyses for macrophages number (F4/80) and activity (i.e., lysozyme 1 and 2) in LLC+Con and LLC+KO tumors shown graphically in relation to LLC-alone tumors set arbitrarily to a value of 1. (D) Real-time PCR analyses for (from top to bottom) the number of CD8 T cells, NK cell number (NK1.1) and activity (granzyme B, perforin), dendritic cell number (Langerin; DC-SIGN), and neutrophils (Ly6G) in LLC alone, with control or KO macrophages. Note the marked attenuation of tumor growth by coinjection of LLC cells with control monocytes, associated with a robust activation of macrophages, T cells, and NK cells. In contrast, coinjection of LLC with KO monocytes failed to activate immune cells and affect tumor growth.

results indicate that heparanase is expressed by macrophages and 0.15 mm vs. 1.3 ± 0.2 mm; P = 0.03) (Fig. 2A). FACS analyses plays a prominent role in their activation, as evidenced by reduced showed that decreased tumor growth was associated with lower cytokine expression, cell motility, and phagocytosis in Hpa-KO numbers of macrophages (P = 0.05; Fig. 2B and Fig. S1C) and T macrophages. cells (P = 0.01) (Fig. 2C and Fig. S1C) in tumors that developed in Hpa-KO mice compared with those in WT mice. In contrast, Heparanase from the Tumor Microenvironment Supports Tumor Growth. To reveal the significance of heparanase contributed the number of neutrophils was not significantly altered (Fig. 2D by the tumor microenvironment for tumor growth, we implanted and Fig. S1C). LLC cells s.c. in WT and Hpa-KO mice and examined tumor To examine the role of bone marrow (BM)-derived cells in the growth. The tumors that developed in the Hpa-KO mice were reduced tumor growth in Hpa-KO mice, WT mice were lethally twofold smaller than those seen in the WT mice (mean, 0.47 ± irradiated, the BM was substituted with BM cells collected from MEDICAL SCIENCES

Gutter-Kapon et al. PNAS | Published online November 14, 2016 | E7811 Downloaded by guest on September 28, 2021 WT (WT-WT) or Hpa-KO (KO-WT) mice, and, following re- in tumor growth (Fig. 3B). In striking contrast, implantation of covery, LLC cells were implanted. Notably, tumor growth was Hpa-KO monocytes together with LLC cells had no effect on tu- reduced in KO-WT compared with WT-WT mice (P = 0.04) mor growth (Fig. 3B). The remarkable decrease in tumor growth by (Fig. 2E), suggesting that the decreased tumor growth in Hpa- the introduction of control monocytes (LLC+Con) was associated KO mice is due in part to the lack of heparanase in BM-derived with a marked activation of macrophages (Fig. 3C; lysozymes 1 and cells that constitute the tumor microenvironment. 2) and natural killer (NK) cells (Fig. 3D; granzyme B, perforine), as well as a significant increase in the recruitment of CD8 T cells and Macrophages Are Trapped at the Periphery of Tumors Developed in dendritic cells (Fig. 3D), which together likely created a strong Hpa-KO Mice. We noted that among the cytokines examined, the antitumor effect, halting tumor growth. expression of macrophage inflammatory protein 2-alpha (MIP-2 = None of these parameters was affected by Hpa-KO macro- CXCL2) was prominently reduced in Hpa-KO macrophages phages (Fig. 3 C and D). Unlike these cell populations, the (Fig. 1B and Fig. S1A). This cytokine is highly implicated in number of neutrophils in the tumor xenografts was not altered macrophage attraction (26), possibly connected to reduced cell significantly by the introduction of monocytes (Fig. 3D, Lower), motility (Fig. 1G) and lower numbers of macrophages in Hpa- as was noted in the MIP-2 overexpression model system (Fig. S3, KO tumors (Fig. 2B). To examine this aspect in the context of Lower). The cytokines most evidently induced in the LLC+Con tumor growth, we transfected LLC cells with MIP-2 gene con- tumors were TNFα and SDF1 (Fig. 4A), possibly suggesting that struct and confirmed a high level of expression by real-time PCR these cytokines mediate the killing effect and elimination of LLC (Fig. S2A). Interestingly, after implantation in WT mice, MIP-2 tumors. Indeed, we found that heparanase (Fig. 4B, Left), as overexpression resulted in a marked decrease in tumor weight well as TNFα, MIP2, and SDF1 (Fig. 4B, Right), increased the (P = 0.004) (Fig. 2F). In striking contrast, tumor weight was not phagocytic capacity of macrophages, thereby facilitating antigen affected by MIP-2 overexpression once cells were implanted in presentation and enhancing antitumor immune responses. Hpa-KO mice (Fig. 2F). FACS (Fig. 2G, Upper and Fig. S1D) Immunostaining also revealed alterations in F4/80-positive mac- and real-time PCR (Fig. 2G, Lower) analyses revealed that the rophage localization in this experimental setting. Whereas mac- number of macrophages in tumors developed by LLC–MIP-2 rophages were localized primarily in the periphery of LLC cells was increased by twofold to threefold compared with con- tumors (Fig. 4C), in agreement with our previous results (Fig. trol cells (LLC-Vo; Fig. 2G), as expected, but appeared to be 2H), the introduction of control, but not Hpa-KO macrophages, similar in magnitude in WT and Hpa-KO mice (Fig. 2G, Upper). reduced the accumulation of macrophages in the tumor pe- Likewise, greater numbers of the classic M1/M2-type macro- riphery and macrophages appeared to populate the entire tumor phages (Fig. S2 B and C), CD4 cells, and CD8 T cells (Fig. S3, mass (Fig. 4C). Notably, when the same experimental procedure Upper and Middle) were quantified following MIP-2 over- was repeated in WT mice, the introduced monocytes had no expression, but again appeared to be comparable in WT and effect on tumor growth (Fig. 4D), strongly implying that tumor Hpa-KO mice, whereas the number of neutrophils was not al- development and elimination critically depend on heparanase tered (Fig. S3, Lower). contributed by the host. More specifically, the differentiation of + In striking contrast, macrophage cytotoxic activity, evidenced CD11b monocytes to kill-type macrophages and the elimination by lysozyme 1 expression (27), was dramatically increased in WT of LLC tumors apparently require heparanase. compared with Hpa-KO macrophages (Fig. 2G, Lower). Thus, whereas MIP-2 overexpression resulted in a >24-fold increase in Cytokine Induction by Heparanase Involves the p38 and JNK Signaling lysozyme 1 levels in tumors developed by LLC–MIP-2 cells vs. Pathways. To further reveal the molecular mechanism underlying control (LLC-Vo) cells in WT mice, only a 7.7-fold increase was cytokine induction by heparanase, we isolated macrophages from induced by the same cells in tumors developed in Hpa-KO mice WT and Hpa-KO mice and exposed them to heparanase added (Fig. 2G, Lower Right; lysozyme 1), a highly statistically signifi- exogenously. As reported previously (25), the addition of heparanase cant difference (P = 0.004). This results suggests that in the stimulated the phosphorylation of Erk in WT macrophages, and absence of heparanase, macrophages are not properly activated even greater increases were seen in Hpa-KO macrophages (Fig. 5A). by MIP-2. Notably, however, the addition of heparanase together with the in- Immunostaining further revealed that in tumors produced hibitor of this signaling pathway resulted in only a modest (i.e., by LLC–MIP-2 cells in WT mice, macrophages were detected twofold) decrease in cytokine expression by heparanase, with the throughout the tumor mass (Fig. 2H, Left). In striking contrast, exception of IL-1β expression, which was reduced more significantly macrophages were detected primarily at the periphery of tumors (Fig. 5B), suggesting that cytokine induction by heparanase is regu- developed in Hpa-KO mice (Fig. 2H, Right). These results in- lated by other signaling pathways. Indeed, we found that heparanase dicate that in the absence of heparanase, macrophages are not enhances the phosphorylation levels of p38 (Fig. 5C and Fig. S4A) fully activated (lysozyme 1) and do not populate the tumors, in and JNK (Fig. 5C and Fig. S4B), as is also evident on immunoflu- accordance with their inability to attenuate tumor growth. orescent staining (Fig. S4C), and exhibits dose-dependency (Fig. S4D). Notably, induction of all of the examined cytokines except IL- Hpa-KO Macrophages Do Not Attenuate Tumor Growth. Given that 1β by heparanase was significantly attenuated by JNK inhibition (Fig. Hpa-KO macrophages failed to penetrate LLC–MIP-2 tumors 5D), whereas inhibition of p38 attenuated the induction of IL-1β,but (Fig. 2H), we decided to implant freshly isolated monocytes/ not of the other cytokines (Fig. 5E). This suggests that Erk, p38, and macrophages together with LLC cells, thereby populating the JNK signaling each regulates the induction of a different set of cy- tumor with macrophages. To this end, cell exudates collected from tokines by heparanase. the peritoneum of WT mice (Fig. 3A, Upper) and Hpa-KO mice Given the numerous cytokines regulated by heparanase (Fig. 1 (Fig. 3A, Lower) after thioglycolate treatment were subjected to A–C), we sought a common transcription factor that mediates + FACS analyses. These exudates contained >90% CD11b cells cytokine gene regulation. Applying nuclear extracts of WT and (i.e., monocytes) (Fig. 3A, Left) (28), but no CD8 T cells (Fig. Hpa-KO macrophages on a transcription factors array revealed 3A, Right). that the DNA-binding capacity of several transcription factors We then mixed WT and Hpa-KO monocytes with LLC cells in was decreased in Hpa-KO macrophages compared with WT a 1:1 ratio and implanted these cells in Hpa-KO mice. LLC cells macrophages, whereas that of other transcription factors was implanted without monocytes served as a control for the effect of increased (Table S2). At the transcriptional level, we could only the introduced monocytes. Implantation of WT monocytes to- validate decreased expression of the AP1 (c-Fos) transcription gether with LLC cells resulted in a prominent (10-fold) decrease factor in Hpa-KO macrophages. Notably, the expression of

E7812 | www.pnas.org/cgi/doi/10.1073/pnas.1611380113 Gutter-Kapon et al. Downloaded by guest on September 28, 2021 PNAS PLUS

Fig. 4. The cytokines TNFα and SDF1 prevail in LLC+Con small tumors. (A) Cytokine expression. Total RNA was extracted from LLC, LLC+Con, and LLC+KO tumors, and corresponding cDNAs were subjected to quantitative real-time PCR analyses using primer sets specific for the indicted cytokines. Expression levels of the cy- tokines in LLC+Con and LLC+KO tumors is shown graphically in relation to their levels in LLC tumors, set arbitrarily to a value of 1. *P < 0.02. (B) Phagocytosis. Cell exudates were collected from the peritoneum of control WT mice, plated on fibronectin-coated 96-well dishes (2 × 104) for 24 h. After washing, zymosan-coated fluorogenic bioparticles (5 μL)wereaddedtoeachwell.Macrophage phagocytosis capacity was quantified in the absence (Con) or presence of heparanase (Hepa; 1 μg/mL), TNF-α (20 ng/mL), MIP-2 (20 ng/mL), or SDF1 (50 ng/mL). (C) Immunostaining. Here 5-μm sections of tumor xenografts produced by LLC, LLC+Con, and LLC+ KO cells were subjected to immunostaining, applying anti-F4/80 antibody. Note that unlike in LLC and LLC+KO tumors, in LLC+Con tumors macrophages populate the entire tumor lesion, correlating with a marked decrease in tumor growth (Fig. 3B). (D) LLC cells (3.5 × 105) were added to an equal number of monocytes collected from the peritoneum of thioglycolate-treated control (WT) or Hpa-KO mice, and cells were implanted in C57BL/6 WT mice. Shown is tumor weightat termination. Note that after implantation in WT mice, the introduction of monocytes has no effect on tumor growth.

c-Fos, but not of c-Jun, was decreased in Hpa-KO macrophages Notably, c-Fos expression was increased in parallel with the in- (Fig. 5F, Left). Likewise, c-Fos expression was increased after the duction of cytokine expression when LLC cells were inoculated exogenous addition of heparanase to Hpa-KO macrophages (Fig. together with control macrophages (Fig. 5H). These results 5F, Right), suggesting that c-Fos is the AP1 transcription factor suggest a linear cascade by which heparanase activates Erk, p38, relevant to the observed cytokine induction by heparanase. and JNK signaling, leading to increased c-Fos levels and induc- We used a c-Fos promoter-luciferase reporter gene and found tion of cytokine expression. induction of luciferase activity after cotransfection of this re- porter and heparanase gene constructs (Fig. S4E). Moreover, we Discussion found that c-Fos levels were increased after the addition of The role of heparanase in tumor initiation, growth, metastasis, heparanase to WT macrophages, but this elevation was only and chemoresistance is emerging, and is encouraging the de- modestly attenuated by the MEK inhibitor (Fig. S4F). In con- velopment of heparanase inhibitors as anticancer drugs (3, 8). trast, c-Fos induction by heparanase was markedly attenuated by Unlike the function of heparanase in cancer cells, very little at- the JNK inhibitor, and appeared to be most prominent when the tention has been given to heparanase contributed by cells com- JNK and p38 inhibitors were combined (P = 0.04) (Fig. 5G). posing the tumor microenvironment. We recently reported that MEDICAL SCIENCES

Gutter-Kapon et al. PNAS | Published online November 14, 2016 | E7813 Downloaded by guest on September 28, 2021 Fig. 5. Cytokine induction by heparanase is mediated by the Erk, p38, and JNK signaling pathways. (A and C) Immunoblotting. Cell exudates were collected from WT and Hpa-KO (KO) mice at 3 d after thioglycolate administration and plated on tissue culture dishes. After 24 h, the dishes were washed, and ad- hering macrophages were incubated under serum-free conditions for 24 h. Heparanase (5 μg/mL) was then added, and cell lysates were prepared at the indicated time points. Lysate samples were subjected to immunoblotting, applying anti–phospho-ERk (A, Upper), phospho-p38 (p-p38; C, Upper), phospho- JNK (pJNK; C, Middle), and anti-actin (C, Lower) antibodies. (B, D, and E) MEK, JNK, and p38 inhibitors. Heparanase (5 μg/mL) was added to adhering Hpa-KO macrophages alone (Hepa) or 30 min after the addition of MEK (B; PD98059; 30 μg/mL), JNK (D; sp600125; 20 μg/mL), or p38 (E; SB203580; 20 μg/mL) inhibitors. Total RNA was extracted after 6 h, and corresponding cDNAs were subjected to quantitative real-time PCR analyses, applying primer sets specific for the indicated cytokines. Cytokine expression is shown graphically in relation to the levels in control untreated macrophages (Con), set arbitrarily to a value of 1. *P < 0.01. (F) Heparanase enhances Fos expression. (Left) Total RNA was extracted from adhering macrophages isolated from thioglycolate-treated WT (Con) and Hpa-KO (KO), mice and corresponding cDNAs were subjected to quantitative real-time PCR analyses applying primer sets specific for c-Fos and c-Jun. Note the decreased c-Fos, but not c-Jun, expression in Hpa-KO macrophages. (Right) C-Fos and c-Jun expression were similarly examined in adhering macrophages following addition of heparanase (5 μg/mL) for 30 min in relation to control untreated macrophages (Con), set arbitrarily to a value of 1. Note that only c-Fos expression was induced by exogenous heparanase. *P = 0.02. (G) c-Fos induction by heparanase involves the Erk, JNK, and p38 signaling pathways. Hep- aranase (5 μg/mL) was added to adhering Hpa-KO macrophages alone (Hepa) or 30 min after the addition of JNK (sp600125), p38 (SB), or both JNK and p38 (Hepa+sp600+SB) inhibitors. Macrophages were also incubated with the inhibitors each alone and in combination without heparanase. Total RNA was extracted after 30 min, and corresponding cDNAs were subjected to quantitative real-time PCR analyses applying primer sets specific for c-Fos. c-Fosex- pression is shown graphically in relation to its level in control untreated macrophages (Con), set arbitrarily to a value of 1. Note that c-Fos induction by heparanase is prevented by the JNK inhibitor and to a lesser extent by the p38 inhibitor. (H) Quantitative real-time PCR analysis of Fos expression in tumors generated by LLC, LLC+Con, and LLC+KO cells. Note that cytokine induction in LLC+Con tumors (Fig. 4A) correlates with increased c-Fos expression.

heparanase-neutralizing monoclonal antibodies attenuate the control mice (Fig. S5A). Reduced LLC tumor growth after re- growth of human lymphoma cells by targeting heparanase ac- placement of WT BM with Hpa-KO BM cells (Fig. 2E)suggeststhat tivity contributed by cells of the tumor microenvironment (14), the attenuation of tumor growth in Hpa-KO mice is mediated in but the nature of these cells has not been characterized. Once part by BM-derived cells that populate the tumor microenvironment. implanted in Hpa-KO mice, LLC cells developed significantly Although T cells likely to play a role in this tumor model (Fig. smaller tumor xenografts compared with those seen in control 2C) (29), in this work we chose to focus on macrophages because WT mice (Fig. 2A), associated with reduced numbers of tumor- previous studies have shown that heparanase, when added ex- associated macrophages (Fig. 2B) and T cells (Fig. 2C and Fig. ogenously or stably transfected, activates macrophages to stim- S1C). Even greater attenuation of tumor growth was noted in ulate cytokine expression (25). Similarly, macrophages isolated El-4 lymphoma cells implanted in Hpa-KO mice compared with from tumor xenografts produced by Panc-1 cells overexpressing

E7814 | www.pnas.org/cgi/doi/10.1073/pnas.1611380113 Gutter-Kapon et al. Downloaded by guest on September 28, 2021 heparanase were found to be more highly activated than mac- Heterogeneity and plasticity have emerged as hallmarks of PNAS PLUS rophages isolated from control tumors (30). Here we found that mononuclear phagocytes (19, 20, 34). Macrophages, like other macrophages exhibited high levels of heparanase activity (Fig. immune effector cells, have multiple subtypes and various phe- 1A), but no heparanase activity was detected in macrophages of notypes depending on the microenvironment. Specifically, mac- Hpa-KO mice. Importantly, the expression of most cytokines rophages can differentiate to distinct entities classically referred examined, including, among others, TNFα,IL-1β, IL-10, and IL-6, to as M1/kill type, which can slow or stop tumor growth, and M2/ was reduced in Hpa-KO macrophages compared with control repair type, which actively stimulate tumor growth (21, 23). This macrophages (Fig. 1B), and reduced cytokine levels were simi- dichotomous phenotype may explain why macrophages can elicit larly quantified in the culture medium conditioned by Hpa-KO a poor prognosis in some tumors and a better prognosis in others macrophages compared with WT macrophages (Fig. S1A). (i.e., non–small-cell lung cancer) (35, 36). In early tumors, TAMs Whereas the expression of heparanase is increased in many appear to have proinflammatory, tumoricidal (M1 or “classically types of tumors, often associated with more aggressive disease activated”) phenotype. These TAMs are phagocytic, present anti- and poor prognosis (4, 5, 31), so far the role of heparanase under gens well, produce Th1-type cytokines (e.g., IL-1β,TNFα), and are normal conditions has not been resolved in settings other than cytotoxic (18). They may also indirectly promote cytotoxicity by ac- autophagy (13). Our present results suggest that heparanase is tivating other cells of the immune system, such as NK cells and T intimately involved in the regulation of cytokine expression by cells (37). Nevertheless, as the tumor becomes established, macro- “ ” macrophages, decisively affecting their function. Likewise, Hpa- phages polarize toward alternatively activated M2 macrophages KO macrophages exhibit reduced motility capacity, critical for that stimulate tumor cell proliferation, migration, angiogenesis, and their surveillance nature (Fig. 1 E–G), in agreement with reduced metastasis (18, 20). The characterization of macrophages as termi- infiltration of Hpa-KO neutrophils and eosinophils to lungs ex- nally differentiated M1 vs. M2 cells is limited, however, because it posed to prolonged smoke exposure or subjected to an allergic describes the extremes of a continuum of functional states, whereas inflammatory model, respectively (32, 33). Most appealingly, the extent of activation is likely to be dynamic, as occurs in the Hpa-KO macrophages exhibited reduced phagocytic capacity (Fig. complex process of tumorigenesis (19). Here, instead we have used S1B), the hallmark of macrophage function as antigen-presenting markers for the functional state of macrophages (lysozymes 1 and 2), cells, whereas heparanase enhanced the phagocytic capacity of as well as NK cells (granzyme, perforine) (Fig. 2G and Fig. 3) to macrophages (Fig. 4B). more accurately assess their competence (27, 38). Localization of macrophages within tumors appears to be criti- We further noted that the expression of MIP-2 (CXCL2), a – chemokine that attracts macrophages to sites of inflammation, cally important for their function and activation state (39 41). In was prominently reduced in Hpa-KO macrophages (Fig. 1B), the MIP-2 overexpression model, reduced macrophage activation in possibly explaining their reduced accumulation in the perito- Hpa-KO tumors (Fig. 2G, Lower) was associated with accumulation H Right neum (Fig. 1G), and also that CXCL1 levels were decreased in of macrophages at the tumor periphery (Fig. 2 , ). Thus, whereas WT and Hap-KO macrophages are attracted to the tumor Hpa-KO macrophages (Fig. S1A, Right). Unexpectedly, over- at similar efficiencies (Fig. 2G, Upper), macrophage penetration and expression of MIP-2 in LLC cells resulted in reduced tumor population of the entire tumor mass requires heparanase. Notably, growth once cells were implanted in WT mice, but not in Hpa- inoculation of control macrophages together with LLC cells into KO mice (Fig. 2F). As expected, overexpression of MIP-2 in Hpa-KO mice was sufficient to attract macrophages from the tumor LLC cells (Fig. S2A) resulted in the recruitment of macrophages periphery to populate the tumor lesion (Fig. 4C). This suggests that (Fig. 2G, Upper and Fig. S1D), as well as CD4 and CD8 T cells the high endogenous activity of heparanase in WT macrophages (Fig. S3), to the resulting tumors, but only at a magnitude (Fig. 1A) is used by the Hpa-KO macrophages to penetrate the comparable to that in WT and Hpa-KO mice, which cannot tumor or that heparanase functions, directly or indirectly (i.e., re- explain the differential tumor growth observed in the WT vs. the lease of HS-bound chemokines) as chemoattractant. Alternatively, Hpa-KO background (Fig. 2F). Similarly, the differential tumor cytokines (e.g., SDF1) secreted by control macrophages may attract growth cannot be explained by the recruitment of M1/M2 mac- peripheral KO macrophages into the tumor. An association be- rophages (Fig. S2 B and C), but may be explained by the mac- tween reduced tumor growth and peripheral localization of mac- rophage activation, as evidenced by lysozyme 1 expression. Thus, rophages has been described in other experimental models (41). > whereas lysozyme levels were induced by 24-fold by MIP-2 in Even more importantly, localization of macrophages to the tumor WT mice, lysozyme induction was threefold lower in Hpa-KO periphery (or tumor front) has been associated with a favorable mice (Fig. 2G, Lower). These results clearly show that Hpa-KO prognosis in patients with colon cancer (39). This finding is in macrophages fail to respond to the antitumor effect of MIP-2, agreement with the idea that the functions of macrophages vary yet the therapeutic significance of MIP-2 as an antitumor agent considerably according to their location within tumors. For exam- clearly requires further in-depth investigation. ple, macrophages are highly proangiogenic in necrotic and hypoxic An even stronger antitumor response was evident when mono- tumor areas, but accumulation of macrophages in close proximity to cytes were implanted together with LLC cells in Hpa-KO mice. well vascularized areas of the tumors correlates with a good prog- Strikingly, tumor growth was halted significantly by coimplanta- nosis (18, 40). Given the highly necrotic feature of LLC tumors in tion of LLC cells with control monocytes (Fig. 3B), correlating our experimental settings (Fig. 2H), localization of macrophages to with marked increases in the numbers and activation of tumor- necrotic areas facilitates their activation toward the M1/kill phe- associated macrophages (e.g., F4/80, lysozymes 1 and 2) (Fig. notype compared with reduced activation when accumulating at 3C). In striking contrast, coimplantation of Hpa-KO monocytes highly vascularized areas of the tumor periphery. Reduced cytokine together with LLC cells had no effect on tumor growth (Fig. 3B) expression was noted not only in macrophages, but also in cells or macrophage recruitment and activation (Fig. 3C). Unlike in isolated from Hpa-KO spleen (Fig. S5B). This finding is in agree- the MIP-2 model, coimplantation of LLC cells with control ment with reduced cytokine expression after heparanase gene si- monocytes resulted in the recruitment and activation of T cells, lencing in T cells (42), whereas cytokine expression was markedly NK cells, and dendritic cells (Fig. 3D), which likely assist in at- induced by the addition of heparanase to macrophages (Fig. 1D)or tenuating tumor growth, correlating with a marked induction of to peripheral blood mononuclear cells (25, 30, 43). TNFα and SDF-1 (Fig. 4A). Taken together, these results in- How heparanase stimulates cytokine expression is not entirely dicate that heparanase is critically important for macrophage clear, but it does not seem to require enzymatic activity. We activation and function; in these experimental settings, macro- conclude this because cytokines were noted to be induced by phages are activated toward a kill phenotype. heparanase also in the presence of a small-molecule inhibitor, MEDICAL SCIENCES

Gutter-Kapon et al. PNAS | Published online November 14, 2016 | E7815 Downloaded by guest on September 28, 2021 OGT2115 (Fig. 1D), at concentrations that completely neutralize Real-Time PCR Analyses. Total RNA was extracted with TRIzol (Sigma-Aldrich), its enzymatic activity (Fig. S5C). In contrast, heparanase activity and RNA (1 μg) was amplified using the One-Step PCR Amplification Kit within macrophages was not affected by OGT2115 (Fig. S5D). (ABgene), according to the manufacturer’s instructions. The PCR primer sets used This may imply that this compound is unable to cross the plasma in this study are listed in Table S3. Cytokine expression was normalized to actin. Data are expressed as the mean level of expression normalized to actin and membrane, enter lysosomes, or function in the acidic environ- ± ment of the lysosome, or some other as-yet unidentified defi- represent the mean SEM of triplicate samples. The results are representative of three independent experiments (43). ciency. The function of enzymatically inactive heparanase is in agreement with our previous report (25), but contradicts results Antibodies and Reagents. Anti-heparanase neutralizing antibodies (1453, presented by others (43). This discrepancy may be due to the use 1023) have been described previously (14). Rat anti- mouse F4/80 antibody of different cells (mouse thioglycolate-stimulated macrophages was purchased from Serotec, and antibodies to phospho-p38 and phospho- vs. human peripheral blood mononuclear cells) or to differing JNK were purchased from Cell Signaling Technology. Anti–phopsho-Erk and assay conditions. For example, we applied relatively low con- the small-molecule heparanase inhibitor OGT 2115 (49) were obtained from centrations of OGT2115 (10 μg/mL) for a short period (6 h) and Santa Cruz Biotechnology. Anti-actin monoclonal antibody was purchased examined gene expression, whereas Goodall et al. (43) applied from Sigma-Aldrich. The selective MEK (PD98059), p38 (SB203580), and JNK much higher concentrations of OGT2115 (200 μg/mL) for a long (sp600125) inhibitors were purchased from Calbiochem and were dissolved period (24 h) and evaluated cytokine release. in DMSO as stock solutions. DMSO was added to the cell culture as a control. Toll-like receptors (TLRs) have been identified to lie up- Cytokine and transcription factor arrays were purchased from R&D Systems stream the signaling cascade that leads to cytokine induction by and Signosis, respectively. MIP2 cDNA was purchased from OriGene. The heparanase (25, 43); however, the underlying molecular mech- c-Fos–luciferase gene construct was kindly provided by Seung Ki Lee, Seoul, anism(s) and relevant transcription factor(s) have not yet been South Korea (50). The luciferase reporter assay was carried out essentially as characterized. We found that heparanase stimulates the phos- described previously (51). Preparation of latent heparanase protein has been phorylation of Erk, p38, and JNK (Fig. 5 A and C) and, more described previously (25). importantly, that inhibition of these pathways practically blocks Tumorigenicity and Immunohistochemistry. LLC cells were detached with cytokine induction by heparanase (Fig. 5 B, D,andE). These trypsin/EDTA, washed with PBS, and brought to a concentration of 3.5 × 106 findings are in agreement with the critical involvement of p38 and cells/mL. Peritoneal exudate cells were collected from WT and Hpa-KO mice JNK signaling in the mediation of TLR responses leading to cy- at 3 d after thioglycolate administration and subjected to FACS analysis. + tokine induction (44, 45). Using a transcription factors array, we Preparations exhibiting >90% CD11b cells (i.e., monocytes) (28) were mixed found that AP1 expression is enhanced by heparanase, and vali- with LLC cells in a 1:1 ratio, and the cell suspension (7 × 105/0.1 mL) was dated that heparanase stimulates the expression of c-Fos, but not inoculated s.c. at the right flank of 6- to 8-wk-old WT and Hpa-KO C57BL/6 of c-Jun (Fig. 5F). Furthermore, c-Fos expression was elevated in mice. Xenograft size was determined by externally measuring tumors in two parallel with the strong cytokine induction that accompanied the dimensions using calipers. inoculation of control macrophages together with LLC cells (Fig. At the end of the experiment, mice were killed, and tumors were removed 5H), whereas c-Fos induction by heparanase was significantly re- and weighed. RNA was extracted from a small portion of each tumor, and the duced by inhibitors of p38 and JNK (Fig. 5G). This suggests a remaining portion was fixed in formalin and embedded in paraffin. Then linear cascade that starts with heparanase-mediated TLR activa- 5-μm formalin-fixed, paraffin-embedded sections were subjected to immu- tion at the cell membrane, continues with Erk/p38/JNK activation, nostaining with the indicated antibodies using the Envision Kit (Dako) ’ and leads to AP1-mediated gene transcription. according to the manufacturer s instructions, as described previously (25). Taken together, our results reveal a role for endogenous hep- aranase in macrophage function. More specifically, our results Flow Cytometry. Freshly isolated macrophages or single-cell suspensions prepared from tumor xenografts were subjected to flow cytometry essen- strongly indicate that heparanase is critically important for tially as described previously (52). The antibodies used for flow cytometry macrophage activation. The outcome of macrophage activation analyses are listed in Table S4. in the experimental settings used in this study was tumor eradi- cation, but it is likely that the same principle holds true in set- BM Transplantation. BM transplantation was performed as described pre- tings where activation of macrophages leads to tumor progression. viously (52). Thus, heparanase inhibitors should be carefully used in cancer types (e.g., pancreatic carcinoma) (6) and conditions where mac- Phagocytosis. Macrophage phagocytosis capacity was evaluated using zymosan- rophages play a protumorigenic function, but not in other types coated IncuCyte pHrodo Bioparticles (Essen) according to the manufacturer’s (e.g., non–small-cell lung carcinoma) (36) to preserve the antitu- instructions. In brief, 3 d after the administration of thioglycolate macrophages mor function of macrophages. were collected from the peritoneum and plated (2 × 104) on fibronectin-coated 96-well plates for 24 h. Cells were then washed, and the zymosan-coated flu- Materials and Methods orogenic bioparticles (5 μL) were added. Once the bioparticles were engulfed by Cells and Cell Culture. LLC cells have been described previously (46). Mouse phagocytosis and entered the acidic phagosome, a substantial increase in peritoneal monocytes/macrophages were harvested from the peritoneal fluorescence was observed and monitored by quantitative live cell imaging fluid of WT or Hpa-KO C57BL/6 mice at 3 d after i.p. injection of thio- (IncuCyte ZOOM Live Cell Analysis System; Essen). glycolate (0.5 mL; 40 mg/mL), essentially as described previously (25). Peri- toneal exudate cells (5 × 106) were plated in a 60-mm dish for 24 h and then Statistics. Data are shown as mean ± SE. The significance of data were de- cultured in DMEM supplemented with glutamine, pyruvate, antibiotics, and termined using the two-tailed Student’s t test. Categorical variables were 2 10% (vol/vol) FCS in a humidified atmosphere containing 5% (vol/vol) CO2 at compared using the χ test or Fisher’s exact test. A P value ≤ 0.05 was con- 37 °C. Nonadherent cells were removed after 24 h by washing, and the cells sidered statistically significant. All experiments were repeated at least three that remained attached were considered macrophages (25). Cell migration times, with similar results obtained. and invasion assays were performed essentially as described previously (47). ACKNOWLEDGMENTS. This study was supported by research grants awarded Cell Lysates, Heparanase Activity, and Protein Blotting. Preparation of cell to I.V. by the Israel Science Foundation (601/14), the United States-Israel lysates, protein blotting, and measurement of heparanase enzymatic activity Binational Science Foundation, the Israel Cancer Research Fund (ICRF), and were carried out as described previously (13, 14, 48). the Rappaport Family Institute Fund. I.V. is a Research Professor at the ICRF.

1. Nakajima M, Irimura T, Di Ferrante D, Di Ferrante N, Nicolson GL (1983) Heparan 2. Vlodavsky I, Fuks Z, Bar-Ner M, Ariav Y, Schirrmacher V (1983) Lymphoma cell- sulfate degradation: Relation to tumor-invasive and metastatic properties of mouse mediated degradation of sulfated proteoglycans in the subendothelial extracel- B16 melanoma sublines. Science 220(4597):611–613. lular matrix: Relationship to tumor cell metastasis. Cancer Res 43(6):2704–2711.

E7816 | www.pnas.org/cgi/doi/10.1073/pnas.1611380113 Gutter-Kapon et al. Downloaded by guest on September 28, 2021 3. Hammond E, Khurana A, Shridhar V, Dredge K (2014) The role of heparanase and 29. Lider O, et al. (1990) Inhibition of T lymphocyte heparanase by heparin prevents T cell PNAS PLUS sulfatases in the modification of heparan sulfate proteoglycans within the tumor migration and T cell-mediated immunity. Eur J Immunol 20(3):493–499. microenvironment and opportunities for novel cancer therapeutics. Front Oncol 30. Hermano E, et al. (2014) Macrophage polarization in pancreatic carcinoma: Role of 4:195. heparanase enzyme. J Natl Cancer Inst 106(12):dju332. 4. Ilan N, Elkin M, Vlodavsky I (2006) Regulation, function and clinical significance of 31. Barash U, et al. (2010) Proteoglycans in health and disease: New concepts for heparanase heparanase in cancer metastasis and angiogenesis. Int J Biochem Cell Biol 38(12): function in tumor progression and metastasis. FEBS J 277(19):3890–3903. 2018–2039. 32. Morris A, et al. (2015) The role of heparanase in pulmonary cell recruitment in re- 5. Vreys V, David G (2007) Mammalian heparanase: What is the message? J Cell Mol Med sponse to an allergic but not non-allergic stimulus. PLoS One 10(6):e0127032. 11(3):427–452. 33. Petrovich E, et al. (2016) Lung ICAM-1 and ICAM-2 support spontaneous intravascular 6. Vlodavsky I, et al. (2012) Significance of heparanase in cancer and inflammation. effector lymphocyte entrapment but are not required for neutrophil entrapment or Cancer Microenviron 5(2):115–132. emigration inside endotoxin-inflamed lungs. FASEB J 30(5):1767–1778. 7. Casu B, Vlodavsky I, Sanderson RD (2008) Non-anticoagulant heparins and inhibition 34. Mantovani A, Sozzani S, Locati M, Allavena P, Sica A (2002) Macrophage polarization: of cancer. Pathophysiol Haemost Thromb 36(3-4):195–203. Tumor-associated macrophages as a paradigm for polarized M2 mononuclear 8. Vlodavsky I, Ilan N, Naggi A, Casu B (2007) Heparanase: Structure, biological func- phagocytes. Trends Immunol 23(11):549–555. tions, and inhibition by heparin-derived mimetics of heparan sulfate. Curr Pharm Des 35. Bingle L, Brown NJ, Lewis CE (2002) The role of tumour-associated macrophages in – 13(20):2057 2073. tumour progression: Implications for new anticancer therapies. J Pathol 196(3): 9. Arvatz G, Shafat I, Levy-Adam F, Ilan N, Vlodavsky I (2011) The heparanase system and 254–265. tumor metastasis: Is heparanase the seed and soil? Cancer Metastasis Rev 30(2): 36. Talmadge JE, Donkor M, Scholar E (2007) Inflammatory cell infiltration of tumors: – 253 268. Jekyll or Hyde. Cancer Metastasis Rev 26(3-4):373–400. 10. Barash U, et al. (2014) Heparanase enhances myeloma progression via CXCL10 37. Gabrilovich DI, Ostrand-Rosenberg S, Bronte V (2012) Coordinated regulation of – downregulation. Leukemia 28(11):2178 2187. myeloid cells by tumours. Nat Rev Immunol 12(4):253–268. 11. Boyango I, et al. (2014) Heparanase cooperates with Ras to drive breast and skin 38. Chen CS, Doloff JC, Waxman DJ (2014) Intermittent metronomic drug schedule is – tumorigenesis. Cancer Res 74(16):4504 4514. essential for activating antitumor innate immunity and tumor xenograft regression. 12. Ramani VC, et al. (2016) Targeting heparanase overcomes chemoresistance and di- Neoplasia 16(1):84–96. minishes relapse in myeloma. Oncotarget 7(2):1598–1607. 39. Forssell J, et al. (2007) High macrophage infiltration along the tumor front correlates 13. Shteingauz A, et al. (2015) Heparanase enhances tumor growth and chemoresistance with improved survival in colon cancer. Clin Cancer Res 13(5):1472–1479. by promoting autophagy. Cancer Res 75(18):3946–3957. 40. Rivera LB, Bergers G (2013) Location, location, location: Macrophage positioning 14. Weissmann M, et al. (2016) Heparanase-neutralizing antibodies attenuate lymphoma within tumors determines pro- or antitumor activity. Cancer Cell 24(6):687–689. tumor growth and metastasis. Proc Natl Acad Sci USA 113(3):704–709. 41. Sun Y, Lodish HF (2010) Adiponectin deficiency promotes tumor growth in mice by 15. Ferrara N, Hillan KJ, Gerber HP, Novotny W (2004) Discovery and development of reducing macrophage infiltration. PLoS One 5(8):e11987. bevacizumab, an anti-VEGF antibody for treating cancer. Nat Rev Drug Discov 3(5): 42. He YQ, et al. (2012) The endoglycosidase heparanase enters the nucleus of T lym- 391–400. phocytes and modulates H3 methylation at actively transcribed genes via the in- 16. Grivennikov SI, Greten FR, Karin M (2010) Immunity, inflammation, and cancer. Cell terplay with key chromatin modifying enzymes. Transcription 3(3):130–145. 140(6):883–899. 43. Goodall KJ, Poon IK, Phipps S, Hulett MD (2014) Soluble heparan sulfate fragments 17. Mantovani A, Sica A (2010) Macrophages, innate immunity and cancer: Balance, generated by heparanase trigger the release of pro-inflammatory cytokines through tolerance, and diversity. Curr Opin Immunol 22(2):231–237. TLR-4. PLoS One 9(10):e109596. 18. Biswas SK, Sica A, Lewis CE (2008) Plasticity of macrophage function during tumor 44. Bode JG, Ehlting C, Häussinger D (2012) The macrophage response towards LPS and its progression: Regulation by distinct molecular mechanisms. J Immunol 180(4): control through the p38(MAPK)-STAT3 axis. Cell Signal 24(6):1185–1194. 2011–2017. 45. Guha M, Mackman N (2001) LPS induction of gene expression in human monocytes. 19. De Palma M, Lewis CE (2013) Macrophage regulation of tumor responses to anti- – cancer therapies. Cancer Cell 23(3):277–286. Cell Signal 13(2):85 94. 20. Biswas SK, Mantovani A (2010) Macrophage plasticity and interaction with lympho- 46. Gingis-Velitski S, et al. (2011) Host response to short-term, single-agent chemotherapy cyte subsets: Cancer as a paradigm. Nat Immunol 11(10):889–896. induces matrix metalloproteinase-9 expression and accelerates metastasis in mice. – 21. Lakshmi Narendra B, Eshvendar Reddy K, Shantikumar S, Ramakrishna S (2013) Im- Cancer Res 71(22):6986 6996. mune system: A double-edged sword in cancer. Inflamm Res 62(9):823–834. 47. Gingis-Velitski S, Zetser A, Flugelman MY, Vlodavsky I, Ilan N (2004) Heparanase in- 22. Lamagna C, Aurrand-Lions M, Imhof BA (2006) Dual role of macrophages in tumor duces endothelial cell migration via protein kinase B/Akt activation. J Biol Chem – growth and angiogenesis. J Leukoc Biol 80(4):705–713. 279(22):23536 23541. 23. Mills CD, Lenz LL, Harris RA (2016) A breakthrough: Macrophage-directed cancer 48. Vlodavsky I (1999) Preparation of extracellular matrices produced by cultured corneal immunotherapy. Cancer Res 76(3):513–516. endothelial and PF-HR9 endodermal cells. Curr Protoc Cell Biol Chapter 10:Unit 10.4. 24. Zcharia E, et al. (2009) Newly generated heparanase knock-out mice unravel co- 49. Li Y, et al. (2013) Suppression of endoplasmic reticulum stress-induced invasion and regulation of heparanase and matrix metalloproteinases. PLoS One 4(4):e5181. migration of breast cancer cells through the downregulation of heparanase. Int J Mol 25. Blich M, et al. (2013) Macrophage activation by heparanase is mediated by TLR-2 and Med 31(5):1234–1242. TLR-4 and associates with plaque progression. Arterioscler Thromb Vasc Biol 33(2): 50. Hwang ES, Hong JH, Bae SC, Ito Y, Lee SK (1999) Regulation of c-fos gene tran- e56–e65. scription and myeloid cell differentiation by acute myeloid leukemia 1 and acute 26. Driscoll KE (1994) Macrophage inflammatory proteins: Biology and role in pulmonary myeloid leukemia-MTG8, a chimeric leukemogenic derivative of acute myeloid leu- inflammation. Exp Lung Res 20(6):473–490. kemia 1. FEBS Lett 446(1):86–90. 27. Doloff JC, Chen CS, Waxman DJ (2014) Anti-tumor innate immunity activated by in- 51. Kehat I, et al. (2006) Inhibition of basic leucine zipper transcription is a major medi- termittent metronomic cyclophosphamide treatment of 9L brain tumor xenografts is ator of atrial dilatation. Cardiovasc Res 70(3):543–554. preserved by anti-angiogenic drugs that spare VEGF receptor 2. Mol Cancer 13:158. 52. Voloshin T, et al. (2015) Blocking IL1β pathway following paclitaxel chemotherapy 28. Gorczyca W, et al. (2011) Immunophenotypic pattern of myeloid populations by flow slightly inhibits primary tumor growth but promotes spontaneous metastasis. Mol cytometry analysis. Methods Cell Biol 103:221–266. Cancer Ther 14(6):1385–1394. MEDICAL SCIENCES

Gutter-Kapon et al. PNAS | Published online November 14, 2016 | E7817 Downloaded by guest on September 28, 2021