ABSTRACT
-
Background/Aims
Lenvatinib resistance and immune exclusion limit outcomes in hepatocellular carcinoma (HCC). We hypothesized that metabolic rewiring orchestrates resistance to lenvatinib and programmed cell death protein 1 (PD-1) blockade.
-
Methods
We established lenvatinib-sensitive/lenvatinib-resistant (LS/LR) HCC models and employed multi-omics (proteomics/RNA-seq), chromatin immunoprecipitation, luciferase, and RNA immunoprecipitation assays to map hexokinase domain containing protein 1 (HKDC1) regulation. Tumor immunity was profiled by scRNA-seq, multiplex fluorescent immunohistochemistry, and flow cytometry. Spermidine (SPD)+lenvatinib efficacy was tested in cell lines and patient-derived organoids/xenografts. Therapeutic effects were tested in an immunocompetent hydrodynamic HCC model with hepatocyte-specific Hkdc1 deletion and were analyzed a postoperative cohort (n=40) treated with lenvatinib+PD-1.
-
Results
HKDC1, upregulated in LR HCC, was transcriptionally activated by upstream stimulatory factor 1 (USF1) and promoted spermine synthase (SMS)-mediated polyamine rewiring. This impaired CD8+ T-cell metabolism, reversible by HKDC1 knockdown or SPD. SPD synergized with lenvatinib, triggering autophagy and suppressing tumor growth in vitro and in vivo. High HKDC1 predicted poor response and survival in patients receiving lenvatinib+aPD-1.
-
Conclusions
A USF1/HKDC1/SMS axis couples polyamine metabolism to immune dysfunction and lenvatinib resistance. HKDC1 is a predictive biomarker and therapeutic node and supports polyamine-axis modulation to sensitize HCC to lenvatinib plus PD-1 therapy.
-
Keywords: Hepatocellular carcinoma; Polyamine metabolism; Lenvatinib resistance; aPD-1 immunotherapy; CD8+ T cells
Study Highlights
• Redefines lenvatinib resistance as a metabolically buffered and immunosuppressive state.
• Reveals a novel resistance mechanism in HCC, linking polyamine metabolic rewiring to the dual processes of immune evasion and lenvatinib resistance.
• Provides translational strategies to sensitize HCC to targeted immunotherapy combinations.
Graphical Abstract
INTRODUCTION
Most primary liver cancers are hepatocellular carcinoma (HCC), which ranks sixth worldwide in incidence and fourth in cancer-related mortality [
1]. Although surgical resection offers the greatest opportunity for 5-year survival, many patients present at intermediate or advanced stages and are ineligible for curative surgery; systemic therapy is therefore the mainstay. Historically, receptor tyrosine kinase inhibitors such as sorafenib and lenvatinib were the only first-line options for advanced disease [
2,
3]. The IMbrave150 phase 3 trial (2020) established immune checkpoint blockade as a therapeutic backbone in HCC, enabling regimens incorporating programmed cell death protein 1 antibody (aPD-1) [
4]. Nonetheless, despite targeted and immune-based therapies, most patients with advanced HCC develop primary or acquired resistance and progress [
5]. Tumor-microenvironmental heterogeneity and cancer-cell metabolic reprogramming are prominent resistance drivers, underscoring the need to define mechanisms and identify tractable sensitization targets.
Hexokinase domain containing protein 1 (HKDC1) is a recently characterized hexokinase-family protein broadly expressed across human tissues [
6]. Emerging studies show that HKDC1 promotes tumor progression by rewiring glucose and lipid metabolism and modulating mitochondrial function [
7–
10]. It can also act as a mRNA-binding protein that drives malignant behavior and chemoresistance in gastric cancer [
10]. In liver cancer, HKDC1 deletion reduces programmed death-ligand 1 (PD-L1) expression and enhances the efficacy of aPD-1 therapy by boosting CD8
+ T-cell cytotoxicity [
11], implicating HKDC1 as a regulator of the HCC immune microenvironment. Given the paucity of targets that simultaneously sensitize tumors to lenvatinib and improve responses to immunotherapy, we hypothesize that therapeutic inhibition of HKDC1, which is elevated in lenvatinib-resistant (LR) HCC cell lines, could sensitize HCC to lenvatinib in combination with aPD-1 therapy.
Polyamines-putrescine, spermidine (SPD), and spermine— are small polycationic metabolites central to cell growth [
12,
13]. Spermine synthase (SMS) converts SPD to spermine. SPD, a natural polyamine, extends lifespan across species and improves hepatic and immune function [
14]. By enhancing protein synthesis and mitochondrial activity, SPD may regulate proliferation [
15,
16]. As a canonical autophagy inducer, SPD triggers autophagy in both normal and malignant cells [
14,
17,
18]. Autophagy is context dependent: insufficient autophagy can be cytoprotective, whereas excessive autophagy can become cytotoxic, precipitating autophagy-dependent cell death and sensitizing tumors to therapy [
19–
22]. Given the role of autophagy in lenvatinib efficacy [
23,
24], SPD likely modulates lenvatinib sensitivity via autophagy. Moreover, SPD directly binds and activates mitochondrial trifunctional protein, increasing fatty-acid oxidation, mitochondrial fitness, and CD8
+ T-cell cytotoxicity, thereby reshaping the tumor immune microenvironment (TIME) and enhancing aPD-1 responses [
25]. Collectively, these observations suggest that combining SPD with lenvatinib and aPD-1 therapy may offer a strategy to improve therapeutic efficacy in HCC.
Here, we show that HKDC1 is upregulated in LR HCC and correlates with prognosis. An upstream stimulatory factor 1 (USF1)→ HKDC1→ SPD axis drives resistance to lenvatinib and immunotherapy. HKDC1 ablation re-sensitizes tumors, partly reverses LR-induced immunosuppression, and augments aPD-1. In 40 recurrent cases on lenvatinib+aPD-1, higher HKDC1 marked nonresponse and stratified PFS, nominating HKDC1 as a biomarker and therapeutic target.
MATERIALS AND METHODS
Human liver cancer specimens
Patient samples from Zhongshan Hospital, Fudan University, included two cohorts: >200 primary HCC resection specimens and 40 primary liver cancer tissues from patients who later recurred and received lenvatinib plus immunotherapy. All patients provided written informed consent, and the study was approved by the Zhongshan Hospital Institutional Review Board (approval No.B2021-248).
Statistical analysis
Statistics were conducted in GraphPad Prism v9. Survival was analyzed by Kaplan–Meier with log-rank testing. Data are presented as mean±standard deviation. Two-group comparisons used two-tailed unpaired Student’s t-tests. Multiple groups were evaluated by one- or two-way ANOVA with Bonferroni post hoc correction. Additional details are provided in the figure legends. Statistical significance was set at P<0.05.
For more detailed descriptions of the methods, please refer to the “MATERIALS AND METHODS” section in the
Supplementary Materials.
RESULTS
HKDC1 is associated with lenvatinib resistance and malignant progression in HCC
Lenvatinib is a first-line therapy for unresectable HCC, yet primary and acquired resistance remains common and clinically consequential [
2]. To model clinically prevalent lenvatinib resistance, we first profiled baseline lenvatinib sensitivity across six HCC cell lines, observing a broad IC50 range and dose-dependent suppression of colony formation (
Fig. 1A,
Supplementary Fig. 1A). Chronic drug exposure generated paired LR/lenvatinib-sensitive (LS) derivatives: LR clones exhibited right-shifted dose–response curves and attenuated growth inhibition under continuous dosing, whereas LS cells retained sensitivity (
Fig. 1B,
Supplementary Fig. 1B, 1C).
In vivo xenograft assays, lenvatinib reduced tumor volume, endpoint weight, and Ki67 in LS tumors but produced minimal effects in LR tumors (
Fig. 1B, 1D,
Supplementary Fig. 1D). Together, these data establish a robust cellular and animal model that recapitulates lenvatinib resistance in HCC.
To delineate LR-associated molecular features, we performed unbiased proteomics on Hep3B-LR versus Hep3B-LS cells, quantifying 4,294 proteins and identifying 252 upregulated and 211 downregulated species (fold change ≥2;
Fig. 1C). Pathway enrichment highlighted prominent metabolic rewiring (
Fig. 1C). Among candidates, HKDC1 was markedly upregulated and coincided with increased epidermal growth factor receptor (EGFR) (
Fig. 1C), a known contributor to lenvatinib resistance [
26]. We therefore focused on HKDC1, given accumulating evidence that this atypical hexokinase supports tumor progression and immune evasion in liver cancer [
11]. Quantitative reverse transcription polymerase chain reaction (qRT-PCR) and immunoblotting showed higher HKDC1 in intrinsically less sensitive lines and sustained elevation in LR versus LS derivatives of Hep3B and HCCLM3, with concordant trends in PLC/PRF/5 and Hep1-6 (
Fig. 1D,
Supplementary Fig. 1D, 1E), supporting HKDC1 upregulation as a hallmark of LR. These results place HKDC1 upregulation among the hallmark features of the LR state.
Clinically, immunohistochemistry (IHC) in the Zhongshan cohort demonstrated higher HKDC1 protein in HCC than in adjacent liver (
Fig. 1E), and elevated HKDC1 predicted worse overall survival (
Fig. 1F). Moreover, high HKDC1 expression was significantly associated with increased tumor number and larger tumor size (
Supplementary Table 1). The Cancer Genome Atlas Liver Hepatocellular Carcinoma (TCGA-LIHC) analysis confirmed that HKDC1 mRNA levels were elevated in tumor tissues, and patients in the high-expression group had shorter survival times (
Supplementary Fig. 1F). Taken together with prior work implicating HKDC1 in HCC biology and immune escape [
11] and the clinical importance of lenvatinib in frontline HCC care [
2], our data nominate HKDC1 as a lenvatinib resistance–associated metabolic node linked to malignant progression in HCC.
To test whether HKDC1 drives lenvatinib resistance, we silenced HKDC1 in independently generated LR HCC cells and verified efficient knockdown by qRT-PCR and Western blot in Hep3B-LR and HCCLM3-LR (
Fig. 2A), with validation in PLC/PRF/5-LR and Hepa1-6-LR (
Supplementary Fig. 2A). Across all cells, HKDC1 loss markedly increased lenvatinib sensitivity: dose–response curves shifted left with reduced IC50, and short-term growth measured by CCK-8 declined more with combined lenvatinib plus shHKDC1 (or shHkdc1; unless otherwise specified, “HKDC1” uniformly denotes the gene in mouse and human) than with either treatment alone (Hep3B-LR, HCCLM3-LR:
Fig. 2B; PLC/PRF/5-LR, Hep1-6-LR:
Supplementary Fig. 2B). Long-term clonogenicity was impaired by HKDC1 knockdown and nearly abolished by the combination in all LR lines (
Fig. 2B,
Supplementary Fig. 2B). DNA synthesis (EdU) decreased after HKDC1 knockdown and further with lenvatinib co-treatment in Hep3B-LR and HCCLM3-LR (
Fig. 2C) and in PLC/PRF/5-LR and Hep1-6-LR (
Supplementary Fig. 2C).
In vivo, HKDC1 silencing curtailed tumor growth of LR cell–derived xenografts, with the lenvatinib combination yielding the strongest suppression of endpoint volume and weight (HCCLM3-LR:
Fig. 2D; Hep1-6-LR:
Supplementary Fig. 2D). IHC confirmed decreased HKDC1 staining in shHKDC1 tumors; proliferation (Ki67) was lowest with the combination (
Supplementary Fig. 2E). Collectively, HKDC1 knockdown suppresses LR HCC cell proliferation and resensitizes tumors to lenvatinib across human and mouse models.
To define HKDC1’s downstream mechanism sustaining lenvatinib resistance, we profiled Hep3B-LR after HKDC1 knockdown. RNA-seq revealed broad shifts enriched for metabolic pathways (
Fig. 3A). Integrating with proteomics (
Fig. 1C) highlighted four overlapping metabolic candidates— CYP3A7, UGT2B10, HPSE, and SMS—with SMS most reduced on HKDC1 silencing (
Fig. 3A). In TCGA-LIHC, SMS mRNA was higher in tumors than normal liver, UGT2B10 decreased, and CYP3A7/HPSE were unchanged; SMS correlated with HKDC1 (
Fig. 3B,
Supplementary Fig. 3A). qRT-PCR validated SMS mRNA elevation across LR lines (
Fig. 3B). SMS protein increased in LR vs. LS cells and stained stronger in LR xenografts (
Fig. 3B,
Supplementary Fig. 3B).
Metabolically, LR cells exhibited the SMS-typical polyamine shift (SPD/spermine)—reduced SPD and increased spermine— in Hep3B-LR and HCCLM3-LR, with similar trends in PLC/PRF/5-LR and Hep1-6-LR (
Fig. 3C,
Supplementary Fig. 3C). IHC analysis of LR tumor tissues showed concurrently decreased SPD and increased spermine levels
in vivo (
Supplementary Fig. 3D). These results align with the canonical role of SMS in converting SPD to spermine in cancer cells (
Supplementary Fig. 3B) [
13]. HKDC1 inhibition lowered SMS protein in Hep3B-LR and HCCLM3-LR (
Fig. 3D) and in PLC/PRF/5-LR and Hep1-6-LR (
Supplementary Fig. 3E). At the metabolite level, HKDC1 loss restored SPD to LS-like levels; co-depleting SMS prevented further SPD increases upon HKDC1 knockdown, positioning SMS downstream of HKDC1 (
Fig. 3D,
Supplementary Fig. 3E).
HKDC1 is known to act as an mRNA-binding protein to promote gastric cancer metastasis and treatment resistance [
10]. Given the significant effect of HKDC1 inhibition on SMS protein levels (
Fig. 3D,
Supplementary Fig. 3E) and the increase of SMS mRNA in LR cells (
Fig. 3B), we hypothesized that HKDC1 may bind to SMS mRNA and regulate its levels. RNA immunoprecipitation (RIP)–qPCR showed selective enrichment of SMS transcripts by anti-HKDC1 vs. IgG in LR cells (
Fig. 3E,
Supplementary Fig. 3F). Transcriptional blockade with actinomycin D revealed accelerated SMS mRNA decay after HKDC1 inhibition (
Fig. 3F,
Supplementary Fig. 3G). We further investigated whether SMS overexpression could reverse the effects of HKDC1 knockdown in PLC/PRF/5-LR cells. Indeed, SMS overexpression counteracted the inhibitory effects of HKDC1 knockdown on tumor cell growth and drug resistance (
Supplementary Fig. 3H). These findings define a post-transcriptional mechanism: HKDC1 binds/stabilizes SMS mRNA, elevates SMS protein, and rewires polyamine metabolism in LR HCC. Given polyamines’ role in growth and immune modulation [
13], targeting HKDC1/SMS may normalize polyamine imbalance.
Given that SPD modulates CD8
+ T cells and augments checkpoint blockade efficacy [
25], we asked whether LR-driven polyamine rewiring reshapes the TIME. In an orthotopic Hep1-6 C57BL/6 model, LR tumors showed higher liver-to-body-weight ratios and increased Ki67 versus LS (
Fig. 4A). Single-cell RNA-seq (16 clusters) revealed a shift toward macrophages and away from NK and cytotoxic CD8
+ T cells, with marker-based feature/violin plots supporting cell-type annotations (
Fig. 4B). We next quantified effector memory (TemScore), activation (ActScore), and exhaustion (Exh-Score) gene-module scores within CD8
+ T-cell clusters. Compared with LR tumors, CD8
+ T cells in LS tumors exhibited markedly higher effector memory and activation scores, whereas canonical exhaustion-marker expression was largely comparable between groups (
Supplementary Fig. 4A), so our subsequent analyses focused on effector memory and activated CD8
+ T cells rather than exhausted subsets. Multiplex IHC and flow cytometry validated denser F4/80
+ CD11b
+ macrophage infiltration and fewer CD8
+ T cells with reduced perforin and granzyme B positivity (PFN
+ and GzmB
+) in LR tumors (
Fig. 4C). Given HKDC1’s reported STAT1-dependent control of PD-L1 in liver cancer [
11], we examined phospho-STAT1 (pSTAT1) and PD-L1 protein levels in LS and LR cells. HKDC1 knockdown reduced PD-L1 and pSTAT1 without altering total STAT1 in four LS lines (
Supplementary Fig. 4B), but not consistently in LR (
Fig. 4D), implicating alternative, polyamine-based TIME modulation.
Considering SPD potentiates the efficacy of aPD-1 therapy by enhancing CD8
+ T-cell mitochondrial function and cytotoxicity [
25], and that LR cells reduce SPD via the HKDC1/SMS axis (
Fig. 3,
Supplementary Fig. 3), we hypothesized that LR cells impair CD8
+ T-cell mitochondrial fitness. CD8
+ T cells exposed to LR-conditioned medium (CM) displayed reduced oxygen consumption rate/extracellular acidification rate (OCR/ECAR), lower mitochondrial, glycolytic and total ATP, diminished spare respiratory capacity, and impaired proliferative capacity (
Fig. 4E, 4F)—a low-glycolysis, low-spare respiratory capacity (SRC) state typical of dysfunctional effector CD8
+ T cells with poor persistence [
27,
28].
In vivo, SPD abolished LS–LR growth differences (
Supplementary Fig. 4C) and partially normalized the LR TIME by restoring CD8
+ T infiltration while macrophage disparities persisted (
Supplementary Fig. 4D). After silencing HKDC1, the CM of Hep1-6 cells (LR vs. LS) brought CD8
+ T cells to equilibrium in terms of OCR, ECAR, ATP distribution and SRC (
Supplementary Fig. 4E). SPD can also restore NK cell function [
29,
30], we further analyzed the relationship between NK cells and CD8
+ T cells. Despite NK cell depletion [
31], significant differences remained between the LS and LR groups in the number of CD8
+ T cells and the proportions of GZMB
+ and perforin
+ CD8
+ T cells (
Supplementary Fig. 4F), indicating that the observed changes in CD8
+ T cells were independent of the presence of NK cells.
Collectively, these data support LR HCC cells, through HKDC1-dependent spermine synthesis, remodeling the TIME—increasing macrophage infiltration, limiting CD8
+ T cell recruitment, and shifting CD8
+ T cell metabolism toward reduced respiratory capacity. Consistent with polyamines’ effects on TME and T cells [
13,
32], SPD supplementation can alleviate LR tumor-driven metabolic immunosuppression and may potentiate responses to PD-1 blockade.
Using an unbiased DNA–protein pull-down with a 2-kb HKDC1 promoter bait in Hep3B-LR cells, we detected a distinct 25–35 kDa band and proteomic enrichment of USF1 (
Fig. 5A,
Supplementary Fig. 5A). Direct USF1–promoter binding was validated by pull-down/immunoblotting and dual-luciferase reporter assays, which showed USF1-dependent transactivation of the HKDC1 promoter (
Fig. 5B,
Supplementary Fig. 5A). Chromatin immunoprecipitation (ChIP) mapped this interaction to promoter “Region 2,” with clear enrichment by ChIP-PCR/Qpcr (
Fig. 5C,
Supplementary Fig. 5B), consistent with the known function of USF1 as an E-box–binding basic helix loop helix leucine zipper (bHLH-LZ) transcription factor implicated in liver cancer progression [
33].
USF1 protein levels were higher in LR than in matched LS cells and USF1 overexpression increased HKDC1 protein (
Fig. 5D,
Supplementary Fig. 5C). Both EdU incorporation and clonogenic growth rose with USF1 but were abrogated when HKDC1 was silenced (
Fig. 5E,
Supplementary Fig. 5D), indicating HKDC1 acts downstream of USF1 to sustain the LR phenotype. Together with prior reports that HKDC1 promotes aggressive liver cancer phenotypes, including immune evasion [
11], these data implicate a USF1–HKDC1 transcriptional axis as a mechanistic driver of lenvatinib resistance.
These effects generalized across human and mouse LR models.
In vivo, IHC revealed stronger USF1 staining in LR tumors than in LS counterparts irrespective of drug exposure (
Fig. 5F). Clinically, TCGA-LIHC analyses showed that USF1 mRNA positively correlates with HKDC1 expression, is elevated in tumors versus normal liver, and associates with worse overall survival (
Supplementary Fig. 5E). Additional experiments were conducted to investigate the impact of USF1 on SMS, SPD, spermine levels, and CD8
+ T cell function in Hep1-6-LR tumor tissue. IHC results indicate that USF1 overexpression significantly increases SMS and spermine levels while decreasing SPD levels, as reflected in the corresponding IHC scores (
Supplementary Fig. 5F). Flow cytometry analysis of CD8
+ T cell function revealed that USF1 overexpression suppresses perforin and granzyme B production in CD8
+ T cells (
Supplementary Fig. 5F). These findings suggest that USF1 regulates the levels of SMS, spermine, and SPD, and its overexpression contributes to inhibition of CD8
+ T cell function in the context of lenvatinib resistance. It nominates USF1–HKDC1–SMS activation as a driver of lenvatinib resistance and a tractable therapeutic target.
Because the USF1/HKDC1/SMS axis markedly suppresses SPD, we tested whether exogenous SPD mitigates lenvatinib resistance. In Hep3B-LR and HCCLM3-LR, Lenvatinib+SPD (Len+SPD) most strongly inhibited proliferation: EdU was lowest and clonogenic growth was nearly abolished (
Fig. 6A). In HCCLM3-LR xenografts, Len+SPD minimized tumor weight/volume and Ki67 (
Fig. 6B). PLC/PRF/5-LR and Hep1-6-LR showed concordant effects: reduced EdU, suppressed clonogenicity (
Supplementary Fig. 6A), and superior control in Hep1-6-LR subcutaneous tumors (
Supplementary Fig. 6B).
To assess translational potential, we generated HCC organoids from two patients. In both, neither SPD nor lenvatinib alone appreciably reduced growth, whereas Len+SPD produced the greatest decrease in mean organoid diameter (
Fig. 6C,
Supplementary Fig. 6C). In the corresponding patient-derived organoid xenograft (PDOX) models (PDOX-1 and PDOX-2), Len+SPD consistently lowered endpoint tumor weight and volume versus either monotherapy (
Fig. 6C,
Supplementary Fig. 6C). IHC analysis revealed elevated expression of HKDC1 in LR PDOX tumor tissues compared with a sensitive PDOX-3 control (
Supplementary Fig. 6D), suggesting that upregulation of HKDC1 may contribute to the development of lenvatinib resistance.
Since SPD is a physiological polyamine and a potent inducer of autophagy [
34], and that excessive autophagy can culminate in cell death [
20,
35], we posited that lenvatinib might elicit autophagy-like death when coupled to SPD. In LS cells, lenvatinib alone induced abundant double-membrane autophagosomes by transmission electron microscopy (TEM) and increased LC3B-II/I ratio, alongside apoptosis-like morphology, but these changes were absent in LR under lenvatinib alone (
Fig. 6D,
Supplementary Fig. 6E). Knockdown of HKDC1 or exogenous supplementation of SPD could restore lenvatinib’s induction of autophagy-like cell death in LR cells (
Fig. 6D,
Supplementary Fig. 6E, 6F). However, after the addition of the autophagy inhibitor SAR405, the restoration of lenvatinib sensitivity mediated by SPD supplementation or HKDC1 knockdown was essentially lost (
Supplementary Fig. 6F), indicating that HKDC1 and SPD are involved in mediating lenvatinib treatment resistance by regulating autophagy-like cell death in LR cells.
To test whether HKDC1 inhibition sensitizes tumors to lenvatinib plus immunotherapy, we generated primary HCC in immunocompetent C57BL/6 gene-edited mice (
Supplementary Fig. 7A) via hydrodynamic tail-vein delivery of sgRNA-p53, SB13, and CTNNB1-Δ90 plasmids (
Fig. 7A). Alb-Cre; Hkdc1
+/+ (control) and Alb-Cre; Hkdc1
f/f (hepatocyte-specific knockout) mice were randomized to placebo, lenvatinib, aPD-1, or the combination (
Fig. 7A). Gross livers and H&E sections showed that Hkdc1 deletion enhanced antitumor efficacy for both monotherapies and the combination relative to matched regimens in wild-type mice (
Fig. 7B). Quantitatively, HKDC1 loss lowered liver-to-body weight ratios and reduced tumor nodule counts across arms, with the greatest reduction under lenvatinib plus aPD-1 (
Fig. 7B). These data position HKDC1 as an intrinsic barrier to TKI–ICI therapy and support genetic or pharmacologic inhibition to sensitize the lenvatinib+aPD-1 regimen.
To assess clinical relevance, we analyzed resected primary tumors (n=40) and paired blood from patients who later received lenvatinib plus aPD-1 at recurrence. Patients were stratified by PFS (sensitive: PFS ≥12 months, n=14; resistant: PFS <12 months, n=26) (
Fig. 7C). MRI demonstrated regression in sensitive cases and progression in resistant ones (
Fig. 7C,
Supplementary Fig. 7B). IHC revealed lower HKDC1 in sensitive tumors and higher in resistant tumors (
Fig. 7C,
Supplementary Fig. 7B). When stratified by primary-tumor HKDC1 IHC score, peripheral blood immune profiles diverged: low-HKDC1 cases had higher NK and CD8
+ T-cell proportions, whereas high-HKDC1 cases showed an immunologically “cold” phenotype (
Fig. 7D). Patients with high HKDC1 exhibited higher AFP and ALT and lower albumin (
Supplementary Fig. 7C). Dichotomization by median IHC score showed HKDC1-Low contained more clinically sensitive cases (11/20, 55%), whereas HKDC1-High was predominantly resistant (17/20, 85%) (
Fig. 7E). Low HKDC1 associated with longer PFS (
Fig. 7E).
These findings dovetail with our earlier evidence that HKDC1 upregulates SMS and remodels polyamine flux, thereby dampening antitumor immunity. In the current models, Hkdc1 deletion synergized with lenvatinib plus aPD-1 therapy to lower tumor burden, likely by permitting better effector T-cell recruitment and activation (
Fig. 7F). Collectively, mouse and patient data nominate HKDC1 as a predictor of poor response and a tractable metabolic target for resensitizing HCC to lenvatinib plus aPD-1.
DISCUSSION
We delineate a tumor-intrinsic USF1/HKDC1/SMS axis coupling metabolic rewiring to immune escape and lenvatinib resistance in HCC: USF1 upregulates HKDC1, enabling its RBP-like function in post-transcriptional cell state programming [
36]. By sustaining SMS, HKDC1 shifts SPD to spermine, suppressing CD8
+ T mitochondrial respiration [
25] and autophagic death, blunting cytotoxic marker expression, and favoring tumor-associated macrophage (TAM) infiltration [
37]— features of an immune-excluded “cold” ecosystem [
13,
32]. HKDC1 loss or SPD restoration reverses effects and improves response.
The immunologic consequences of polyamine reprogramming unify several observations. Conditioned media from LR tumors impaired T-cell mitochondrial function, whereas SPD supplementation normalized these deficits, in line with SPD’s capacity to enhance CD8
+ T fatty-acid oxidation and mitochondrial fitness [
25]. HKDC1 silencing collapsed LS–LR metabolic and immune disparities, implicating the HKDC1/SMS node as a necessary driver of the immunosuppressive TME. While prior work linked HKDC1 to STAT1/PD-L1 regulation in HCC [
11], our data indicate that in the LR state the dominant immune-modulatory route is polyamine-dependent, aligning with a broader literature that places polyamine metabolism as a master regulator of T-cell lineage and function [
13,
25,
32].
Clinically, lenvatinib is a frontline standard in unresectable HCC, yet adaptive resistance is frequent. Our data show that lenvatinib alone elicits robust autophagy-like cell death in sensitive models but not in LR cells; critically, SPD co-treatment reinstates this death program in LR cells. Mechanistically, SPD is a potent autophagy inducer— through EP300 inhibition and histone hypoacetylation [
38]. We interpret our ultrastructural and LC3B evidence as “autophagy-like cell death” heeding the caveats articulated by Kroemer and Levine [
35] and recognizing precedents in which oncogenic stress culminates in autophagy-associated lethality [
20]. Together, these data support that HKDC1–SMS depletes SPD, blunting lenvatinib-induced autophagic death; SPD restoration reinstates lethal autophagy and tumor control.
A kinome-wide CRISPR screen identified EGFR activation as a driver of lenvatinib resistance and showed EGFR inhibition synergizes with lenvatinib [
26]. Consistently, our LR model exhibits elevated EGFR (
Fig. 1C), validating the model and highlighting heterogeneous resistance. Beyond EGFR, we uncover a metabolism-anchored USF1/HKDC1/SMS axis tied to polyamine metabolism, supporting combinations that target polyamines alongside lenvatinib. The polyamine pathway governs tumor growth and T-cell function [
13,
32], and SPD serves as an immunometabolism adjuvant: in mice, SPD activates mitochondrial trifunctional protein and enhances antitumor immunity, including checkpoint responses [
25].
Using acquired LR models, we show the HKDC1/SMS/SPD axis as a central mechanism of lenvatinib resistance. SPD synergizes with lenvatinib in LR HCC cells and patient-derived organoids, restoring growth inhibition. Given SPD’s roles in metabolism and immunity, we propose polyamine metabolism as the mechanistic nexus underlying resistance to lenvatinib plus aPD-1. Although anti-VEGF/FGF agents can transiently normalize vasculature and favor immune infiltration [
39], clinical results are mixed—activity in single-arm studies [
40] but no overall survival (OS) benefit in a phase 3 trial [
41]—indicating resistance to targeted–immunotherapy combinations. HKDC1, a pleiotropic effector that assembles ribonucleoprotein complexes to rewire metabolism [
10], stabilizes SMS mRNA, depletes SPD, and confers resistance to lenvatinib plus aPD-1. Thus, HKDC1 could be a therapeutic target to overcome resistance to lenvatinib and PD-1 blockade.
Clinical data link higher HKDC1 with poor PFS and “cold” peripheral immunity, identifying HKDC1 as predictive and prognostic in recurrent HCC on Len+aPD-1. Given lenvatinib’s multi-kinase spectrum and heterogeneous outcomes of its ICI combinations [
2,
40,
41], stratifying patients by tumor HKDC1—and potentially circulating SPD/spermine ratios— offers a rational trial framework. Therapeutically, three non-exclusive strategies merit testing: (i) HKDC1 inhibition (as agents emerge); (ii) SMS inhibition/rebalanced polyamine catabolism; and (iii) SPD supplementation to restore T-cell metabolism, amplify lenvatinib-induced autophagic death, and enhance responses to RTK inhibitors and PD-1 blockade; careful dosing/safety are essential.
Several limitations merit discussion. First, while we mapped USF1 binding to a defined promoter region and demonstrated reporter transactivation and dependency, USF1 often cooperates with partner factors [
42,
43]. Comprehensive chromatin-level profiling (e.g., CUT&RUN/CUT&Tag with cofactor mapping) would refine mechanism and therapeutic tractability [
44]. Reliance on genetic perturbations and exogenous SPD in preclinical models raises pharmacologic-fidelity concerns. Future work should deploy small-molecule modulators of the USF1/HKDC1/SMS axis and dose-ranging SPD studies in immunocompetent HCC to strengthen causality and safety. Although murine orthotopic and hydrodynamic systems capture immunity, species differences in polyamine transport and T-cell biology limit generalizability; prospective, HKDC1-stratified cohorts with standardized immune-metabolic correlatives are needed to validate predictive utility. Because “autophagic cell death” remains contested, orthogonal assays—genetic blockade of core autophagy genes and real-time flux reporters— should disentangle lethal autophagy from apoptosis-coupled autophagy under Len+SPD [
45]. LR cells drive an immunosuppressive microenvironment affecting CD8
+ T cells, macrophages, and NK cells, yet SPD selectively restores CD8
+ T and NK cell function [
25,
29,
30]. Although SPD has been reported to affect both CD8
+ T cells and NK cells, our NK-cell-depletion analysis supports that the CD8
+ T-cell impairment observed in LR tumors is not secondary to the loss of NK cells. These findings further support a CD8
+ T-cell-centered immunometabolic mechanism. However, the observed changes in macrophages may require further investigation through additional experiments.
By defining a USF1/HKDC1/SMS circuit linking polyamine metabolism to T-cell bioenergetics and lenvatinib responsiveness, we recast lenvatinib resistance as metabolically buffered and immunosuppressive. SPD reinstates lenvatinib-induced tumor killing, and HKDC1 loss boosts Len+aPD-1 responses—potentially converting intrinsically “cold” advanced HCC into combination-responsive disease with deeper, more durable benefit.
FOOTNOTES
-
Authors’ contribution
Methodology, Junxian Du, Jinglei Wan, Guiqi Zhu, and Lina Song; investigation, Yufan Cai, Yi Yang, Jialiang Cai, and Yujie Shao; writing–manuscript, Shiping Chen; review and revising writing, Zhi Dai; visualization, Peiling Zhang, Shiping Chen, and Bing Quan; project administration, Yang Zhang and Biao Wang; conceptualization and funding acquisition, Zhi Dai.
-
Acknowledgements
This research was supported by the National Natural Science Foundation of China (82372946, 82072670), the Shanghai Municipal Science and Technology Commission (21Y21900100), and the Beijing Bethune Charitable Foundation (ZLZX019).
-
Conflicts of Interest
The authors have no conflicts to disclose.
SUPPLEMENTARY MATERIAL
Supplementary material is available at Clinical and Molecular Hepatology website (
http://www.e-cmh.org).
Supplementary Figure 1.
HKDC1 is associated with lenvatinib resistance and malignant progression in HCC. (A) Colony formation assay showing the number of colonies in Hep3B-WT, Huh7-WT, PLC-WT, HCCLM3-WT, MHCC97H-WT, and Hep1-6-WT cells treated with different concentrations of lenvatinib, n=3. (B) Morphological changes of PLC/PRF/5 and Hep1-6 cells before and after induction of resistance to lenvatinib in vitro. (C) IC50 values, CCK-8 proliferation, and clonogenic assays were used to evaluate lenvatinib response in sensitive (LS), and resistant (LR) PLC/PRF/5 and Hep1-6 cells (n=3). (D) Lenvatinib efficacy in Hep1-6 LS and LR subcutaneous xenografts: tumor images/volume/weight and Ki67 IHC staining, with DMSO as control (n=6/group). (E) qRT-PCR and Western blot assays were used to detect the relative expression levels of HKDC1 in PLC/PRF/5 and Hep1-6 (LR vs. LS) cells, n=3. (F) HKDC1 expression in TCGALIHC (tumor, n=374; normal, n=50) and its correlation with OS. Data are presented as mean±standard deviation, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001, ns means not significant. HKDC1, hexokinase domain containing protein 1; HCC, hepatocellular carcinoma; LS, lenvatinib-sensitive; LR, lenvatinib-resistant; qRT-PCR, quantitative reverse transcription polymerase chain reaction; TCGA-LIHC, The Cancer Genome Atlas Liver Hepatocellular Carcinoma; OS, overall survival.
cmh-2025-1269-Supplementary-Fig-1.pdf
Supplementary Figure 2.
Reducing HKDC1 expression can significantly inhibit the proliferation of LR cells and enhance lenvatinib sensitivity. (A) The knockdown efficiency of HKDC1 by shRNA was confirmed via qRT-PCR and Western blot in PLC/PRF/5-LR and Hep1-6- LR cells, with shNC as the negative control. (B) IC50 values, short-term proliferation (CCK-8), and long-term clonogenic capacity were compared in PLC/PRF/5-LR and Hep1-6-LR cells after HKDC1 knockdown (shHKDC1 vs. shNC). (C) The effect of HKDC1 knockdown on lenvatinib efficacy was assessed by EdU (Azide 555) proliferation assay in PLC/PRF/5-LR and Hep1-6-LR cells treated with 25 μM or 40 μM lenvatinib, respectively (n=3). (D) Tumor volume, endpoint weight, and images of Hep1-6-LR xenografts (shHkdc1 vs. shNC) from nude mice treated with lenvatinib or vehicle (DMSO) are shown (n=6 per group), scale bars, 1 cm. (E) Tumor proliferation (Ki67) and HKDC1 expression were assessed by IHC in xenografts from mice treated with lenvatinib or vehicle (DMSO). Data are presented as mean±standard deviation, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001, ns means not significant. HKDC1, hexokinase domain containing protein 1; LR, lenvatinib-resistant; qRT-PCR, quantitative reverse transcription polymerase chain reaction.
cmh-2025-1269-Supplementary-Fig-2.pdf
Supplementary Figure 3.
HKDC1 regulates polyamine metabolism by binding to and enhancing SMS mRNA stability. (A) Expression of CYP3A7, UGT2B10, and HPSE in HCC vs. normal liver (TCGA-LIHC) and Pearson correlation between HKDC1 and SMS mRNA. (B) Schematic representation of the polyamine metabolic pathway, SMS protein levels in PLC/PRF/5 and Hep1-6 cells (LR vs. LS), and IHC staining of SMS in subcutaneous tumors. (C) The polyamines spermidine and spermine were quantified by ELISA in PLC/PRF/5 and Hep1-6 cells (LR vs. LS; n=3), and the ratio of spermidine to spermine was calculated for LR cells. (D) Spermidine and spermine levels were detected by IHC in tumor tissues from HCCLM3 and Hep1-6 models (LS vs. LR). n=6 per group. (E) HKDC1 regulates SMS to control spermidine levels in PLC/PRF/5-LR and Hep1-6-LR cells, n=6. (F) RIP-qPCR analysis of HKDC1 binding to SMS mRNA in PLC/PRF/5-LR and Hep1-6-LR cells, n=3. (G) Effect of HKDC1 knockdown on SMS mRNA stability in PLC/PRF/5-LR and Hep1-6-LR cells, n=3. (H) Effects of SMS overexpression and HKDC1 knockdown on cell growth and lenvatinib resistance. Data are presented as mean±standard deviation, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001, ns means not significant. HKDC1, hexokinase domain containing protein 1; SMS, spermine synthase; LR, lenvatinib-resistant; LS, lenvatinib-sensitive; ELISA, enzyme-linked immunosorbent assay.
cmh-2025-1269-Supplementary-Fig-3.pdf
Supplementary Figure 4.
LR cells can form a suppressive immune microenvironment by regulating polyamine metabolism (A) Compare the effector memory (TemScore: Cd44, Cx3cr1, Klrg1, Il7r, Gzmb, Prf1, Ifng, Tnf), activation (ActScore: Cd69, Il2ra, Icos, Ifng, Tnf, Gzmb, Prf1, Nkg7, Mki67), and exhaustion (ExhScore: Pdcd1, Havcr2, Lag3, Tigit, Ctla4, Tox, Eomes) gene module scores within CD8+ T cell clusters in LS and LR tumors. (B) To investigate the effects of HKDC1 inhibition on PD-L1, pSTAT1, and STAT1 protein levels in LS cells. (C) Representative images and liver/body weight ratios of LS/LR tumors, and Ki67 IHC staining, in C57BL/6 mice treated with SPD (n=6/ group). (D) Immune cell infiltration (F4/80+ macrophages, CD8+ T cells) in LS/LR tumors with SPD treatment, analyzed by mIHC and flow cytometry (n=6/group). (E) Metabolic function (OCR, ECAR, ATP production, SRC) of CD8+ T cells treated with supernatants from Hep1- 6-LS-shHkdc1 and Hep1-6-LR-shHkdc1 cells (n=3/group). (F) Flow cytometry was used to analyze the number of NK1.1+ cells and CD8+ T cells in Hep1-6-LS and LR tumor-bearing C57BL/6 mice treated with either anti-NK1.1 antibody or IgG control (NK cells depletion method as previously reported31). n=6/group. Data are presented as mean±standard deviation, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001, ns means not significant. LR, lenvatinib-resistant; LS, lenvatinib-sensitive; HKDC1, hexokinase domain containing protein 1; PD-L1, programmed death-ligand 1; pSTAT1, phospho-STAT1; IHC, immunohistochemistry; SPD, spermidine; mIHC, multiplex fluorescent IHC; OCR, oxygen consumption rate; ECAR, extracellular acidification rate; SRC, spare respiratory capacity.
cmh-2025-1269-Supplementary-Fig-4.pdf
Supplementary Figure 5.
USF1 facilitates lenvatinib resistance in HCC through transcriptionally activating HKDC1. (A) The binding of USF1 to the HKDC1 promoter and its regulatory role were verified by DNA pulldown combined with Western blot and dual-luciferase reporter assays, respectively, in PLC/PRF/5-LR and Hep1-6-LR cells (n=3). (B) ChIP-PCR and ChIP-qPCR were used to detect and quantify, respectively, the enrichment of the HKDC1 promoter ‘region 2’ mediated by USF1 in PLC/PRF/5-LR and Hep1-6-LR cells (n=3). (C) Western blot was used to determine HKDC1 expression in the LR vs. LS state and its regulation by USF1 in PLC/PRF/5-LR and Hep1-6- LR cells, respectively. (D) EdU and colony formation assays evaluated the effect of the USF1-HKDC1 axis on proliferation of PLC/PRF/5- LR and Hep1-6-LR cells under lenvatinib treatment (25 μM and 40 μM, respectively; n=3). (E) Correlation between USF1 and HKDC1 mRNA expression, comparison of USF1 levels between tumor (n=374) and normal (n=50) tissues, and OS curves for patients with high or low USF1 expression in the TCGA-LIHC dataset. (F) The effects of USF1 on downstream factors such as SMS, SPD, spermidine levels, and CD8+ T cell function were examined. Data are presented as mean±standard deviation, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001, ns means not significant. USF1, upstream stimulatory factor 1; HCC, hepatocellular carcinoma; HKDC1, hexokinase domain containing protein 1; LR, lenvatinib-resistant; ChIP, chromatin immunoprecipitation; PCR, polymerase chain reaction; LS, lenvatinib-sensitive; OS, overall survival; SMS, spermine synthase; SPD, spermidine.
cmh-2025-1269-Supplementary-Fig-5.pdf
Supplementary Figure 6.
SPD augments the efficacy of lenvatinib by promoting autophagy-like cell death. (A) Combined lenvatinib and SPD synergistically inhibit proliferation and clonogenicity of Hep1-6-LR cells (n=3/group). (B) Combined lenvatinib and SPD suppress tumor growth and proliferation (Ki67 IHC staining) in Hep1-6-LR xenografts (n=6/group), scale bars, 1 cm. (C) Representative images and mean diameter of patient-derived organoids (Patient 2) under different treatments. n=3/group; tumor images, weight, and volume of the Patient 2-derived organoid xenograft (PDOX-2) model under different treatments (n=6/group). (D) Representative images of HKDC1 IHC staining in PDOX-3, PDOX-1, and PDOX-2 tumor models (n=6/group). (E) Autophagic responses in PLC/PRF/5 and Hep1-6 cells (LS vs. LR) under different treatments for 24 h, shown by TEM ultrastructure (red arrows indicate autophagosomes) and LC3B-II Western blot. (F) Hep3B-LR cells subjected to HKDC1 knockdown and SPD treatment, with or without co-treatment of the autophagy inhibitor SAR405 (5 μM, 24 h), were subsequently assessed for lenvatinib-induced autophagic cell death. Data are presented as mean±standard deviation, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001, ns means not significant. SPD, spermidine; Len, lenvatinib; IHC, immunohistochemistry; LR, lenvatinib-resistant; HKDC1, hexokinase domain containing protein 1; LS, lenvatinib-sensitive; TEM, transmission electron microscopy.
cmh-2025-1269-Supplementary-Fig-6.pdf
Supplementary Figure 7.
HKDC1 loss enhances HCC response to lenvatinib plus aPD-1 therapy via TIME modulation. (A) Schematic diagram of the C57BL/6 hepatocyte-specific Hkdc1 gene-knockout mouse. (B) Pre- and post-treatment MRI and corresponding HKDC1 IHC staining in sensitive (Case 3) and resistant (Case 4) HCC patients. (C) Quantification of serum biomarkers and liver function parameters in HCC patients stratified by HKDC1 expression level, including AFP, CA199, TB, DB, Alb, ALT, AST, and TNFα. Data are presented as mean±standard deviation. HKDC1, hexokinase domain containing protein 1; HCC, hepatocellular carcinoma; aPD-1, programmed cell death protein 1 antibody; IHC, immunohistochemistry; AFP, alpha-fetoprotein; CA199, carbohydrate antigen 199; TB, total bilirubin; DB, direct bilirubin; Alb, albumin; ALT, alanine aminotransferase; AST, aspartate aminotransferase; TNFα, tumor necrosis factor alpha.
cmh-2025-1269-Supplementary-Fig-7.pdf
Figure 1HKDC1 is associated with lenvatinib resistance and malignant progression in HCC. (A) Baseline lenvatinib IC50 in wild-type HCC lines (n=3/group), colony formation under lenvatinib, schematic of in-vitro induction of LR lines, and morphology of Hep3B/HCCLM3 before vs. after LR induction. (B) Comparative IC50 and colony assays for LS vs. LR Hep3B/HCCLM3; CCK-8 proliferation at 5 and 40 μM; in vivo HCCLM3 LS/LR xenograft tumor volume/weight under lenvatinib (n=6/group). (C) Proteomics (Hep3B LR vs. LS): 4,294 proteins quantified; 463 differentially expressed (252 up, 211 down); KEGG enrichment of the 252 upregulated (DAVID). (D) HKDC1 expression by qRT-PCR/Western in five wild-type HCC lines and in Hep3B/HCCLM3 (LR vs. LS; n=3); IHC for Ki67 and HKDC1 in HCCLM3 LR/LS xenografts with/without lenvatinib. (E) IHC images (left) and score analysis (right) showing HKDC1 levels in normal liver (n=209) vs. HCC (n=231). (F) Using an R-derived optimal HKDC1 IHC cutoff (92.5933), 230 HCC patients (one missing survival) were dichotomized for overall-survival analysis. Data are presented as mean±standard deviation, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001, ns means not significant. HKDC1, hexokinase domain containing protein 1; HCC, hepatocellular carcinoma; LR, lenvatinib-resistant; LS, lenvatinib-sensitive; qRT-PCR, quantitative reverse transcription polymerase chain reaction; IHC, immunohistochemistry; Len, lenvatinib; WT, wildtype.
Figure 2Reducing HKDC1 expression can significantly inhibit the proliferation of LR cells and enhance lenvatinib sensitivity. (A) shRNA-mediated HKDC1 knockdown was validated by qRT-PCR and Western in Hep3B-LR and HCCLM3-LR (shNC control). (B) Post-knockdown, lenvatinib IC50, short-term proliferation (CCK-8), and long-term clonogenicity were compared in both LR lines. (C) EdU (Azide 555) assays assessed proliferation under 5 μM (Hep3B-LR) or 40 μM (HCCLM3-LR) lenvatinib (n=3). (D) In nude mice with HCCLM3-LR xenografts (shHKDC1 vs. shNC), tumor volume, endpoint weight, and gross images were recorded under lenvatinib or vehicle (DMSO) (n=6/group), scale bars, 1 cm. Data are presented as mean±standard deviation, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001, ns means not significant. HKDC1, hexokinase domain containing protein 1; LR, lenvatinib-resistant; qRT-PCR, quantitative reverse transcription polymerase chain reaction.
Figure 3HKDC1 regulates polyamine metabolism by binding to and enhancing SMS mRNA stability. (A) RNA-seq of Hep3B-LR (shHKDC1 vs. shNC) identified differentially expressed genes (DEGs) and KEGG enrichment was performed. The study focused on overlapping between metabolic genes (n=136;
Fig. 3A) and prior proteomics (LR vs. LS; n=38;
Fig. 1C). (B) SMS was elevated in HCC vs. normal (TCGA-LIHC, n=374/50); SMS mRNA was higher in LR vs. LS across multiple lines and protein was elevated in LR; IHC confirmed increased SMS in HCCLM3-LR xenografts. (C) ELISA quantified spermidine or spermine in HCCLM3 and Hep3B (LR vs. LS; n=3). (D) HKDC1 regulates SMS to control spermidine in Hep3B-LR/HCCLM3-LR (n=6). (E) RIP–qPCR showed HKDC1 binding to SMS mRNA (n=3). (F) HKDC1 knockdown reduced SMS mRNA stability in Hep3B-LR/HCCLM3-LR (n=3). Data are presented as mean±standard deviation, *
P<0.05, **
P<0.01, ***
P<0.001, ****
P<0.0001, ns means not significant. HKDC1, hexokinase domain containing protein 1; DEGs, differentially expressed genes; LR, lenvatinib-resistant; LS, lenvatinib-sensitive; SMS, spermine synthase; HCC, hepatocellular carcinoma; TCGA-LIHC, The Cancer Genome Atlas Liver Hepatocellular Carcinoma; IHC, immunohistochemistry; ELISA, enzyme-linked immunosorbent assay.
Figure 4LR cells can form a suppressive immune microenvironment by regulating polyamine metabolism. (A) Orthotopic Hep1-6-LR and -LS tumor images, liver/body weight ratio, and Ki67 IHC staining (n=6/group). (B) Single-cell transcriptomic analysis of LS and LR tumors reveals distinct cell populations and marker gene expression (n=1/group). (C) Analysis of tumor immune infiltration (F4/80+ macrophages, CD8+ T cells) in LS vs. LR groups by mIHC and flow cytometry (n=6/group). (D) Effects of HKDC1 inhibition on PD-L1, pSTAT1, and STAT1 protein levels in LR cells. (E) Metabolic function (OCR, ECAR, ATP production, SRC) of CD8+ T cells treated with LS or LR tumor cell supernatants (n=3/group). (F) The effect of Hep1-6-LR conditioned medium on CD8+ T cell proliferation in vitro. n=3/group. Data are presented as mean±standard deviation, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. LR, lenvatinib-resistant; LS, lenvatinib-sensitive; IHC, immunohistochemistry; mIHC, multiplex fluorescent IHC; HKDC1, hexokinase domain containing protein 1; OCR, oxygen consumption rate; ECAR, extracellular acidification rate; ATP, adenosine triphosphate; SRC, spare respiratory capacity.
Figure 5USF1 facilitates lenvatinib resistance in HCC through transcriptionally activating HKDC1. (A) Proteomic comparison of biotin-labeled (DNP) vs. unlabeled (NC) DNA pulldown from the 2-kb HKDC1 promoter in Hep3B-LR identified differentially enriched proteins (red: increased; blue: decreased, DPN/NC); samples were validated by silver staining with whole-cell lysate as input control. (B) USF1 binding and regulation were confirmed by DNA pulldown/Western and dual-luciferase assays in Hep3B-LR and HCCLM3-LR (n=3). (C) ChIP-PCR/qPCR quantified USF1-mediated enrichment at HKDC1 promoter “region 2” in HCCLM3-LR and Hep3B-LR (n=3). (D) Western blots assessed HKDC1 levels in LR vs. LS and their USF1 dependence in Hep3B/HCCLM3. (E) EdU and colony assays tested USF1–HKDC1 axis’ effect on cell proliferation under lenvatinib (HCCLM3-LR, 40 μM; Hep3B-LR, 5 μM; n=3). (F) IHC measured USF1 in HCCLM3 LR/LS xenografts±lenvatinib. Data are presented as mean±standard deviation, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001, ns means not significant. USF1, upstream stimulatory factor 1; HCC, hepatocellular carcinoma; HKDC1, hexokinase domain containing protein 1; ChIP, chromatin immunoprecipitation; PCR, polymerase chain reaction; LR, lenvatinib-resistant; LS, lenvatinib-sensitive; IHC, immunohistochemistry.
Figure 6SPD augments the efficacy of lenvatinib by promoting autophagy-like cell death. (A) Combined lenvatinib and SPD synergistically inhibit proliferation and clonogenicity of LR cells (n=3/group). (B) Combined lenvatinib and SPD suppress tumor growth and proliferation (Ki67) in HCCLM3-LR xenografts (n=6/group), scale bars, 1 cm. (C) Representative images and mean diameter of patient-derived organoids (Patient 1) under different treatments. n=3/group; tumor images, weight, and volume of the Patient 1-derived organoid xenograft (PDOX-1) model under different treatments (n=6/group), scale bars, 1 cm. (D) Autophagic responses in Hep3B and HCCLM3 cells (LS vs. LR) under different treatments for 24 h, shown by TEM ultrastructure (red arrows indicate autophagosomes) and LC3B-II Western blot. Data are presented as mean±standard deviation, **P<0.01, ***P<0.001, ****P<0.0001, ns means not significant. SPD, spermidine; LR, lenvatinib-resistant; LS, lenvatinib-sensitive; TEM, transmission electron microscopy.
Figure 7HKDC1 loss enhances HCC response to lenvatinib plus aPD-1 therapy via TIME modulation. (A) Schematic of the primary HCC model established by hydrodynamic injection of oncogenic plasmids and the subsequent treatment schedule. (B) Liver gross morphology, H&E staining, liver/body weight ratio, and tumor nodule count in treated Alb-Cre; Hkdc1f/f vs. Alb-Cre; Hkdc1+/+ mice (n=5/group). (C) Workflow of clinical sample analysis (by Figdraw, TTPPRa3441) and representative data: pre-/post-treatment MRI and HKDC1 IHC staining in sensitive vs. resistant HCC patients, with IHC score quantification (sensitive, n=14; resistant, n=26). (D) Quantification of peripheral B lymphocytes, T lymphocytes, NK cells, CD8+ T cells, and CD4+ T cells in HCC patients stratified by HKDC1 expression median (n=6/group). (E) Association of HKDC1 expression with treatment response distribution (n=20/group) and progression-free survival in Zhongshan Hospital HCC patients (cutoff=85.63). (F) Schematic diagram (by Figdraw, IOASIf30f4) illustrates the role of HKDC1 in the tumor immune microenvironment and in the response to combined lenvatinib plus aPD-1 therapy. Data are presented as mean±standard deviation, *P or #P<0.05, **P or ##P<0.01, ####P<0.0001; “#” indicates a statistically significant difference compared to the placebo group; ns means not significant. HKDC1, hexokinase domain containing protein 1; HCC, hepatocellular carcinoma; aPD-1, programmed cell death protein 1 antibody; TIME, tumor immune microenvironment; Len, lenvatinib.
Abbreviations
programmed cell death protein 1 antibody
chromatin immunoprecipitation
extracellular acidification rate
epidermal growth factor receptor
hexokinase domain containing protein 1
programmed death-ligand 1
patient-derived organoid xenograft
progression-free survival
quantitative reverse transcription polymerase chain reaction
spare respiratory capacity
tumor-associated macrophage
The Cancer Genome Atlas Liver Hepatocellular Carcinoma
tumor immune microenvironment
upstream stimulatory factor 1
REFERENCES
- 1. Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global Cancer Statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2021;71:209-249.
- 2. Kudo M, Finn RS, Qin S, Han KH, Ikeda K, Piscaglia F, et al. Lenvatinib versus sorafenib in first-line treatment of patients with unresectable hepatocellular carcinoma: a randomised phase 3 non-inferiority trial. Lancet 2018;391:1163-1173.
- 3. Llovet JM, Ricci S, Mazzaferro V, Hilgard P, Gane E, Blanc JF, et al. Sorafenib in advanced hepatocellular carcinoma. N Engl J Med 2008;359:378-390.
- 4. Cheng AL, Qin S, Ikeda M, Galle PR, Ducreux M, Kim TY, et al. Updated efficacy and safety data from IMbrave150: atezolizumab plus bevacizumab vs. sorafenib for unresectable hepatocellular carcinoma. J Hepatol 2022;76:862-873.
- 5. Ladd AD, Duarte S, Sahin I, Zarrinpar A. Mechanisms of drug resistance in HCC. Hepatology 2024;79:926-940.
- 6. Zapater JL, Lednovich KR, Khan MW, Pusec CM, Layden BT. Hexokinase domain-containing protein-1 in metabolic diseases and beyond. Trends Endocrinol Metab 2022;33:72-84.
- 7. Khan MW, Terry AR, Priyadarshini M, Ilievski V, Farooq Z, Guzman G, et al. The hexokinase “HKDC1” interaction with the mitochondria is essential for liver cancer progression. Cell Death Dis 2022;13:660.
- 8. Ludvik AE, Pusec CM, Priyadarshini M, Angueira AR, Guo C, Lo A, et al. HKDC1 is a novel hexokinase involved in whole-body glucose use. Endocrinology 2016;157:3452-3461.
- 9. Wang MQ, Chen YR, Xu HW, Zhan JR, Suo DQ, Wang JJ, et al. HKDC1 upregulation promotes glycolysis and disease progression, and confers chemoresistance onto gastric cancer. Cancer Sci 2023;114:1365-1377.
- 10. Zhao P, Yuan F, Xu L, Jin Z, Liu Y, Su J, et al. HKDC1 reprograms lipid metabolism to enhance gastric cancer metastasis and cisplatin resistance via forming a ribonucleoprotein complex. Cancer Lett 2023;569:216305.
- 11. Zhang Y, Wang M, Ye L, Shen S, Zhang Y, Qian X, et al. HKDC1 promotes tumor immune evasion in hepatocellular carcinoma by coupling cytoskeleton to STAT1 activation and PD-L1 expression. Nat Commun 2024;15:1314.
- 12. Arruabarrena-Aristorena A, Zabala-Letona A, Carracedo A. Oil for the cancer engine: the cross-talk between oncogenic signaling and polyamine metabolism. Sci Adv 2018;4:eaar2606.
- 13. Casero RA, Murray Stewart T, Pegg AE. Polyamine metabolism and cancer: treatments, challenges and opportunities. Nat Rev Cancer 2018;18:681-695.
- 14. Madeo F, Eisenberg T, Pietrocola F, Kroemer G. Spermidine in health and disease. Science 2018;359:eaan2788.
- 15. Liang Y, Piao C, Beuschel CB, Toppe D, Kollipara L, Bogdanow B, et al. EIF5A hypusination, boosted by dietary spermidine, protects from premature brain aging and mitochondrial dysfunction. Cell Rep 2021;35:108941.
- 16. Park MH, Wolff EC. Hypusine, a polyamine-derived amino acid critical for eukaryotic translation. J Biol Chem 2018;293:18710-18718.
- 17. Hofer SJ, Daskalaki I, Bergmann M, Friščić J, Zimmermann A, Mueller MI, et al. Spermidine is essential for fasting-mediated autophagy and longevity. Nat Cell Biol 2024;26:1571-1584.
- 18. Hofer SJ, Simon AK, Bergmann M, Eisenberg T, Kroemer G, Madeo F. Mechanisms of spermidine-induced autophagy and geroprotection. Nat Aging 2022;2:1112-1129.
- 19. Ban W, Sun M, Huang H, Huang W, Pan S, Liu P, et al. Engineered bacterial outer membrane vesicles encapsulating oncolytic adenoviruses enhance the efficacy of cancer virotherapy by augmenting tumor cell autophagy. Nat Commun 2023;14:2933.
- 20. Elgendy M, Sheridan C, Brumatti G, Martin SJ. Oncogenic Ras-induced expression of Noxa and Beclin-1 promotes autophagic cell death and limits clonogenic survival. Mol Cell 2011;42:23-35.
- 21. Green DR, Levine B. To be or not to be? How selective autophagy and cell death govern cell fate. Cell 2014;157:65-75.
- 22. Wang X, Li M, Ren K, Xia C, Li J, Yu Q, et al. On-demand autophagy cascade amplification nanoparticles precisely enhanced oxaliplatin-induced cancer immunotherapy. Adv Mater 2020;32:e2002160.
- 23. Fernández-Palanca P, Payo-Serafín T, San-Miguel B, Méndez-Blanco C, Tuñón MJ, González-Gallego J, et al. Hepatocellular carcinoma cells loss lenvatinib efficacy in vitro through autophagy and hypoxia response-derived neuropilin-1 degradation. Acta Pharmacol Sin 2023;44:1066-1082.
- 24. Zheng Y, Huang C, Lu L, Yu K, Zhao J, Chen M, et al. STOML2 potentiates metastasis of hepatocellular carcinoma by promoting PINK1-mediated mitophagy and regulates sensitivity to lenvatinib. J Hematol Oncol 2021;14:16.
- 25. Al-Habsi M, Chamoto K, Matsumoto K, Nomura N, Zhang B, Sugiura Y, et al. Spermidine activates mitochondrial trifunctional protein and improves antitumor immunity in mice. Science 2022;378:eabj3510.
- 26. Jin H, Shi Y, Lv Y, Yuan S, Ramirez CFA, Lieftink C, et al. EGFR activation limits the response of liver cancer to lenvatinib. Nature 2021;595:730-734.
- 27. Siska PJ, Beckermann KE, Mason FM, Andrejeva G, Greenplate AR, Sendor AB, et al. Mitochondrial dysregulation and glycolytic insufficiency functionally impair CD8 T cells infiltrating human renal cell carcinoma. JCI Insight 2017;2:e93411.
- 28. van der Windt GJ, Everts B, Chang CH, Curtis JD, Freitas TC, Amiel E, et al. Mitochondrial respiratory capacity is a critical regulator of CD8+ T cell memory development. Immunity 2012;36:68-78.
- 29. O’Brien KL, Assmann N, O’Connor E, Keane C, Walls J, Choi C, et al. De novo polyamine synthesis supports metabolic and functional responses in activated murine NK cells. Eur J Immunol 2021;51:91-102.
- 30. He H, Song Z, Lin S, Wang Y, Wang G. Exploring the effect of polyamines on NK cell function in colorectal cancer process based on glycolysis. Int Immunopharmacol 2023;117:109944.
- 31. Qi C, Hao H, Zhang W, Fu Y, Han Y, Li J, et al. Targeting formyl peptide receptor 2 to suppress neuroinflammation in neuromyelitis optica spectrum disorder. Theranostics 2025;15:4495-4506.
- 32. Puleston DJ, Baixauli F, Sanin DE, Edwards-Hicks J, Villa M, Kabat AM, et al. Polyamine metabolism is a central determinant of helper T cell lineage fidelity. Cell 2021;184:4186-4202e20.
- 33. Liu S, Qiu J, He W, Geng C, He G, Liu C, et al. TUG1 long non-coding RNA enlists the USF1 transcription factor to overexpress ROMO1 leading to hepatocellular carcinoma growth and metastasis. MedComm 2020;1:386-399.
- 34. Eisenberg T, Knauer H, Schauer A, Büttner S, Ruckenstuhl C, Carmona-Gutierrez D, et al. Induction of autophagy by spermidine promotes longevity. Nat Cell Biol 2009;11:1305-1314.
- 35. Kroemer G, Levine B. Autophagic cell death: the story of a misnomer. Nat Rev Mol Cell Biol 2008;9:1004-1010.
- 36. Hentze MW, Castello A, Schwarzl T, Preiss T. A brave new world of RNA-binding proteins. Nat Rev Mol Cell Biol 2018;19:327-341.
- 37. Sun Y, Zhou P, Qian J, Zeng Q, Wei G, Li Y, et al. Spermine synthase engages in macrophages M2 polarization to sabotage antitumor immunity in hepatocellular carcinoma. Cell Death Differ 2025;32:573-586.
- 38. Pietrocola F, Lachkar S, Enot DP, Niso-Santano M, Bravo-San Pedro JM, Sica V, et al. Spermidine induces autophagy by inhibiting the acetyltransferase EP300. Cell Death Differ 2015;22:509-516.
- 39. Martin JD, Seano G, Jain RK. Normalizing Function of Tumor Vessels: Progress, Opportunities, and Challenges. Annu Rev Physiol 2019;81:505-534.
- 40. Finn RS, Ikeda M, Zhu AX, Sung MW, Baron AD, Kudo M, et al. Phase Ib study of lenvatinib plus pembrolizumab in patients with unresectable hepatocellular carcinoma. J Clin Oncol 2020;38:2960-2970.
- 41. Llovet JM, Kudo M, Merle P, Meyer T, Qin S, Ikeda M, et al. Lenvatinib plus pembrolizumab versus lenvatinib plus placebo for advanced hepatocellular carcinoma (LEAP-002): a randomised, double-blind, phase 3 trial. Lancet Oncol 2023;24:1399-1410.
- 42. Bernardini A, Lorenzo M, Chaves-Sanjuan A, Swuec P, Pigni M, Saad D, et al. The USR domain of USF1 mediates NF-Y interactions and cooperative DNA binding. Int J Biol Macromol 2021;193:401-413.
- 43. Zhou Z, Li X, Deng C, Ney PA, Huang S, Bungert J. USF and NF-E2 cooperate to regulate the recruitment and activity of RNA polymerase II in the β-globin gene locus. J Biol Chem 2010;285:15894-15905.
- 44. Zambanini G, Nordin A, Jonasson M, Pagella P, Cantù C. A new CUT&RUN low volume-urea (LoV-U) protocol optimized for transcriptional co-factors uncovers Wnt/β-catenin tissue-specific genomic targets. Development 2022;149:dev201124.
- 45. Kimura S, Noda T, Yoshimori T. Dissection of the autophagosome maturation process by a novel reporter protein, tandem fluorescent-tagged LC3. Autophagy 2007;3:452-460.
