Targeting phase separation: a promising treatment option for hepatocellular carcinoma

Abstract

The spontaneous phenomena known as liquid–liquid phase separation (LLPS) is caused by weak interactions between substances. Under specific circumstances, macromolecules like proteins and nucleic acids can dynamically aggregate to form biomolecular condensates. This phenomenon offers a novel perspective on the intricate spatiotemporal coordination within living cells. Recent research has shown that LLPS is crucial for the initiation and progression of cancer, mainly by influencing multiple cellular activities such as metabolism, autophagy, stress responses, immune reactions, transcriptional regulation and intracellular signaling pathways, etc. Dysregulation of LLPS significantly affects the proliferation, metastasis, and therapeutic resistance of hepatocellular carcinoma (HCC) cells. Here, we introduce recent advances in understanding how LLPS regulates HCC-associated signaling pathways. Furthermore, we discuss the molecular mechanisms underlying the LLPS of oncogenic signaling molecules and its potential implication. Finally, we summarize several feasible approaches for treating HCC by targeting LLPS. These findings have the potential to establish a novel theoretical framework and therapeutic hypothesis for cancer treatment, thus providing more precise and individualized clinical strategies and significantly enhance patient prognosis and overall survival rates.

Introduction

Liver cancer represents a major global health problem because of its high incidence and poor prognosis. HCC, as the most common histological subtype of liver cancer, accounts for about 90% of all cases [1]. Increasing evidence suggests that aberrant transduction of HCC-related signaling pathways can induce tumor cell proliferation and metastasis, such as Hippo signaling [2], Wnt/β-catenin signaling [3], and Nuclear Factor kappa B (NF-κB) [4, 5]. Recently, molecular targeted therapy and immunotherapy have brought new hope to HCC patients, however, their therapeutic efficacy remains limited by certain constraints, such as drug resistance or immune escape in some patients [6]. Therefore, a deeper comprehension of the molecular processes behind HCC and the identification of novel, more effective therapeutic targets are critical to advancing treatment strategies and improving patient outcomes.

Elucidating the mechanisms of spatiotemporal coordination within biological pathways is critical for understanding the molecular underpinnings of disruptions in cancer-associated signaling pathways. In recent years, LLPS has emerged as a novel framework for elucidating the precise spatiotemporal regulation of molecular events within living cells. This biophysical process facilitates the spontaneous segregation of biological macromolecules, such as proteins and nucleic acids, from a uniformly distributed liquid state into two or more coexisting phases: a dilute phase and a dense phase [7]. The dense phase typically manifests as droplet-like structures capable of fusion, dripping, or dynamic reorganization, thereby serving as compartments for segregation, enrichment, and regulation within the cell. These structures provide a unique microenvironment conducive to complex cellular activities [8]. Biomolecular condensates formed through LLPS are now recognized as a pervasive and essential cellular mechanism, playing a critical role in the regulation of macromolecular metabolism and biochemical reactions. Consequently, they have become a focal point in contemporary studies of cell biology and disease pathogenesis. Emerging evidence indicates that HCC progression is strongly linked to phase-separated condensates, including p62 bodies [9], paraspeckles [10] and stress granules (SGs) [11]. These condensates display aberrant dynamic properties and functional alterations within tumor cells, contributing to tumorigenesis either directly or indirectly. Thus, LLPS has been identified as a potential cancer biomarker and a promising target for therapeutic intervention [12]. As reported, the detection of elevated levels of p62 bodies [9], paraspeckles [10], and RING finger protein 214 (RNF214) [13] LLPS-associated complexes holds significant potential as biomarkers and therapeutic targets for intervention in HCC.

Therefore, understanding the functions and mechanisms of LLPS in these processes is essential to develop novel treatment approaches for HCC. In this article, we introduce the regulatory mechanisms of LLPS, review recent research on how biomolecular condensates affect signaling pathways involved in hepatocarcinogenesis, and elucidate the application of LLPS in metabolic reprogramming, autophagy, and immune response in HCC. Finally, we summarize potential approaches for HCC treatment by targeting LLPS. In summary, this review provides novel perspectives for the diagnosis and treatment of HCC based on LLPS.

Characteristics of LLPS

Eukaryotic cells are characterized by the presence of both traditional membrane-bound organelles (such as the nucleus, mitochondria, and endoplasmic reticulum) and a diverse array of membrane-less organelles (MLO) and dynamic biomolecular assemblies. In contrast to membrane-bound organelles, which utilize lipid bilayers for spatial compartmentalization, MLOs are typically formed through LLPS. This is a dynamic and reversible process driven by weak multivalent interactions among biomolecules [14]. A key insight into this mechanism arose from the observation that P granules in Caenorhabditis elegans exhibit liquid-like phase transition behavior [15]. Unlike the structurally defined and membrane-enclosed nature of traditional organelles, LLPS-mediated condensates exhibit significant dynamicity, facilitating rapid assembly, disassembly, and molecular exchange in response to cellular signals. This dynamic nature provides a flexible platform for spatiotemporal regulation and plays critical roles in various biological processes, including gene expression regulation, signal transduction, and stress responses. For instance, SGs protect mRNA from degradation during cellular stress [16].

In recent years, the role of LLPS in the assembly of MLO is becoming more widely acknowledged. LLPS involves a broad range of biomolecules, including proteins, nucleic acids, and lipids. Upon reaching a critical concentration, these biomolecules can either act individually or interact with each other to aggregate into dense, membrane-less condensates occurring LLPS form variable-sized, freestanding droplets, which appear as oil droplets under the microscope and are isolated from their surroundings. The droplets are dynamic spherical structures, and their size is comparable to that of organelles, so they are also called MLO in certain studies, such as nucleoli, cajal bodies, paraspeckles localized in the nucleus, P-bodies, SGs localized in the cytoplasm [17]. Nucleoli is the largest and most easily detectable MLO in eukaryotic cells, which plays a central role not only in ribosome biogenesis but also in critical cellular functions including DNA replication, DNA repair, and the cellular stress response [18]. P-bodies and SGs are ribonucleoprotein (RNP)-based cytoplasmic biomolecular condensates that assemble under stress to dynamically segregate translationally inactivated mRNPs into distinct compartments. Aberrant aggregation of disease-associated proteins, such as TAR DNA-binding protein 43(TDP-43), disrupting LLPS-dependent SG dynamics, represents a key mechanism underlying various neurodegenerative diseases [19]. Importantly, LLPS is increasingly recognized as a critical mechanism in cancer biology. In HCC, the formation of SG downregulates Myc proto oncogene protein expression by suppressing the translation of c-Myc [20]. Modulating the formation and disassembly of SGs offers novel avenues for advancing disease diagnosis and therapeutic strategies. Furthermore, Wang S et al. [10] demonstrated that the formation of paraspeckles in silenced HCC cells inhibited HCC progression, highlighting the critical role of paraspeckles in hepatocarcinogenesis. In summary, compared to traditional membrane-bound organelles, LLPS-derived biomolecular condensates offer distinct advantages such as reversibility, responsiveness, and rapid dynamic remodeling. These features enable more agile regulation of cellular processes and are increasingly recognized as key factors in the progression of various diseases, including cancer.

Regulation of LLPS

LLPS is usually driven by weak, dynamic multivalent interactions between proteins, or between proteins and RNAs, associated with intrinsically disordered regions (IDRs) and hydrophobic amino acid regions [21, 22]. Most RNA-binding proteins (RBP) have IDRs and low-complexity regions (LCRs), called prion-like structural domains (PLDs), that permit the formation of biomolecular condensates in an overcrowded nuclear environment [23,24,25]. For example, Fused in Sarcoma protein (FUS), a protein linked to mRNA splicing and transcription, has been shown to undergo phase separation due to its prion-like LCR in its sequence [26]. Furthermore, a 2023 study found that Y box-binding protein 1 (YBX1), a highly disordered RBP with a fixed structure in the middle, and nuclear circASH2 enhanced the degradation of Tropomyosin 4 (TPM4) transcripts by augmenting the LLPS of YBX1, which inhibited HCC metastasis [27]. In fact, its IDR is a key feature of the protein LLPS.

Under normal conditions, most LLPS processes are reversible, and their formation is influenced by various factors, which are mainly categorized into two aspects: first, the nature of the protein sequence, such as the effect of post-translational modifications (PTMs); and second, by the microenvironment of the system, such as component concentration, salt concentration, and temperature (Fig. 1).

Regulation of the microenvironment

Numerous studies have shown that direct regulation of component concentration in solutions directly affects the occurrence of phase separation. For example, in different cancer cells, Bromodomain containing 4 short isoforms (BRD4S) undergo LLPS in the nucleus to form nuclear puncta, which recruits transcription-associated proteins and promotes transcriptional activity and thus promotes cancer cell proliferation. Under in vitro conditions, an increase in BRD4S concentration was observed to correlate with a rise in the quantity of droplets formed via LLPS, exhibiting a clear concentration dependence [28]. Furthermore, temperature has been shown to influence the assembly of aggregates. A well-characterized example is the germ granule protein DEAD-box helicase 4 (DDX4), whose intrinsically disordered N-terminus region undergoes LLPS to form droplets both in cells and in vitro. Intriguingly, DDX4 condensates exhibit high sensitive to temperature fluctuations, dissociating at elevated temperatures and reassembling at lower temperatures [29]. Moreover, the concentration of salt ions exerts a critical effect in phase separation systems. Elbaum-Garfinkle S et al. [30] demonstrated that LAF-1 was isolated in vitro into P granule-like droplets, and as the salt concentration increased, the mobility of particles within these droplets also increased. This phenomenon occurs because salt ions modulate weak intermolecular interactions, thereby influencing the phase separation. Interestingly, elevated salt concentrations can also inhibit LLPS. For instance, high salt inhibited the phase separation of SERPINE1 mRNA Binding Protein 1 and DDX4 [31, 32]. Therefore, for different phase separation systems, salt concentration under specific conditions may lead to different phenomena, and the specifics still need to be experimentally verified based on the properties of biomolecules. In addition, pH also affects the stability of specific proteins, further modulating their intermolecular interactions and subsequently impacting LLPS. For instance, low pH promotes the LLPS of Sup35 [33], α-Syn [34]. Molecular chaperones represent another major component influencing phase separation. These proteins are ubiquitously present in cells, ensuring proper proteins folding and functionality by either providing a conducive environment or directly participating in the folding process. Small heat shock protein 27 (Hsp27) is a typical chaperone localized to SGs. Research has shown that Hsp27 inhibits FUS LLPS by interacting weakly with the LCRs of FUS [35]. In patients with atypical fibrolamellar carcinoma of liver oncogenic DnaJB1-PKAcat loses myristoylation and Hsp70 binding, inhibits the regulatory subunit Iα (RIα) LLPS and leads to tumorigenesis [36]. Thus, the above examples illustrate the important role of molecular chaperones in maintaining intracellular homeostasis and the significance of normal LLPS for organisms.

Regulation of PTM

PTM represents a primary mechanism regulating phase separation. Common types of PTMs, including protein phosphorylation, methylation, acetylation, and ubiquitination, have been extensively investigated for their impact on LLPS. These PTMs modulate LLPS, either positively or negatively, by altering the physicochemical properties, spatial conformation, charge state, and hydrophobicity of proteins. Here, we present examples of PTMs influencing phase separation.

In terms of signal transduction, phosphorylation of transcriptional coactivator with PDZ-binding motif (TAZ) inhibits its LLPS, preventing the recruitment of transcriptional regulators to the phase-variable region thus affecting downstream gene expression [37]. Interestingly, phosphorylation at distinct locations on the same protein can exert differential effects on LLPS. For instance, the phase separation of the glutamate receptor component GluN2B and the auxiliary protein stargazin is inhibited by phosphorylation of Ser78 on postsynaptic density protein 95, but stargazin LLPS is induced by phosphorylation of Ser116 [38]. Ubiquitination modifications are essential for regulating the formation and stabilization of LLPS. A recent study showed that phase separation of the ubiquitin ligase RNF214 promotes HCC proliferation, migration and metastasis [13]. Targeting RNF214 may offer the possibility of developing new therapies for HCC. Acetylation is also crucial for LLPS, for example, the acetyltransferase KAT8 promotes interferon regulatory factor 1(IRF1) acetylation through LLPS, thereby facilitating PD-L1 transcription and contributing to tumor immune escape [39]. Interestingly, different PTMs of the same protein can have distinct effects on LLPS. For instance, hyperacetylation of Tau inhibits LLPS and prevents microtubule assembly [40], ubiquitination of Tau stabilizes droplets [41], and phosphorylation of Tau promotes LLPS [42]. Methylation is an effective way to regulate different subcellular microenvironments. For instance, methylation of arginine in DDX4 altered the phase transition of DDX4 and significantly reduced droplet stability, suggesting that methylation of DDX4 attenuates its LLPS [29]. It is worth researching in-depth how to use the PTM regulatory mechanism in LLPS to comprehend illness pathophysiology and thus develop targeted drugs.

Overall, the formation of LLPS is a highly intricate process governed by a variety of factors. These factors influence the propensity of LLPS by altering intermolecular interactions, solution conditions, and molecular structure. Elucidating the interplay among these factors will enhance our understanding of the mechanism underlying LLPS, thereby laying the groundwork for early diagnosis and therapeutic intervention.

Fig. 1

Factors regulating liquid-liquid phase separation. A variety of factors can promote or inhibit liquid-liquid separation, including temperature, component concentration, salt concentration, chaperones, and PTM

Potential role of LLPS in HCC

LLPS is crucial in HCC, acting as both a tumor promoter and suppressor depending on molecular interactions and intracellular environments involved. For example, the Myc-Associated Zinc finger protein (MAZ), recruited by G-quadruplex (G4) structures, forms MAZ-CCND1-G4 condensates that activate CCND1, driving HCC progression [43]. In contrast, Speckle-type POZ protein (SPOP) forms film-free clusters by LLPS that suppress hepatocarcinogenesis [44]. SPOP is generally downregulated in HCC patients, and it expression is negatively associated with tumor grade [45, 46].

LLPS provides new perspectives and concepts for key genes expression and signaling during the development of HCC. Table 1 summarizes the main examples of aberrant LLPS involved in HCC signaling pathways. Here, we will elucidate the relationship between dysregulated phase separation and oncogenic signaling pathways (Fig. 2) and the therapeutic potential of targeting oncogenic molecules.

LLPS in metabolic reprogramming

Metabolic reprogramming is prevalent in the genesis and progression of HCC, conferring tumor cells with a unique metabolic advantage that facilitates sustained malignant proliferation [47]. The role of LLPS in metabolic regulation is mainly to provide microenvironmental support for cellular metabolic activities by facilitating the dynamic aggregation and segregation of key metabolic enzymes, metabolic intermediates and signaling molecules. The emerging understanding of HCC glucose metabolism shows that tumor cells may metabolically catabolize glucose by storing glucose as glycogen for energy, rather than in the form of anaerobic glycolysis. Importantly, the accumulated glycogen undergoes LLPS, and the glycogen-binding protein Laforin interacts with mammalian STE20-like protein kinase 1/2 (Mst1/2), thus partially encapsulating Mst1/2 in glycogen droplets to inhibit the Hippo signaling pathway and activate the downstream Yes-associated protein (YAP), which contributes to the development of the cancer [48]. Furthermore, in humans and mice, Glucose-6-Phosphatase or the glycogenolysis enzyme-liver glycogen phosphorylase deficiency leads to glycogen accumulation, promoting hepatomegaly and tumorigenesis in a YAP-dependent manner [49]. YAP is a key oncogenic factor in HCC, which forms a liquid condensate in the nucleus via LLPS that compartmentalizes the transcription factor TEA Domain transcription factor 1 (TEAD1) and other YAP-associated coactivators and subsequently induces transcription of YAP-specific proliferative genes [50]. Several studies have demonstrated that YAP modulates the involvement of cell proliferation, metabolism and metastasis, and promotes resistance to targeted therapies in HCC [51, 52]. Disruption of YAP and its LLPS represent promising strategies for inhibiting hepatocarcinogenesis.

SG is a RNP condensate formed in response to external stresses and serves as a signaling hub that regulates gene expression and signal transduction, which is crucial for cancer cell survival under unfavorable conditions [53]. By reallocating energy and material resources, tumor cells can adapt to stresses such as oxygen and nutrient deficiencies in the microenvironment while maintaining rapid cell proliferation. This metabolic reprogramming sets the stage for invasion, metastasis, and drug resistance. For instance, in the context of glutamine deprivation, upregulated levels of LncRNA GIRGL interact with CAPRIN1 and glutaminase-1 (GLS1) mRNA, facilitating the LLPS of CAPRIN1 and driving the assembly of SGs. This process inhibits GLS1 translation and allows tumor cells to survive under glutamine deprivation stress [54]. Circular Vesicle Associated Membrane Protein 3 (CircVAMP3) can drive CAPRIN1 LLPS, resulting in SGs formation, which suppresses c-Myc translation and HCC cells proliferation and metastasis [20]. Moreover, sorafenib, a Raf1/Mek/Erk kinase inhibitor, is clinically used to treat advanced HCC. Research has demonstrated that sorafenib therapy triggers the unfolded protein response and causes eukaryotic Initiation Factor 2α to be phosphorylated, which in turn causes the development of SG [55, 56]. Therefore, SGs play a key role in metabolic reprogramming through LLPS and microenvironmental adaptation. Investigating the mechanism of SGs contributes to an in-depth understanding of metabolic adaptation strategies in HCC and provides an important direction for developing novel therapies.

LLPS in autophagy

Under normal conditions, autophagy effectively recognizes and removes abnormal biomolecular condensates formed via LLPS. However, when stress or pathological factors lead to abnormal phase separation of proteins and escape autophagic surveillance, biomolecular condensates may accumulate and trigger cellular dysfunction [57]. P62 is a pivotal autophagy receptor protein that has been reported to exhibit overexpression in precancerous liver diseases and HCC [58]. In HCC, p62 promotes Nuclear Factor Erythroid 2-Related Factor 2 (Nrf2) activation through interaction with the Nrf2 repressor protein Keap1, thereby enhancing antioxidant gene expression [59]. Notably, the phase separation of p62 is intricately dependent on its self-aggregation, its interactions with ubiquitin, and the valence of polyubiquitin chains [60]. P62 bodies formed through LLPS can serve as sites for autophagosomes to regulate oxidative stress and thus induce tumor development [61, 62]. For instance, Modulator of Apoptosis 1 (MOAP-1) interacted with p62 and interfered with its auto-oligomerization activity, altering its LLPS and increasing the dissociation of p62 bodies to negatively impact p62-Keap1-Nrf2 signaling. In a diethylnitrosamine-induced hepatocarcinogenesis model, mice lacking MOAP-1 exhibit a higher tumor load accompanied by markedly increased levels of p62 bodies and heightened activation of the Nrf2 signaling [9]. MOAP-1 is the first regulator to exhibit catabolic enzyme-like activity and may be a new target for drug therapy in HCC. In addition, Hepatic Mallory-Denk bodies, commonly observed in various liver diseases [63], require p62 for their maturation [64], which can inhibit HCC migration and invasion by knocking out p62 [65]. Recent studies found that vault, a non-membrane organelle, is degraded by selective autophagy that is dependent on p62 phase-separation, and that its damage is associated with nonalcoholic steatohepatitis-derived HCC [66]. In summary, the above results imply that targeting the LLPS behavior of p62 and restoring normal autophagy represents a promising therapy.

LLPS in immune microenvironment

LLPS is an emerging molecular regulatory mechanism crucial for modulating tumor immune-related signaling within the dynamic immune microenvironment [67]. The cGAS-STING pathway detects exogenous or aberrant DNA to trigger innate immunity. Upon recognition of pathogenic DNA, cyclic GMP-AMP synthase (cGAS) catalyzes the synthesis of cyclic-GMP-AMP (cGAMP), a second messenger that subsequently activates stimulator of interferon genes (STING). Activated STING recruits and phosphorylates the TANK-binding kinase 1 (TBK1), which stimulates the IRF3. IRF3 then translocates the nucleus and promotes interferon (IFN) transcription as well as the expression of inflammatory cytokines [68]. These immune molecules can activate immune cells in the tumor microenvironment, such as dendritic cells (DCs), Natural killer (NK) cells and CD8 + T cells, and enhance the body’s anti-tumor ability [69]. Studies have demonstrated that immune cell infiltration within HCC tissues is improved and hepatitis B virus (HBV) replication is markedly inhibited in vivo when the cGAS-STING signaling pathway is activated [70, 71]. Moreover, research indicated that DNA binds to cGAS and induces its dimer to undergo phase separation, greatly enhancing cGAMP production and promoting innate immune signaling [72]. Regulating LLPS in the cGAS-STING pathway is key to maintaining immune homeostasis and offers a novel direction and application for enhancing anti-tumor immunotherapy.

Furthermore, LLPS is essential in the immune evasion of HCC. For instance, paraspeckles in HCC cells inhibit IFN-γ-IFNGR1 signaling by sequestering IFNGR1 mRNA to help tumor cells evade immune surveillance and thus avoid T-cell killing [73]. Clinically, HCC patients with overexpression of nuclear paraspeckle assembly transcript 1 (NEAT1), a component of paraspeckles, showed greater resistance to T-cell-based therapies. Thus, NEAT1 represents a potential therapeutic target for HCC immunotherapy. These findings highlight the dual role of LLPS in both promoting immune activation and facilitating immune evasion, underscoring its complexity in HCC progression.

LLPS in cAMP/PKA signaling pathway

Research has indicated that the cAMP/PKA pathway is crucial for the initiation and advancement of hepatic cancer. CAMP is synthesized through the catalytic synthesis of transmembrane AC mediated by the G protein-coupled receptor. It regulates the proliferation, migration and metabolism of cancer cells by modulating PKA activity and influencing the phosphorylation status of downstream target proteins [74]. A 2020 study showed that the type I regulatory subunit of PKA, RIα, undergoes LLPS to form a biomolecular condensate with high cAMP and PKA activity, which is a major factor in cAMP compartmentalization. Notably, deletion of RIα LLPS and disruption of cAMP region compartmentalization promotes hepatocarcinogenesis. In atypical hepatocellular carcinoma fibrous lamellar carcinoma, DnaJB1-PKAcat inhibits RIα LLPS, dysregulating the cAMP/PKA pathway, leading to increased cell proliferation and transformation [36]. In conclusion, disrupting RIα phase separation results in altered cAMP compartmentalization, which is linked to hepatocarcinogenesis. Targeting this pathway could present a potential therapeutic approach for HCC.

LLPS in RAS/MAPK signaling pathway

According to earlier research, the Ras/MAPK pathway is triggered in 50–100% of human HCC cases and is associated with dismal prognosis [75]. Several studies have demonstrated that aberrant phase separation contributes to the MAPK signaling activation. SHP2, a non-receptor protein tyrosine phosphatase, is a key mediator of hepatocyte-driven tumorigenic signaling and is critical in the RAS/MAPK signaling cascade [76, 77]. The Zhu research team found that SHP2 mutants undergo aberrant phase separation and recruit wild-type SHP2 to the phase separation system, activating the phosphatase activity of SHP2 to promote hyperactivation of the downstream MAPK signaling pathway [78]. Moreover, LAT-Grb2-SOS-mediated phase transition can enhance RAS activation [79], and EML4-ALK formed phase-separated condensates that activate RAS to drive RAS/MAPK pathway [80]. In HCC, fetal TGF-β Activated Kinase 1 (TAK1) tightly interacts with TAK1-binding protein 3 (TAB3), forming liquid‐like condensates that sustain MAPK signaling activation to promote tumor cell proliferation and migration [81]. Therefore, TAK1 and its LLPS-forming condensate may serve as promising biomarkers for HCC.

LLPS in IL-6/STAT3 signaling pathway

IL-6 and STAT3 are crucial in the development of HCC [10, 82, 83]. Wang S et al. [10] found that increased NEAT1 expression in HCC samples based on TCGA data. Silencing NEAT1 inhibited IL-6-induced STAT3 phosphorylation, thereby inhibiting HCC progression. The finding suggests that IL-6 signaling promotes the formation of paraspeckles in HCC, which activates STAT3 and promotes HCC progression. Therefore, paraspeckles represent a promising therapeutic target for HCC.

LLPS in NF-κB signaling pathway

The link between the NF-κB signaling pathway and LLPS reveals a novel molecular regulatory mechanism. A 2024 study found that cellular prion protein (PrPc) generates LLPS and recruits TRAF2, TAB3 and TAK1 to form a dynamic condensate that activates the NF-κB signaling to specifically enhance IL-8 expression, thereby promoting tumor progression [5]. Precise regulation against NF-κB signaling lays the foundation for combined anti-inflammatory and anti-tumor therapeutic strategies.

In conclusion, biomolecular condensates are essential for tumor cell signaling and other key cellular biological processes. These condensates support tumor cell proliferation, migration and drug resistance by dynamically regulating signaling pathway activity, gene expression and protein stability. Targeting the formation or depolymerization of these biomolecular condensates may not only inhibit the aberrant activity of tumor-related signaling pathways but also restore normal cellular functions, potentially reducing the risk of tumor progression and metastasis.

Fig. 2

Role of LLPS in oncogenic signaling in HCC. cGAS-dsDNA condensates promote cGAMP production and innate immune signaling. Laforin-Mst1/2 condensates disrupt WW45-Mst1/2 complex and activate Yap to promote HCC development. Fetal TAK1 interacts with TAB3, forming liquid-like condensates and maintaining MAPK signaling activation to promote tumor proliferation. PrPc generates LLPS and recruits TRAF2, TAB3 and TAK1 to form a dynamic condensate that activates the NF-κB signaling pathway to accelerate tumor progression. DnaJB1-PKAcat inhibits RIα’s LLPS, disrupting the cAMP compartment and dysregulating the cAMP/PKA signaling pathway. High levels of p62 bodies activate Nrf2 signaling and accelerate cancer development. The formation of paraspeckles activates STAT3 and promotes HCC progression

Table 1 Oncogenic signaling in HCC associated condensates that were involved in LLPS

LLPS-based targeted therapy for HCC

With the increasing research on the mechanism of LLPS involvement in HCC, it has become increasingly evident that targeting the aberrant activation and phase-separation processes of these signaling pathways may emerge as one of the most promising strategies for treating HCC (Fig. 3). Table 2 summarizes the drugs or molecules that interfere with hepatocarcinogenesis, and progression based on LLPS.

Table 2 Table 2 Summary of strategies and drugs interfering with hepatocarcinogenesis based on LLPS

Targeted therapy for condensates

The potential for LLPS-based therapies targeting aberrant condensates is enormous. For example, Steroid Receptor Coactivator-1 (SRC-1) is a crucial co-activator within the YAP/TEAD transcriptional complex. Research has shown that SRC-1 interacts with YAP/TEAD and enhances YAP transcriptional activity by compartmentalizing the SRC-1/YAP/TEAD condensate. The anti-HIV medication elvitegravir specifically binds to the SRC-1 protein to disrupt SRC-1/YAP/TEAD condensate SRC-1 LLPS and hence antagonize YAP oncogenic transcriptional activity [84]. In addition, verteporfin disrupts the formation of YAP-TEAD complex and inhibits cell growth in HCC [85]. Therefore, the potential for LLPS-based therapies targeting aberrant condensates is enormous. Similarly, SHP2 mutants promote MAPK activation through aberrant phase separation, as described previously. Zhu G et al. [78] found that SHP2 allosteric inhibitors limit the phase separation ability of disease-associated SHP2 mutants by stabilizing the self-closed conformation of SHP2. Modulating the tumor microenvironment with LLPS inhibitors is a new method to limit tumor development. For example, TAK1 inhibitors modulate immune cell infiltration in HCC and offer a promising approach to inhibit tumor growth [81].

Furthermore, certain transcriptional coactivators undergo phase separation at super enhancers, forming droplets that concentrate the transcriptional machinery and regulate gene expression [86]. Transcription factors such as Twist1 and YY1, which are key inducers of EMT in tumors, are extensively expressed across various malignancies and often correlate with poor clinical prognosis. J M et al. [87] found that Twist1 and YY1 combine with p300 at the super-enhancer of miR-9 to generate a phase-separated condensate that exerts a transcriptional activation function and facilitates hepatoma cells migration and invasion. Notably, the authors further demonstrated that metformin disrupts the condensate separating the Twist1-YY1-p300 phases, thereby decreasing miR-9 expression and thus inhibiting HCC progression. This effect may be attributed to the guanidine moiety in metformin, which interferes with the charged amino acid interaction between Twist1 and YY1. Therefore, the Twist1-YY1-P300 complex offers a potential therapeutic target for HCC, and metformin, as an inhibitor of phase separation, presents a novel approach for its anti-tumor activity. Additionally, Ubiquitin-Specific Peptidase 10 (USP10), a highly conserved deubiquitinating enzyme, has been identified to be crucial to the progression of several cancers. In HCC, USP10 accelerates hepatoma cells growth by directly interacting with YAP/TAZ to stabilize YAP/TAZ and remove its polyubiquitin chains. High USP10 levels predict a bad prognosis for HCC patients [88]. Therefore, USP10 represents a potential therapeutic target for HCC patients with long-term activation of YAP/TAZ. The development of small-molecule inhibitors targeting USP10 is critical for advancing tumor therapy.

In summary, LLPS targeted therapies provide promising insights into developing strategies for HCC treatment.

Phase separation-related NcRNA

As molecular biology and sequencing technologies have advanced, it has become clear that ncRNAs are crucial for many diseases, including cancer. ncRNAs encompass miRNAs, lncRNAs, circRNAs, snoRNAs, and others. These ncRNAs act as scaffolds for protein complexes that regulate signaling pathways and engage in signal transduction partnerships with proteins. Recent studies have revealed that ncRNAs play a regulatory role in phase-separation processes. For example, nuclear circASH2 inhibits HCC invasion and metastasis by mediating YBX1 phase separation and assembling a complex with hnRNP to target and regulate TPM4, thus changing the skeleton structure of tumor cells [27]. Another study found that a circRNA highly down-regulated in HCC tissues, circVAMP3, interacted with CAPRIN and G3BP1 to promote SGs formation by acting as a molecular scaffold that triggers CAPRIN1 LLPS, thereby inhibiting c-Myc translation and HCC progression [20]. Further interference experiments using circVAMP3 across various tumor cell lines from different tissues demonstrated its ability to suppress tumor cells growth, indicating that circVAMP3 plays a broad tumor-suppressive role [20]. Therefore, circVAMP3 may be an ideal target for the diagnosis and therapeutic of HCC and other cancers.

Furthermore, URB1-antisense RNA 1 (AS1), a lncRNA linked to ferroptosis, is significantly upregulated in sorafenib-resistant HCC samples. URB1-AS1 promotes ferritin phase separation, reducing free iron levels and inhibiting ferroptosis, thereby mediating sorafenib resistance in HCC. Notably, the study demonstrated that N-acetylgalactosamine (GalNAc)-siRNA targeting URB1-AS1 improved the therapeutic efficacy of sorafenib in HCC [89]. Additionally, a recent study identified ZNF32-AS2, another lncRNA associated with LLPS in HCC, promotes proliferation, migration, and sorafenib resistance in hepatoma cells [12]. ZNF32-AS2 may be a new potential biomarker in HCC, with potential utility in evaluating patient prognosis and predicting therapeutic efficacy in clinical settings.

The non-phase-separated mode of ncRNAs has been widely reported, but their phase-separated regulatory mode in HCC has not been extensively studied. It is believed that more ncRNAs will be discovered in the future, and in-depth investigation of the intrinsic mechanism of ncRNAs as novel LLPS regulatory elements in hepatocarcinogenesis will bring new ideas for disease therapy and drug development.

Phase separation-related immunotherapy

Sorafenib monotherapy has long been the standard of care for first-line treatment of advanced HCC; however, resistance to sorafenib remains a significant challenge. Understanding the molecular mechanisms underlying drug resistance in HCC and identifying more effective therapeutic targets are crucial for improving treatment outcomes. In recent years, anti-PD-1/PD-L1 therapy has received much attention among many anticancer immunotherapies for its anti-tumor effects by blocking T-cell depletion and enhancing immune surveillance [90]. A study revealed that KAT8-IRF1 forms biomolecular condensates with transcription-promoting function to upregulate PD-L1 expression in tumor cells upon IFN-γ stimulation. Notably, the team identified a peptide that blocks condensates formation, reduces the level of PD-L1 and enhances anti-tumor immunity [39]. Designing specific blocking peptides to target HCC-associated condensates is a promising strategy. Another study found that IFN-γ increases YAP LLPS in cancer cells following anti-PD-1 therapy, thereby promoting the transcription of immunosuppressive factors. Disrupting YAP-mediated LLPS inhibits cancer cells proliferation, strengthens immune responses, and enhances the sensitivity of cancer cells to anti-PD-1 therapy [91]. Similarly, dihydroartemisinin enhances anti-PD-1 efficacy by inhibiting YAP expression, thereby breaking the tumor immunosuppressive microenvironment [92]. These results imply that developing medicines that modulate YAP LLPS might be a hopeful approach for cancer treatment with aberrant condensate formation, and that combining LLPS-based therapies with immunotherapy may result in a synergistic effect and improved therapeutic efficacy.

Clinically, immunotherapy targeting PD-L1 has demonstrated promising therapeutic efficacy in some cancer patients. However, the phase separation process in HCC still faces poorly understood, presenting challenges such as unclear mechanism, unidentified targets, and low response rates. Therefore, systematic and comprehensive investigations into the regulatory mechanisms of PD-L1 can help to design novel immunotherapeutic strategies targeting phase separation processes.

Other

A recent study suggests that five LLPS-related genes, BMX, FYN, KPNA 2, PFKFB 4, and SPP 1, may be effective prognostic markers for HCC [93]. This study marks the first instance of constructing a prognostic profile for HCC patients based on LLPS-related genes. As research delves deeper into the mechanisms of LLPS, its prognostic relevance in HCC is expected to become increasingly evident. Technologies such as CRISPR-Cas9 and RNAi may provide effective therapeutic strategies for targeting key genes involved in condensates. Additionally, in another novel study, Sun Y et al. [94] demonstrated the development of artificial hydrogel scaffolds by LLPS between polyethylene glycol and dextran. The resulting macroporous hydrogel material structurally mimics the extracellular matrix of hepatocytes, providing an optimal microenvironment for hepatocyte attachment and ensuring high cell survival rates. These hydrogels offer promising applications in liver tissue engineering and research.

Challenges and strategies in LLPS-based targeted therapies

While LLPS-based targeted therapies present promising opportunities for the treatment of HCC, a significant challenge persists in enhancing drug specificity and selectivity to mitigate off-target effects. This challenge is particularly pronounced given that many proteins involved in phase separation also perform essential physiological functions. To address this issue, several promising strategies can be employed: Firstly, the selective partitioning and concentration of drugs within transcriptional condensates significantly influence their pharmacodynamics [95]. Consequently, incorporating considerations of condensate type, molecular properties, and cellular context into drug design or screening processes can facilitate precise targeting of disease-associated LLPS structures, thereby enhancing drug specificity and selectivity. Secondly, employing disease-relevant models and patient-derived cells can assist in identifying potential off-target effects and refining the therapeutic window. Furthermore, integrating condensate partitioning assays into early drug development pipelines can elucidate drug distribution patterns and potential side effects across different condensates. Third, the development of proteolysis-targeting chimeras (PROTACs) that selectively bind LLPS-related proteins and simultaneously recruit E3 ubiquitin ligases provides a precise approach to degrade pathological targets while sparing normal proteins. For instance, ARV-825 is a PROTAC that recruits BRD4 to the E3 ubiquitin ligase cereblon for its efficient degradation, leading to the downregulation of c-Myc expression [96]. BRD4 is a prominent target in anticancer drug development, and utilizing PROTACs to target BRD4 condensates introduces an innovative approach for cancer therapy and drug discovery [97]. Furthermore, the development of allosteric inhibitors that target the allosteric sites of LLPS-associated proteins offers a promising strategy to disrupt aberrant LLPS while maintaining normal physiological functions. As previously mentioned, SHP2 allosteric inhibitors can mitigate the LLPS of SHP2 mutants, thereby enhancing SHP2 phosphotyrosine phosphatase (PTP) activity [78]. Targeting allosteric sites in drug design presents a novel and promising challenge. For instance, garsorasib binds specifically to an allosteric pocket of KRAS G12C, enabling selective inhibition of the mutant protein without affecting wild-type KRAS signaling, and exhibits strong antitumor activity [98]. Similarly, the design of specific blocking peptides to target condensates represents an effective strategy. For instance, the designed peptide ReACp53 selectively targets mutant p53 to rescue p53 function in cancer cell lines, without affecting wild type p53 [99]. Another peptide, 2142-R8, as previously discussed, disrupts KAT8–IRF1 condensates, thereby downregulating PD-L1 expression and enhancing antitumor immunity [39]. Finally, the development of sophisticated drug delivery systems, such as liposomes and nanoparticles, shows significant potential [100]. Recent research has demonstrated that DNA and histones can form coacervate vesicles through LLPS, which exhibit superior kinetic stability and markedly improve drug delivery efficiency [101]. Collectively, these strategies underscore a multifaceted approach to enhancing the specificity and selectivity of LLPS-based therapies in HCC. In the future, combining mechanistic studies, advanced technologies and translational medicine modeling will be the key to fully exploiting the therapeutic potential of biomolecular condensates.

Fig. 3

Prospects for HCC treatment by regulating LLPS. A Metformin disrupts the Twist1-YY1-p300 condensates, thereby decreasing the expression of miR-9 and thus inhibiting HCC progression. B Elvitegravir specifically binds to the SRC-1/YAP/TEAD condensate and hence antagonize YAP oncogenic transcriptional activity. C 2142-R8 peptide disrupts KAT8–IRF1 condensates and consequently down-regulate PD-L1 expression and enhance antitumor immunity. D Development of hydrogel scaffolds mimics the structural organization of liver by phase separation between polyethylene glycol and dextran. E CircVAMP3 interacted with CAPRIN1 and G3BP1 inhibits the translation of c-Myc through the formation of SGs to negatively regulate HCC cells proliferation and migration. F CircASH2 regulates TPM4 to alter tumor cytoskeleton for inhibiting HCC invasion and metastasis by promoting YBX1 LLPS. G GalNAc-si URB1-AS1 increases sorafenib resistance in HCC by inhibiting ferritin phase separation

Discussion and future perspectives

As research advances, the functions of LLPS in cancer are gradually being revealed, and its important roles in transcriptional regulation, signaling, metabolic adaptation, stress response and immune escape demonstrate the great potential of this field. Understanding the molecular events underlying biomolecular condensates formation in HCC may offer novel insights for developing drugs that specifically target-phase-separated condensates. Drugs targeting LLPS can be developed through two primary strategies: first, if phase separation promotes cancer progression, as observed in Laforin-Mst1/2, YAP-TEAD, MAZ-CCND1-G4 and SHP2 condensates, inhibitors of LLPS could be designed to competitively bind to the phase separation sites, thereby disrupting key protein interactions or directly preventing condensate formation. Conversely, if the occurrence of phase separation is essential to the cancer suppression pathway, as in the cases of RIα, SPOP, and YBX1 above, phase separation can be maintained through the development of drugs that modulate droplet properties.

Despite advancements in the field, several challenges persist that require further attention. Primarily, the identification and validation of specific condensates as therapeutic targets are complicated by their dynamic characteristics and functional heterogeneity. While 1,6-hexanediol is frequently employed to disrupt LLPS, its effects lack specificity and often result in indiscriminate cytotoxicity [102]. Techniques such as CRISPR-Cas9 gene editing and RNA interference (RNAi) are also widely utilized for targeting proteins; however, they present limitations when applied to LLPS. Gene editing induces irreversible DNA modifications, thereby limiting the study of transient protein states. RNAi, on the other hand, requires relatively long onset time, rendering it unsuitable for observing the rapid and dynamic nature of LLPS [103]. Consequently, there is a need for more sophisticated, high-resolution methodologies to monitor LLPS behavior in living cells in real-time. Emerging technologies, including high-content imaging, magnetic resonance-based probes, X-ray crystallography, and organoid culture systems, hold potential for providing deeper insights into the formation, stability, and disease-associated dynamics of condensates. Concurrently, the integration of multi-omics data with artificial intelligence (AI) is being investigated to predict condensate formation and validate their involvement in disease. For instance, a recently proposed strategy combines AI-based therapeutic target discovery through the PandaOmics platform with phase separation propensity analysis using FuzDrop, enabling the prioritization of disease-relevant phase-separating proteins for therapeutic intervention [104]. These cutting-edge technologies are crucial for elucidating the roles and regulatory mechanisms of LLPS in HCC, as well as for identifying novel therapeutic targets. However, the selection of appropriate LLPS-related biomarkers in HCC remains a significant challenge. Although certain LLPS-associated genes or molecules have been suggested as potential biomarkers, their clinical applicability necessitates further validation, particularly concerning their correlation with patient survival and disease progression. Biomarkers should aid in stratifying patients who are most likely to benefit from LLPS-based therapies and in guiding rational combination strategies to overcome resistance mechanisms. Lastly, the design of precise therapeutic interventions targeting intrinsically disordered proteins and LLPS-driven condensates is challenging, partly due to the absence of well-defined binding pockets and the dynamic, multivalent nature of these structures. Current strategies aim to disrupt multivalent interactions within IDRs using novel modalities such as peptides, PROTACs, and allosteric inhibitors [105]. In addition, it has been shown that some antineoplastic drugs can exhibit selective partitioning into specific condensates, such as those formed by MED1 or BRD4 at super-enhancers [95]. Optimizing the physicochemical properties of small molecules can promote their preferential enrichment in disease-relevant condensates, thereby improving target selectivity and minimizing off-target effects. Incorporating condensate partitioning assays into early drug development and combining disease-relevant models with multi-omics data may further support the rational design of LLPS-based precision therapeutics.

In conclusion, LLPS-based therapies exhibit significant potential in HCC therapy and targeting LLPS-mediated oncoproteins and oncogenic signaling pathways represent promising avenues for the development of new therapies. Although the field is still in its early stages, with limited understanding of LLPS target identification, its mechanistic roles, and its relationship to disease progression, we anticipate the discovery of novel biomarkers and therapeutic targets in the future. Continued research in this area is likely to yield transformative breakthroughs in cancer diagnosis and treatment.