Post-translational modifications in hepatocellular carcinoma: Linking senescence, metabolic reprogramming, and immune evasion for therapeutic innovation

Abstract

Hepatocellular carcinoma remains a leading cause of cancer mortality. Although immune checkpoint inhibitors have improved outcomes for a subset of patients, primary and acquired resistance are common. Post-translational modifications (PTMs) provide a rapid and reversible regulatory layer that links oncogenic signaling, metabolism, and chromatin state to cellular senescence and the tumor microenvironment. Here we synthesize evidence showing how ubiquitination, phosphorylation, acetylation, methylation, SUMOylation, O-GlcNAcylation, and lactylation modulate core senescence programs (p53/retinoblastoma protein, DNA-damage response) and the senescence-associated secretory phenotype, thereby shaping myeloid recruitment, T-cell dysfunction, and immune evasion in hepatocellular carcinoma. We further discuss how metabolism-coupled PTMs rewire glycolysis-epigenetics crosstalk and generate spatially confined senescence-metabolic-immune niches that can be resolved by single-cell and spatial multi-omics. The current evidence base is dominated by mechanistic studies and correlative clinical datasets, underscoring the need for prospective validation and standardized PTM/senescence biomarkers. Finally, we propose a sequential “induce-remodel-clear” therapeutic concept in which senescence induction is paired with PTM-targeted modulation and immune or senolytic clearance to improve response durability.

Key Words: Hepatocellular carcinoma; Post-translational modifications; Cellular senescence; Tumor immune microenvironment; Metabolic reprogramming; Immunotherapy; Single-cell multi-omics; Spatial transcriptomics; Therapy resistance; Precision medicine

Core Tip: Posttranslational modifications (PTMs) constitute a dynamic regulatory network that critically shapes cellular senescence and the immunosuppressive microenvironment in hepatocellular carcinoma. Key metabolism-linked PTMs such as O-GlcNAcylation and histone lactylation couple glycolytic reprogramming with epigenetic remodeling, driving therapy resistance and immune evasion. Emerging single-cell and spatial multi-omics reveal spatially organized senescence-immune-metabolic niches, providing a basis for multidimensional stratification. This review proposes a sequential “induce-remodel-clear” therapeutic framework, integrating precise senescence induction, PTM-targeted modulation, and enhanced immune clearance to overcome resistance and advance precision hepatocellular carcinoma therapy.

INTRODUCTION

Hepatocellular carcinoma (HCC) is among the most lethal malignancies worldwide and typically arises on a background of chronic liver disease, including chronic viral hepatitis, metabolic dysfunction-associated steatotic liver disease (MASLD), and alcohol-associated liver disease. Global surveillance data continue to show a disproportionate burden in Asia and a rising incidence in regions where metabolic risk factors are increasing[1,2]. These epidemiologic trends, together with late presentation and limited liver reserve in many patients, make durable disease control difficult and highlight the need for mechanism-informed combination therapy. Systemic therapy for advanced HCC has shifted from kinase inhibitor monotherapy to immunotherapy-based combinations. Contemporary guidelines recommend atezolizumab plus bevacizumab or the single tremelimumab regular interval durvalumab regimen as preferred first-line options for patients with preserved liver function and good performance status, with tyrosine kinase inhibitors (TKIs) or other immunotherapy regimens used when contraindications or prior exposure limit these choices[3-6]. However, objective responses remain limited and resistance is frequent, reflecting the strong immunosuppressive pressures of the cirrhotic liver and HCC microenvironment. Post-translational modifications (PTMs) are reversible chemical changes that alter protein activity, stability, localization, and molecular interactions. In HCC, PTMs reshape senescence decision points by tuning the DNA-damage response, p53/retinoblastoma protein (pRb) signaling, chromatin accessibility, and stress-adaptation pathways. Functionally, PTM control can be viewed across three coupled layers: (1) Checkpoint and repair signaling that determines whether senescence is initiated; (2) Transcriptional and epigenetic programs that maintain or enable escape from senescence; and (3) Secretory and surface phenotypes that regulate immune recognition and clearance[7-9]. A key translational insight is the inseparability of metabolic rewiring from PTM remodeling in HCC. Metabolism-linked PTMs such as O-GlcNAcylation and histone lactylation connect nutrient availability to transcriptional and epigenetic states, thereby influencing therapy-induced senescence (TIS), senescence escape, and the establishment of an immunosuppressive tumor microenvironment (TME). In parallel, multiple PTM cascades centered on nuclear factor-κB (NF-κB) and related regulators shape the magnitude and composition of the senescence-associated secretory phenotype (SASP), which can either support immune-mediated clearance or, when persistent, drive inflammation, angiogenesis, and resistance[10-14]. Single-cell and spatial multi-omics are now resolving HCC heterogeneity across malignant cells, immune subsets, endothelium, and fibroblasts, and mapping how these compartments interact in space and time. These approaches uncover a spatially organized architecture within the microenvironment, where senescence programs, metabolic states, and immune exclusion consistently co-localize. This helps explain differential therapeutic responses in histologically similar tumors[15-17]. Importantly, much of this evidence is observational; prospective studies linking these spatial patterns (including PTM and senescence signatures) directly to treatment response and survival are still limited. Therapeutically, senescence is a double-edged sword: While it can arrest tumor growth and enhance antigen presentation, persistent senescent cells and a chronic SASP may promote immunosuppression and tissue remodeling. Against this background, a practical objective is to induce senescence in a controlled manner, reshape PTM-governed senescence and metabolic outputs to favor immune activation, and then remove senescent cells to prevent chronic toxicity. We use this logic to organize the review and to motivate a sequential “induce-remodel-clear” concept that integrates senescence induction, PTM-targeted modulation, and immune or senolytic clearance. In the sections below, we first outline mechanistic principles linking major PTM types to senescence regulation in HCC, then discuss how metabolism-coupled PTMs reshape the immune landscape, and finally summarize therapeutic opportunities and the most important gaps that currently limit clinical translation. Figure 1 provides an overview of the PTM-senescence-immune network and highlights where current evidence is strong vs primarily hypothesis-generating.

Figure 1 Post-translational modifications orchestrate cellular senescence and immune microenvironment in hepatocellular carcinoma. ATR: Ataxia telangiectasia and Rad3-related; CHK1/2: Checkpoint kinase 1/2; HDAC6: Histone deacetylase 6; MDM2: Mouse double minute 2 homolog; SASP: Senescence-associated secretory phenotype; OGT: O-GlcNAc transferase; LDHA: Lactate dehydrogenase A; EZH2: Enhancer of zeste homolog 2; PD-1: Programmed cell death protein 1; PD-L1: Programmed death-ligand 1; p53: Tumor protein p53; p300: Histone acetyltransferase p300; CBP: CREB-binding protein; NF-κB: Nuclear factor-κB; YBX1: Y-box binding protein 1; ATM: Ataxia telangiectasia mutated; UBA2: SUMO-activating enzyme subunit 2; UBC9: Ubiquitin-conjugating enzyme E2 I.

FUNDAMENTALS OF PTM-SENESCENCE INTERPLAY IN HCC

With the widespread application of immunotherapy and targeted therapy in HCC, cellular senescence and its accompanying PTMs have been identified as key regulators of therapeutic response and resistance. Building on the background outlined earlier, this chapter systematically elucidates the molecular mechanisms underlying HCC senescence and its regulatory network within the PTM framework.

Epidemiology and current therapeutic landscape of HCC

Cellular senescence is a persistent state of proliferative arrest triggered by diverse stimuli, including DNA damage, telomere dysfunction, oxidative or metabolic stress, and cancer therapy induced pressure. Its defining characteristics encompass irreversible cell cycle arrest, chromatin remodeling, and the formation of a senescence associated SASP. The SASP comprises a complex array of components, such as inflammatory cytokines [e.g., interleukin (IL)-6, IL-8], chemokines of the C-C motif chemokine ligand (CCL) family, matrix metalloproteinases (MMPs), and various growth factors. By remodeling the TME, the SASP exerts dual effects: On one hand, it can establish a tumor suppressive barrier through immune mediated clearance and stromal reorganization; on the other hand, if senescent cells are not promptly eliminated, sustained SASP secretion may exacerbate chronic inflammation, promote immunosuppression, and ultimately accelerate tumor progression[18].

In HCC, the chronic liver disease background, driven by factors such as viral hepatitis and MASLD, fuels a self-perpetuating cycle that sequentially involves inflammation, fibrosis, and regeneration. This cycle not only amplifies senescence inducing stimuli but also reinforces the immune tolerant properties of the TME, thereby significantly increasing the risk of prolonged SASP persistence following TIS[18]. Multiple investigations utilizing cohort data and multi-omics approaches have shown that scoring systems based on genes linked to cellular senescence, such as the cellular senescence score (CSS), along with associated classification models, can effectively predict patient prognosis, immune infiltration patterns, and drug sensitivity in HCC. Consequently, senescence burden is increasingly recognized as a valuable metric for risk stratification and treatment decision making in HCC[19-21].

PTMs as dynamic regulators of cellular senescence

PTMs play central regulatory roles in key decisions governing whether cells enter senescence, how they maintain the senescent state, and whether they are effectively cleared. Their functional network can be categorized into three interconnected layers: (1) Cell cycle checkpoint and DNA damage response; (2) Inflammatory transcriptional regulation; and (3) Metabolic epigenetic coupling. These layers collectively form the core framework for senescence fate determination[8,10].

At the DNA damage response level, multisite phosphorylation and acetylation enhance and stabilize the transcriptional activity of p53, while poly-ubiquitination mediated by murine double minute 2 (MDM2) promotes its degradation. The dynamic balance between these PTMs determines whether cells undergo senescence, apoptosis, or survival escape[22,23]. Similar PTM based regulation operates in the p16/pRb axis, where the phosphorylation status of Rb directly modulates E2F transcriptional activity and G1/S transition, thereby influencing cell cycle progression[24,25].

In inflammatory transcriptional regulation, NF-κB serves as a master transcription factor driving and sustaining the SASP. Multiple PTMs, such as phosphorylation, acetylation, methylation, and SUMOylation, coordinately regulate its nuclear translocation, DNA binding, and interactions with coactivators, thereby finely tuning the intensity and composition of the SASP[26,27]. These PTM mediated changes not only affect the expression of inflammatory cytokines such as IL-6 and tumor necrosis factor-α, but also modulate the production of chemokines and immunosuppressive molecules, ultimately determining whether senescent cells are cleared by the immune system or contribute to an immunosuppressive TME[28,29].

Within the metabolic epigenetic coupling layer, PTMs act as a bridge linking metabolic signals to epigenetic regulation, profoundly influencing the senescence threshold, SASP characteristics, and immune clearance efficiency. Among these, O-GlcNAcylation and histone lactylation are particularly prominent in HCC[30,31]. O-GlcNAcylation senses glucose and glutamine flux through the hexosamine biosynthesis pathway, converting nutrient status into broad transcriptional and protein interaction changes. Elevated global O-GlcNAc levels in HCC are closely associated with enhanced tumor invasiveness, metabolic adaptation, and therapy resistance[32]. Prior studies have established that O-GlcNAcylation of Y-box binding protein 1 drives a feedforward loop linking glycolysis to histone lactylation, thereby amplifying prosurvival and metastatic transcriptional programs in preclinical HCC models[10,33]. Complementary evidence shows that O-GlcNAcylation stabilizes other oncogenic factors like S-phase kinase-associated protein 2 and Progranulin, further supporting tumor progression[34]. In patient cohorts, histone lactylation [e.g., histone H3 lysine 18 lactylation (H3K18 La)] is closely associated with adverse clinicopathological features, including tumor migration, multidrug resistance, and altered lipid metabolism[35-37]. While these mechanistic and correlative findings robustly support the oncogenic role of this PTM axis and justify exploring O-linked O-GlcNAc transferase (OGT) or lactate pathway inhibition, a critical translational gap persists. Direct evidence that pharmacologically disrupting this specific axis can reverse immunosuppression and improve therapeutic outcomes in human HCC remains absent.

From a senescence regulation perspective, lactylation can influence the expression of genes associated with senescence and the composition of the SASP. This enhances both the inflammatory and immunosuppressive features of senescent cells and elevates their metabolic adaptability and properties resembling stem cells. Collectively, these changes reduce the susceptibility of senescent cells to clearance mediated by immune and pharmacological interventions[38-40]. Thus, in-depth investigation of lactate production, transport, and the associated enzymatic systems holds important implications for designing combined therapeutic strategies that sequentially induce and then clear senescent cells.

Emerging insights from single-cell and spatial multi-omics

The rapid development of single cell and spatial multi-omics technologies has provided clear clues for unraveling the complex spatial relationships between cellular senescence and PTMs in HCC. By integrating single cell transcriptomics, spatial transcriptomics, and spatial proteomics data, researchers can simultaneously resolve cell subtype distribution, functional states, and local microenvironmental contexts within tissue architecture, thereby systematically mapping the spatiotemporal heterogeneity of HCC[41].

Spatial multi-omics technologies are mapping previously unrecognized architectures within the HCC microenvironment. For instance, senescent endothelial cells [high Cyclin-dependent kinase inhibitor 2A (CDKN2A)] localize to regions of high-grade malignancy and immunosuppression, while metabolic hotspots co-occur with immune-excluded zones[17,41-43]. These findings provide high-value, hypothesis-generating insights that refine our understanding of tumor heterogeneity. A critical caveat is that these descriptive maps, while illuminating, do not establish causality. The observed spatial correlations do not prove that one element (e.g., a lactate-rich zone) drives the immune exclusion phenotype. The essential next step is spatial functional validation, experimentally perturbing these niches in vivo to test causal relationships. Such studies are currently lacking.

Despite the need for such validation, these spatial patterns already offer critical mechanistic clues. Further spatial co-localization analyses reveal that tumor cells, myeloid immune cells, and metabolic hotspots (e.g., lactate-enriched zones) often form spatially defined functional clusters. Such regions typically exhibit upregulated immune-checkpoint expression, restricted T cell infiltration, and other features of immune exclusion, potentially marking active lactylation-modification zones. These observations provide important clues for designing locally combined immunotherapeutic approaches[17,41,43,44].

From a translational standpoint, integrating senescence associated gene signatures with immune, metabolic, and PTM activity patterns derived from single cell and spatial multi-omics can help construct more robust prognostic and treatment response prediction models. Such integrated models are expected to improve the identification of patients likely to benefit from combination regimens involving immune checkpoint inhibitors (ICIs), epigenetic drugs, and metabolism targeted therapies, thereby providing a more precise molecular basis for personalized treatment decisions in HCC.

Toward multimodal stratification systems for senescence burden quantification in HCC

Quantifying senescence burden into clinically applicable metrics is a prerequisite for the precise integration of PTM and senescence features into HCC therapy[45]. In recent years, several senescence associated scoring models have been established and validated. For example, the CSS has been shown in The Cancer Genome Atlas and multiple independent cohorts to correlate significantly with overall survival, immune filtration patterns, and drug sensitivity in HCC patients, enabling the identification of individuals with a “highsenescence/highimmunosuppression” phenotype and guiding the selection of immunotherapy and combination strategies[46,47]. Similarly, the senescence risk score (SRS) and other signature models based on transcriptomic data integrate gene expression profiles related to senescence to reveal differences across senescence subtypes in mutational background, immune microenvironment, and activity of metabolic pathways, offering new insights for molecular classification and therapeutic stratification in HCC[48].

Building on these advances, integrating CSS/SRS based transcriptional features with metabolism linked PTM modules and spatial immune markers allows the construction of a multimodal stratification system. This system aims to identify patient subgroups suitable for sequential “induce-remodel-clear” therapy and to evaluate the durability of response and risk of resistance across different treatment combinations and sequencing schedules by dynamically monitoring changes in CSS/SRS, SASP profiles, and PTM associated indicators[49].

In summary, the establishment and evolution of cellular senescence in HCC are influenced by a synergistic regulatory framework consisting of three interconnected PTM layers. These include cell cycle and DNA damage response, inflammatory transcriptional regulation, and metabolic epigenetic coupling. Among them, O-GlcNAcylation and histone lactylation, as PTMs representative of metabolic linkage, are closely associated with key phenotypes in tumor cells such as senescence threshold, SASP characteristics, immune evasion, and therapy resistance[35,50,51]. Single cell and spatial multi-omics technologies are progressively enabling the spatial visualization of senescence associated cellular states and PTM activity patterns, thereby providing critical technical support for building a multimodal stratification framework that integrates CSS/SRS, spatial immune features, and PTM functional modules. Looking forward, efforts should focus on: (1) Advancing methodological integration of multi-omics and spatial analyses; (2) Strengthening standardization and reproducibility in evidence generation; and (3) Exploring multimodal stratification guided combination therapies and dynamic monitoring tools in clinical translation.

PTM-SENESCENCE NETWORKS IN KEY SIGNALING PATHWAYS

The initiation and progression of cellular senescence in HCC are tightly regulated by multiple key signaling pathways, whose core node activities largely depend on the dynamic balance of PTMs. Through modifications such as phosphorylation, acetylation, ubiquitination, and SUMOylation, PTMs precisely control DNA damage response, signal transduction, transcriptional activation, and metabolic reprogramming, thereby influencing cell fate decisions under stress, directing cells toward senescence, apoptosis, or aberrant proliferation[45,52]. Systematic dissection of PTM regulatory networks within these critical pathways not only helps clarify the molecular basis of HCC senescence but also provides theoretical support for therapeutic strategies targeting the PTM-senescence axis.

p53/pRb axis: PTM-mediated senescence decisions

The p53 protein is a central switch directing cells toward senescence or apoptosis. Its activity is regulated by a well defined repertoire of PTMs, including phosphorylation, acetylation, and ubiquitination, a mechanism solidly established across cancer models, including HCC. For example, MDM2 mediated ubiquitination promotes p53 degradation, an axis linked to therapy resistance in both cellular and patient derived contexts[53-58]. Nevertheless, translating this universal knowledge into effective HCC therapy is challenging. Key unresolved questions include: (1) Which upstream signals specify p53 PTM patterns within HCC’s unique inflammatory and metabolic microenvironment; and (2) Why drugs targeting the p53 MDM2 axis show limited and variable efficacy in HCC patients. These gaps highlight that, while the p53 pathway is undoubtedly critical, targeting its PTMs for HCC treatment requires a deeper, disease specific understanding of its regulatory network and accompanying biomarkers. This regulation, in turn, dictates cell fate through canonical effectors. Through p21 Cyclin-dependent kinase (CDK) inhibition, activated p53 signaling converges on the pRb-E2F axis to reinforce durable cell cycle arrest and establish the senescent phenotype[59-61]. Detailed, PTM type-specific mechanisms governing this and other p53 functions are discussed in section 3.

Beyond p53 itself, epigenetic and deacetylase networks modulate the p53-pRb program and influence treatment responses. Multiple histone deacetylase (HDAC) isoforms (including HDAC6, HDAC2, HDAC11, and HDAC5) have been implicated in HCCrelevant signaling, malignant phenotypes, and therapy sensitivity, supporting continued interest in deacetylasetargeted strategies and p53-MDM2/MDMX interaction blockade where p53 is intact[62-67].

PTM states along the p53 axis may also intersect with immune regulation. Experimental studies suggest that p53 activation can influence immune-checkpoint signaling [including programmed death-1 (PD-1)/PD-ligand 1 (PD-L1)-related pathways], raising the possibility that p53-centric interventions could modulate immunotherapy responsiveness in selected contexts[68,69].

Mechanistic target of rapamycin/mitogen-activated protein kinase pathways: Coupling stress, metabolism, and senescence

The mechanistic target of rapamycin complex 1 (mTORC1) and mitogen-activated protein kinase signaling pathways play central roles in TIS and its escape, broadly participating in stress response, DNA damage repair, autophagy regulation, and metabolic reprogramming[70]. Typically, transient and moderate cellular stress can activate senescence programs through the p38 and p53 axis, leading to cell cycle arrest. In contrast, sustained pro proliferative signaling and metabolic activation, such as chronically elevated mTORC1 activity, inhibit autophagy and promote protein and lipid synthesis. This in turn destabilizes the senescent state and increases the risk of senescence reversal, resumption of proliferation, or entry into reversible drug resistant states[71,72].

Recent studies reveal a direct functional coupling between the DNA damage response and cellular metabolism. For instance, the DNA damage-sensing kinase ataxia telangiectasia and Rad3-related (ATR) not only initiates canonical DNA damage response but also upregulates squalene epoxidase via phosphorylation, thereby promoting de novo cholesterol synthesis and enhancing mTORC1 activity along with its lysosomal localization[73]. Cholesterol further amplifies mTORC1 signaling through a positive feedback loop mediated by the FAS-associated factor 2-small nucleolar RNA host gene 6 complex, enabling cancer cells to survive under stress and acquire therapeutic tolerance. HCC, where metabolic burden is typically high, such pathways translate DNA damage signals into cholesterol biosynthesis and mTORC1 activation[74]. This mechanism may substantially raise the threshold for inducing stable senescence and increase the risk of reversible resistance to TKIs or radiotherapy[75,76].

TIS is increasingly recognized across modalities relevant to HCC, including radiotherapy, chemotherapy, TKIs, and loco-regional interventions such as transarterial chemoembolization (TACE)[77,78]. While TIS can transiently restrain tumor growth and promote immune activation, persistence of senescent cells may sustain SASP signaling, amplify chronic inflammation, and contribute to immune escape, recurrence, and resistance[79-82]. The immunological and translational implications of this duality are addressed in section 4.

Based on these mechanisms, temporally combining mTOR/mitogen-activated protein kinase pathway inhibition with TIS and immunotherapy is emerging as a promising therapeutic strategy. On one hand, short term coadministration of ATR/mTORC1 inhibitors or metabolic drugs targeting cholesterol or glycolysis after TACE, radiotherapy, or TKI induced TIS may limit damage repair and metabolic compensation, preventing cells from entering reversible senescent states[83-85]. On the other hand, ICIs (e.g., PD-1/PD-L1 antibodies) or senolytic/senomorphic agents can be superimposed during the TIS peak phase to enhance the clearance of senescent cells[86,87]. Current preclinical evidence supports the potential of rationally designed sequential treatment regimens that involve first inducing senescence, followed by inhibiting cellular repair mechanisms, and culminating in the clearance of senescent cells. Such strategies may enhance antitumor efficacy and delay the emergence of resistance while avoiding a substantial increase in toxicity[86,88].

NF-κB-SASP pathway: Fine-tuning of secretory phenotypes by PTMs

The composition and intensity of SASP are tightly coupled to NF-κB activity, and PTMs on the p65 subunit are important determinants of transcriptional output[89]. In HCC models, SUMOylation pathway alterations frequently cooccur with NF-κB activation, and SUMO modification of the p65 subunit is linked to proinflammatory transcription and adverse outcomes[90-96]. These correlations nominate the SUMO NF-κB axis as a candidate node for SASP modulation. However, the translational rationale for targeting this axis remains primarily correlative. Current evidence largely documents cooccurrence (e.g., parallel upregulation of SUMO enzymes and NF-κB targets), while direct causal proof, that inhibiting p65 SUMOylation remodels the SASP and improves the immune microenvironment in HCC, is missing. Moreover, the functional consequences of SUMOylation are exquisitely site specific, yet the precise modification sites on p65 that govern inflammatory responses in HCC are undefined. This lack of mechanistic resolution currently limits the translational feasibility of targeting this node. Notably, PTM effects on p65 are context dependent and site specific: Depending on the modification pattern and cellular state, NF-κB signaling may be amplified or restrained, with distinct consequences for inflammation, survival signaling, and immune interactions[29,92,97].

Within the tumor immune microenvironment, SASP can produce divergent effects. Chemokines such as C-X-C motif chemokine ligand 10 may enhance CD8+ Tcell recruitment and immune surveillance, whereas sustained IL-6/IL-8-dominant programs can favor myeloid-derived suppressor cells (MDSCs) and regulatory T cells (Tregs) accumulation and reinforce immune suppression[98-100]. These differences motivate a shift from indiscriminate SASP suppression toward selective tuning of SASP composition, including via PTM-targeted modulation of upstream transcriptional control.

Accordingly, SASP reprogramming has emerged as a therapeutic concept in which NF-κB-linked outputs are reshaped to reduce tumorpromoting inflammatory signals while preserving or enhancing immune-stimulatory cues. In HCC, interventions that influence upstream PTM networks (e.g., SUMO/deSUMOylation and acetylation control) may provide practical entry points to couple senescence induction with subsequent immune-mediated clearance in a time-structured manner[101-105].

Mechanistic roles of major PTM types in HCC senescence

PTMs play a central regulatory role in the initiation, maintenance, and escape of cellular senescence in HCC through reversible, dynamic, and site specific chemical modifications. Different PTM types interact and crosstalk to form a complex regulatory network, collectively determining cell fate under stress. This chapter systematically outlines the molecular mechanisms, functional evidence, and translational potential of major PTM types, including ubiquitination/SUMOylation, acetylation/deacetylation, O-GlcNAcylation, lactylation, and methylation, in HCC senescence, providing a theoretical foundation for precision interventions targeting the PTM senescence axis.

Ubiquitination/SUMOylation in senescence regulation

Ubiquitination and deubiquitination regulate senescence by controlling the stability of key stressresponse and cell-cycle proteins. In HCC, the MDM2-p53 axis remains a prototypical example in which ubiquitindependent turnover can weaken senescence enforcement and facilitate proliferative escape, particularly under therapeutic pressure[53-58]. In TIS, stabilizing senescence forcing nodes may lower the threshold for durable growth arrest. However, the clinical value of enhancing senescence depends on the subsequent clearance of senescent cells, as discussed in section 4.

SUMOylation provides an additional regulatory layer that rapidly reshapes inflammatory transcription programs and SASP output. Rather than reiterating pathway-level details, section 3 emphasizes how SUMO circuitry intersects with senescence-associated inflammation and immune remodeling in HCC[90-96]. Collectively, ubiquitin- and SUMO-dependent control points influence both the integrity of senescence barriers and the inflammatory programs that determine whether senescent cells are cleared or persist to fuel relapse.

Acetylation/deacetylation: HDAC/sirtuin-p300/CREB-binding protein axis

Cellular senescence in HCC is orchestrated by an intricate network of post-translational modifications that extend far beyond individual modification types. Recent advances have illuminated multiple interconnected regulatory axes that collectively determine senescence outcomes and therapeutic responses. The development of MDM2 inhibitors has emerged as a promising strategy not only to reactivate p53 but also to overcome resistance to anticancer immunotherapy, with various small-molecule inhibitors targeting the p53-MDM2/MDMX interaction currently under investigation[106,107]. The complexity of MDM2 biology extends further, as MDM2 independently sustains signal transducer and activator of transcription 5 stability to control T cell-mediated antitumor immunity, revealing a p53-independent function that links ubiquitination to immune surveillance[108]. Beyond the tumor cell intrinsic mechanisms, senescence in the tumor microenvironment plays a critical role: Senescent cancer-associated fibroblasts actively restrict CD8+ T cell activation and limit immunotherapy responsiveness through SASP-mediated mechanisms[109]. Metabolic reprogramming adds another layer of regulation, with transcription factors such as forkhead box K2 promoting nucleotide synthesis and DNA repair, thereby enabling cancer cells to withstand chemotherapy-induced stress and potentially bypass therapy-induced senescence[110]. Concurrently, SUMOylation circuitry contributes to HCC progression, with SUMO-activating enzyme subunit 1 emerging as a key driver of HCC metastasis and a promising diagnostic metabolic biomarker[111,112]. Within this multifaceted PTM landscape, acetylation of p53, NF-κB, and various chromatin-associated proteins represents another crucial regulatory mechanism in the initiation and maintenance of cellular senescence in HCC. p300/CREB-binding protein (CBP)-mediated acetylation of p53 significantly enhances its binding affinity to target gene promoters and transcriptional activity, upregulating cell cycle inhibitors such as p21 and thereby driving long-term, relatively stable proliferative arrest[67,68]. The acetylation plays an essential role in cell fate determination through the activation of p53, and acetylation modification augments the DNA-binding stability and transcriptional function of p53[113]. Conversely, members of the HDAC family, particularly sirtuin 1 (a class III HDAC), can deacetylate p53 and cooperate with other signaling axes to reduce p53 stability and transcriptional output, thereby undermining the maintenance of the senescent state[68,114,115]. Mechanistically, MIER2 facilitates p53 deacetylation by binding to HDAC1 (a class I HDAC), and deacetylation of p53 hinders its transcriptional activity[113]. Sirtuin 1-mediated deacetylation maintains homeostasis by inhibiting oxidative stress-induced premature senescence[115].

The regulatory landscape of acetylation in HCC extends beyond the p53 axis to encompass a broader network of deacetylases and their substrates. Among these, HDAC6 has emerged as a multifaceted regulator, modulating inflammatory signaling through signal transducer and activator of transcription 3 acetylation and influencing diverse cancer-related processes across solid tumors[116,117]. The therapeutic potential of targeting deacetylases is further exemplified by the development of multi-target HDAC inhibitors, which have shown promise in HCC treatment by activating the p53 signaling pathway[118]. Complementing these strategies, novel agents such as the CDK9 inhibitor oroxylin A have been demonstrated to promote wild-type p53 stability by simultaneously disrupting both MDM2 and Sirtuin 1 signaling, thereby enhancing p53-mediated tumor suppression[119]. These interconnected regulatory mechanisms highlight the complexity of the acetylation network and its potential as a therapeutic target in HCC.

It is crucial to recognize that the biological effects of acetylation are not linearly correlated with its global level; rather, they are strongly context dependent and site specific. At the level of acetyltransferases, p300/CBP inhibitors (e.g., B0292) in HCC preclinical models can remodel metabolic epigenetic networks and suppress malignant phenotypes. This suggests that dysregulated or nonspecific acetylation in particular genetic and metabolic backgrounds may paradoxically enhance the adaptive plasticity of tumor cells[120,121]. This observation further reveals the context dual nature of p300/CBP regulation: On one hand, site specific enhancement of p53 acetylation can strengthen the senescence barrier; on the other hand, in chromatin states dominated by proliferative transcriptional programs, moderate inhibition of p300/CBP may help reduce aberrant chromatin accessibility, thereby restricting tumor cell phenotypic plasticity.

O-GlcNAcylation: Metabolic-epigenetic coupling in senescence

O-GlcNAc modification is dynamically regulated by OGT and O-GlcNAcase. This modification acts as a crucial PTM that integrates metabolic signals, such as glucose and glutamine flux through the hexosamine biosynthesis pathway, into protein function and transcriptional networks. Multiple studies have demonstrated that global O-GlcNAcylation levels are commonly elevated in HCC and show significant correlation with increased tumor invasiveness, enhanced metabolic adaptability, and resistance to diverse therapies. Consequently, OGT and O-GlcNAcase are regarded as promising druggable targets[122,123].

Mechanistically, O-GlcNAcylation of Y-box binding protein 1 has been shown to drive a positive feedback loop that couples enhanced glycolysis with elevated histone lactylation. Specifically, O-GlcNAcylated transcription factors upregulate glycolytic genes and lactate production; concomitantly, the accumulated lactate serves as a substrate for histone lactylation modifications (e.g., H3K18 La), which in turn further activates genes associated with cell survival, migration, and immunosuppression[10,35,124]. This cascade amplification, spanning from metabolism-linked PTMs to epigenetically linked PTMs, provides a novel mechanistic explanation for how HCC tumor cells sustain malignant phenotypes and achieve immune escape under conditions of high metabolic stress.

Based on these mechanisms, two potential interventional strategies can be proposed: First, inhibiting OGT activity or reducing hexosamine biosynthesis pathway flux to lower O-GlcNAc burden may attenuate proinflammatory components of the SASP, alleviate myeloid derived immunosuppression in the TME, and thereby enhance immune responsiveness; second, combining O-GlcNAc targeted modulation with existing immune checkpoint inhibitors or antiangiogenic therapies may improve T cell metabolic fitness and antigen presentation efficiency, increasing the likelihood that therapy induced senescent cells are effectively cleared by the immune system. From an integrative senescence regulation perspective, O-GlcNAcylation not only modulates the activity of key signaling pathways such as p53/NF-κB, but also influences SASP composition and intensity through synergistic interactions with epigenetic modifications like histone lactylation. Thus, it constitutes a central molecular hub linking metabolic reprogramming, epigenetic remodeling, and the regulation of cellular senescence fate in HCC.

Lactylation: A glycolytic epigenetic marker in senescence

Lactylation is an emerging epigenetic PTM that uses lactate as a donor to modify histone lysine residues, typically exemplified by H3K18 La, thereby rapidly converting the high glycolytic metabolic signals of tumor cells into alterations in chromatin structure and transcriptional programs. In HCC models and clinical specimens, elevated H3K18 La levels are significantly associated with unfavorable prognostic features, including increased metastasis risk after ablation or embolization and enhanced ferroptosis resistance, indicating a pivotal role for lactylation in tumor invasion, metastasis, and therapy tolerance[125,126].

Studies have found that lactylation can enhance the activity of the phosphoinositide 3-kinase/protein kinase B and glycolytic signaling pathways through regulation of the neural precursor cell expressed developmentally down-regulated protein 4-phosphatase and tensin homolog axis, promoting resistance to chemotherapeutic agents such as oxaliplatin and 5-fluorouracil, thereby establishing a bidirectional amplification loop between metabolic reprogramming and drug tolerance[12,127]. These mechanistic insights offer a novel direction for resensitization strategies following TACE or first line systemic therapy. During the window of therapy induced senescence, a combination approach employing inhibitors of lactate production or transport, for example agents targeting lactate dehydrogenase or the monocarboxylate transporter (MCT) family MCT1/4, can be implemented. Inhibiting lactate dehydrogenase reduces lactate synthesis and lowers substrate availability for histone lactylation, while blocking MCT limits lactate release into the TME, alleviating its metabolic suppressive effects on immune and stromal cells[12,128,129]. Future therapeutic strategies may also focus on specifically targeting the enzymes that catalyze or remove lactylation modifications. Combining such targeting with immunotherapy or epigenetic modulating agents may attenuate the immunosuppressive microenvironment and curb the survival and expansion of drug resistant cell clones[130,131].

From the perspective of senescence regulation, lactylation exerts dual effects: On one hand, it alters senescence related gene expression and SASP composition, skewing the secretory profile of senescent cells toward immunosuppressive, proinflammatory, and prometastatic directions; on the other hand, it enhances tumor cell stemness and metabolic adaptability, reducing the efficiency with which senescent cells are cleared by the immune system or therapeutic agents[38,40,132,133]. Therefore, interventions aimed at reducing lactylation levels not only modulate metabolic pathways but also reset the epigenetically embedded memory of metabolic and epigenetic coupling. This strategy can be temporally coordinated with modulation of O-GlcNAcylation. The former primarily targets chromatin remodeling mediated by the glycolysis and lactate axis, while the latter systemically regulates the interactions between metabolic and epigenetic networks[40,50,134]. Together, they can establish a more favorable microenvironmental basis for subsequent immune mediated clearance of senescent cells.

Methylation: Histone and nonhistone targets in senescence

During senescence, repressive histone methylation marks such as histone H3 lysine 27 trimethylation (H3K27me3) and histone H3 lysine 9 trimethylation play important roles in modulating chromatin accessibility of key senescence associated genes including p16 and p21[135-137]. Enhancer of zeste homolog 2 (EZH2), the primary methyltransferase catalyzing H3K27me3, is frequently overexpressed or hyperactive in HCC subtypes associated with dedifferentiation, immune evasion, and multidrug resistance[138]. Multiple studies confirm that EZH2 inhibition reduces H3K27me3 levels, reverses epigenetic silencing of senescence loci, and thereby induces cellular senescence while suppressing tumor proliferation[136,138].

In terms of combination therapy strategies, EZH2 inhibitors show promise for synergistic use with immune checkpoint inhibitors or senolytic agents, not only enhancing senescence induction but also improving the efficiency of senescent cell clearance through modulation of antigen presentation and immune infiltration. Such approaches may functionally complement acetylation/deacetylationbased regulatory networks[138-140].

In addition, methylation of nonhistone proteins, including key transcription factors or signaling molecules such as p53 and p65, also participates in regulating inflammatory and stress related transcriptional networks. This process may influence senescence stability and SASP composition. However, current evidence largely derives from other tumor types or non-hepatocyte models; within the specific context of HCC senescence, systematic and causally validated functional evidence linking defined methylation sites to senescence phenotypes remains scarce[138-140].

A more feasible current research pathway involves the integration of multimodal data, encompassing transcriptomics, proteomics, PTM omics, and spatial omics, to infer the role of methylation events in HCC senescence progression. This would be followed by functional screening to identify key modification sites and pathways that hold potential for patient stratification and possess drug ability. Such an approach could establish an empirical basis for incorporating the interplay between methylation and senescence into prospective clinical studies.

Collectively, these PTM classes form an interconnected regulatory layer that shapes both senescence programs and immune regulation in HCC. Table 1 summarizes the PTM classes most consistently implicated in this context, highlights their dominant senescence axes and immune effects, and indicates the typical level of supporting evidence to aid interpretation and prioritization[141-156].

Table 1 Major post-translational modification classes linking senescence regulation to immune features in hepatocellular carcinoma.

PTM class	Key regulators/substrates (examples)	Primary senescence axis	Immune/TME-relevant effect	Evidence in HCC (typical)	Translational implication	Ref.
Ubiquitination/deubiquitination	MDM2-p53; NEDD4-PTEN; USP family	Checkpoint control (p53/pRb), senescence escape	Shapes antigen presentation and myeloid/T-cell balance via p53–inflammation and PI3K-AKT signaling	Strong mechanistic; clinical links mainly correlative	Target E3/DUBs to stabilize p53 and sensitize to ICI or senolytics	[141-143]
SUMOylation/deSUMOylation	SUMO1/2/3; PIAS; SENP enzymes	Stress signaling; transcription factor activity	Can promote immune evasion through TF reprogramming and inflammatory signaling	Mechanistic; limited prospective clinical validation	SENP/PIAS modulation as an immuno-sensitizing adjunct	[144,145]
Acetylation/deacetylation	p300/CBP; SIRT1/2; HDACs	p53 activity; chromatin accessibility; SASP tuning	Controls inflammatory transcription and exhaustion programs (NF-κB, STATs)	Strong in vitro/in vivo; growing clinical correlative data	HDAC/SIRT targeting to balance senescence induction vs chronic SASP; combine with IO	[146-148]
O-GlcNAcylation	OGT/OGA; YBX1; c-Myc (examples)	Metabolism–chromatin coupling; senescence tolerance	Links nutrient state to immune suppression (myeloid bias, T-cell dysfunction)	Emerging mechanistic; early clinical correlative signals	OGT/OGA inhibition to disrupt metabolic tolerance and recondition TME	[149-151]
Histone lactylation	Lactate; p300; H3K18 La (marker)	Epigenetic remodeling of senescence and stemness programs	Promotes pro-tumor inflammation and immune suppression in lactate-rich niches	Emerging; evidence largely preclinical/omics	Target lactate production/transport or lactylation writers in combination regimens	[152,153]
Histone/non-histone methylation	EZH2; SETD2; KMT/KDM enzymes	Stable chromatin states; senescence maintenance/escape	Regulates antigen presentation and myeloid polarization via epigenetic programs	Mechanistic + correlative cohorts; limited intervention trials in HCC	Epigenetic therapy to restore immune visibility and constrain senescence escape	[154-156]

PTM: Posttranslational modification; TME: Tumor microenvironment; HCC: Hepatocellular carcinoma; MDM2: Mouse double minute 2 homolog; p53: Tumor protein p53; NEDD4: Neural precursor cell expressed developmentally downregulated protein 4; PTEN: Phosphatase and tensin homolog; USP: Ubiquitin-specific protease; pRb: Retinoblastoma protein; PI3K: Phosphoinositide 3-kinase; AKT: Protein kinase B; E3: E3 ubiquitin-protein ligase; DUB: Deubiquitinating enzyme; ICI: Immune checkpoint inhibitor; SUMO: Small ubiquitinlike modifier; PIAS: Protein inhibitor of activated signal transducer and activator of transcription; SENP: Sentrin/small ubiquitin like modifier-specific protease; TF: Transcription factor; p300: Histone acetyltransferase p300; CBP: CREB-binding protein; SIRT: Sirtuin; HDAC: Histone deacetylase; SASP: Senescence-associated secretory phenotype; NF-κB: Nuclear factor-κB; STAT: Signal transducer and activator of transcription; IO: Immunotherapy; OGT: O-GlcNAcylation transferase; OGA: O-GlcNAcase; YBX1: Y-box binding protein 1; c-Myc: Cellular myelocytomatosis oncogene; H3K18 La: Histone H3 lysine 18 lactylation; EZH2: Enhancer of zeste homolog 2; SETD2: SET domain containing 2; KMT: Lysine methyltransferase; KDM: Lysine demethylase.

IMMUNOLOGICAL IMPLICATIONS AND CLINICAL TRANSLATION OF SENESCENCE IN HCC

Cellular senescence in HCC is not an isolated phenomenon but is deeply embedded within the immune microenvironment shaped by chronic liver disease and continuously interacts with the immune system via the SASP. The outcome of this interaction, namely whether senescent cells are cleared or persist, directly governs tumor progression and therapeutic response. Therefore, a deeper understanding of the regulatory network linking senescence, the SASP, and immunity, coupled with the development of stratification strategies based on senescence features and temporally optimized treatments, is essential for enhancing immunotherapy efficacy and overcoming resistance.

Dual immunoregulatory functions of senescence and SASP in HCC

Throughout HCC development, a self-perpetuating pathological cycle of inflammation, fibrosis, and parenchymal regeneration, driven by etiologies such as chronic viral hepatitis, MASLD, and alcohol-associated liver disease, operates continuously. This cycle ultimately culminates in the malignant transformation of hepatocytes. This process drives a substantial number of hepatocytes and hepatic stellate cells into a senescent state, accompanied by sustained release of a complex SASP. The SASP comprises IL-6, IL-8, CCL-2, CCL-5, tumor necrosis factor-β, various angiogenic factors, and matrix remodeling proteases, which act in a paracrine manner to profoundly influence the tumor immune microenvironment[157-159].

The role of senescence and SASP in the tumor microenvironment is inherently dualistic. As comprehensively reviewed, senescent cells actively remodel their surroundings through SASP, exerting both tumor-suppressive and tumor-promoting effects depending on the persistence and composition of secreted factors[160]. This duality is particularly evident in the context of therapy-induced senescence, where the initial growth arrest can be beneficial for tumor suppression, but the accumulation of persistent senescent cells may drive cancer resistance and recurrence through sustained SASP-mediated inflammation and immunosuppression[161]. Understanding this delicate balance is essential for deciphering how the chronic SASP release in the cirrhotic liver microenvironment contributes to HCC progression and therapeutic resistance.

Recent multi-omics studies have further confirmed a close correlation between senescence burden and immune features, patient prognosis, and immunotherapy response in HCC. Senescence associated transcriptomic scores (e.g., CSS, SRS), derived from large retrospective cohorts like The Cancer Genome Atlas, demonstrate robust prognostic value and association with immune phenotypes across independent validation sets[48,162]. This constitutes well established clinical correlative evidence. Their utility in predicting response to immunotherapy, however, is less substantiated. Current supportive data primarily stem from post hoc bioinformatic analyses of clinical trial datasets or small retrospective studies, representing a lower level of evidence. A pivotal translational gap is the absence of prospective trials that pre-stratify patients based on these scores to test their predictive power. Consequently, these signatures currently serve better as tools for biological insight and hypothesis generation than for guiding clinical decisions.

Given this context, the clear biological impact but immature predictive tools, it is therefore imperative that the biology underpinning these scores be formally integrated into clinical development. Consequently, the crosstalk between cellular senescence and the SASP in HCC not only directly participates in tumor progression but also significantly influences immunotherapy sensitivity and long-term patient outcomes, making it a critical contextual factor that should be incorporated into the design of clinical treatment strategies[163].

Senescence-associated immune mapping and patient stratification

To quantify senescence burden and facilitate its clinical translation, numerous studies have focused on developing senescence associated scoring systems based on transcriptomic and clinical data. For example, a senescence scoring model developed by Gao et al[164] demonstrated that patients with low CSS had longer overall survival and showed more abundant intratumoral infiltration of CD8+ T cells and natural killer (NK) cells in immune infiltration analyses. In contrast, patients with high CSS generally exhibited increased enrichment of immunosuppressive cells such as Tregs and MDSCs, upregulated expression of immune checkpoint genes, and functional enrichment analyses revealing aberrant activation of glycolysis and lipid metabolism pathways[165-167]. Furthermore, prognostic models based on senescence associated genes or noncoding RNAs constructed across multiple HCC cohorts have confirmed that such molecular features derived from senescence related gene expression correlate closely with immune infiltration status, immune functional activity, and response rates to TACE or systemic therapies, highlighting their potential value in guiding individualized treatment[168].

In terms of immune cell composition, HCC with high senescence burden typically exhibits the following characteristics: Reduced numbers of CD8+ T cells, most of which display a PD1+ T-cell immunoglobulin and mucin-domain containing-3+ terminally exhausted phenotype; significantly elevated proportions of Tregs, MDSCs, and M2 type tumor associated macrophages; impaired dendritic cell maturation and antigen presentation function; and concurrent endothelial cell dysfunction, which collectively restrict the infiltration of effector immune cells into the tumor core[169,170]. At the metabolic and PTM levels, high senescence burden is often accompanied by enhanced aerobic glycolysis, lactate accumulation, and increased levels of metabolism linked PTMs such as O-GlcNAcylation and histone lactylation. These observations align with the previously described OGT-H3K18 La regulatory axis and further corroborate, in clinical cohorts, the close interconnection among metabolic reprogramming, PTM modifications, and immune suppression[169,170].

Therefore, building on senescence scores such as CSS and integrating immune infiltration maps, metabolic markers, and specific PTM features can enable the construction of an operational and verifiable multimodal stratification framework that incorporates senescence, immune, and metabolic dimensions. Such a system would provide the foundation for precisely identifying patients most likely to benefit from therapeutic strategies that combine the induction of senescence with subsequent immune-mediated clearance.

Immune surveillance and escape of senescent cells

Within the tumor immune microenvironment, the effective clearance of senescent cells by the immune system hinges on two critical and PTM-regulated steps: Initial recognition and subsequent lethal attack. At the recognition stage, classic studies have established that diverse senescence inducing triggers, such as replicative senescence, oncogene activation, and DNA damage, lead to the upregulation of ligands for the NK cell activating receptor NK group 2D (NKG2D), including the major histocompatibility complex class I polypeptide-related sequence A/B and UL16-binding protein families, on the cell surface. Consequently, this enhances NK cell mediated recognition and clearance of senescent cells, which constitutes a fundamental mechanism of senescence dependent immune surveillance[171,172]. Murine models further confirm that tumor cells can recruit NK cells via secretion of chemokines such as CCL2 and clear senescent clones in an NKG2D dependent manner, indicating a tight functional link between cellular senescence, SASP, and NKG2D signaling[171]. Published reports note that the expression level, glycosylation status, and membrane anchoring stability of these NKG2D ligands are finely tuned by multiple PTMs, including O-GlcNAcylation, phosphorylation, and proteolytic cleavage, with persistent DNA damage response and SASP also contributing to their regulation[171,173].

Nevertheless, persistent senescent cells can also achieve immune escape via PTM related mechanisms. Muñoz et al[174] found that MMPs, which are highly expressed in senescent cells, cleave NKG2D ligands and promote their shedding from the cell membrane, generating soluble ligands that interfere with NK/T cell recognition of senescent cells. Broad-spectrum MMP inhibition, exemplified by the agent GM6001, can block this process and restore NKG2D-dependent immune clearance[175]. Further studies reveal that senescent cells can promote major histocompatibility complex class I polypeptide-related sequence A/B shedding through a long noncoding RNA-a disintegrin and metalloproteinase domain-containing protein 10 (ADAM10) axis, similarly impairing immune surveillance. Previous work has shown that combining low-dose chemotherapy with ADAM10 inhibitors can restore NK cell killing capacity[176,177]. Collectively, these findings suggest that targeting PTM processes involved in NKG2D ligand cleavage or glycosylation represents a potential strategy to enhance the immunorecognition of senescent cells in HCC.

At the effector stage, PD-L1 expression and its glycosylation status are key facilitators of immune escape by TIS cells. Studies demonstrate that TIS upregulates PD-L1 transcription and promotes its glycosylation in tumor cells, a process dependent on the endoplasmic reticulum associated protein ribophorin 1. Interfering with reticulum associated protein ribophorin 1 disrupts PD-L1 glycosylation, leading to its degradation and thereby sensitizing senescent cells to T cell mediated killing[79,178,179]. This finding, however, provides only a partial and context specific answer. It is derived from engineered TIS models, leaving its applicability to the spectrum of senescence in untreated human HCC tumors unknown. Consequently, bridging this gap requires direct clinical correlation, detecting PD-L1 modification states in patient-derived senescent cells, and interventional testing of glycosylation targeting combinations in trials. If this axis operates in patients, it carries a significant therapeutic implication. The observation implies that even when TIS suppresses tumor proliferation in the short term, concomitant upregulation and stabilization of PD-L1 may still confer protection to senescent cells through PD-L1 dependent immune escape mechanisms, thereby compromising clearance efficacy. Consequently, timely combination of PD-1/PD-L1 inhibitors or interventions targeting PD-L1 glycosylation after senescence induction may synergistically enhance the clearance of senescent cells[79,179,180].

A more forward-looking approach is to leverage the senescence phenotype as an entry point for targeted therapy. Given that senescent cells often exhibit high expression of NKG2D ligands, recent studies have begun developing NKG2D chimeric antigen receptor T cells (CAR T) cells. These CAR T cells exhibit specific recognition and killing of tumor cells under diverse stress or senescence-associated conditions. Their promising antitumor activity in xenograft models points to a novel direction for developing adoptive cell therapies aimed at senescent cells[171,181].

In HCC, senescent cells are dual-natured entities: They are amenable to immune clearance yet also mediate immune escape. On one hand, elevated expression of surface NKG2D ligands and secretion of SASP-derived chemokines foster an environment for clearance. On the other hand, ligand shedding, heightened PD-L1 expression and modification, and SASP-driven immunosuppression promote their survival and potential role in recurrence. Therefore, a major translational focus going forward should be on precise interventions targeting specific PTM processes, such as inhibiting MMP- or ADAM10-mediated ligand cleavage or modulating PD-L1 glycosylation, combined rationally with immunotherapy to tip the balance toward the total elimination of senescent cells.

TIS and immunotherapy synergies

Radiotherapy, chemotherapy, TKIs, and loco-regional therapies (including TACE and ablation) can induce TIS in HCC across tumor and stromal compartments[85,182]. Evidence from preclinical studies indicates that the net effect of TIS is governed by timing and clearance: When senescent cells persist, sustained SASP can promote angiogenesis, epithelial-mesenchymal transition, immunosuppression, and metabolic rewiring, which may ultimately facilitate recurrence and resistance[161,183]. Emerging work also highlights that senescence programs differ by inducing modality, with variability in markers, SASP composition, and immune consequences, underscoring the importance of defining an actionable therapeutic window in which senescence induction is paired with immune activation and/or senolytic-enabled clearance[161,183,184].

In mouse models, TIS has been shown to remodel the tumor myeloid niche, leading to a significant increase in MDSCs and immunosuppressive macrophages, thereby attenuating the efficacy of subsequent immune checkpoint inhibitors. Conversely, coadministration of senolytic agents to clear TIS cells can partially reverse this immunosuppressive state and restore the proliferation and effector function of CD8+ T cells and other immune effectors, providing key functional validation for sequential “induce-remodel-clear” strategies[185].

At the systemic therapy level, combination regimens integrating immune checkpoint inhibition with antiangiogenic therapy have become a first line standard of care for advanced HCC. One prominent example is the regimen of atezolizumab combined with bevacizumab, known as the IMbrave150 regimen[186]. Their core mechanism lies in simultaneously targeting two critical nodes in the tumor immune microenvironment: PD-L1 inhibitors relieve T cell immunosuppression, while anti-vascular endothelial growth factor (VEGF) agents block pathological angiogenesis, improve vascular function, and mitigate myeloid derived immunosuppression. This strategy has significantly improved survival outcomes in advanced HCC patients and also provides a clinical rationale for sequential treatment models that first remodel the vascular and myeloid immune environment before activating effector T cell function[186,187]. Although existing studies were not specifically designed to target the interplay between senescence and the SASP, mechanistic insights reveal that anti-VEGF therapy achieves multifaceted immunomodulatory effects. It not only corrects vascular abnormalities, restores sinusoidal endothelial cell function, and enhances tumor perfusion and T cell migration, but also partially reverses VEGF-mediated dysfunction in dendritic cells and mitigates the abnormal accumulation of myeloid immune cells. Consequently, this therapy establishes a more favorable immune microenvironment, thereby providing a stronger foundation for implementing subsequent or concurrent therapeutic strategies that aim to first induce senescence and then promote its immune-mediated clearance[186,188]. Consequently, incorporating antiangiogenic therapy into the systemic treatment framework for HCC not only directly suppresses tumor blood supply but also provides important immunological support for subsequent senescence-based precision interventions.

Therefore, future clinical trial designs could investigate combining PD-1/PD-L1 inhibitors, SASP modulators, agents that target metabolism-linked PTMs, and selective senolytics during TIS phases elicited by TACE, radiotherapy, or specific TKIs. The goal is to establish an integrated, sequential treatment paradigm comprising three core actions: Inducing senescence, remodeling the TME, and clearing the resulting senescent cells.

Clinical translation pathways and future study design

Integrating the research advances outlined above, the regulatory axis linking senescence, the SASP, and immunity in HCC demonstrates substantial plasticity and therapeutic potential, and is progressively advancing from mechanistic exploration toward clinical translation. To enable systematic implementation of senescence directed therapeutic strategies, this study proposes the following clinically relevant pathways and research priorities across three dimensions: Patient stratification, temporally optimized therapy, and study design.

The precise application of senescence-guided therapies necessitates a stratification framework that integrates multidimensional biomarkers. Essential components of this framework include indicators of senescence burden, features of the immune microenvironment, and markers reflecting metabolism-linked PTMs. For instance, senescence burden can be assessed using scores such as CSS and SRS; immune microenvironment features encompass the spatial distribution of CD8+ T cells, Tregs, and MDSCs; and relevant metabolic-PTM markers comprise circulating lactate levels, tissue levels of histone lactylation and O-GlcNAcylation, as well as PD-L1 glycosylation status[189,190]. Such an integrated system would aid in identifying high-risk patient subgroups exhibiting a co-occurrence of high senescence burden, profound immunosuppression, and intense metabolic drive. Concurrently, spatial multi-omics technologies could map the spatial relationships between regions enriched for senescent cells and areas with immune or metabolic aberrations within tumors, thereby providing a rationale for designing locally targeted combination therapies.

Second, a sequential treatment strategy spanning from the induction of senescence to its clearance must be established. Given the dynamic nature of senescence, a therapeutically effective approach likely requires a temporally coordinated combination framework. The initial step involves inducing TIS using modalities such as radiotherapy, TACE, TKIs, or CDK4/6 inhibitors. Senescence induction should be confirmed through biopsy and liquid biomarkers; examples include elevated p16 or p21, the DNA damage marker γH2AX, and characteristic SASP profiles. Subsequently, during the peak window of TIS, immunotherapy combined with TME modulators should be deployed to enhance immune-mediated clearance. Immunotherapeutic agents may include PD-1 or PD-L1 inhibitors and NKG2D-targeting CAR T cells, while microenvironment modulators could encompass Janus kinase/signal transducer and activator of transcription pathway inhibitors, OGT inhibitors, or inhibitors of lactate transport. Finally, during the subsequent disease stabilization phase, short-course, intermittent administration of senolytic agents can be implemented to reduce the residual pool of senescent cells and thereby mitigate long-term recurrence risk[48,189-191].

Third, a robust methodological and longitudinal follow-up framework must be designed. Advancing clinical translation requires a prospective study design incorporating data collection at multiple timepoints and across multiple modalities. At key treatment intervals, peripheral blood, tumor tissue, and imaging data should be collected in a synchronized manner to construct a dynamic atlas that delineates the co-evolution of cellular senescence, immune activity, metabolic state, and PTMs. This integrated approach enables the identification of composite biomarkers predictive of treatment response and resistance. Furthermore, surrogate endpoints reflecting biological changes, such as immune microenvironment remodeling and SASP dynamics, should be incorporated into clinical endpoint definitions to accelerate the optimization and validation of novel therapeutic strategies.

Building on a sequential therapeutic logic, senescence induction, reshaping of SASP and coupled metabolic/PTM programs, and then immunemediated clearance, an integrated treatment framework can be organized by clinical stage, intervention steps, and monitoring indicators (Table 2)[192-198]. In HCC, senescence-linked signaling is dynamic and can remodel the tumor ecosystem; therefore, effective strategies are expected to combine: (1) Precise senescence induction; (2) Control of SASP and metabolism-linked PTM circuits to limit immunosuppressive drift; and (3) Timely activation of clearance mechanisms. Multi-omics-informed stratification and longitudinal monitoring can support rational timing and patient selection, improving the likelihood that senescence is converted from a transient state into a therapeutically exploitable vulnerability.

Table 2 Combined senescence-post-translational modification-immunotherapy strategy in hepatocellular carcinoma based on an “induce-remodel-clear” framework.

Clinical stage	Step 1: Induce senescence	Step 2: Remodel PTM/metabolism/SASP (emphasizing immune regulation)	Step 3: Immune/senolytic clearance	Key monitoring indicators (including immune infiltration and function)	Potential beneficiary population	Ref.
Pre-systemic therapy (prior to IO ± anti-VEGF initiation)	TACE/ablation or short-course TKI ± local radiotherapy to induce TIS	Anti-VEGF to improve vascular perfusion and T-cell infiltration; low-intensity JAK/STAT or NF-κB inhibition to reduce IL-6/IL-8type SASP; exploratory OGT/MCT modulation to alleviate metabolic suppression	Initiate PD-1/PD-L1 monoclonal antibody; add shortcourse senolytics during stable phase	Tissue: P16/p21, γH2AX, Ki-67, H3K18 La, O-GlcNAc; spatial immune: CD8+ T/Treg/MDSC/TAM spatial distribution; plasma: IL-6/IL-8/VEGF, lactate; dynamic CSS/SRS	Child-Pugh A patients with prominent vascular abnormalities, myeloid enrichment, and partially inflamed TIME	[26,192-194]
Post-TACE/radiotherapy/ablation (TIS window)	Local therapy to induce definite TIS (elevated p16/p21, γH2AX)	Short-course ATR/mTORC1 inhibition to restrict repair; JAK/STAT or NF-κB inhibition to reduce pro-tumor SASP; LDH/MCT inhibition to lower lactate and lactylation	Intensify PD-1/PD-L1 therapy during TIS peak; experimental addition of senolytics	Paired biopsies: Senescence markers, SASP profile, CD8+ T-cell re-infiltration; imaging: Necrotic area and “inflammatory rim”; dynamic cytokine profiling	Patients with moderate-to-high recurrence risk after local therapy, preserved liver function, and accessible tissue sampling	[85,195]
TKI or IO maintenance phase with acquired resistance	Long-term TKI/IO selects reversible senescence/drug-resistant clones	MDM2 inhibition + p300/CBP modulation to stabilize p53-dependent senescence; OGT/MCT and SUMO pathway targeting to reshape SASP and metabolism	Escalate IO regimen (e.g., PD-1 + CTLA-4) or explore NKG2D-CAR T; short-course senolytics to eliminate persistent senescent clones	PTM: O-GlcNAc, H3K18 La, EZH2; immune phenotype: PD-L1 glycosylation, TCR clonality, T cell exhaustion markers; metabolism: Lactate, glycolysis score	Patients initially responsive to IO/TKI who later progress slowly, with high senescence burden, immunosuppressive SASP, and strong metabolic reprogramming	[103,184,196]
Perioperative/early-intermediate HCC (neoadjuvant/adjuvant phase)	Short-course neoadjuvant TACE/radiotherapy or CDK4/6 inhibitor-induced TIS; or gentle postoperative senescence induction	Low-dose JAK/STAT/NF-κB inhibition to balance regenerative-phase SASP; moderate OGT/MCT inhibition to avoid immunosuppressive niche formation	Medium- to long-term PD-1/PD-L1 ± anti-VEGF as immune surveillance; short-course senolytic trial in very-high-risk subgroups	Resection specimens: CSS/SRS, PTM (O-GlcNAc, H3K18 La, EZH2), spatial immune mapping; follow-up: SASP factors, ctDNA, recurrence patterns	BCLC early-intermediate stage, high-risk of recurrence post-resection/transplantation (MVI, satellite nodules, molecular high-risk), with preserved liver function	[45,197,198]

PTM: Post-translational modification; SASP: Senescence-associated secretory phenotype; HCC: Hepatocellular carcinoma; IO: Immunotherapy; TACE: Transarterial chemoembolization; TKI: Tyrosine kinase inhibitor; TIS: Therapy-induced senescence; VEGF: Vascular endothelial growth factor; JAK: Janus kinase; STAT: Signal transducer and activator of transcription; NF-κB: Nuclear factor-κB; IL-6: Interleukin-6; IL-8: Interleukin-8; OGT: O-GlcNAcylation transferase; MCT: Monocarboxylate transporter; PD-1: Programmed cell death protein 1; PD-L1: Programmed death-ligand 1; H3K18 La: Histone H3 lysine 18 lactylation; Treg: Regulatory T cell; MDSC: Myeloid-derived suppressor cell; TAM: Tumor-associated macrophage; TIME: Tumor immune microenvironment; CSS: Cellular senescence score; SRS: Senescence-related signature; γH2AX: Phosphorylated histone H2AX; Ki-67: Marker of proliferation Ki-67; ATR: Ataxia telangiectasia and Rad3-related; mTORC1: Mechanistic target of rapamycin complex 1; LDH: Lactate dehydrogenase; CBP: CREB-binding protein; SUMO: Small ubiquitinlike modifier; CTLA-4: Cytotoxic T-lymphocyte-associated protein 4; NKG2D: Natural killer group 2D; CAR T: Chimeric antigen receptor T cell; EZH2: Enhancer of zeste homolog 2; TCR: T cell receptor; CDK4/6: Cyclin-dependent kinase 4/6; MVI: Microvascular invasion; BCLC: Barcelona Clinic Liver Cancer; ctDNA: Circulating tumor DNA.

SINGLE-CELL AND SPATIAL MULTI-OMICS: MAPPING SENESCENCE, PTMS, AND IMMUNE INTERACTIONS IN HCC

Single-cell and spatially resolved omics add resolution that bulk profiling cannot provide, allowing senescence states, PTM-linked programs, and immune phenotypes to be assigned to specific cell populations and tissue regions in HCC. This section focuses on how these technologies refine mechanism, reveal spatial constraints on immune access, and inform patient stratification and trial design.

Technological advances in single cell and spatial multi-omics for HCC senescence

Single-cell RNA sequencing and spatial transcriptomic/proteomic platforms have enabled reconstruction of cell-state trajectories and their tissue localization in HCC cohorts. Spatial mapping can distinguish immune-inflamed from immune-excluded regions and link these patterns to local chemokine programs and suppressive niches, providing a practical basis for integrating senescence biology and PTM-targeted interventions with immunotherapy selection and scheduling[199-201].

In HCC immune microenvironment research, the integration of single cell and spatial data now allows systematic deconvolution of patient samples across three dimensions: Cell type, functional state, and spatial location. As exemplified by the study of Zhu et al[202], combined single cell and spatial transcriptomic analyses identified the mitochondrial flavin adenine dinucleotide synthesis related gene flavin adenine dinucleotide synthetase 1 as a regulator of the HCC immune microenvironment: Regions with high flavin adenine dinucleotide synthetase 1 expression exhibited distinct metabolic signatures and immune infiltration patterns, nominating it as a candidate biomarker for combining metabolic/mitochondrial targeting with immunotherapy[202]. These methodological advances establish a technical foundation for incorporating senescence and PTM information into single cell and spatial models.

Concurrently, single-cell analytical frameworks for senescence research are being progressively refined. Systematic investigations based on large-scale single-cell atlases have comprehensively characterized the molecular features of senescence across multiple tissues and cell types in mammals. For instance, by integrating cross-tissue single-cell data, studies have proposed transcriptomic signatures for identifying cellular senescence states and highlighted substantial discrepancies between in vivo senescence transcriptomes and in vitro senescence induction models, underscoring the necessity of redefining senescence features within disease-specific contexts[203]. Related advances include the construction of the Human Cellular Aging Transcriptome Atlas, which integrates approximately 92 million cells spanning over 50 tissue types and an age range of 0-110 years, providing a critical resource for systematically exploring senescence transcriptomes at single-cell resolution[204].

At the platform and resource level, knowledge bases such as Aging Annotation database (AgeAnno) consolidate single-cell RNA-seq and ATAC-seq data from over 1.67 million cells across 28 healthy tissue samples, encompassing dynamic functional annotations for 5580 senescence-associated genes and supporting analyses of differential expression, cell-cell communication, and transcriptional regulatory networks[205]. Furthermore, computational tools like the SenePy platform integrate over 60 mouse and human single-cell senescence signatures, enabling senescence scoring across diverse tissues and disease contexts. This provides a directly applicable analytical tool for precisely identifying senescence-associated cell subpopulations in HCC single-cell datasets[147]. Additionally, the Pan-species Single-cell Aging Atlas single-cell transcriptomic atlas, derived from more than 20 million cells across 623 mouse tissues, reveals over 3000 distinct cell states and more than 200 senescence-related cell populations[206]. The Aging Fly Cell Atlas, through whole-organism single-nucleus transcriptomic profiling of aged Drosophila, identifies 163 distinct cell types[207]. Together, these efforts advance the establishment of cross-species senescence cell atlases.

These studies further elucidate cell type-specific alterations during aging. Senescence is often accompanied by a loss of cellular identity, manifested as increased transcriptional noise in aged tissues[208]. To this end, studies have developed single-cell transcriptome-based “aging clocks” capable of quantifying cell type-specific senescence and rejuvenation dynamics[209]. In tissues such as the testis, comparative single-cell transcriptomic analyses between young and aged mice have systematically revealed senescence-associated molecular signatures[210].

Multi-omics integration strategies further deepen the understanding of senescence mechanisms. For example, integrated analyses of transcriptomic, methylomic, single-cell sequencing, and metabolomic data have systematically delineated senescence-associated regulatory networks in adipose and muscle tissues[211]. In brain aging research, single-cell omics technologies enable high-resolution characterization, uncovering age-dependent cellular alterations and remodeling of intercellular interactions[212].

In summary, through the systematic application of single-cell and spatial multi-omics technologies, significant progress has been made in the systematic characterization of senescence molecular features, the dissection of cell type-specific changes, and integrated cross-omics analyses. These advancements establish a methodological and knowledge foundation for precisely defining senescence features, constructing spatiotemporal atlases, and developing stratified diagnostic and therapeutic frameworks in disease contexts such as HCC.

Spatiotemporal characterization of senescence-associated cell subpopulations

At the tumor cell level, multiple single cell transcriptomic studies of HCC have identified tumor cell subpopulations displaying a transcriptionally defined senescence prone phenotype. These cells typically exhibit downregulation of cell cycle genes, upregulation of DNA damage response genes and p53/p21 signaling, and elevated expression of certain SASP related chemokines and inflammatory factors[42,213]. While these transcriptomic profiles are informative, they define a phenotype, not a functionally validated state. True senescence requires confirmation of growth arrest and SASP secretion, and current single cell methods cannot directly measure the PTM landscapes (e.g., lactylation) hypothesized to drive these transcriptional programs. Thus, inferences remain correlative. Spatially resolving these transcriptionally defined populations, however, has proven highly revealing. Spatial transcriptomic analyses further reveal that such senescent like tumor cells are often localized at the tumor-stroma interface or in regions with prominent vascular remodeling. Their spatial proximity to areas enriched in myeloid derived suppressor cells, M2 type tumor associated macrophages, and regions with sparse CD8+ T cell infiltration suggests that local SASP signals may cooperate with myeloid polarization, immune exclusion, and vascular abnormalities to shape an immunosuppressive microenvironment[214]. In the vascular compartment, integrated single-cell and spatial profiling has identified senescent endothelial states and suggested their association with aberrant angiogenesis and immune exclusion, highlighting a spatial dimension that may influence drug delivery and immune access in HCC[215,216].

Regarding hepatic nonparenchymal cells, multiple studies from the perspective of liver disease progression and fibrosis have demonstrated that liver sinusoidal endothelial cells and hepatic stellate cells are prone to senescence during chronic liver injury. Their senescence associated secretome promotes capillarization, fibrogenesis, and increased risk of HCC. For example, Lv et al[217] found that upregulation of CDKN2A/p16 in hepatic stellate cells inhibits their activation, thereby partially delaying fibrosis progression. Meanwhile, Wilkinson et al[218] showed in a chronic liver disease model that SASP released by senescent like cells elevates expression in liver sinusoidal endothelial cells, alters pore structure, and impairs barrier function, creating a microenvironment conducive to tumorigenesis and immune escape. These findings related to fibrosis and chronic liver disease align with the spatial distribution of CDKN2A+ senescent endothelial cells reported by Ma et al[16] in HCC[219]. Collectively, they delineate a continuum of cellular senescence affecting endothelial and mesenchymal cell lineages that extends from pre-neoplastic through neoplastic disease stages. This integrated perspective provides crucial insights for understanding the remodeling of the immune microenvironment during hepatocarcinogenesis[215].

It should be noted that direct detection of protein PTMs in HCC single cell data remains technically challenging. Nevertheless, by analyzing the transcriptional activity of senescence associated gene sets, metabolic programs, and PTM related enzymes, functional links between senescence, PTMs, and immunity can be inferred. For instance, several HCC multi-omics studies have shown that tumor cell clusters with highly active glycolysis and lactate metabolism pathways often exhibit upregulation of OGT and related metabolic regulatory genes. Spatially, these regions frequently coincide with features of immune exclusion and are enriched for myeloid cells. This spatial correlation aligns with the previously proposed mechanistic axis, derived from in vitro and animal models, which links OGT activity, histone lactylation, the SASP, and the establishment of an immunosuppressive microenvironment[220].

Precision stratification and multimodal integration of senescence signatures

Across datasets, senescence-related transcriptional signatures have been developed for prognostic assessment and immune stratification in HCC. While these signatures support the clinical relevance of senescence-associated programs, their translational value will depend on harmonized definitions, external validation, and prospective evaluation alongside treatment response endpoints[48,182,221,222].

Although these studies primarily rely on bulk tissue transcriptomics, they provide a foundational framework for integrating single cell and spatial information into senescence-based stratification. First, tools such as SenePy or HCC specific senescence signatures can be applied to single cell HCC datasets to perform single cell senescence scoring for tumor, endothelial or stromal, and immune cells, thereby identifying senescence enriched regions and cell subpopulations[147]. Second, these senescence scores can be integrated with spatial transcriptomic maps depicting the distribution of immune cells, vascular density, and metabolic features. This integration can be achieved, for example, by incorporating established patterns of spatial immune heterogeneity or utilizing defined spatial immune scoring systems. The combined analysis enables the selection of specific tumor regions and patient cohorts characterized by a high burden of senescent cells, a profoundly immunosuppressive microenvironment, and a consequently elevated risk of tumor recurrence[214,223]. Finally, further integration of metabolic and PTM related transcriptional features enables the construction of a multimodal evaluation system that combines senescence scores, spatial immune characteristics, and metabolic PTM indicators. This integrated approach would offer more precise patient stratification and enrollment criteria for future clinical trials exploring combined senescence induction, immune clearance, and PTM targeted interventions in HCC[224,225].

Clinical translation: From spatiotemporal maps to actionable stratification systems

Translating the integrated network of senescence, PTMs, and immunity, which has been elucidated by single cell and spatial omics studies, into a clinically useful decision support tool necessitates establishing systematic pathways encompassing sampling methodologies, detection systems, and trial design. We propose a multidimensional framework to address this. Firstly, sampling should be longitudinal and dynamic, centered on key clinical timepoints such as before and after TACE or radiotherapy, as well as during the initiation and evaluation phases of systemic therapy when combined with immunotherapy and antiangiogenic agents. Spatially, priority should be given to biologically informative regions including the tumor core, invasive front, vessel-dense areas, and fibrous septa or necrotic margins. This comprehensive strategy enables the systematic capture of dynamic changes in senescence burden, immune infiltration patterns, and features involving metabolism and PTMs, all at single cell and spatial resolution[199,226].

Second, for detection system development, integrated systems should combine transcriptomics-based senescence scores with key spatial immune metrics: The differentiation state of CD8+ T cells, spatial distributions of Tregs, MDSCs, tumor associated macrophages, and CDKN2A-high endothelial senescence hotspots. Concurrently, essential PTM and metabolic markers must be incorporated, including histone lactylation (H3K18 La), OGT expression, lactate metabolism genes, and core SASP factors like IL-6, IL-8, CCL2, and tumor necrosis factor-β). This will yield a unified biomarker panel serving stratification and companion diagnostic purposes, ready for validation in prospective cohorts[45,215,227,228].

Third, in terms of clinical trial design, prospective randomized substudies can be integrated into ongoing immunotherapy and antiangiogenic combination regimens during phases that trigger senescence, for instance with local therapy, targeted agents, or CDK4/6 inhibitors. These studies would assess the effectiveness of a combined approach involving senescence induction, immune clearance, and metabolic/PTM intervention within precisely stratified patient subgroups characterized by specific senescence-immune signatures. Throughout, dynamic single-cell and spatial multi-omics data must be analyzed alongside standard clinical endpoints like progression-free survival and overall survival, to verify that biological responses correlate with tangible clinical outcomes[45,70,78,195,229].

Finally, all senescence related scores and multimodal stratification models require prospective registration and external validation in independent cohorts, followed by iterative refinement into simplified indicator sets that can be robustly assayed on routine clinical platforms. With the continuing accumulation of HCC single cell and spatial data resources, along with ongoing advances in senescence and PTM targeted interventions, precision stratification systems integrating multilayered information are poised to become pivotal tools for advancing personalized treatment and therapeutic monitoring in HCC.

DISCUSSION

This review delineates a dynamic, interconnected layer of PTMs that couples cell-intrinsic senescence programs with the extrinsic tumor immune microenvironment in HCC. This PTM-senescence-immune axis not only elucidates key mechanisms underlying resistance to current immunotherapies but also unveils novel targets for sequential therapeutic intervention. To translate this mechanistic understanding into clinical practice, a critical appraisal of the robustness of the evidence and the specific barriers to translation is imperative. We synthesize these insights below, with a detailed evaluation of evidence levels and translational roadblocks provided in Supplementary Table 1.

An evidence landscape: From established mechanisms to clinical hypotheses

Current evidence reveals a clear hierarchy. The PTM-mediated regulation of DNA damage and the p53/pRb pathways, through phosphorylation, acetylation, and ubiquitination, constitutes a conserved, mechanistically solid core governing senescence decisions[53-58]. Within the unique context of chronic inflammation and metabolic stress characteristic of HCC, however, two metabolism-coupled PTMs emerge as the most relevant bridges linking core HCC pathophysiology to immune evasion: O-GlcNAcylation and histone lactylation. O-GlcNAcylation, a nutrient-sensing modification, drives a feedforward loop linking glycolysis to histone lactylation, a mechanism clearly established in preclinical models[10,33]. Complementarily, lactylation directly incorporates the end-product of the Warburg effect, lactate, into chromatin (e.g., at H3K18 La) and is robustly associated with metastasis, therapy resistance, and an immunosuppressive microenvironment in clinical HCC samples[35-37,43]. In contrast, while roles for SUMOylation, methylation, and other PTMs in HCC senescence have been reported, their precise functional specificity and hierarchical importance require further causal validation[90-96,138].

Central translational barriers and prioritized pathways

Despite fruitful mechanistic discoveries, clinical translation faces three overarching challenges. First, a broken causal chain. The association of metabolic PTMs with poor prognosis largely stems from correlative multi-omics analyses. Whether pharmacological perturbation of a specific PTM can reverse immune suppression and improve outcomes in patients remains an untested hypothesis. For instance, while inhibiting O-GlcNAcylation or lactate production shows promise in preclinical models, their safety, pharmacodynamics, and synergy with existing therapies like ICIs in humans are unknown[10,128]. Second, a lack of spatiotemporal dynamics. TIS and its associated PTM landscape are dynamic processes. Current understanding, derived mostly from static biopsies, leaves the peak timing of TIS, the evolution of its PTM signature, and the optimal therapeutic window for implementing senolytic or immune-mediated clearance strategies largely unexplored in patients[77,85]. Third, an absence of predictive biomarkers. Senescence transcriptomic scores (e.g., CSS/SRS) show robust prognostic and immune correlative value in retrospective cohorts[45-48]. However, their efficacy as prospectively validated, actionable tools for guiding individualized treatment decisions awaits confirmation in rigorously designed clinical trials.

Toward a refined “induce-remodel-clear” therapeutic framework

Given this evidence landscape, the proposed “induce-remodel-clear” strategy is best viewed as a conceptual roadmap for priority validation, not a mature clinical protocol. Its translation demands closed-loop verification of each step: (1) Precision induction. While leveraging radiotherapy, TACE, or CDK4/6 inhibitors to induce TIS, parallel development of non-invasive monitoring via liquid biopsies including cell-free DNA and SASP factors, along with serial imaging, is essential to quantify individual senescent burden[78,197]; (2) Targeted remodeling. This step represents the key intervention point tailored to HCC biology. Priority should be given to testing agents targeting the metabolic-PTM axis (e.g., OGT, lactate dehydrogenase, or MCT inhibitors) for their ability to reprogram the SASP from an immunosuppressive (IL-6/IL-8-dominant) to an immune-stimulatory (C-X-C motif chemokine ligand 9/10-dominant) profile[98,100]. Validation requires physiologically relevant models (e.g., humanized mice, patient-derived organoid co-cultures) followed by early-phase clinical trials with embedded biomarker analyses; and (3) Timely clearance. The timing of clearance strategies (e.g., senolytics, NKG2D-CAR T, or PD-1/PD-L1 blockade) is critical. Preliminary evidence suggests combining clearance modalities post-TIS peak may reverse immunosuppression[171,185]. This must be systematically tested in well-designed trials (e.g., neoadjuvant or adjuvant settings) using paired pre- and post-treatment biopsies.

CONCLUSION

The PTM network establishes an essential molecular regulatory architecture that systematically orchestrates the dynamic interplay between cellular senescence and the tumor immune microenvironment in HCC. This review demonstrates that PTM modifications, particularly the core modifications of O-GlcNAcylation and histone lactylation, function as critical regulatory hubs within the metabolism-epigenetics-immunity axis, representing the most promising therapeutic targets within the unique pathophysiological context of HCC. The field is currently undergoing a pivotal transition from observational correlation to interventional causal validation. To advance clinical translation, future research should prioritize longitudinal, patient-centered studies to systematically validate the functional significance of these PTM nodes. Of particular importance is the integration of multidimensional biomarkers, including PTM-specific modification signatures, senescence-associated scores, and spatially resolved immune profiles, into prospective clinical trial designs, thereby transforming senescence from a passive bystander state into a precisely targetable therapeutic vulnerability. Ultimately, targeting the PTM-senescence-immune regulatory axis will provide a novel paradigm for overcoming immunotherapy resistance in HCC and advancing precision oncology. The integrated evidence presented herein not only provides a mechanistic foundation but also delineates a translational roadmap for converting the senescence process into a controllable therapeutic dimension, thereby opening new avenues for developing innovative combination therapies and improving clinical prognosis.

ACKNOWLEDGEMENTS

We sincerely thank all members of our research teams for their valuable discussions and technical support.