HMGA2 promotes epithelial-mesenchymal transition in intrahepatic cholangiocarcinoma via transforming growth factor-β/small mothers against decapentaplegic 2

Abstract

BACKGROUND

Intrahepatic cholangiocarcinoma (ICC) is a highly malignant biliary tract tumor with strong invasiveness, high metastatic potential and poor prognosis. HMGA2 contributes to tumorigenesis and malignant progression in multiple cancers, but its clinical significance, biological functions, and molecular mechanisms in ICC remain unclear.

AIM

To investigate the clinical significance, biological functions, and molecular mechanisms of HMGA2 in ICC progression.

METHODS

HMGA2 expression was evaluated by immunohistochemistry in postoperative tumor specimens from 61 patients with ICC and its associations with clinicopathological features and survival were analyzed. ICC cell lines with HMGA2 knockdown or overexpression were established. Cell proliferation, migration, and invasion were assessed using the cell counting kit-8, 5-ethynyl-2’-deoxyuridine incorporation, wound-healing, and Transwell assays. Xenograft experiments evaluated in vivo tumor growth. Western blotting was conducted to detect the expression of HMGA2, transforming growth factor-β (TGF-β)/small mothers against decapentaplegic 2 (Smad2) pathway-related proteins, and epithelial-mesenchymal transition (EMT)-associated markers.

RESULTS

HMGA2 positivity was associated with poor overall survival (P = 0.009) and remained an independent prognostic factor [hazard ratio (HR) = 2.271, 95% confidence interval (CI): 1.163-4.437, P = 0.016], along with vascular invasion (HR = 2.040, 95%CI: 1.083-3.844, P = 0.027). Cell counting kit-8 assays showed that HMGA2 overexpression significantly increased ICC cell viability at 24 hours, 48 hours, and 72 hours, whereas HMGA2 knockdown reduced cell viability. 5-ethynyl-2’-deoxyuridine assays further confirmed enhanced proliferation after HMGA2 overexpression (21.20% ± 1.27% vs 17.23% ± 0.95%, P = 0.0073) and reduced proliferation after HMGA2 knockdown (47.77% ± 1.16% vs 55.47% ± 1.33%, P = 0.0006). HMGA2 also promoted migration, invasion, xenograft growth, and TGF-β/Smad2-mediated EMT.

CONCLUSION

HMGA2 predicts poor prognosis and promotes ICC progression by activating the TGF-β/Smad2/EMT axis, supporting its potential as a biomarker and therapeutic target.

Key Words: Intrahepatic cholangiocarcinoma; HMGA2; Transforming growth factor-β; Small mothers against decapentaplegic 2 signaling; Epithelial-mesenchymal transition

Core Tip: This study demonstrates that HMGA2 is overexpressed in intrahepatic cholangiocarcinoma (ICC) and independently predicts poor prognosis. Functional experiments showed that HMGA2 promotes ICC cell proliferation, migration, invasion, and tumor growth. Mechanistically, HMGA2 activates the transforming growth factor-β/small mothers against decapentaplegic 2 signaling pathway and induces epithelial-mesenchymal transition, thereby driving ICC progression. These findings highlight HMGA2 as a potential prognostic biomarker and therapeutic target for ICC.

INTRODUCTION

Intrahepatic cholangiocarcinoma (ICC) is a malignant tumor originating in the epithelial lining of the secondary and peripheral intrahepatic bile ducts. Its global incidence is rising, particularly in regions with a high prevalence of chronic liver disease and biliary tract infections[1,2]. As a highly aggressive malignancy, ICC is characterized by diagnostic challenges, rapid progression, and a high metastatic potential. Consequently, approximately 80% of patients present with lymph node or distant metastasis at the time of diagnosis, rendering them ineligible for radical surgery[1]. Although surgical resection remains the primary treatment, and R0 resection offers superior overall survival (OS) compared with R1 resection, the prognosis remains poor. Even following R0 resection, the median OS is only 10.9 months, with 5-year survival rates ranging from 7% to 21.4%[3]. Palliative chemotherapy and targeted therapies for advanced disease show limited efficacy. Therefore, elucidating the molecular mechanisms driving malignant progression and identifying biomarkers with prognostic and therapeutic potential are essential for optimizing clinical management and improving patient outcomes.

Epithelial-mesenchymal transition (EMT) is a critical biological process in cancer progression that enables tumor cells to acquire aggressive phenotypes. This process is typically characterized by the downregulation of E-cadherin and the concomitant upregulation of mesenchymal markers, including N-cadherin, vimentin, and Slug, which collectively enhance migration, invasion, and apoptosis resistance[4]. The transforming growth factor-β/small mothers against decapentaplegic (TGF-β/Smad) signaling pathway is a classical regulator of the EMT process. Abnormal activation of this pathway induces the expression of EMT-related transcription factors (TFs) by phosphorylating Smad2/3 and other downstream molecules, thereby driving the development of malignant phenotypes[5]. Previous studies have confirmed that platelet-derived TGF-β1 induces EMT and promotes ICC metastasis by activating the Smad2 pathway, suggesting that the TGF-β/Smad-mediated EMT axis constitutes a critical regulatory network in ICC progression[6]. However, the upstream regulators and specific mechanisms underlying the abnormal activation of this pathway in ICC remain poorly understood, hindering the development of targeted therapeutic strategies.

HMGA2, a non-histone chromatin-binding protein, functions as a critical transcriptional regulator of diverse biological processes, including cell proliferation, differentiation, and EMT[4,7]. Accumulating evidence indicates that dysregulated HMGA2 is implicated in the progression of various malignancies and closely correlated with patient prognosis[8,9]. In hilar and distal cholangiocarcinoma, elevated HMGA2 expression is associated with vascular invasion, neural invasion, and a poorly differentiated phenotype, serving as an independent prognostic factor for poor survival[10]. Furthermore, research on hepatocellular carcinoma has demonstrated that genetic alterations in the TGF-β signaling pathway can upregulate HMGA2 by modulating the PJA1/Smad3 axis[11], suggesting a potential interaction between HMGA2 and the TGF-β signaling pathway. However, the specific role of HMGA2 in ICC remains poorly defined. The current study was designed to characterize HMGA2 expression in surgically resected ICC tissues and determine its functional involvement in ICC cell migration, invasion, and EMT, thereby investigating the potential mechanisms mediating HMGA2-regulated EMT.

MATERIALS AND METHODS

Patients and specimen collection

Ethical approval for this retrospective study was granted by the Ethics Committee of Lanzhou University Second Hospital (approval No. 2023A-561). A total of 61 surgical specimens were obtained from patients with ICC who underwent curative radical partial hepatectomy at the Department of General Surgery between January 2018 and September 2022. None of the included patients received preoperative chemotherapy or radiotherapy. Each specimen was histologically confirmed and classified by two independent pathologists.

Immunohistochemical staining

Paraffin-embedded ICC sections (4 μm) were deparaffinized in xylene and rehydrated through a graded ethanol series. Antigen retrieval was performed by heating sections in citrate buffer at 96 °C for 15 minutes, followed by natural cooling to room temperature. Endogenous peroxidase activity was inhibited by treatment with 3% hydrogen peroxide for 10 minutes. After washing, the sections were incubated overnight at 4 °C with a primary anti-HMGA2 antibody (ab207301, 1:500; Abcam, Cambridge, United Kingdom), followed by incubation with a secondary antibody. Signal detection was achieved using 3,3’-diaminobenzidine, and nuclei were counterstained with hematoxylin (Beyotime, Shanghai, China).

Immunoreactivity was independently evaluated by two senior pathologists in a double-blinded manner. A semi-quantitative scoring system was employed based on the product of staining intensity (0, absent; 1, light yellow; 2, yellowish-brown; and 3, dark brown) and the proportion of positive cells (1, 1%-25%; 2, 26%-50%; 3, 51%-75%; and 4, 76%-100%). Final composite scores (0-12) were categorized as HMGA2-negative (score < 6) or HMGA2-positive (score ≥ 6). Discrepancies in scoring were resolved through joint review and consensus.

Cell lines and cell culture

Human intrahepatic biliary epithelial cells (HIBEpiCs) and the ICC cell line HCCC-9810 were obtained from Procell (Wuhan, Hubei Province, China), while the RBE and HUCCT1 ICC cell lines were purchased from Servicebio (Wuhan, Hubei Province, China). HIBEpiCs were maintained in a specialized medium (CM-H042, Procell), whereas ICC cell lines were cultured in RPMI-1640 (Sunview, Shenzhen, Guangdong Province, China) supplemented with 10% fetal bovine serum and 1% penicillin. All cells were incubated at 37 °C in a humidified atmosphere containing 5% carbon dioxide (CO₂).

Reverse transcription quantitative polymerase chain reaction

Total RNA was extracted using TRIzol reagent (Ambion, Austin, TX, United States), and reverse transcription was performed to synthesize complementary DNA using a commercial kit (Yeasen, Shanghai, China). Gene expression levels were quantified via real-time polymerase chain reaction (PCR) using a SYBR Green assay kit (Yeasen) on a real-time amplification system (Rocgene, Beijing, China). Target gene expression was normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) levels, and relative quantification was calculated using the 2^-ΔΔCt method. The primer sequences are summarized in Supplementary Table 1.

Establishment of stable HMGA2 cell lines

Western blot (WB) analysis was initially performed to assess baseline HMGA2 expression in HIBEpiCs and three ICC cell lines (HUCCT1, RBE, and HCCC-9810). Based on these profiles, ICC cell lines with relatively high or low HMGA2 levels were selected for functional studies. Lentiviral vectors for HMGA2 short hairpin RNA (shRNA) (knockdown) and the full-length HMGA2 sequence (overexpression) were constructed and synthesized by HANBIO (Shanghai, China). Cells were seeded at a density of 1 × 10⁵ cells/well and allowed to reach approximately 80% confluence before transfection. ICC cell lines were transfected at a uniform multiplicity of infection. After 48 hours, stable clones were selected using 1.5 μg/mL puromycin (Invitrogen, Carlsbad, CA, United States). Transduction efficiency and stable HMGA2 modulation were confirmed via reverse transcription quantitative PCR (RT-qPCR) and WB analysis.

Cell counting kit-8 assay

Cell proliferative capacity was evaluated using a cell counting kit-8 (CCK-8) assay kit (Sunview). Cells were seeded into 96-well plates at a density of 1 × 10⁴ cells/well, with five replicate wells per group and incubated at 37 °C. At 24 hours, 48 hours, and 72 hours post-seeding, 10 μL of CCK-8 reagent was added to each well. Following incubation, absorbance was measured at 450 nm using a microplate reader (Detie Bio, Nanjing, Jiangsu Province, China).

5-ethynyl-2’-deoxyuridine incorporation assay

DNA synthesis activity was assessed using a Click-iT 5-ethynyl-2’-deoxyuridine (EdU) assay kit (Yeasen). Cells were seeded into 12-well plates (2 × 10⁵ cells/well) and cultured for 24 hours, followed by incubation with 10 μM EdU for 2 hours. Subsequently, the cells were fixed, permeabilized, and processed for the Click-iT reaction according to the manufacturer’s instructions. Nuclei were counterstained with Hoechst 33342. Fluorescence images were captured from five random fields per well using a fluorescence microscope. Proliferative capacity was quantified as the percentage of EdU-positive cells relative to the total number of nuclei.

Wound healing assay

Cells were plated into six-well plates (5 × 10⁵ cells/well) and cultured until confluent. A linear scratch wound was created using a sterile pipette tip. After removing detached cells by washing with phosphate-buffered saline, fresh medium was added, and the cells were incubated at 37 °C in a 5% CO₂ atmosphere. Wound healing was monitored and photographed at 0 hours, 24 hours, and 48 hours to assess cell migration.

Transwell invasion assay

Serum-starved cells were seeded into the upper compartments of Matrigel-coated Transwell chambers (Corning, NY, United States). Complete medium containing fetal bovine serum was added to the lower compartments as a chemoattractant. After 24 hours of incubation at 37 °C in a 5% CO₂ atmosphere, non-invading cells on the upper membrane surface were removed. Cells that invaded through the Matrigel barrier were fixed, stained, and imaged microscopically. Cell invasion was quantified by counting the number of invading cells in five random fields using ImageJ software.

WB analysis

Total cell lysates were prepared using ice-cold radio immunoprecipitation assay buffer (Solarbio, Beijing, China) and subjected to sonication for protein extraction, followed by centrifugation at 12000 rpm for 10 minutes at 4 °C. The resulting supernatants were collected, and equal amounts of protein, as determined by the bicinchoninic acid assay, were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Solarbio, Beijing, China) and transferred onto polyvinylidene fluoride membranes (Millipore, Billerica MA, United States). Following blocking with 5% non-fat milk, the membranes were incubated overnight at 4 °C with primary antibodies, followed by incubation with secondary antibodies at room temperature. Signals were detected using an enhanced chemiluminescence kit (Yeasen). The following primary antibodies were used for WB analysis: HMGA2 (20795-1-AP, 1:5000, Proteintech), E-cadherin (60335-1-Ig, 1:2000, Proteintech), N-cadherin (66119-1-Ig, 1:2000, Proteintech), vimentin (60330-1-Ig, 1:2000, Proteintech), SNAI/Slug (12129-1-AP, 1:2000, Proteintech), Smad2 (ET1604-22, 1:2000, HUABIO), phospho-Smad2 (YP1185/B8501, 1:2000, Immunoway), TGF-β (YT4632/B3201, 1:2000, Proteintech), and GAPDH (60004-1-Ig, 1:2000, Proteintech). The corresponding secondary antibodies were horseradish peroxidase-labeled anti-rabbit immunoglobulin G (IgG) (RS0002, 1:5000, Immunoway) and anti-mouse IgG (RS0001, 1:5000, Immunoway). The relative intensity of the protein bands was analyzed using ImageJ software.

Rescue assay

To determine whether the function of HMGA2 is dependent on the TGF-β/Smad2 signaling pathway, rescue experiments were performed. To inhibit the TGF-β/Smad2 pathway, cells stably overexpressing HMGA2 were treated with 20 μM SB525334 (HY-12043, MedChemExpress, Princeton, NJ, United States) for 48 hours. In stable HMGA2-knockdown cells, the pathway was activated by treatment with 15 μM SRI-011381 (HY-1000347A, MedChemExpress) for 48 hours. Appropriate control groups, including negative controls (NCs) and drug controls, were established. Transwell and WB assays were performed to evaluate cell invasive ability and the expression of pathway-related proteins, respectively, to elucidate the functional relationship between HMGA2 and the TGF-β/Smad2 pathway.

Animal experiments

Male BALB/c nude mice (4-6 weeks old) were obtained from the Lanzhou Veterinary Research Institute, Chinese Academy of Agricultural Sciences (Lanzhou, Gansu Province, China). All animals were housed in a specific pathogen-free environment under controlled conditions and randomly allocated into three groups (n = 5 per group). ICC cells expressing shRNA-mediated HMGA2 knockdown (sh-HMGA2), HMGA2 overexpression plasmid (OE-HMGA2), or control vectors were resuspended in a 1:1 ratio with Matrigel (354248, Corning) and subcutaneously injected into the right axillary region. Tumor volumes were measured regularly for 3 consecutive weeks to monitor growth kinetics. At the study endpoint, all mice were euthanized, and the xenograft tumors were excised, weighed, and fixed in paraffin for immunohistochemical (IHC) staining. All procedures were approved by the Animal Welfare and Ethics Committee of Lanzhou University Second Hospital (approval No. D2023-382).

Statistical analysis

All experiments were performed in triplicate. Statistical evaluations were conducted using GraphPad Prism 10.1 (San Diego, CA, United States) and SPSS 22.0 (Chicago, IL, United States). Quantitative data are presented as the mean ± SD. Intergroup differences among multiple groups were analyzed using one-way analysis of variance followed by Tukey’s post hoc test. Survival curves were estimated using the Kaplan-Meier method and compared via the log-rank test. A P value < 0.05 was considered statistically significant. Significance annotations: Ns, P > 0.05; ^aP < 0.05; ^bP < 0.01; ^cP < 0.001; ^dP < 0.0001.

RESULTS

HMGA2 is an independent prognostic factor for poor survival in ICC

To evaluate the clinical significance of HMGA2, IHC analysis was performed on postoperative specimens from 61 patients with ICC. HMGA2 protein was localized exclusively within the nuclei of tumor cells. Based on the scoring system defined in materials and methods, patients were stratified into HMGA2-positive (score ≥ 6) and HMGA2-negative (score < 6) groups (Figure 1).

Figure 1 Immunohistochemical staining of HMGA2 in intrahepatic cholangiocarcinoma patient tissues. A: Negative expression of HMGA2; B: Positive expression of HMGA2. Original magnification, × 40; scale bar, 300 μm.

The study cohort comprised 61 patients with ICC (41 males and 20 females), with a mean age of 59.67 ± 7.30 years. As shown in Table 1, no significant differences in baseline clinicopathological features were observed between the HMGA2-positive and HMGA2-negative groups (P > 0.05). To evaluate the prognostic value of HMGA2, Kaplan-Meier survival curves were generated for OS and disease-free survival (DFS). Patients in the HMGA2-positive group exhibited a significantly shorter median OS than those in the HMGA2-negative group [15.0 months 95% confidence interval (CI): 13.161-16.839 vs 18.0 months 95%CI: 11.578-24.422; log-rank χ² = 6.727, P = 0.009; Figure 2]. Similarly, DFS analysis indicated that HMGA2-positive patients had a significantly higher risk of postoperative recurrence than HMGA2-negative patients (log-rank χ² = 5.936, P = 0.015). To identify independent prognostic factors, variables with P < 0.1 in the univariate Cox regression were included in a multivariate model. Univariate analysis (Table 2) indicated that HMGA2 overexpression, positive surgical margins, poor tumor differentiation, and vascular invasion were significantly correlated with poor survival (P < 0.05). Subsequent multivariate analysis (Table 3) confirmed that HMGA2 overexpression (P = 0.016) and vascular invasion (P = 0.027) served as independent risk factors for poor prognosis in patients with ICC.

Figure 2 Prognostic value of HMGA2 expression in intrahepatic cholangiocarcinoma patients. A: Kaplan-Meier analysis of overall survival; B: Kaplan-Meier analysis of disease-free survival.

Table 1 Clinicopathological characteristics of intrahepatic cholangiocarcinoma patients stratified by HMGA2 expression, n (%).

Characteristic		Patients	HMGA2 (-)	HMGA2 (+)	χ2 value	P value
Gender	Female	21 (34.43)	10	11	0.131	0.717
	Male	40 (65.57)	21	19
Age (year)	< 60	30 (49.18)	14	16	0.407	0.523
	≥ 60	31 (50.82)	17	14
TBIL (μmol/L)	< 26	19 (31.15)	11	8	0.553	0.457
	≥ 26	42 (68.85)	20	22
GGT (U/L)	< 60	8 (13.11)	4	4	0.000	1.000
	≥ 60	53 (86.89)	27	26
ALP (U/L)	< 125	8 (13.11)	5	3	0.109	0.742
	≥ 125	53 (86.69)	26	27
CEA (ng/mL)	< 3.4	36 (59.02)	19	17	0.135	0.714
	≥ 3.4	25 (40.98)	12	13
CA125 (U/mL)	< 35	28 (45.90)	15	13	0.157	0.692
	≥ 35	33 (54.10)	16	17
CA199 (U/mL)	< 27	25 (40.98)	12	13	0.135	0.714
	≥ 27	36 (59.02)	19	17
Hepatitis B	Negative	51 (83.61)	27	24	0.162	0.687
	Positive	10 (16.39)	4	6
Hepatitis C	Negative	57 (93.44)	30	27	0.304	0.581
	Positive	4 (6.56)	1	3
Bile duct stones	No	31 (50.82)	17	14	0.407	0.523
	Yes	30 (49.18)	14	16
Differentiation degree	Well/moderately	40 (65.57)	22	18	0.812	0.367
	Poorly	21 (34.43)	9	12
Tumor size (cm)	< 3	22 (36.07)	10	12	0.396	0.529
	≥ 3	39 (63.93)	21	18
Bile duct type	Small	34 (55.74)	19	15	0.788	0.375
	Large	27 (44.26)	12	15
Surgical margin	Negative	57 (93.44)	29	28	0.000	1.000
	Positive	4 (6.56)	2	2
Vascular invasion	No	32 (52.46)	16	16	0.018	0.893
	Yes	29 (47.54)	15	14
Perineural invasion	No	30 (49.18)	18	12	1.991	0.158
	Yes	31 (50.82)	13	18
Lymph node metastasis	No	49 (80.33)	27	22	1.828	0.176
	Yes	12 (19.67)	4	8
Postoperative chemotherapy	No	20 (32.79)	9	11	0.403	0.525
	Yes	41 (67.21)	22	19

TBIL: Total bilirubin; GGT: Gamma-glutamyl transferase; ALP: Alkaline phosphatase; CEA: Carcinoembryonic antigen; CA125: Carbohydrate antigen 125; CA199: Carbohydrate antigen 199.

Table 2 Univariate analyses of overall survival in intrahepatic cholangiocarcinoma patients.

Variables	Factors	β	SE	Wald	P value	HR (95%CI)
Gender	Female/male	-0.221	0.312	0.502	0.479	0.802 (0.435-1.478)
Age	< 60/≥ 60	0.092	0.297	0.096	0.756	1.097 (0.612-1.964)
TBIL (μmol/L)	< 26/≥ 26	0.011	0.322	0.001	0.972	1.011 (0.538-1.901)
GGT (U/L)	< 60/≥ 60	0.430	0.477	0.810	0.368	1.537 (0.603-3.915)
ALP (U/L)	< 125/≥ 125	0.547	0.478	1.310	0.252	1.728 (0.677-4.406)
CEA (ng/mL)	< 3.4/≥ 3.4	0.119	0.299	0.158	0.691	1.126 (0.626-2.025)
CA125 (U/mL)	< 35/≥ 35	-0.373	0.304	1.506	0.220	0.689 (0.380-1.249)
CA199 (U/mL)	< 27/≥ 27	0.335	0.314	1.140	0.286	1.398 (0.756-2.585)
Hepatitis B	Negative/positive	-0.294	0.443	0.440	0.507	0.745 (0.313-1.776)
Hepatitis C	Negative/positive	0.299	0.535	0.313	0.576	1.349 (0.473-3.846)
Bile duct stones	No/yes	-0.208	0.296	0.493	0.483	0.812 (0.454-1.452)
Differentiation degree	Well, moderately/poorly	0.834	0.322	6.706	0.010	2.302 (1.225-4.328)
Tumor size	< 3/≥ 3	0.279	0.317	0.776	0.378	1.322 (0.711-2.460)
Bile duct type	Small/large	0.302	0.296	1.043	0.307	1.353 (0.757-2.417)
Surgical margin	Negative/positive	1.114	0.544	4.194	0.041	3.045 (1.049-8.840)
Vascular invasion	No/yes	0.684	0.310	4.869	0.027	1.982 (1.079-3.639)
Perineural invasion	No/yes	0.072	0.296	0.059	0.808	1.075 (0.602-1.920)
Lymph node metastasis	No/yes	0.282	0.361	0.610	0.435	1.326 (0.653-2.693)
Postoperative chemotherapy	No/yes	0.147	0.327	0.203	0.652	1.159 (0.610-2.199)
HMGA2	Negative/positive	0.790	0.323	5.977	0.014	2.203 (1.170-4.148)

HR: Hazard ratio; CI: Confidence interval; TBIL: Total bilirubin; GGT: Gamma-glutamyl transferase; ALP: Alkaline phosphatase; CEA: Carcinoembryonic antigen; CA125: Carbohydrate antigen 125; CA199: Carbohydrate antigen 199.

Table 3 Multivariate analyses of overall survival in intrahepatic cholangiocarcinoma patients.

Variables	Factors	β	SE	Wald	P value	HR (95%CI)
Differentiation degree	Well, moderately/poorly	0.610	0.349	3.045	0.081	1.840 (0.928-3.650)
Surgical margin	Negative/positive	0.416	0.588	0.501	0.479	1.516 (0.479-4.800)
Vascular invasion	No/yes	0.713	0.323	4.869	0.027	2.040 (1.083-3.844)
HMGA2	Negative/positive	0.820	0.342	5.769	0.016	2.271 (1.163-4.437)

HR: Hazard ratio; CI: Confidence interval.

HMGA2 is highly expressed in human ICC cell lines

WB analysis revealed that HMGA2 levels were higher in RBE cells and lower in HCCC-9810 cells relative to HIBEpiCs (Figure 3A). Consequently, we established stable HMGA2 knockdown and overexpression cell lines and confirmed their efficacy via RT-qPCR and WB analysis. In HCCC-9810 cells, transfection with an OE-HMGA2 significantly increased both messenger RNA and protein levels of HMGA2 compared with empty vector or NCs (Figure 3B and C). In RBE cells, all three shRNA targets (113, 114, and 115) reduced HMGA2 messenger RNA and protein levels compared with the non-targeting shRNA negative control (sh-NC) and NC cell groups, with target 115 exhibiting the highest silencing efficiency (Figure 3D and E). Consequently, RBE cells stably transduced with the lentivirus corresponding to target 115 (sh-HMGA2) were used for all subsequent functional validation.

Figure 3 HMGA2 regulates the proliferative phenotype of intrahepatic cholangiocarcinoma cells. A: HMGA2 protein expression in human intrahepatic biliary epithelial cells and intrahepatic cholangiocarcinoma cell lines (RBE, HCCC-9810, HUCCT1); B and C: Efficiency of HMGA2 overexpression verified by Western blot (WB) and reverse transcription quantitative polymerase chain reaction (RT-qPCR); D and E: Efficiency of HMGA2 knockdown validated by WB and RT-qPCR; F and G: Cell viability detected by cell counting kit-8 assay; H: Cell proliferation assessed by 5-ethynyl-2’-deoxyuridine staining. ^bP < 0.01. ^cP < 0.001. ^dP < 0.0001. NS: Not significant; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; NC: Negative control; OE-HMGA2: HMGA2 over-expression; mRNA: Messenger RNA; sh-NC: Short hairpin RNA negative control; sh-HMGA2: Short hairpin RNA-mediated HMGA2 knockdown; EdU: 5-ethynyl-2’-deoxyuridine.

HMGA2 promotes the proliferation of ICC cells in vitro

According to the CCK-8 assay results, HMGA2-overexpressing cells exhibited significantly increased proliferative activity at 24 hours, 48 hours, and 72 hours compared with the vector control group (P < 0.0001; Figure 3F). Conversely, HMGA2 knockdown significantly reduced proliferative capacity at 48 hours and 72 hours compared with the sh-NC group (P < 0.001; Figure 3G). These findings were further validated by EdU incorporation assays. Overexpression of HMGA2 resulted in a significantly higher proportion of EdU-positive cells than in control groups (P = 0.0073), indicating enhanced DNA synthesis and proliferative activity. In contrast, HMGA2 knockdown significantly decreased the number of EdU-positive cells compared with the sh-NC group (P = 0.0006; Figure 3H). Together, these results demonstrate that HMGA2 is a key regulator of ICC cell proliferation.

HMGA2 regulates the metastatic phenotype of ICC cells

To evaluate the role of HMGA2 in metastasis, cell scratch and Transwell invasion assays were performed. Cell scratch assays demonstrated that HMGA2 knockdown in RBE cells significantly inhibited wound closure compared with the control group, resulting in a wider residual scratch. Conversely, as shown in Figure 4, HMGA2 overexpression in HCCC-9810 cells significantly accelerated migration, as evidenced by a narrower scratch width (Figure 4A and C). The Transwell invasion assay was employed to determine the influence of HMGA2 on invasive capacity. As shown in Figure 4B and D, HMGA2 knockdown reduced the number of cells penetrating the Matrigel-coated membrane. In contrast, HMGA2 overexpression enhanced the invasive properties of HCCC-9810 cells compared with the control group. These results collectively indicate that HMGA2 is a critical regulator of the metastatic phenotype in ICC.

Figure 4 HMGA2 regulates the metastatic phenotype and epithelial-mesenchymal transition in intrahepatic cholangiocarcinoma cells. A-D: Cell migration determined by cell scratch assay (A and C) and cell invasion evaluated by Transwell invasion assay (B and D); E-H: Expression of epithelial-mesenchymal transition-related markers analyzed by Western blot (WB) (E and G) and expression levels of transforming growth factor-β and phospho-small mothers against decapentaplegic 2 measured by WB (F and H). ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. ^dP < 0.0001. NC: Negative control; OE-HMGA2: HMGA2 over-expression; sh-NC: Short hairpin RNA negative control; sh-HMGA2: Short hairpin RNA-mediated HMGA2 knockdown; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; TGF: Transforming growth factor; Smad2: Small mothers against decapentaplegic 2; p-Smad2: Phospho-small mothers against decapentaplegic 2.

HMGA2 promotes EMT by activating the TGF-β/Smad2 signaling pathway

To elucidate the mechanism by which HMGA2 promotes metastasis, we examined the expression of EMT markers and pathway-associated proteins. WB analysis revealed that HMGA2 overexpression in HCCC-9810 cells significantly downregulated the epithelial marker E-cadherin, while upregulating mesenchymal markers including N-cadherin, vimentin, and Slug (Figure 4E). Conversely, an opposite expression pattern was observed in RBE cells following HMGA2 knockdown (Figure 4G). Mounting evidence indicates that HMGA2 is critically involved in EMT regulation and promotes tumor cell invasion by targeting the TGF-β signaling pathways. Therefore, we assessed whether HMGA2 regulates the TGF-β/Smad2 signaling axis. HMGA2 overexpression led to a marked increase in TGF-β and phospho-Smad2 protein levels in HCCC-9810 cells relative to controls, whereas HMGA2 knockdown significantly decreased the expression of these proteins in RBE cells (Figure 4F and H). No significant differences were observed in total Smad2 or GAPDH expression levels. Collectively, these results demonstrate that HMGA2 promotes the EMT phenotype by activating the TGF-β/Smad2 pathway in ICC cells.

HMGA2 regulates EMT and metastatic phenotypes in ICC cells via the TGF-β/Smad2 signaling pathway

To verify whether the TGF-β/Smad2 axis is the primary mediator of HMGA2-driven malignancy, rescue experiments were performed using a TGF-β inhibitor (SB525334) and an activator (SRI-011381). WB analysis revealed that SB525334 significantly blocked the upregulation of TGF-β and phospho-Smad protein levels induced by HMGA2 overexpression (Figure 5A). Conversely, SRI-011381 effectively restored the expression of these proteins in HMGA2-silenced cells (Figure 5B). These findings validate that the TGF-β/Smad2 axis acts as a key downstream effector of HMGA2 and that the pharmacological interventions employed exhibit clear target specificity. On this basis, cell invasion assays and WB analysis were performed to evaluate whether the TGF-β/Smad2 pathway mediates the malignant progression driven by HMGA2 in ICC.

Figure 5 HMGA2 drives epithelial-mesenchymal transition and metastasis in cholangiocarcinoma cells via the transforming growth factor-β/small mothers against decapentaplegic 2 pathway. A and B: Western blot (WB) analysis of transforming growth factor-β/small mothers against decapentaplegic 2 (TGF-β/Smad2) pathway intervention efficiency; C and D: Transwell assays showing the TGF-β/Smad2 pathway mediates HMGA2-egulated cell invasion; E and F: WB demonstrating that the TGF-β/Smad2 pathway mediates HMGA2-induced epithelial-mesenchymal transition. ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. ^dP < 0.0001. NC: Negative control; OE-HMGA2: HMGA2 over-expression; sh-HMGA2: Short hairpin RNA-mediated HMGA2 knockdown; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; TGF: Transforming growth factor; Smad2: Small mothers against decapentaplegic 2; p-Smad2: Phospho-small mothers against decapentaplegic 2.

As presented in the Transwell invasion assay (Figure 5C and D), the increased invasiveness induced by HMGA2 overexpression was markedly reversed by inhibiting the TGF-β signaling pathway, while activating this pathway partially rescued the invasive capacity reduced by HMGA2 knockdown. Consistently, WB analysis (Figure 5E and F) revealed that the EMT phenotype promoted by HMGA2 overexpression was abolished by blocking TGF-β signaling, whereas activation of this pathway restored the EMT program that was suppressed upon HMGA2 depletion. Collectively, our results support the notion that the TGF-β/Smad2 signaling pathway serves as a crucial mediator underlying HMGA2-induced EMT and metastatic progression in ICC.

HMGA2 promotes the growth and proliferation of ICC in vivo

BALB/c nude mice were used to establish a subcutaneous ICC xenograft model. The HMGA2 knockdown group exhibited markedly smaller tumor volumes and lighter tumor weights than the overexpression and control groups (Figure 6A and B). IHC analysis further demonstrated that in RBE cell-derived xenografts, HMGA2 knockdown significantly reduced the proportion of HMGA2-positive cells; conversely, in HCCC-9810 cell-derived xenografts, HMGA2 overexpression markedly increased the number of HMGA2-positive cells (Figure 6C).

Figure 6 HMGA2 promotes intrahepatic cholangiocarcinoma cell growth in nude mice. A: Images of xenograft tumors and tumor growth curves 3 weeks after implantation; B: Resected tumor tissues and tumor weight analysis; C: Immunohistochemical staining of HMGA2 in xenograft tumors. Original magnification, × 40; scale bar, 300 μm. ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. ^dP < 0.0001. OE-HMGA2: HMGA2 over-expression; sh-HMGA2: Short hairpin RNA-mediated HMGA2 knockdown.

DISCUSSION

ICC originates from the epithelium of the intrahepatic bile ducts, predominantly presenting as a highly malignant adenocarcinoma. The majority of patients present with intrahepatic and lymph node metastases at the time of diagnosis, a clinical challenge further compounded by profound tumor heterogeneity and a distinct immune microenvironment[12]. Accumulating evidence suggests that HMGA2 contributes substantially to tumor growth, invasion, and the EMT phenotype, involving multiple signaling pathways and interacting RNA signaling axes[4]. However, the specific role of HMGA2 in ICC progression and the molecular mechanisms governing its activity remain largely uncharacterized.

In the current study, IHC analysis of 61 postoperative ICC specimens revealed that HMGA2 expression significantly correlates with poor survival outcomes. Multivariate analysis further identified both HMGA2 expression and vascular invasion as independent risk factors for poor prognosis in patients with ICC. These findings align with emerging evidence designating HMGA2 as a robust predictor of shortened survival in hilar and distal cholangiocarcinoma[10]. Previous studies have similarly linked HMGA2 levels to aggressive clinicopathological features, including advanced disease status, tumor grade, vascular invasion, serosal invasion, and lymph node metastasis[13-16]. Beyond biliary malignancies, HMGA2 is a well-established driver of progression across a broad spectrum of human cancers, such as bladder, papillary renal, colorectal, and breast carcinomas[8,9,13,17]. By integrating our clinical data with these established reports, our study further substantiates the pivotal role of HMGA2 in driving malignancy. Consequently, HMGA2 represents a clinically valuable prognostic biomarker for ICC and a potential therapeutic target for clinical intervention.

Furthermore, the current study demonstrated that HMGA2 promotes malignant phenotypes in ICC, specifically enhancing proliferation, migration, and invasion. Importantly, these functional findings are consistent with our clinical observation that HMGA2 positivity was associated with poor prognosis, suggesting that HMGA2 is not merely a prognostic indicator but may actively contribute to ICC progression. In RBE cells, HMGA2 knockdown increased E-cadherin expression while reducing N-cadherin, vimentin, and Slug levels. Conversely, HMGA2 overexpression in HCCC-9810 cells significantly induced EMT. EMT is considered a primary driver of tumor metastasis[18]. Given that EMT is closely associated with tumor invasion, metastasis, stemness, and therapeutic resistance, these results indicate that HMGA2 may facilitate EMT-associated phenotypic remodeling in ICC cells, thereby contributing to aggressive tumor behavior and poor survival[19,20]. Therefore, our findings indicate that HMGA2 is not merely a prognostic indicator, but may also actively contribute to ICC malignant progression.

A range of signals from the tumor microenvironment can trigger EMT, a process controlled by a coordinated network of EMT-related TFs. Papoutsoglou et al[21] outlined that TGF-β dysfunction is a critical contributor to tumor initiation and subsequent progression. The desmoplastic microenvironment of ICC is marked by elevated TGF-β expression, a factor significantly associated with reduced patient survival. Tumor cells undergoing TGF-β-induced EMT may acquire sustained survival, self-renewal capacity, and differentiation plasticity, making them major drivers of tumor recurrence and distant metastasis[22]. Due to the distinctive anatomical setting of cholangiocarcinoma, the biliary microenvironment, and the prominent desmoplastic stroma, TGF-β signaling exhibits a complex dual role in this disease.

The present findings are particularly relevant because ICC is characterized by a dense desmoplastic microenvironment, in which TGF-β signaling has been implicated in tumor progression, invasion, and metastatic dissemination[23,24]. In addition, cancer-associated fibroblast-derived factors, including TGF-β, may contribute to EMT activation and enhanced invasive potential in cholangiocarcinoma, further supporting the biological relevance of this pathway in the ICC microenvironment[25,26]. Existing research has confirmed that HMGA2 exhibits significant cell-type specificity in regulating tumor EMT processes, with the TGF-β/Smad signaling pathway serving as the core regulatory axis in various contexts. In nasopharyngeal carcinoma cells, loss of HMGA2 significantly suppresses EMT progression by interfering with TGF-β/Smad3 pathway activity[27]. Cycloheximide experiments have confirmed that the rapid upregulation of SNAIL by TGF-β is directly regulated by the TGF-β/Smad pathway[28], while its sustained induction depends on the de novo synthesis of proteins such as HMGA2, the expression of which is also upregulated by the TGF-β/Smad pathway[29]. Notably, Slug expression patterns are closely correlated with the initiation and progression of numerous malignancies. In colorectal cancer[30], high Slug expression correlates with enhanced tumor cell stemness characteristics and is also linked to poor prognosis and increased metastatic lymph node risk in patients with gastric cancer[31]. Slug may act as a key regulator of tumor invasion and metastasis in the gastrointestinal tract, possibly by regulating EMT and maintaining tumor stem cell identity. Rather than simply confirming the general role of EMT in cancer, our results suggest that HMGA2 may represent an important molecular contributor connecting EMT-associated transcriptional remodeling with TGF-β/Smad2 signaling activation in ICC. Nevertheless, the current data mainly demonstrate a functional association based on protein expression changes and rescue experiments.

This study has several limitations. First, only 61 patients with ICC from a single center were included, and no a priori sample size estimation or power analysis was performed, which may limit the statistical robustness and generalizability of the clinical findings. Larger multicenter cohorts with predefined power calculations are required for further validation. Second, although protein expression analyses and rescue experiments suggest a functional link between HMGA2 and TGF-β/Smad2 pathway, the direct regulatory mechanism remains unclear. Future studies should further investigate the molecular interaction between HMGA2 and this pathway using more in-depth mechanistic approaches.

CONCLUSION

The present study indicates that HMGA2 functions as an independent prognostic biomarker for adverse outcomes in ICC and promotes EMT and metastatic behavior through activation of the TGF-β/Smad2 signaling pathway. Our findings provide a molecular basis for ICC progression and suggest that the HMGA2/TGF-β/Smad2 axis is a promising therapeutic target. Future studies are warranted to translate these insights into clinical strategies.