4-methylumbelliferone-mediated polarization of M1 macrophages correlate with decreased hepatocellular carcinoma aggressiveness in mice

Hepatocellular carcinoma (HCC) arises in the setting of advanced liver fibrosis, a dynamic and complex inflammatory disease. The tumor microenvironment (TME) is a mixture of cellular components including cancer cells, cancer stem cells (CSCs), tumor-associated macrophages (TAM), and dendritic cells (DCs), which might drive to tumor progression and resistance to therapies. In this work, we study the effects of 4-methylumbelliferone (4Mu) on TME and how this change could be exploited to promote a potent immune response against HCC. First, we observed that 4Mu therapy induced a switch of hepatic macrophages (Mϕ) towards an M1 type profile, and HCC cells (Hepa129 cells) exposed to conditioned medium (CM) derived from Mϕ treated with 4Mu showed reduced expression of several CSCs markers and aggressiveness. HCC cells incubated with CM derived from Mϕ treated with 4Mu grew in immunosuppressed mice while presented delayed tumor progression in immunocompetent mice. HCC cells treated with 4Mu were more susceptible to phagocytosis by DCs, and when DCs were pulsed with HCC cells previously treated with 4Mu displayed a potent antitumoral effect in therapeutic vaccination protocols. In conclusion, 4Mu has the ability to modulate TME into a less hostile milieu and to potentiate immunotherapeutic strategies against HCC.


Results
In vivo 4Mu therapy induces hepatic macrophages polarization towards a M1 profile. Macrophages (Mϕ) are major components of TME, and they have a pivotal role in promoting HCC progression 26 . We have studied the effects of 4Mu on hepatic Mϕ population in our experimental model of HCC with associatedfibrosis 25 . After 4 weeks of TAA administration, mice were inoculated orthotopically with 1.25 × 10 5 Hepa 129 cells (day 0). On day 5, animals received daily saline or 4Mu orally (200 mg/kg; Fig. 1a, Left). On day 9 and 15 mice were sacrificed, and liver sample collected. The fraction of non-parenchymal cells from tumor, peri-tumor, and non-tumor liver regions (Fig. 1a, Right) were obtained and analyzed by flow cytometry. The percentage of F4/80 + CD206 + and F4/80 + CD86 + cells was evaluated and the M1/M2 proportion was calculated as log10 (CD86 + /CD206 + ). 4Mu therapy induced a polarized M1 profile in tumor and non-tumor sections in comparison with saline group (Fig. 1b, Right) on day 9. In addition, we observed a M1 Mϕ profile induced by 4Mu in peri-tumor and non-tumor sections on day 15; Fig. 1b also showed the reduced percentage of F4/80 + CD206 + cells (44.0 ± 0.63% vs. 72.0 ± 4.41%; ***p < 0.001 and 60.0 ± 1.90% vs. 78.2 ± 1.43%, *p < 0.05 4Mu vs. saline, respectively. This effect is accompanied by an increase in the percentage of F4/80CD86 + cells in tumor pieces (63.5 ± 4.63% vs. 32.6 ± 1.24%; *p < 0.05 and 85.2 ± 5.22% vs. 71.8 ± 4.12%; *p < 0.05 4Mu vs. saline) induced by 4Mu on day 9 and day 15, respectively. We also found that the ratio of mRNA levels of iNOS/Arg1 in hepatic Mϕ isolated from tumor sections of 4Mu treated mice was enhanced both at day 9 and at day 15 (**p < 0. 01 on day 9; *p < 0.05 on day 15; Fig. 1c). Although the levels of cytokines modulated in tumor hepatic Mϕ on day 9 did not show clearly a definitive profile, there was a significant increase in the mRNA levels of IL1-B and TNF-α together with a decrease of TGF-β and IL-10 on day 15, which means a clear polarization induced in tumor hepatic M towards an M1 profile after 4Mu treatment (Fig. 1c).

Effect of M1 macrophages polarization induced by 4Mu on hepatocarcinogenesis.
We next aimed to evaluate if the M1 phenotype of Mϕ induced by 4Mu has effects on the capability of HCC cells to establish and growth. To this end, s.c. tumors were developed in C3H/He mice with Hepa129 cells cultured in the presence of CM derived from pMϕ treated or not with 4Mu (4Mu-treated pMϕ derived CM). Remarkably, we found a high index of tumor development and progression in Hepa129 control tumors generated by cells incubated with either RPMI or with CM derived from untreated pMϕ. On the other hand, the growth of tumors established from Hepa129 cells pre-treated with 4Mu-treated pMϕ derived-CM was significantly reduced (Fig. 3a) Then, we wanted to study if tumor aggressiveness was modified by 4Mu-treated pMϕ derived CM and untreated pMϕ-derived CM). We observed that after 24 h and 48 h of pre-conditioning, 4Mu-treated pMϕ derived-CM reduced the expression levels of cancer stemness markers (TLR4 and CD47) 30,31 , and the totipotency factor Sox2 32 in cells (*p < 0.05, **p < 0,01 and ****p < 0,001; Kruskal-Wallis test) (Fig. 3b). As we described, HCC are composed of subpopulations of tumor cells with diverse tumorigenic abilities 33 . Then, we magnetically isolated CD133 + and CD133 -Hepa129 cells to compare if CM mediate a differential effect mainly on CSCs. Figure 3c showed that 24 h of pre-conditioning with 4Mu treated-pMϕ derived-CM has no effect on the CSCs markers levels while 48 h of pre-conditioning with 4Mu treated-pMϕ derived-CM strongly reduces levels of TLR4, CD47 and Sox2 expression on CSCs. In addition, CM derived from 4Mu-treated pMϕ showed a reduced amount of IL-6 in comparison with untreated pMϕ (Fig. 3d).
We next administrated Hepa 129 cells exposed to CM from 4Mu treated-pMϕ in athymic Nu/Nu mice. The results showed that tumor progression in immunosuppressed mice with Hepa129 cells pretreated with pMϕ derived-CM or 4Mu treated-pMϕ derived-CM were similar to controls (Fig. 4a).
We previously demonstrated that 4Mu does not show a significant impact on Hepa129 cell growth and survival. However, 4Mu can modulate the expression of CSCs markers, particularly CD47, contributing to the phagocytosis of CD133 + CSCs 25 . Based on this, modulation of CD47 expression might have a dual effect, on one side induces a less aggressive tumor phenotype and, on the other side, HCC cells might become more susceptible to the recognition by the immune system. Figure 4b,c reveal that CD133 + cells could generate tumors in both immunocompetent and immunocompromised mice. In contrast, when the expression of CD47 is inhibited by 4Mu, tumors progressed only in immunocompromised animals while were potently controlled in immunocompetent mice.

4Mu stimulates antigen presentation by dendritic cells and improves their capability to phagocyte cancer cells. We assessed whether 4Mu have effects on the function of dendritic cells (DCs). First, we
performed an in vitro phagocytosis assay using BM-derived DCs from C3H/He mice. Hepa129 HCC cells were labeled with DAPI, co-cultured with DCs for 2 h, and incubated with MHC II and CD86 antibodies. We quantified the presence of MHC II + CD86 + DCs, gated them, and then identified MHCII + CD86 + DAPI + phagocyted Hepa 129 cells (Fig. 5a, Right). Interestingly, phagocytosis was significantly increased in Hepa129 + 4Mu cells compared with untreated Hepa129 cells (Fig. 5a, Left; 36.4 ± 3.69 vs. 24.8 ± 3.40; *p < 0.05, Mann-Whitney test). We also observed that 4Mu could facilitate the maturation of DCs since the percentages of CD11 + MHCII + CD86 + DCs were higher when DCs were exposed to 4Mu for 72 h ( Supplementary Fig. S3). This result suggests that the ability to recognize tumor cells by DCs is increased by 4Mu.

Discussion
In the last years, the study of the TME components on carcinogenesis has gained an special interest due their potential as therapeutic targets 34 . Different cellular components of the TME could act in synergy to facilitate cancer progression and to avoid the immune system recognition. In particular, TAMs population and their activation status are involved in tumor aggressiveess 15 . The classical Mϕ activation M1 type is characterized by the expression of the co-stimulatory molecule CD86, an increase in the activity of iNOS, and in the production of pro-inflammatory cytokines such as IL-1β and TNF-α. On the other hand, the alternative type 2 polarized Mϕ (M2) is known by the expression of CD206, an increased activity of arginase I, and by the release of the anti-inflammatory cytokines IL-10 and TGF-β 17 .
Hepatic Mϕ play an essential role in the pathogenesis of chronic inflammatory liver diseases, and in the progression of HCC 26,35 . TAMs are one of the more abundant immune cells that infiltrate tumors. Yeung et al. described a direct effect M2 Mϕ on HCC growth. They showed that the expansion of TAMs with M2 phenotype promote tumor growth and invasiveness through the release of CCL22. Moreover, the peritumoral accumulation of M2 Mϕ observed in patient samples correlates with poor prognosis, and constitutes a predictor of patient survival 15 .
The role of circulating stem cell-like tumor cells phenotypes in HCC was addressed by Sun et al. They observed that the presence of circulating HCC cells with a stem-like phenotype (Epcam + /CD133 + ) was associated with recurrence in HCC patients after surgery 36 . In addition to CCL22, it has been demonstrated that M2 TAMs also secrete TGF-β and IL-6 promoting EMT and acquisition of CSCs-like properties 24,37 . IL-6 detected in human HCC samples correlated with the presence of CSCs markers, and IL-6 expressed by TAMs induces the expansion of CD44 + cells in culture 24 .
Our previous results showed that 4Mu mitigates thioacetamide-induced liver chronic injury by reducing hyaluronan deposition, and hepatic stellate cells activation. 4Mu therapy also inhibited angiogenesis and tumor growth in vivo 28,38 . In the present work, we demonstrated that 4Mu modified the hepatic Mϕ phenotype. and their cytokine secretion profile. While a typical M2 activation was found in Mϕ from tumor, peri-tumor, and non-tumor tissues of untreated HCC, there was a significant M1 polarization induction in tumoral Mϕ after 4Mu treatment. When HCC cells were cultured in presence of supernatant from 4Mu-induced M1 Mϕ, and challenge immunocompetent mice with this preconditioned HCC cells, a delayed tumor progression was observed. This data suggests that 4Mu modulates the ability of hepatic Mϕ to promote tumor growth. Remarkably, CM derived from Mϕ treated with 4Mu induces a significant reduction of stemness-related markers on both whole HCC cells and more significant on isolated CSCs.
Recently, it has been reported that increased expression of CD47 on HCC cells was positively correlated with the density of hepatic Mϕ, and with poor clinical prognosis 39 . In this work, the authors also suggested that IL-6 derived from hepatic TAMs was involved in the up regulation of CD47 expression on HCC cells. In this line, we observed in our model that CM derived from Mϕ treated with 4Mu showed lower levels of IL-6, and that 4Mu reduced the stemness-related phenotype on HCC cells and their capacity to growth in vivo after the exposure to CM derived from Mϕ treated with 4Mu.
Conditioned media from Mϕ treated with 4Mu induced a less aggressive HCC phenotype and facilitated immune system recognition. We also demonstrated that CD133 + CSCs generate tumors in both immunocompetent and immunocompromised mice, but when the expression of stemness-related markers is inhibited by 4Mu, tumors grew only in athymic animals while were susceptible to the immune system control in immunocompetent mice.
It has been proposed that hepatic M2 Mϕ interact with cytotoxic CD8 + T cells, and induce resistance to immunotherapy. In addition, a positive association was demonstrated between the expression levels of M2 Mϕ markers, decreased CD8 + T cell infiltration, and PD-L1 levels in tumor pieces from patients with HCC 19 . In our hands, M1 polarization was induced in HCC tumors after 4Mu treatment, the percentage of CD3 + CD8 + tumor infiltrating T cells was increased, and by the reduction of CD47 expression on CSCs they might result more "visible" to the immune system, open the possibility to combine 4Mu with other immunotherapies like checkpoints inhibitors 40 .
Drug resistance was reported in advanced HCC patients treated with standard tyrosine kinase inhibitors (TKI) 3,41 . Sorafenib-resistant clones derived from HCC cell lines showed CSCs properties, including up-regulation of CD47. In addition, CD47 expression was found to be regulated by nuclear factor kappa B (NF-κB), and human HCC samples showed a positive correlation between NF-κB and the presence of CD47 42 . Modulation of CSCs markers, particularly CD47, directly by 4Mu or indirectly through the induction of M1 hepatic Mϕ  www.nature.com/scientificreports/ generated by 4Mu could be considered as an approach to sensitize cancer cells to TKI therapy, particularly in patients who have been previously treated with sorafenib. We showed that 4Mu decreases CD47 expression on HCC cells and facilitates phagocytosis by Mϕ, which is also associated with antitumor immune response in mice. It has been reported a role of SIRP-expressing DC in antitumor responses, including in HCC; it is possible that CD47 down regulation by 4Mu may affect the response mediated by both macrophages and DC. Here, we have also described that 4Mu has the ability to turn DCs more active to recognize and engulf tumor cells. It have been reported that DCs migrate into tumor-draining lymph nodes and prime CD8 + or CD4 + T cells to induce antitumor responses in mouse models, although their manipulation in cancer vaccination protocols has not reached a potent clinical impact 43,44 . Tumor progression was significantly inhibited in mice that receive DCs pulsed with 4Mu-treated HCC lysate in comparison with mice DCs pulsed with HCC lysate alone, suggesting that 4Mu therapy increases the potential of DCs to generate immunity against HCC. Dendritic cell-based vaccines have been tested as therapeutic tool for HCC 45 . A recently reported meta-analysis aimed to investigate the efficacy of DCs alone or combined with conventional treatments illustrated that cellular immunotherapy improve prognosis by increasing overall survival and reducing recurrence in patients with advanced HCC 46 .
All in all, our results suggest that 4Mu exerts a significant antitumoral effect: (i) by inducing a switch of hepatic Mϕ into a M1 profile, (ii) by reducing their capacity to secrete IL-6, (iii) by increasing HCC recognition by the immune system upon incubation of tumor cells with CM derived from Mϕ treated with 4Mu; (iv) by reducing the expression of several CSCs markers on HCC cells; and (v) by increasing the ability of DCs to inhibit HCC tumor growth in therapeutic vaccination protocols. In conclusion, our data highlight the potential of 4Mu to modulate the TME facilitating the induction of an immune response against HCC.

Materials and methods
Animals. Six-to-eight-week-old male C3Hj/He, BALB/c and athymic N:NIH(S)-nu mice (Nu/Nu) mice were purchased from Centro Atómico Ezeiza (Buenos Aires, Argentina). Animals were maintained at our Animal Resources Facilities in accordance with the experimental ethical committee and the NIH guidelines on the ethical use of animals. The Animal Care Committee from School of Biomedical Sciences, Universidad Austral, approved the experimental protocol (protocol #2018-05) which was based on the essential points of the ARRIVE guidelines. Drugs. 4 Methylumbelliferone (4Mu) sodium salt was purchased from Sigma-Aldrich (USA) 28 .

Cell lines. Hepa129 cells (HCC cells syngeneic with
In vivo experiments. Experimental model of HCC associated with fibrosis. C3H/He mice received 200 mg/ kg of thioacetamide (TAA) (Sigma-Aldrich, USA) intraperitoneally (i.p.) 3 times a week for 4 weeks to develop liver fibrosis. TAA-treated livers from tumor-bearing mice showed the extensive appearance of portal-portal and central-portal fibrous septae and distortion of liver architecture ( Supplementary Fig. S1). On day 28, mice were anesthetized and orthotopic tumors were established by subcapsular inoculation of 1.25 × 10 5 Hepa129 cells directly into the left liver lobe by laparotomy 25 (day 0). Five days after tumor implantation, mice were distributed in groups (n = 8/group) and received: (i) saline (control), or (ii) 4Mu 200 mg/kg in the drinking water ad libitum. TAA was administrated until mice were used in ex vivo studies. On day 9 and 15 mice were sacrificed, tumor volume was measured using a caliper, and liver samples were collected. In some experiments, livers were perfused with collagenase and separated in 3 sections (peri-tumoral, tumoral, and non-tumoral tissue). Isolation of non-parenchymal cells from each tissue section was carried out using Histodenz.
Mice vaccination with DCs. HCC tumors were induced s.c. in C3H/He mice (n = 4-5/group) by inoculating 1 × 10 6 Hepa129 cells, or in BALB/c mice (n = 4/group) by inoculating BNL cells. Tumors were detected approximately 10 days after injection. Then, 1 × 10 5 matured DCs, or DCs loaded with lysates were injected peritumorally into tumor-bearing mice. Tumor growth was measured 3 times per week, and volumes were calculated as described above.
Ex vivo experiments. Macrophages profile assessment. Histodenz isolated liver cells were cultured for 30 min. Then, cells adhered to plastic were collected, and used for flow cytometry analysis or stored at − 80 °C with Trizol (Invitrogen, USA) for further RNA isolation. Flow cytometry of F4/80+, CD206+ and CD86+ were performed using a FACS Aria (BD Biosciences, USA). For FACS, potential autofluorescence of 4Mu was ex- Isolation of intraperitoneal macrophages. Macrophages were isolated from the peritoneal cavity (pMϕ), incubated in serum-free medium for 30 min in a 24-well tissue-culture plate, and treated with 0.5 mM 4Mu or RPMI for additional 72 h. Then, cells were collected and used for flow cytometry analysis or maintained with Trizol (Invitrogen, USA) at − 80 °C. In addition, conditioned medium (CM) from 4Mu-treated pMϕ (4Mutreated pMϕ derived-CM) or pMϕ derived-CM was stored at − 80 °C until its use. Flow cytometry of F4/80+, CD206+ and CD86+ were performed using a FACSAria (BD Biosciences, USA), macrophages type1/type2 proportion was calculated as above and quantitative PCR analysis were performed as we described above. Hepa129 cells were incubated with RPMI (Hepa129 + RPMI), CM from isolated pMϕ (Hepa129 + pMϕ derived-CM) and conditioned media from isolated pMϕ in vitro treated with 4Mu (Hepa129 + 4Mu treated-pMϕ derived-CM) for 48 h. Preconditioned 1 × 10 6 Hepa129 cells were used for in vivo experiments or for western blot. Intraperitoneal Mϕ also were obtained on day 9 and day 15 from mice with fibrosis-associated HCC treated with saline or 20 mg/kg 4Mu. Flow cytometry of F4/80+ CD86+ and CD206+ cells also were analyzed and the M1/M2 proportions were calculated as above.

Quantification of CD3+ cells on HCC tumors.
Tumor lysates from mice treated with 4Mu or saline were obtained after liver perfusion with collagenase (Sigma-Aldrich, USA). Cell suspensions were then treated with RBC lysis buffer and stained with anti-CD4, anti-CD8, and anti-CD3 antibodies (BD Biosciences, USA). Then, lymphocytic infiltrate on tumor tissue was analyzed by flow cytometry.
Westen blotting. expression of TLR4, Sox2 and CD47 was detected in extracts from preconditioned whole, CD133+ or CD133-Hepa129 cells by immunobloting. Briefly, cells were collected and incubated in lysis buffer Table 1. Specific forward and reverse primers.
IL-6 quantification. To characterize conditioned medium, supernatants of pMϕ were collected 72 h after being treated with 4Mu, replaced with RPMI serum-free and collected after 24 and 48 h of culture. The concentration of IL-6 was measured by ELISA from BD (BD Biosciences, USA) according to the manufacture's guideline.
Generation of bone marrow-derived DCs. DCs were generated from murine bone marrow cells as described previously 54 . Briefly, bone marrow of C3H/He mice or BALB/c were obtained from femurs and tibias, subjected to mechanic disruption, cell suspensions obtained and cultured with RPMI 1640 with 10% FBS, and IL-4 and GM-CSF (20 ng/ml; PeproTech, Germany). On day 3 and 5, the medium was withdrawn and replaced by new fresh RPMI. On day 7, suspension cells were collected (DCs) and used for experiments.
Phenotypic analysis of DCs. DCs were cultured with 0.5 mM 4Mu during 72 h. Then, DCs were stained with CD11c-PE, CD86-APC, and MCH-II-FITC (BD Biosciences, USA) and analyzed by flow cytometry.
Antigen loading. Hepa129, BNL cells or tumor extracts from HCC bearing-mice were obtain as previously 54 .
Briefly, DCs were pulsed with cells or tumor lysates alone (200 μg/10 6 cells/ml) at 37 °C for 18 h. Cells were then centrifuged, characterized by flow cytometry, and used for in vivo experiments.

In vitro assays. Cell isolation by magnetic-activated cell sorting (MACS). Hepa129 cells were labeled with
primary CD133/1 antibody (Miltenyi Biotec, Germany), and the CD133 + was subsequently magnetically isolated using MACS Columns.

Statistical analysis.
All experiments were repeated at least 2 or 3 times on different occasions. Values were expressed as the mean ± SEM. Mann-Whitney, Tukey's or Kruskall-Wallis (ANOVA) multiple comparison tests were used to evaluate the statistical differences between groups. Mice survival was analyzed by a Kaplan-Meier curve. P value < 0.05 was considered as significant. Prism software (Graph Pad, San Diego, CA, USA) was employed for the statistical analysis 25 .