Translocated HMGB3 is involved in papillary thyroid cancer progression by activating cytoplasmic TLR3 and transmembrane TREM1

ABSTRACT

The family of high mobility group box (HMGB) proteins participates in various biological processes including immunity, inflammation, as well as cancer formation and progression. However, its role in thyroid cancer remains to be clarified. We performed quantitative RT-PCR (qRT-PCR), western blot, enzyme-linked immunosorbent, immunohistochemistry, and immunofluorescence assays to evaluate the expression level and subcellular location of HMGB3. The effects of HMGB3 knockdown on malignant biological behaviors of thyroid cancer were determined by cell proliferation assays, cell cycle and apoptosis assays, and transwell chamber migration and invasion assays. Differential expression genes (DEGs) altered by HMGB3 were analyzed using the Ingenuity Pathway Analysis (IPA) and TRRUST v2 database. HMGB3 correlated pathways predicted by bioinformatic analysis were then confirmed using western blot, co-immunoprecipitation, dual-luciferase reporter assay, and flow cytometry. We found that HMGB3 is overexpressed and its downregulation inhibits cell viability, promotes cell apoptosis and cell cycle arrest, and suppresses cell migration and invasion in thyroid cancer. In PTC, both tissue and serum levels of HMGB3 are elevated and are correlated with lymph node metastasis and advanced tumor stage. Mechanistically, we observed the translocation of HMGB3 in PTC, induced at least partially by hypoxia. Cytoplasmic HMGB3 activates nucleic-acid-mediated TLR3/NF-κB signaling and extracellular HMGB3 interacts with the transmembrane TREM1 receptor in PTC. This study demonstrates the oncogenic role of HMGB3 cytoplasmic and extracellular translocation in papillary thyroid cancers; we recommend its future use as a potential circulating biomarker and therapeutic target for PTC.

Introduction

The incidence of thyroid cancer continues to increase, which has been one of the top 10 cancers in the world in terms of new cases for decades [1,2]. Around 95% of thyroid cancers are differentiated thyroid cancer (DTC), including papillary thyroid cancer (PTC) and follicular thyroid cancer (FTC), which has a good overall prognosis with the treatment of surgery and adjuvant radioiodine therapy. However, recurrence of DTCs occurs in 15–30% of cases leading to unfavorable outcomes [3,4]. Tumor immune microenvironment (TIME) has been implicated in DTC progression in light of the correlation of immune infiltration with histopathological aggressiveness [5,6]. Although progress has been made in profiling the immune contexture of thyroid cancer, the molecular connections bridging the network between cancer cells and immune cells remain poorly understood [5,6].

The high mobility group box (HMGB) proteins, including HMGB1, HMGB2, and HMGB3, are a family of nonhistone chromatin-associated proteins. They interact with both DNA and proteins and have multiple functions depending on their subcellular position [7]. In the nucleus, HMGB proteins participate in DNA damage repair, telomere maintenance, and transcriptional regulation via binding to chromosomal DNA and interacting with transcriptional factors [8,9]. In the cytoplasm, these proteins bind to nucleic acids derived from damaged cells, bacteria, or viruses and activate nucleic-acid-sensing receptors participating in immune responses [10]. Cytoplasmic HMGBs are also shown to mediate autophagy modulation, which is important in tumorigenesis [11–14]. Being released into extracellular space, HMGB proteins interact with different receptors regulating various biological processes including immunity [15,16], inflammation [17–19], cell proliferation [20–22], and angiogenesis [23]. However, whether HMGBs participate in thyroid carcinogenesis remains unknown. Here, we found a high expression of HMGB3 in thyroid cancers and demonstrated that translocated HMGB3 exerts an oncogenic role in PTC through activating the cell-surface triggering receptor expressed on myeloid cells 1 (TREM1) receptor and the cytoplasmic toll-like receptor 3 (TLR3)/Nuclear factor-κB (NF-κB) pathway.

Materials and methods

Clinical specimens

PTCs and the corresponding non-neoplastic tissues (n = 38) were obtained from patients who underwent surgery. Patients did not undergo radiotherapy or chemotherapy before the surgery. In addition, preoperative serum samples from 29 PTC patients and 14 blood samples from normal healthy controls were collected. All materials used in this study were approved by the Ethics Committee of the First Affiliated Hospital of Xi’an Jiaotong University.

Cell culture and treatments

The PTC cell lines K1 and BCPAP were grown at 37°C in RPMI 1640 medium (HyClone, Logan, UT) with 10% fetal bovine serum (Gibco, USA). The FTC cell line FTC133 was cultured in DMEM/HAM’S F-12 (HyClone, Logan, UT) medium. And the anaplastic thyroid cancer (ATC) cell line CAL62 was incubated with DMEM high glucose (Gibco, USA).

For hypoxia treatment, K1 cells were cultured in a hypoxic incubator (Bugbox-M, Ruskinn, UK) at the condition of 1% O2, 5% CO2, and 94% N2 for the indicated time. Induction of chemical hypoxia was performed by adding cobalt chloride (CoCl2, 100 μM, Sigma-Aldrich, Cat. #C8661) to the K1 cell culture medium for an indicated time.

For the treatment of exogenous recombinant HMGB3, time- and dose-dependent experiments were designed by treating K1 cell lines with recombinant HMGB3 (Signalway Antibody, Cat. #AP75401) at different concentrations (0, 10, and 50 nM) for different time periods (4, 12, and 24 h).

Lentiviral vector transduction

Lentiviral particles encoding small hairpin RNA (shRNA) targeting HMGB3 (sh-HMGB3, target sequence: 5”-CGTAATTGACACATCTCTT-3‘) and control shRNA (sh-Control, target sequence: 5’-TTCTCCGAACGTGTCACGT-3”) were obtained from GeneChem (Shanghai, China). Cells at a density of 60% were transduced with 10 multiplicity of infection (MOI) of the viral particles. To obtain maximal transduction efficiency, enhanced Infection solution (ENi.S., GeneChem, Cat. REVG0002) and polybrene (GeneChem, Cat. REVG0001) were used according to the manufacturer’s instructions. The expression of HMGB3 after 72 h of transduction was detected by qRT-PCR and western blot.

Inhibition of TREM1 receptor by LP17 peptide

The LP17 peptide (LQVTDSGLYRCVIYHPP), a synthetic TREM1 inhibitor, was synthesized by BankPeptide Biological Technology (Hefei, China). To verify whether extracellular HMGB3 stimulates TREM1 receptors, recombinant HMGB3 protein (50 nM) in combination with LP17 peptide (100 ng/ml) was added to K1 cell lines for 24 h. Whole-cell extracts were then prepared for western blot and qRT-PCR.

Stimulation of TLR3/NF-κB signaling by Poly(I:C)

As a substitute for viral dsRNA, TLR3 agonist Poly(I:C) (InvivoGen, Cat. 31852-29-6) has been shown to activate the transcription factor NF-κB through the TLR3 receptor [24,25]. To determine whether HMGB3 activates NF-κB pathway through the TLR3 receptor, K1 cells were challenged with 10 μg/ml TLR3 agonist Poly(I:C) for 12 h after sh-HMGB3 or sh-Control transduction. The effects of Poly(I:C) on the nuclear translocation of NF-κB P65 and the transcriptional expression of its downstream cytokines were detected by western blot and qRT-PCR, respectively.

Cell malignant biological behavior assays

Detailed procedures for cell proliferation analysis, cell cycle and apoptosis assays, transwell chamber migration, and invasion assays are described in the Supplementary Methods.

Quantitative RT-PCR (qRT-PCR)

qRT-PCR analyses were performed as previously described and 18S served as the parallel controls [26]. The primer sequences used in the study are listed in Supplemental Table S1.

Protein extraction and western blot analysis

Total proteins, nuclear proteins, and cytoplasmic proteins were extracted, respectively, according to the methods described previously [27,28]. The primary antibodies against HMGB3, NF-κB p65, TREM1, TLR3, AKT (A444), p-AKT (S473), p-AKT (T308), p-Erk1/2, Erk1/2, GAPDH, and PCNA were used for western blot. The catalog numbers of the above antibodies are shown in Supplemental Table S2. Western Bright ECL detection system (Advansta, Menlo Park, CA) was used to detect the immunoblotting signals.

Dual-luciferase reporter assay

The pNF-κB-TA-Luc firefly plasmid (Cat. P0456) and pGL4.74 [hRluc-TK] renilla plasmid (Cat. P0873) were purchased from MiaoLing Plasmid Platform. The Dual-Luciferase® Reporter Assay System (Promega) was used in our study. Briefly, K1 cells transduced with sh-Control or sh-HMGB3 were transfected with pNF-κB-TA-Luc firefly plasmid and pGL4.74 [hRluc-TK] renilla plasmid. Cells were then lysed in a passive lysis buffer (Promega) after 48 h of transfection. The reporter activity was normalized using the firefly luciferase value divided by the renilla luciferase value. Three replicates were done for each sample in luciferase assays.

Co-immunoprecipitation

To determine whether HMGB3 interacts with TLR3, lysis solutions of K1 cells were incubated with either anti-HMGB3 polyclonal rabbit antibody (dilution 1:200, Cat. ab75782, Abcam) or recombinant rabbit IgG monoclonal antibody (dilution 1:200, Cat. EPR25A, Abcam) for negative control overnight at 4°C. Ten microliters of Protein A/G PLUS-Agarose beads (sc-2003, Santa Cruz) was added to the cell lysate overnight, which were then mixed slowly with a rotator at 4°C for 3 h. The supernatants of the cell lysate were removed by centrifugation at 4,000 rpm for 3 min after washing thrice with ice-cold RIPA buffer. To separate the proteins from the beads, the immunoblotting loading buffer was added to the cell lysate for 5 min at 95°C. The supernatants were then collected for subsequent immunoblotting analysis.

Enzyme-linked immunosorbent assay

The serum HMGB3 levels of 29 PTC patients and 14 healthy controls were determined using HMGB3 ELISA kit (SED773Hu, Cloud-Clone Corp) according to the manufacturer’s instructions. During each assay, blank holes, standard holes, and sample holes were set up, respectively. One hundred microliters of standard solution or sample solution was added to each hole. All experiments were performed in triplicate.

Flow cytometry

To detect the expression of TREM1 receptor on the surface of PTC cell lines, K1 and BCPAP were stained with TREM1 antibody (proteintech 11,791–1-AP, 1:200) or isotype anti-rabbit IgG antibody (Abcam, ab172730, 1:1000) for 60 min at 4°C in FACS buffer. After washing twice with pre-cooled PBS, cells were then stained with the goat anti-rabbit IgG (H+L)-PE (Cat. Ab72465, Abcam) pre-adsorbed secondary antibody for 60 min at 4°C.

To detect the cytoplasmic expression of TLR3, permeabilized and non-permeabilized K1 cells were stained with TLR3 antibody (Santa Cruz, Cat. sc-32,232, 1:20) for 60 min at 4°C. A FACScan flow cytometer (BD Biosciences) was used for cytometric analysis.

Immunohistochemistry and immunofluorescence

The expression and subcellular localization of the HMGB3 protein in PTC tissues (n = 28) were determined by immunohistochemistry and immunofluorescence. For immunohistochemistry, the tissue sections were incubated with anti-HMGB3 (dilution 1:100, Cat. ab75782, Abcam), and protein expression levels were evaluated using the immunoreactive score (IRS) [29]. For immunofluorescence, the tissue sections were incubated with primary antibodies against anti-HMGB3 (dilution 1:50, Cat. ab75782, Abcam) or HIF1α (dilution 1:400, Cat. #36169, Cell Signaling Technology) overnight at 4°C. After washing thrice with PBS, incubate the sections with FITC or Cy3-conjugated secondary antibodies for 1 h at room temperature in the dark. DAPI was then used for counterstaining, and a fluorescence microscope was used for final analysis.

mRNA microarray procedures

To identify differentially expressed genes (DEGs) altered by HMGB3, the total RNA of K1 cells transduced with sh-HMGB3 or sh-Control was extracted. RNA integrity and purity were assessed by Nanodrop 2000 (Thermo Scientific, Shanghai, China) and Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA). Samples with RNA integrity number (RIN) ≥7 were performed for downstream analyses according to GeneChip® Expression Analysis technical manual. Each chip was scanned by Affymetrix GeneChip Scanner 3000 7 G (Affymetrix, Santa Clara, CA). DEGs were defined as adjusted p-value <0.05 and absolute fold change ≥1.5.

Bioinformatics analysis

Meta-analysis of HMGB DNA microarray data was performed using Oncomine database [30] (Compendia Bioscience, Ann Arbor, Michigan, USA) with the primary filters for “thyroid gland carcinoma” and “cancer vs. normal analysis”. To ensure the reliability of the screening results, we chose “Human Genome U133 or U133 Plus 2.0 Array platform” as the screening condition. Eight analyses with 111 specimens including 24 normal thyroid tissues and 101 thyroid cancer tissues were enrolled in the analysis.

The Cancer Genome Atlas (TCGA) database covers molecular characterization data of matched normal samples for more than 20,000 primary cancers and 33 cancer types. TCGA Thyroid Cancer (THCA) gene expression RNAseq data (n = 572) were downloaded for analysis of HMGB3 gene mRNA expression level. The Kaplan–Meier (KM) plotter (https://kmplot.com/analysis/) was used to analyze the correlation between HMGB3 expression and patients’ recurrence-free survival (RFS) in PTC (n = 502). Stage I-II (low stage) or stage III-IV (high stage) PTC patients with high and low HMGB3 expression were split by auto-select best cutoff. Hazard ratios (HR) with 95% confidence intervals and logrank p-value were determined.

The list of DEGs was imported into the Ingenuity Pathway Analysis (IPA) software (Ingenuity® Systems, CA) to relate changes to potential molecular signal networks [31]. The TRRUST database (https://www.grnpedia.org/trrust/, version 2.0) was used to predict the transcriptional factors regulating DEGs. STRING database (https://string-db.org/, version 11.0) was used to identify proteins interacting with HMGB3. Confidence score was set at 0.4.

Statistical analysis

The SPSS statistical package (20.0, Chicago, IL) was used for statistical analysis. Statistical significance was defined as p-value <0.05. All values were presented as mean±SD. The Student’s t-test and the Mann–Whitney U test were used to compare the normal distribution or non-normal distribution of continuous variables between the two groups, respectively. The relationships between HMGB3 expression and the clinicopathological characteristics were analyzed by Fisher’s exact test.

Results

HMGB3 is upregulated in thyroid cancer

To determine whether the transcriptional levels of HMGBs in thyroid cancers were altered, a meta-analysis of publicly available transcriptomic profiling data using the Oncomine database was performed. As shown in Figure 1(a), HMGB3 was overexpressed in 6 of the 8 datasets of thyroid cancer tissues compared with noncancerous tissues (gene median rank 899.0, p = 0.003); whereas up-regulation of HMGB1 and HMGB2 were only observed in 3 and 2 of the datasets, respectively.

Figure 1.

HMGB3 is upregulated in thyroid cancers. (a) Meta-analysis of gene expression profiling for HMGB1, HMGB2, and HMGB3 using oncomine database. Tissue type: 1. Follicular variant PTC vs. Normal. Giordano Thyroid; 2. Tall cell variant PTC vs. Normal. Giordano Thyroid; 3. FTC vs. Normal. Giordano Thyroid; 4. Oncocytic FTC vs. Normal. Giordano Thyroid; 5. PTC vs. Normal. Giordano Thyroid; 6. ATC vs. Normal. Giordano Thyroid; 7. PTC vs. Normal. he Thyroid; 8. PTC vs. Normal. Vasko Thyroid. (b) Relative mRNA expression levels of HMGB3 in PTCs compared with matched non-malignant thyroid tissues by qRT-PCR (n = 38). (c) Western blot of HMGB3 in PTCs (T, n = 38) and adjacent normal tissues (N, n = 38). (d) Kaplan–Meier survival curves for recurrence-free survival by the median value of HMGB3 expression in high stage (stage III-IV) PTC. (e) Representative images of HMGB3 staining in PTCs (T) and adjacent normal thyroid tissues (N) by immunohistochemistry (left). Scatter plot of HMGB3 IHC score in PTCs (T, n = 28) and adjacent normal thyroid tissues (T, n = 28) (right). (f-g) Representative images and scatter plot of HMGB3 immunohistochemical staining in PTC low clinical stage (stage I and II, n = 16) and high clinical stage (stage III and IV, n = 12). (h) the serum level of HMGB3 in PTC patients (T, n = 29) and normal healthy people (N, n = 14) by ELISA. (i) The serum level of HMGB3 in PTC patients without lymph node metastasis (N0, n = 14) and PTC with lymph node metastasis (N1, n = 15) by ELISA. Data are presented as mean ± SD and statistically significant differences are indicated as follows: ***p < 0.001, **p < 0.01, *p < 0.05.

To verify the high expression of HMGB3, we first collected 38 primary PTC samples (T) and their matched non-cancerous (N) thyroid tissues and performed qRT-PCR analysis. As presented in Figure 1(b), a significantly higher expression of HMGB3 was observed in PTC samples compared with non-cancerous samples (2.52 ± 0.20 vs. 1.89 ± 0.12, p < 0.01). Western blot results also found that the content of HMGB3 protein in PTC tissues (T) was increased compared with that in their corresponding adjacent normal tissues (N) (Figure 1(c), p < 0.001). To evaluate the clinical significance of high HMGB3 expression in PTC, the KM survival curves were used to analyze the correlation between HMGB3 expression and patient’s RFS in PTC. Although the KM curves were not affected by HMGB3 expression in stage I-II (low stage) PTC (Supplemental Figure S1a), high HMGB3 expression was identified to be connected with poor RFS in stage III-IV (high stage) PTC (Figure 1(d), p < 0.05).

The expression of HMGB3 in paraffin sections of 28 PTC patients was further determined by immunohistochemistry, and its correlation with clinical features was analyzed. As shown in Figure 1(e), the HMGB3 expression level was significantly elevated in PTC tissues (T), compared with that in adjacent normal thyroid tissues (N) (3.46 ± 2.37 vs. 1.32 ± 1.47, p < 0.001). In addition, the IHC score of HMGB3 expression in stage III-IV (high stage, n = 12) PTC was significantly higher than that in stage I-II (low stage, n = 16) PTC (5.08 ± 2.58 vs. 2.25 ± 1.24, p<0.01; Figure 1(f-g)). To access the association of HMBG3 expression with clinicopathological characteristics, we divided the 28 paired samples into a high-expression group (n = 7) and a low-expression group (n = 21) according to the median of HMBG3 expression. As shown in Table 1, HMGB3 expression was not correlated with gender, age, and tumor invasion, while its high expression was correlated with lymph node metastasis (p = 0.016) and tumor stage (p = 0.001).

Table 1.

The correlation of HMGB3 expression with clinicopathological characteristics of PTC.

Variable		HMGB3 expression
	n	Low level	High level	p-value
Age
≤55	24	18	6	0.747
>55	4	3	1
Gender
Male	6	5	1	0.522
Female	22	16	6
Invasion
Yes	1	1	0	0.750
No	27	20	7
Lymph node metastasis
Yes	17	10	7	0.016
No	11	11	0
Clinical stages
I, II	16	16	0	0.001
III, IV	12	5	7

Furthermore, we collected serum samples of 29 PTC patients and 14 healthy controls and detected the free HMGB3 protein level in serum by ELISA. As presented in Figure 1(h), the serum level of HMGB3 in PTC patients (T) was significantly elevated than that in healthy controls (N; 0.42 ± 0.39 vs. 0.18 ± 0.14 ng/mL, p < 0.05). Interestingly, the serum level of HMGB3 in PTC patients with lymph node metastasis (N1, n = 15) was higher than that in those without (N0, n = 14; 0.50 ± 0.40vs. 0.34 ± 0.37 ng/mL, p < 0.05; Figure 1(i)).

HMGB3 knockdown inhibits malignant biological behaviors of thyroid cancer cells

To evaluate the role of HMGB3 in thyroid cancer cells, loss-of-function experiments using PTC cell lines (K1 and BCPAP), FTC cell line (FTC133), and ATC cell line (CAL62) transduced with sh-HMGB3 or sh-Control were carried out. The efficiency of HMGB3 knockdown was verified by qRT-PCR and western blot (Supplemental Figure S1b-c). Cell proliferation analysis using Celigo Imaging Cytometer was first performed, and a remarkable inhibition of cell proliferation was observed in the HMGB3 knockdown group, in comparison with the control group (Figure 2(a)). Also, we determined the effect of HMGB3 down-regulation on cell apoptosis. The fraction of apoptotic cells was significantly expanded in HMGB3 knockdown cells compared to that in the control cells (Figure 2(b), K1: 4.34 ± 0.17% vs. 8.28 ± 0.54%, p < 0.001; BCPAP: 3.73 ± 0.25% vs. 5.70 ± 0.50%, p < 0.001; FTC133: 7.97 ± 0.12% vs. 10.13 ± 0.25%, p < 0.001; CAL62: 7.13 ± 0.15% vs. 8.37 ± 0.32%, p < 0.01). Next, we evaluated its effect on cell cycle phases and found G1 or S phase arrest with HMGB3 down-regulation in thyroid cancer cell lines (Figure 2(c), K1 at G1 phase: 45.29 ± 0.48% vs. 52.90 ± 0.95%, p < 0.001; K1 at S phase: 34.4 ± 0.8 to 26.1 ± 1.3%, p < 0.001; BCPAP at S phase: 42.95 ± 0.52% vs. 46.85 ± 0.47%, p < 0.01; FTC133 at G1 phase: 47.41 ± 1.49% vs. 51.53 ± 0.92%, p < 0.01; CAL62 at G1 phase: 47.15 ± 1.19% vs. 51.89 ± 0.70%, p < 0.05). In addition, HMGB3 knockdown significantly inhibited migration in K1, BCPAP, FTC133, and CAL62 (Figure 2(d), K1: 164.75 ± 9.81 vs. 94.75 ± 7.93, p < 0.001; BCPAP: 216.25 ± 8.50 vs. 85.00 ± 4.76, p < 0.001; FTC133: 119.50 ± 7.05 vs. 56.25 ± 2.63, p < 0.001; CAL62: 345.25 ± 13.84 vs. 172.75 ± 5.25, p < 0.001) and invasion in BCPAP, FTC133, and CAL62 (Figure 2(e), BCPAP: 205.50 ± 4.20 vs. 75.25 ± 8.54, p < 0.001; FTC133: 94.00 ± 4.97 vs. 56.50 ± 3.87, p < 0.001; CAL62: 462.75 ± 25.64 vs. 102.25 ± 9.07, p < 0.001). These results indicate that HMGB3 down-regulation inhibits proliferation, migration, and invasion ability of thyroid cancer cells and promotes G1/S cell cycle arrest and apoptosis, suggesting an oncogenic role in thyroid cancer.

Figure 2.

Knockdown of HMGB3 inhibits malignant biological behaviors of thyroid cancer cells. (a) Cell proliferation analysis of thyroid cancer cell lines (K1, BCPAP, FTC133, CAL62) transduced with sh-control or sh-HMGB3 using celigo imaging cytometer. Cell images were captured twice a day for 8 days using a fluorescence microscope. (b) Cell apoptosis analysis of thyroid cancer cell lines transduced with sh-control or sh-HMGB3 for 72 h using Annexin V-FITC/7-AAD assay. (c) Cell cycle analysis of K1, BCPAP, FTC133, and CAL62 cell lines transduced with sh-control or sh-HMGB3 by flow cytometry. Cell population at each phase were indicated. (d-e) migration (d) and invasion (e) abilities of K1, BCPAP, FTC133, and CAL62 cell lines transduced with sh-control or sh-HMGB3 using transwell chamber. Data were presented as mean ± SD of three independent experiments. Statistically significant differences were indicated as follows: ***p < 0.001, **p < 0.01, *p < 0.05.

HMGB3 translocation and its association with hypoxia in PTC

The biological functions of HMGBs are closely related to their cellular localization. By immunofluorescence, we found that HMGB3 was mainly distributed in the nucleus in normal thyroid tissues (N), with a high overlap rate with the nuclear stain DAPI (Figure 3(a), left). While in PTC tissues (T), HMGB3 rarely overlapped with the nuclear stain DAPI and was mainly distributed in the cytoplasm (Figure 3(a), right). Meanwhile, the distribution of hypoxia-inducible factor HIF1α, which was significantly elevated in PTC, overlapped with that of HMGB3 (Figure 3(a)). Consistently, cBioPortal database analysis found that HMGB3 mRNA expression in PTC was positively correlated with three tumor mRNA-based hypoxia scores (Buffa, Winter, and Ragnum scores, Figure 3(b)).

Figure 3.

HMGB3 translocation and its association with hypoxia in PTC. (a) The expression of HMGB3 and HIF1α in PTCs and normal thyroid tissues by Immunofluorescence. In normal thyroid tissue (N), HMGB3 (red fluorescence) was mainly expressed in the nucleus (blue fluorescence) and the expression level of HIF1α (green fluorescence) was low. In PTCs (T), HMGB3 was mainly distributed in the cytoplasm and overlapped with HIF1α that was higher than that in normal thyroid tissues. Magnification: 2000×; scale bar: 10 μM. (b) Relationship between HMGB3 mRNA expression and hypoxia scores (Buffa, Winter, and Ragnum) in well-differentiated thyroid cancer. (c-e) the effects of hypoxia on the expression and translocation of HMGB3 in K1 cells. After cobalt chloride (CoCl2) induction or hypoxia incubator induction for 0 h, 6 h, 12 h, 18 h, and 24 h, the protein changes of HMGB3 in total cells (c) and the cytoplasm of the cells(d) were detected by Western Blot. Changes of extracellular HMGB3 (e) after hypoxia induction were detected by ELISA. Data were presented as mean ± SD of three independent experiments. Statistically significant differences were indicated as follows: ***p < 0.001.

To determine whether hypoxia affects the expression and translocation of HMGB3 in thyroid cancer cells, K1 cells with relatively high expression of HMGB3 (Supplemental Figure S1d) were induced by hypoxic incubator and cobalt chloride. As presented in Figure 3(c), the total protein level of HMGB3 in the K1 cell line did not change significantly under hypoxia and normoxia, while the cytoplasmic protein level of HMGB3 increased gradually with the time of cobalt chloride-induction (CoCL2) and hypoxic incubator induction (Figure 3(d)). Furthermore, the release of HMGB3 in the cell supernatant was significantly elevated after hypoxia inductions for 24 h by ELISA (Figure 3e).

DEGs altered by HMGB3 knockdown in PTC cells and the ingenuity pathway analysis

To identify the transcriptomic alterations governed by HMGB3 in thyroid cancer, we used microarray expression profiles of K1 cells transduced with sh-HMGB3 and sh-Control. As described above, DEGs were identified using the criteria of fold change ≥1.5 and p-value <0.05. We found 824 (521 down-regulated and 303 up-regulated) genes that were differentially expressed upon HMGB3 down-regulation. The heat map of the top 50 DEGs is shown in Figure 4(a), and a list of all DEGs is presented in Supplemental Tables S3 and S4.

Figure 4.

DEGs altered by HMGB3 knockdown in PTC cells and the ingenuity pathway analysis (IPA). (a) Heatmap of the top 50 DEGs in K1 cells transduced with sh-control or sh-HMGB3. DEGs were identified by using at least a 1.5-fold change and a p < 0.05 threshold. (b) Top 20 diseases and functions annotation of the identified DEGs altered by HMGB3 knockdown by IPA. (c) The top 7 canonical pathways enriched for HMGB3-related DEGs by IPA.

Diseases or function enrichment analysis by IPA was performed on all the DEGs. Knockdown of HMGB3 activated cell death, apoptosis, and necrosis, while inhibited proliferation, cell transformation, migration, and angiogenesis (Figure 4(b)), consistent with its tumor promoting role. In addition, the migration of granulocytes and phagocytes, and the mobilization of leukocytes and phagocytes were also significantly inhibited. These data suggest that HMGB3 not only affects tumor cell malignant behaviors in PTC but may also play a role in the migration and mobilization of tumor-infiltrating immune cells.

We next used IPA to determine the most significant biological pathways related to HMGB3. Ninety-five canonical pathways connected with cell growth, cell cycle regulation, apoptosis, and immune response were identified with a p-value <0.05 (Supplemental Table S5). The top seven pathways with a |Z-score|>2.0 is shown in Figure 4(c). It is worth noting that TREM1 signaling, which compromises tumor immunity by enhancing Treg cells infiltration in PTC according to our previous study [32], was the top down-regulated canonical pathway after HMGB3 knockdown (Z-score = −2.646, p = 0.012).

To identify the candidate downstream effectors of HMGB3 in PTC progression, we performed IPA regulator analysis of the DEGs. Sixty-six activated and 122 inhibited regulators covering all molecular types, including transcriptional factors, cytokines, microRNAs, and receptors, were picked out by a threshold of |Z-score|>2.0 and p-value <0.05 (Supplemental Table S6). The top 20 activated and inhibited regulators are shown in Table 2. Transcription factors NF-κB, TBX2, and cytokines IL1A and IL1B were predicted to be significantly inhibited after HMGB3 knockdown; transcription factors HNF4A, NUPR1, and CDKN2A, miR-124-3p, and let-7 were predicted to be significantly activated after HMGB3 down-regulated.

Table 2.

Top 20 activated and inhibited regulators analyzed by IPA of DEGs altered by HMGB3 knockdown.

Upstream Regulator	Predicted State	Z-score	p-value of overlap
HNF4A	Activated	3.968	0.000071100
CLDN7	Activated	3.963	0.000003760
Let-7	Activated	3.464	0.000003160
HMOX1	Activated	3.185	0.001460000
Epigallocatechin-gallate	Activated	3.075	0.000757000
Calcitriol	Activated	3.009	0.000000000
NUPR1	Activated	2.967	0.000000307
Nifedipine	Activated	2.778	0.001010000
miR-124-3p (and other miRNAs w/seed AAGGCAC)	Activated	2.769	0.029700000
CDKN2A	Activated	2.735	0.001190000
TBX2	Inhibited	−3.051	0.000001630
Salmonella minnesota R595 lipopolysaccharides	Inhibited	−3.067	0.000000154
TRADD	Inhibited	−3.148	0.000000153
Peptidoglycan	Inhibited	−3.229	0.001260000
Cocaine	Inhibited	−3.234	0.067800000
MYD88	Inhibited	−3.304	0.000251000
E. coli B4 lipopolysaccharide	Inhibited	−3.411	0.000090800
IL1A	Inhibited	−3.515	0.000000517
NFkB (complex)	Inhibited	−3.548	0.000000000
IL1B	Inhibited	−3.988	0.000000000

HMGB3 activates cell-surface TREM1 receptor on PTC cells

Our previous findings found that the TREM1 receptor was mainly expressed on malignant epithelial cells in PTC tissues, instead of myeloid cells [32]. In the current study, we first used flow cytometry to detect the location of TREM1 receptor on PTC cell lines. As shown in Figure 5(a), the TREM1 receptor was detected on the surface of K1 and BCPAP cells, consistent with our previous findings.

Figure 5.

HMGB3 activates cell-surface TREM1 receptor on PTC cells. (a) Detection of TREM1 expression on the surface of K1 and BCPAP cells by flow cytometry. (b-c) The effects of HMGB3 knockdown on the expression of TREM1 downstream signals, including p-AKT and p-Erk1/2 by Western Blot (b) and cytokines by qRT-PCR (c). (d) The effects of extracellular HMGB3 on TREM1 signaling were determined by the expression of p-AKT (T308), p-AKT (S473), and p-ERK1/2 in K1 cells treated with recombinant HMGB3 at the concentrations of 0 nM, 10 nM, and 50 nM (left). To verify the activation of TREM1 receptor by extracellular HMGB3, K1 stimulated by recombinant HMGB3 were then treated with TREM1 antagonist LP17 (10 μM) for 24 h and changes of TREM1 downstream signals were detected by Western blot. ***p < 0.001.

To assess the relationship between HMGB3 and TREM1 signaling, the effects of HMGB3 knockdown on TREM1 downstream signals (p-AKT and p-Erk1/2) and downstream cytokines (TNF, IL1B, IL8, CXCL2, CXCL3, CCL2, and CCL20) were determined in K1 cells [33,34]. As shown in Figures 5(b,c), the phosphorylation levels of AKT (T308), AKT (S473), and ERK1/2 and the mRNA expression levels of TNF, IL1B, IL8, CXCL2, CXCL3, CCL2, and CCL20 dropped significantly after HMGB3 downregulation.

Next, to test the effects of extracellular HMGB3 on TREM1 signaling, we add human recombinant HMGB3 protein at concentrations of 0 nM, 10 nM, and 50 nM to K1 cell medium for 24 h. As shown in Figure 5(d), 10 nM and 50 nM recombinant HMGB3 protein significantly stimulated the expression of p-AKT (T308), p-AKT (S473), and p-ERK1/2. To further verify the activation of TREM1 receptor on PTC cells by extracellular HMGB3, we used the TREM1 antagonist LP17 polypeptide (10 μM) to block the TREM1 pathway. As a soluble TREM1 (sTREM1) mimetic, LP17 functioned as a decoy receptor to bind with TREM1 ligands, preventing the ligands from binding to TREM1 on the membrane surface, thereby blocking the TREM1 pathway [35]. As shown in Figure 6(d), LP17 inhibited the activation of p-AKT (T308), p-AKT (S473), and p-ERK1/2 in K1 cells by exogenous recombinant HMGB3. These results suggest that extracellular HMGB3 activates the TREM1 receptor expressed on the surface of PTC cells.

Figure 6.

HMGB3 activates cytoplasmic TLR3/NF-κB signaling in PTC cells (a) proteins known or predicted to be interacted with HMGB3 by STRING database retrieved with a confidence score > 0.4. (b) The effect of HMGB3 downregulation on NF-κB P65 nuclear translocation in K1 cells by western blot. (c) The effect of HMGB3 downregulation on NF-κB signaling in K1 cells by luciferase assay. (d) The nuclear expression of NF-κB P65 was detected by western blot after the addition of recombinant human HMG3 protein (10 nM or 50 nM) to K1 cells for the indicated time. (e) The STRING database predicted the top 20 proteins and networks that might interact with HMGB3 protein. (f) Expression of TLR3 receptor on the surface of papillary thyroid cancer K1 cell lines by flow cytometry. (g-h) The TLR3 agonist poly(I:C) was added to K1 cells transduced with sh-control or sh-HMGB3. Western blot (g) was performed to detect the expression of NF-κB P65 in nuclear proteins and TLR3 in total proteins. qRT-qPCR (h) was performed to analyze the transcriptional expression of its downstream cytokines. (i) The interaction of HMGB3 with TLR3 was detected by co-immunoprecipitation (co-IP). ***p < 0.001, **p < 0.01.

HMGB3 activates cytoplasmic TLR3/NF-κB pathway in PTC cells

In IPA regulator analysis of the DEGs altered by HMGB3, transcription factor NF-κB was predicted to be significantly inhibited after HMGB3 knockdown (Table 2; Z-score = −3.548, p < 0.001). To verify the association, we used TRRUST v2 to predict the transcriptional factors regulating the top 500 DGEs. The list of transcriptional factors was intersected with the DGEs to build a network using Cytoscape 3.7.2. The interactions showed that NF-κB was a key transcriptional factor regulating the DEGs altered by HMGB3 (Figure 6(a)).

To further confirm the results of the bioinformatics analysis, we performed a western blot of NF-κB expression on K1 cells with HMGB3 knockdown. As shown in Figure 6(b), NF-κB p65 in the nucleus decreased significantly after the HMGB3 knockdown (p < 0.01). However, there was no significant change in the cytoplasmic level of NF-κB p65. The luciferase study also showed that HMGB3 knockdown significantly downregulated NF-κB p65 promoter-mediated firefly luciferase activity in K1 cells (Figure 6(c), p < 0.001). To determine whether the extracellular HMGB3 plays a role in its association with NF-κB signaling, we treated the K1 cells with exogenous HMGB3 at concentrations of 10 nM or 50 nM for 4, 12, and 24 h. However, no significant effect on the nuclear expression of NF-κB P65 was observed with the treatment of exogenous HMGB3 (Figure 6(d)). These results indicate that NF-κB regulated by intracellular but not by extracellular, HMGB3 in PTC cells.

To further study how HMGB3 regulates NF-κB, protein–protein interaction network analysis was performed on the STRING database. As shown in Figure 6(e), 20 proteins known or predicted to be interacted with HMGB3 were retrieved with a confidence score >0.4. Among these proteins, TLR3, a known NF-κB activator [36], has been shown to be upregulated and related to the poor prognosis in PTC [37–39]. Using flow cytometry, we found that the TLR3 staining signal was rarely detected on the surface of K1 cells. After permeabilizing K1 cells, the staining signal of the TLR3 antibody was significantly enhanced compared to that of the isotype control antibody. These results demonstrate that TLR3 receptors are mainly distributed intracellularly in PTC cells (Figure 6f).

We thus performed western blot and qRT-PCR for TLR3/NF-κB signaling in K1 cells transduced with sh-HMGB3 and sh-Control. As shown in Figure 6(g), with the treatment of Poly(I:C), NF-κB p65 expression was significantly increased in K1 cells transduced with sh-Control, indicating an activation of TLR3/NF-κB signaling by Poly(I:C). However, K1 cells transduced with sh-HMGB3 inhibited the expression of NF-κB p65 and TLR3 protein level with or without the stimulation of Poly(I:C). Consistently, transcriptional levels of RELA (NF-κB p65) and genes downstream of NF-κB, including CCL2, CCND1, MYC, CXCL1, IL-1, TNF, and Bcl-2 were significantly decreased after HMGB3 knockdown, even with the stimulation of Poly(I:C) (Figure 6(h)). Furthermore, Co-IP found that HMGB3 protein was precipitated by HMGB3 antibody and TLR3 protein was also precipitated, confirming that HMGB3 can interact directly with TLR3 (Figure 6(i)). These results indicate that HMGB3 regulates NF-κB signaling through the cytoplasmic nucleic-acid-mediated TLR3 pathway.

Discussion

The family of HMGB proteins has been shown to participate in various biological processes of cancer formation and progression, including proliferation, angiogenesis, metastasis, as well as drug resistance [40]. However, its role in the tumorigenesis and progression of thyroid cancer remains unclear. We have found that HMGB3, instead of HMGB1 and HMGB2, is overexpressed and exerts an oncogenic role in thyroid cancer. In addition, it is relocated to the cytoplasm and extracellular milieu in PTC, and this relocation is induced at least partially by hypoxia. To the best of our knowledge, this is the first time that HMGB3 translocation has been described. We thus provide evidence that the cytoplasmic HMGB3 activates nucleic-acid mediated TLR3/NF-κB signaling and that the extracellular HMGB3 interacts with the transmembrane TREM1 receptor in PTC (Figure 7). TLR3/NF-κB pathway is known to regulate inflammatory responses that are linked to tumorigenesis [41]. TREM1 signaling has been shown to promote an immunosuppressive microenvironment by enhancing the infiltration of regulatory T (Treg) cells in PTC according to our previous study [32]. Thus, HMGB3 translocation with its activation of TLR3/NF-κB and TREM1 signaling may serve as one of the molecular mechanisms contributing to the cross-talk between cancer cells and the immune system.

Figure 7.

Schematic model of extracellular and cytoplasmic HMGB3 in PTC progression. cytoplasmic HMGB3 activates nucleic-acid mediated TLR3/NF-κB signaling, leading to the nuclear translocation of NF-κB p65. Extracellular HMGB3 interacts with transmembrane TREM1 receptors expressed on PTC cells, leading to the activation of its downstream signals, including p-AKT and p-ERK, that promote proliferation and migration of cancer cells.

The family members of HMGB proteins share 80% homology in their amino acid sequence, containing a unique DNA-binding domain, the HMG-box [42]. HMGB1 and HMGB2 are ubiquitously expressed, while HGMB3 is mainly restricted in embryos and hematopoietic stem cells [42]. In thyroid cancer, HMGB1 has been shown to increase the expression of oncogenic miR-221 and miR-222 in PTC [43] and ATC [44]. HMGB1-mediated autophagy was involved in vemurafenib resistance [45] and regulated sodium/iodide symporter protein degradation in thyroid cancer cells [14]. Compared with HMGB1, HMGB2 [46] and HMGB3 [47] are relatively less well studied. Here, we found that transcriptional expression of HMGB3, not of HMGB1 and HMGB2, in thyroid cancer was remarkably elevated compared to that in normal tissues by Oncomine meta-analysis. HMGB3 knockdown significantly decreases cell viability, promotes cell apoptosis and cell cycle arrest, and inhibits cell migration and invasion in thyroid cancer cells. This indicates an oncogenic role of HMGB3 in thyroid cancer, which is consistent with a recent study showing that ubiquitinated upregulation of HMGB3 fosters thyroid cancer progression [47]. We also observed that HMGB3 high expression was positively correlated with lymph node involvement and advanced tumor stages and was related to poor RFS in stage III-IV PTC, further suggesting its clinical significance in PTC.

The biological functions of HMGB proteins are largely dependent on their subcellular locations [42]. Induced by external stimulus, HMGB1 and HMGB2 commute from the nucleus to the cytoplasm, and eventually outside the cells by posttranslational modifications [7,40]. Extracellular HMGB1 has been shown to promote tumor cell proliferation [21,22] and angiogenesis [23] and its blockade inhibits tumor growth by remodeling TIME [48]. Cytoplasmic HMGB1 interacts with mitochondrial DNA, inducing liver cancer growth [49] and regulates autophagy-mediated degradation of sodium/iodide symporter in thyroid cancer [14]. Similar to HMGB1, HMGB2 is also active both inside and outside the cells, while its biological roles in different subcellular locations in carcinogenesis are poorly understood [40]. For HMGB3, its translocation and the corresponding function are currently not reported. We here show that HMGB3 is mainly distributed in the cytoplasm of PTC tissues, rather than the nucleus as in noncancerous tissues. In addition, the HMGB3 serum level was significantly increased in PTC patients, particularly those with lymph node metastasis, than in healthy controls. Together, these results demonstrate that HMGB3 is relocated from the nucleus to the cytoplasm and eventually outside of the cells, playing a role in PTC progression. HMGB3 can be a potential circulating biomarker for predicting PTC prognosis.

Hypoxia, the main feature in late-stage solid cancers, has been shown to compromise anti-cancer immune responses by promoting immunosuppressive TIME and impairing the functions of effector immune cells [50,51]. Hypoxia-induced activation of HIF1α signaling collaborates with pro-inflammatory NF-κB pathways encouraging malignant progression in various cancer types [52,53], including thyroid cancers [54]. Although progress has been made in understanding the activation of these two pathways, the molecular mechanisms behind their crosstalk remain to be clarified [53]. Here, we observed that the cytoplasmic HMGB3 overlapped with the elevated HIF1α signals in PTC tissues. In addition, cytoplasmic and extracellular HMGB3 were both increased by hypoxia in PTC cell lines. These results indicate that HMGB3 translocation was triggered, at least partially, by hypoxia. Consistently, hypoxia has been shown to induce the translocation of HMGB1 from the nucleus to the cytoplasm in liver cancer [49]. That study further showed that cytoplasmic HMGB1 interacted with damaged mitochondrial DNA, promoting liver cancer growth by activating TLR9/NF-κB signaling [49]. This finding was in fact comparable to our data showing that the cytoplasmic, rather than extracellular, HMGB3 interacts with TLR3 regulating Poly(I:C)-mediated TLR3/NF-κB pathway. During the activation of innate immune responses, HMGB proteins exerting as DAMP are essential for the recognition of nucleic acids by TLRs [55]. TLR3, TLR7, TLR8, and TLR9 are commonly located on endosomes in order to detect intracellular nucleic acids derived from damaged cells, viruses, or bacteria [56]. Among these nucleic-acid mediated TLRs, functional overexpression of TLR3 has been demonstrated in PTC [37,38]. Although the interaction of HMGB3 with other intracellular TLRs cannot be excluded, our findings that hypoxia-induced cytoplasmic translocation of HMGB3 regulates nucleic-acid mediated TLR3/NF-κB signaling may be one of the mechanisms bridging hypoxia and inflammation in PTC progression.

We also evaluated the role of secreted HMGB3 in PTC and found that extracellular HMGB3 activates the TREM1 receptor expressed on the surface of PTC cells. In addition, IPA analysis of DEG altered by HMGB3 showed that HMGB3 not only affected tumor cell malignant behaviors in PTC but also participated in the migration and mobilization of tumor-infiltrating immune cells. TREM1, an immunoglobulin (Ig) superfamily receptor, is primarily expressed on myeloid cells regulating its differentiation through the release of inflammatory mediators [57]. Recently, studies have shown that TREM1 signaling compromises TIME in various cancer types [58–61]. In our previous study, TREM1 was overexpressed and was correlated with poor outcomes in PTC. Interestingly, it was mainly expressed on malignant epithelial cells, rather than on macrophages, and fosters an immunosuppressive microenvironment by enhancing the enrichment of Treg cells in PTC [32]. Together, these data indicate that extracellular HMGB3 may act as an endogenous TREM1-ligand to foster an immunosuppressive tumor microenvironment in PTC progression. However, more evidence is needed to demonstrate the proposed HMGB3-TREM1 interaction and its role in PTC immune microenvironment.

In summary, we demonstrate an oncogenic role of HMGB3 in thyroid cancers and show that HMGB3 is translocated to the cytoplasm activating nucleic-acid mediated TLR3/NF-κB pathway in PTC cells and to the extracellular space interacting with TREM1 receptor expressed on PTC cells. Our results promote the potential value of HMGB3 as a circulating biomarker to monitor PTC progression and open up the possibility for the future use of HMGB3 as a therapeutic target for PTC.

Funding Statement

The research was supported by grants from the National Science Foundation of China [grant No. 81500597], Natural Science Basic Research Plan in Shaanxi Province of China [grant No. 2017JM8051].

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15384101.2024.2302244