HMGB1 Promotes Lymphangiogenesis through the Activation of RAGE on M2 Macrophages in Laryngeal Squamous Cell Carcinoma

Background Receptor for advanced glycation end products (RAGE) is implicated in tumor biology. Released high mobility group box protein 1 (HMGB1) ligand binding to RAGE receptor in tumor cells promotes tumor progression. The mechanisms of HMGB1-RAGE signaling in M2 macrophages involved in lymphangiogenesis in laryngeal carcinoma remain poorly understood. Here, we assessed the effect of HMGB1-RAGE signaling on M2 macrophages in lymphangiogenesis. Methods HMGB1, CD163, and D2-40 in laryngeal squamous cell carcinoma (LSCC, n = 123), laryngeal precursor lesions (LPLs, n = 102), and vocal polyp (VP, n = 55) were analyzed by immunohistochemistry. THP-1 cell-expressed RAGE gene was knocked down and then polarized to M0 macrophages and M2 macrophages. IL-23, TNF-α, TGF-β, and IL-10 were measured by ELISA; IL-1β, IL-12, IL-10, and CCL-13 were evaluated by RT-qPCR, and CD206, CD163, and RAGE were evaluated by western blot to evaluate whether classical M2 macrophages were obtained. Conditioned media from RAGE+/- M0 macrophages and RAGE+/- M2 macrophages incubated in the presence or absence of HMGB1, anti-Toll-like receptor (TLR)2, anti-TLR4 antibodies, and anti-VEGF-C antibodies were collected separately for human dermal lymphatic endothelial cells (HDLEC) for proliferation, migration, lymphangiogenesis assay, and VEGF-C concentration analysis. Results HMGB1 and M2 macrophage densities were increased in LSCC (P < 0.01). HMGB1 and M2 macrophage densities were significantly correlated with lymphatic vessel density (LVD) in LSCC (P < 0.01). The HMGB1 overexpression and higher M2 macrophage density were involved in lymph node metastasis (P < 0.01) and poor prognosis (P < 0.05). In vitro, conditioned medium from HMGB1-stimulated RAGE+ M2 macrophages activated lymphangiogenesis by upregulating the VEGF compared to controls (P < 0.05). On the contrary, RAGE knockdown obviously decreased the corresponding effects of HMGB1-preconditioned M2 macrophages upon HDLEC (P < 0.05). HMGB1-TLR pathway does not significantly increase HDLEC proliferation, migration, and lymphangiogenesis on M2 macrophages. Conclusions HMGB1 promotes lymphangiogenesis by activation of RAGE on M2 macrophages. Targeting RAGE may provide an effective therapeutic strategy against M2 macrophages in LSCC patients with lymph node metastasis.


Introduction
Despite advances being made in the detection and treatment of laryngeal squamous cell carcinoma (LSCC), the 5-year survival rate has not improved, particularly in patients who suffer from lymph node metastases [1]. Lymphangiogenesis is important in the pathogenesis of lymph node metastasis. Thus, new therapies to block the formation of new lymphatic vessels are urgently required.
HMGB1 is a nuclear protein involved in crucial biological processes [2]. Increased amounts of HMGB1 in tissues have been closely associated with the proliferation, invasion, metastasis, and prognosis of many tumors [3,4]. More significantly, HMGB1 has been described as a damageassociated molecular pattern (DAMP) molecule in a number of infectious diseases and cancer [5,6]. After being released from inflammatory cells, necrotic cells, or tumor cells, extracellular HMGB1 can bind to pattern recognition receptors [2,7] and induce inflammation or promote tumor progression. The receptor of advanced glycation end products (RAGE) is a multiligand cell-surface receptor overexpressed in inflammation, cancer, and atherosclerosis [8][9][10]. Furthermore, RAGE is the major receptor for HMGB1 on macrophages [11]. RAGE has a greater affinity than Toll-like receptor (TLR) for HMGB1, although the HMGB1 pathway is also mediated by TLR2 and TLR4, and the cooperation between RAGE and TLRs has also been reported [7]. The HMGB1/RAGE signaling pathway regulates chemokines, cytokines, and adhesion molecules, which ultimately regulate cell proliferation, differentiation, and migration [12][13][14].
Macrophages are generated from monocytes, which have remarkable plasticity that allows them to efficiently respond to environmental signals and alter their phenotype. One phenotype is M2 macrophages, which when present in the tumor microenvironment can promote tumor cell proliferation, invasion, metastasis, and carcinogenesis. Classical M2 macrophages express CD163, CD206, and RAGE [15,16]. Increasing evidence has suggested that HMGB1 and M2 macrophages are involved in lymph node metastasis [7].
Compelling evidence has indicated that the contribution of RAGE to tumor biology is not only its expression on cancer cells but also its specific enhancement of the inflammatory milieu in the tumor microenvironment. Although RAGE protein as a tumor cell receptor has been investigated [10,17], the contribution of HMGB1/RAGE signaling to tumor lymphangiogenesis on M2 macrophages has not yet been investigated, at least to the best of our knowledge.

Patients and Tissue Preparation.
A total of 280 paraffinembedded samples were selected from the Department of Pathology, Beijing Tongren Hospital, Capital Medical University, between November 2009 and June 2011. Three experienced pathologists simultaneously confirmed the diagnosis and graded the hematoxylin-eosin-stained sections when neoplastic according to the 2017 4th Edition of the World Health Organization Classification of Head and Neck Tumors [18]. None of the patients had been treated with radiotherapy or chemotherapy. This study was approved by the Ethics Committee of Beijing Tongren Hospital of Capital Medical University, Beijing, China. As these samples were from samples obtained in the past, exemption from patient consent was obtained by the same ethics committee. The demographic and clinicopathological characteristics of the 280 patients are presented in Table 1. In total, 265 of the cases were male and 15 females, with a mean age of 59.8 years (range, 29-84 years). 101 of all patients had drunk alcohol; smoking index (which was the number of cigarettes per day multiplied by years of smoking) <400 was found in 166 patients. Smoking index ≥ 400 was found in 114 patients. A total of 55 patients had vocal polyp (VP), 102 had laryngeal precursor lesions (LPLs), 51 with low-grade dysplasia and 51 with high-grade dysplasia, and 123 had LSCC. Among the 123 LSCC cases, 10 were TNM stage I, 15 stage II, 55 stage III, and 43 stage IV. Lymph node metastasis was found in 57 cases. Vocal polyp (55) and LPLs (102) do not have TNM stage, lymph node metastasis, and differentiation. Seventy-eight patients died of laryngeal carcinoma. The 5-year survival rate was 63.4%. The overall survival was determined from the time of diagnosis to September 2019.
2.2. Immunohistochemistry. The slides were deparaffinized and dehydrated with descending grades of alcohol wash. Endogenous peroxidase activity was blocked, and antigen retrieval was conducted for 2.5 min in 0.01 mol/l (pH 6.0) trisodium citrate buffer. The slides were incubated with primary antibody to HMGB1 (ab18256, diluted 1 : 1.000), D2-40 (ab77854, diluted 1 : 40), and CD163 (ab182422, prediluted) (all from Abcam, No. 1 Kendall Square, Suite B2304 Cambridge, MA 02139-1517, USA) in an incubator at 37°C for 30 min. After being rinsed with phosphate-buffered saline, the slides were incubated with biotinylated goat anti-mouse/rabbit secondary antibody (ab6789/ab6721, diluted 1 : 1.000, Abcam) at 37°C for 15 min and then with streptavidin-biotin-peroxidase complex (ZSGB-BIO, Beijing, China). Slides processed with phosphate-buffered saline in place of the primary antibody were used as a negative control and esophageal squamous cell carcinoma as a positive control.
According to the published criteria [21], D2-40-positive staining (single endothelial cell or cell clusters) was interpreted as evidence of a lymphatic vessel. The vascular-rich area in peritumoral, intratumoral, and normal tissue was defined, and 10 fields of highly D2-40-positive vessels (hotspots) were counted under a magnification of ×200. The mean value of 10 fields of highly D2-40-positive vessels was calculated as the lymphatic vessel density (LVD). Taking the mean LVD value (5:42 ± 2:49 per 200 fields) as the cutoff, the cases were divided into high LVD and low LVD cases for statistical analysis.  [15]. To obtain a RAGE-THP-1 cell line, the transfection of RAGE shRNA lentiviral particles was used to knock down RAGE gene expression. In brief, the THP-1 cells were resuspended in RPMI-1640 medium with 10% FBS. A total of 1 × 10 3 THP-1 cells were transfected with RAGE shRNA lentiviral particles (sc-36374-V, Santa Cruz Biotechnology, Inc., 10410 Finnell Street Dallas, Texas 75220, USA), the nontargeting control sequence (sc-108080), or copGFP control plasmid (sc-108084) in 96-well plates, as recommended by the manufacturer (Santa Cruz Biotechnology, Inc., 10410 Finnell Street Dallas, Texas 75220, USA). Following 1.5 months of puromycin selection, stable cultures of THP-1 cells with RAGE-targeted knockdown were selected and cloned for the induction of macrophages. The efficiency of RAGE-targeted knockdown in the macrophages was examined by western blot analysis.
2.12. Statistical Analysis. Quantitative data are presented as the means ± SD. Significant differences between the 2 groups were analyzed using a two-sided Student's t-test, and differences between multiple groups were analyzed with a twosided ANOVA with Dunnett's post hoc test. For the overall survival rate, the Kaplan-Meier test with the log rank test was used. The correlation analysis was performed using Spearman's correlation analysis. The data of the HMGB1 expression and CD163 + M2 macrophage density were analyzed using the χ 2 test and Student's t-test, respectively (except for lesion variables). The data of lesion variables were analyzed using ANOVA with Dunnett's post hoc test. Qualitative data were representative of three independent experiments. All analyses were conducted using SPSS20 software. A P value < 0.05 was considered to indicate a statistically significant difference.

Association between the Patient Clinicopathological
Characteristics and HMGB1 Expression and CD163+ M2 Macrophage Density. The HMGB1 expression and clinicopathological variables are presented in Table 2. A high HMGB1 expression was significantly associated with an advanced stage (stages III and IV; P < 0:001) and lymph node metastasis (P < 0:001). Among the LSCC samples with lymph node metastasis, the HMGB1 expression was high in 41 samples (33.4%). Survival was poorer in patients with a high HMGB1 expression in LSCC than in those with a low HMGB1 expression (P < 0:05, Figure 2(a)). On the whole, the results indicated that HMGB1 was overexpressed in the majority of patients and was associated with lymph node metastasis and a poor prognosis.

Classic Markers of M2 Macrophages and M0
Macrophages. The RAGE expression in the THP-1 cells has previously been documented [15] and was confirmed in the present study. Therefore, we also used RAGE shRNA lentiviral particle transfection to knock down RAGE gene expression and generated a RAGE +/-THP-1 cell line, which was used for macrophage differentiation. We achieved M0 macrophages and M2 macrophages by the use of a previously reported protocol [22]. The expression levels of IL-1β, IL-12, IL-10, and CCL-13 were evaluated by RT-qPCR. The protein levels of CD163, CD206, and RAGE were assessed with western blot analysis. CD163 and CD206 were strongly expressed in M2 macrophages; the expression levels of RAGE were low in the M2 macrophages in which RAGE was knocked down (M2 macrophage shRAGE) (Figure 3). The M0-differentiated macrophages expressed higher levels of IL-1β and IL-12 than did the differentiated M2 macrophages. Moreover, the M2 macrophages exhibited a classic pattern as regards the levels of IL-23, TNF-α, TGF-β, and IL-10 measured in the cell culture supernatants (Figure 4). Thus, we successfully obtained classical M2 macrophages and M0 macrophages.

HMGB1 Promotes HDLEC Proliferation, Migration, and Lymphangiogenesis by the Activation of RAGE on M2
Macrophages In Vitro. Lymphangiogenesis is a major biological function of M2 macrophages in promoting tumor progression. The process is described as lymphatic endothelial cell proliferation, migration and remodeling, and preservation of lymphatic vessels. Thus, in order to assess whether the HMGB1-induced activation of RAGE on CD163 + M2 macrophages promotes lymphangiogenesis, we evaluated HDLEC proliferation with CCK-8 assay after incubation with 6 different conditioned media or ECBM (control) in vitro. We observed that HDLEC incubated with conditioned medium from HMGB1-stimulated RAGE + M2 macrophages had the highest proliferation ability (P < 0:05). In addition, when the HDLEC were cultured with conditioned medium from M0 macrophages treated with HMGB1, a greater proliferation of HDLEC was observed compared with that of the controls (P < 0:05). RAGE knockdown decreased the proproliferative effects of HMGB1 preconditioned M2 macrophages on HDLEC (P < 0:05, Figure 5).
To determine whether the activation of RAGE + M2 macrophages by HMGB1 can potentiate the HDLEC migratory activity, we performed a cell migration assay. The results revealed that conditioned medium from the RAGE + M2 macrophages treated with HMGB1 significantly promoted HDLEC migration in comparison with that of the other conditioned media (P < 0:05; Figure 6); however, RAGEtargeted knockdown in M2 macrophages treated with HMGB1 markedly reduced the promigratory effects of the M2 macrophage-conditioned medium (P < 0:05; Figure 6). Moreover, conditioned medium from M0 macrophages stimulated with HMGB1 also enhanced the HDLEC migratory activity compared with that of the control (P < 0:05, Figure 6).
To evaluate lymphangiogenesis, 6 different media or the ECBM control were added to HDLEC placed in Matrigel-coated wells. Lymphangiogenesis was significantly increased when the HDLEC were incubated with medium from RAGE + M2 macrophages treated with HMGB1 (P < 0:05; Figure 7); RAGE knockdown decreased the prolymphangiogenic effects of HMGB1-preconditioned M2 macrophages on the HDLEC (P < 0:05; Figure 7). In addition, conditioned medium from M0 macrophages preconditioned with HMGB1 led to an increase in the lymphatic vessel network compared with that of the control (P < 0:05, Figure 7).
The importance of the HMGB1-induced activation of RAGE on M2 macrophages was further confirmed by treating the M2 macrophages with anti-TLR2 antibody and anti-TLR4 antibody, none of which inhibited the proliferative ability of the HDLEC; the HMGB1-TLR pathway did not significantly increase HDLEC migration on M2 macrophages. Similar results were observed with the migratory ability and lymphangiogenesis (Figure 8).
3.6. VEGF-C in Collection of Conditioned Medium. Cytokines are important in the functionality and phenotypic polarization of macrophages. VEGF-C is a major cytokine involved in the lymphangiogenesis. Thus, in this study, we examined the VEGF-C concentration in 6 different conditioned media and ECBM (control). We found that HMGB1-induced VEGF-C production was significantly higher in RAGE + M2 macrophage-conditioned medium (P < 0:01), and RAGE knockdown decreased the VEGF-C concentration in the conditioned medium from HMGB1 preconditioned M2 macrophages (P < 0:05, Figure 9). Of note, the VEGF-C level in the conditioned medium from M0 macrophages stimulated with HMGB1 was not increased compared with that from the control (P >0.05).

Disease Markers
The importance of the VEGF-C induction after the activation of RAGE on M2 macrophages was further confirmed by the addition of anti-VEGF-C antibody to the conditioned medium. The results revealed that HDLEC incubated with conditioned medium from RAGE + M2 macrophages, preconditioned with HMGB1 and anti-VEGF-C antibody, exhibited a proliferation ability similar to that of the HDLEC incubated with the other conditioned media (Figure 10(a)). Moreover, we observed that the RAGE-targeted knockdown in the M2 macrophages treated with HMGB1 and anti-VEGF-C antibody did not markedly reduce the promigratory effects of the M2 macrophage-conditioned medium (Figures 10(b) and 10(c)). Similar results were observed with lymphangiogenesis (Figures 10(d) and 10(e)). Of note, the HDLEC incubated with conditioned medium from M2 mac-rophages or M0 macrophages treated with HMGB1 exhibited a greater proliferative activity and lymphangiogenesis capacity compared with that of the controls (P < 0:05, Figures 10(a), 10(d), and 10(e)).

Discussion
For many patients with LSCC, early evidence of tumor spread is regional draining lymph node metastasis, which leads to the main cause of cancer-related mortality [1]. Over the past decade, the understanding of the complex molecular mechanisms involved in lymphangiogenesis has markedly improved [25,26]; however, no antilymphangiogenic compounds have yet been approved for use in clinical practice, at least to the best of our knowledge. New or enhanced

10
Disease Markers therapies to block the formation of new lymphatic vessels are urgently required. An increased HMGB1 expression has been reported to be closely associated with the proliferation, invasion, metastasis, and prognosis of tumors [3,4]. Tumor cells secrete cytokines to recruit monocytes infiltrated in the tumor stroma and promote their differentiation and polarization to M2 macrophages [27]. A compelling body of evidence has indicated the involvement of HMGB1 and M2 macrophages in lymphatic metastasis [28,29].
In this study, in immunohistochemistry experiments, we found that HMGB1 and CD163 + M2 macrophage densities were increased with the development of the disease (from VP, LPL to LSCC) and that the HMBG1 expression was significantly higher in LSCC (P < 0:05). Other studies [30,31] have made similar observations in other tumors: In the carcinogenesis of gastric and cervical tumors, the expression level of HMGB1 has been found to be increased in the sequence of epithelial metaplasia-dysplasia-cancer. In addition, HMGB1 protein overexpression has been shown to be involved in lymph node metastasis and to be associated with a poor prognosis of patients with LSCC [32,33], as we have found. It has been previously reported [17] that HMGB1 interaction with RAGE activates the NF-κB/STAT3 pathway, which are molecular effector mechanisms linked to tumor cell proliferation and invasion, in a mouse model of lung cancer, and the blockade of HMGB1 can be targeted to suppress tumor development and metastasis. The results of this study revealed that the density of CD163 + M2 macrophages was significantly higher in tissue samples from

Disease Markers
patients who had a poor prognosis (P < 0:05) or with lymph node metastasis (P < 0:05). Lin et al. [28] reported that increased CD163 + tumor-associated macrophage (TAM) infiltration in LSCC can be a marker of metastasis and prognosis, but the authors did not provide density counts of the infiltration. In this study, we provided the mean density values of CD163 + M2 macrophage infiltration, which is a more objective prediction of the prognosis of patients with LSCC.
In complementary experiments, we found that HMGB1 protein expression and CD163 + M2 macrophage infiltration were positively associated with LVD (P < 0:01). A previous study reported that plentiful lymphangiogenesis was observed in patients with pancreatic cancer with nodal metastasis [34], which was consistent with our findings. Moreover, Kurahara et al. [34] found that the higher LVD was associated with higher densities of CD163 + TAMs in pancreatic tumors, which also was similar to the present findings in LSCC (P < 0:05). Thus, we hypothesized that HMGB1 promotes lymphangiogenesis through the activation of RAGE on M2 macrophages and established an in vitro model to examine this hypothesis. We found that the treatment of RAGE + M2 macrophages with HMGB1 promoted lymphangiogenesis by inducing HDLEC proliferation, migration, and vessel formation, as evidenced by the RAGE-targeted knockdown leading to decrease lymphangiogenesis (P < 0:05).
RAGE is a type I transmembrane protein, with carboxylation N-glycosylation, which promotes the binding of HMGB1 and signal transduction [7,8]. RAGE has a greater affinity than TLR for HMGB1, although the HMGB1 pathway is also mediated by TLR2 and TLR4, and the cooperation between RAGE and TLRs has also been reported [7]. Cells of various types and the pathophysiological context seem to determine this partnership and determine which receptor is dominant [35]. Moreover, in this study, when TLR2 and TLR4 were blocked, the proliferative ability and lymphangiogenesis of HDLEC were not significantly inhibited, although RAGE knockdown decreased the effects of HMGB1 stimulation. Thus, this suggests that the HMGB1-TLR pathway is not crucial for lymphangiogenesis in HDLEC, as TLR2 needs to bind to HMGB1-containing nucleosomes to induce cytokine production, whereas HMGB1 alone does not [36]. Therefore, the contribution of RAGE seems to be predominant in HMGB1-induced prolymphangiogenesis. M2 macrophages produce a number of potent vascular growth factors, such as VEGF-C, which stimulate lymphatic vessel formation [37]. The secretion of VEGF-C by macrophages, in particular, seems to be important for proangiogenesis in tissue repair [38], and decreased VEGF-C secretion in granulation tissue is followed by significantly reduced angiogenesis [39]. The recruitment and abundant infiltration of macrophages and the secretion of VEGF-C stimulate lymphangiogenesis, and this has been proven by a number of studies [40,41]. Suzuki et al. proposed that TGF-β upregulated VEGF-C expression in macrophages, thus enhancing lymphangiogenesis [42]. TNF receptor 1 activation in macrophages by TNF-α has been shown to promote the expression of VEGF-C, which in turn induces VEGFR3 on lymphatic endothelial cells [43]. The data of the present study on the HDLEC lymphangiogenesis assay revealed that RAGE + M2 macrophages preconditioned with HMGB1 produced significant higher levels of VEGF-C, which could promote lymphangiogenesis (P < 0:05). However, anti-VEGF-C antibody decreased the prolymphangiogenic effects of HMGB1-preconditioned RAGE + M2 macrophages on HDLEC. Thus, it is suggested that the secretion of VEGF-C is a key factor for the stimulation of lymphangiogenesis during the process of RAGE activation on M2 macrophages.
The results of this study also demonstrated that although HMGB1-treated RAGE + M0 macrophages exhibited an increase in lymphangiogenesis, lymphangiogenesis was significant less than that of the HMGB1-treated RAGE + M2 macrophages, even though RAGE was equally expressed in the M0 macrophages and M2 macrophages. As RAGE gene has extreme polymorphism, RAGE mRNA is one of the main reprogramming changes in the transcriptome during the macrophage polarization process [44,45]. Polymorphisms may affect gene transcriptional activity and result in differences in binding affinity to ligands. Thus, RAGE on M2 macrophages may have a higher affinity for HMGB1 than do M0 macrophages. Although RAGE activation of classic proinflammatory responses has been studied extensively, the data from the present study illustrate that the macrophage status is important in altering RAGE traditional proinflammatory responses. Moreover, we used a reduced PMA concentration, which may also improve the response of macrophages to HMGB1 during THP-1 cell differentiation to macrophages [22]. The data of this study also revealed that HMGB1-treated M0 macrophages promoted HDLEC proliferation compared with control (P < 0:05), but not in parallel with increments in VEGF-C levels, and HDLEC proliferation did not differ between the cells in