Biomolecular Condensates in Telomere Maintenance of ALT Cancer Cells

Abstract

Alternative Lengthening of Telomeres (ALT) pathway is a telomerase-independent mechanism that utilizes homology-directed repair (HDR) to sustain telomere length in specific cancers. Biomolecular condensates, such as ALT-associated promyelocytic leukemia nuclear bodies (APBs), have emerged as critical players in the ALT pathway, supporting telomere maintenance in ALT-positive cells. These condensates bring together DNA repair proteins, telomeric repeats, and other regulatory elements. By regulating replication stress and promoting DNA synthesis, ALT condensates create an environment conducive to HDR-based telomere extension. This review explores recent advancements in ALT, focusing on understanding the role of biomolecular condensates in ALT and how they impact telomere dynamics and stability.

1. Alternative lengthening of telomeres (ALT) pathway

1.1. Telomere elongation via ALT

Telomeres are the protective caps at the ends of chromosomes. Telomeres are non-coding G-rich (5’-TTAGGG-3’) DNA repeats associated with the Shelterin complex [1–4]. The Shelterin complex, composed of six proteins, TRF1, TRF2, POT1, RAP1, TIN2, and TPP1, plays a crucial role in preventing telomeres from being mistaken for DNA double-strand breaks (DSBs), thus preventing unwanted DNA repair [1, 5–7]. The primary function of the telomere is to maintain chromosomal stability and prevent chromosomal degradation [1, 4]. In normal somatic cell division, the telomere progressively shortens within each cell cycle due to the lagging strand of the replication fork being unable to replicate the ends of telomeres fully [1, 5, 8]. As a result, the accumulation of excessively shortened telomeres triggers cellular senescence, apoptosis, or permanent cell cycle arrest and further serves as an obstacle for tumorigenesis [2, 9, 10]. A key feature of cancer is its ability to sustain limitless cell division, a capability that heavily depends on telomere elongation [11, 12].

Most cancers achieve telomere elongation by activating telomerase. Telomerase is an RNA-dependent DNA polymerase responsible for synthesizing telomeric DNA sequences [1, 2, 5]. It comprises the catalytic enzyme telomerase reverse transcriptase (TERT) and a non-coding human telomerase RNA (hTR) component. By binding to the 3’-single-stranded telomeric DNA overhang, the telomerase can elongate this overhang by utilizing its reverse transcriptase function of TERT, with the telomerase RNA as a template [1, 5, 13]. Telomerase activity is silenced in the majority of normal human somatic cells but is detected in over 90% of cancerous cells [9, 14].

While most cancers maintain their telomeres through telomerase activation, a small group, around 10%, lacks telomerase and instead relies on the alternative lengthening of telomeres (ALT) pathway [15–17]. ALT was first discovered in the telomerase-deficient mutant of S. cerevisiae by Lundblad, Szostak, and Blackburn, who found that certain cells lacking the essential telomerase component EST1 were able to bypass cellular senescence [18, 19]. This mechanism was later identified and characterized in human cancer cells [20, 21]. The ALT phenotype is particularly enriched in specific cancer types, such as grade II and III astrocytoma, osteosarcomas, and pancreatic neuroendocrine tumors [22–24].

ALT uses HDR to extend telomere length following replication stress-induced DNA damage at telomeres [17, 25, 26]. Telomeres are hard-to-replicate regions due to the presence of G-quadruplexes formed by the G-rich telomeric repeats. In addition, some telomeres are transcribed into telomere repeats containing RNA (TERRA) by RNA Polymerase II [27–29]. TERRA can localize to other telomeres by hybridizing with telomere DNA to form DNA–RNA hybrid structures known as R-loops, which act as another source of replication stress at telomeres [28, 30, 31]. Replication stress at telomeres can interfere with DNA replication, leading to stalled or collapsed replication forks and the formation of one-ended DSBs, which subsequently activate the ALT pathway [25, 32–34] (Fig. 1).

Figure 1. Overview of the ALT pathway.

In ALT-positive cells, replication stress in the S phase, resulting from R-loops and G4 on telomeres, leads to replication fork stalling. This stalling often progresses to replication fork collapse, resulting in DNA damage that activates the ALT pathway for telomere maintenance. Telomeres are elongated within APBs via BIR mechanism during the G2 and early M phases, leading to highly heterogeneous telomere lengths, frequent telomere sister chromatid exchanges, and abundant extrachromosomal telomeric DNA repeats.

1.2. ALT hallmarks

Except for the lack of telomerase, ALT cancer cells do not have a universal genetic mutation, though a lot of them share loss in ATRX/DAXX [35–38]. Instead, ALT cells are defined by distinct hallmarks (Fig. 1). First, they have a unique nuclear structure called ALT-associated promyelocytic leukemia (PML) bodies (APBs) [16, 39]. APBs are nuclear condensates composed of the PML protein, clusters of multiple telomeres, and various proteins involved in DNA replication and repair [16, 25, 39–42]. APBs, detected as overlapping structures using telomere fluorescence in situ hybridization (FISH) combined with PML immunofluorescence (IF), have been established as diagnostic markers for ALT [43]. Additionally, research in liposarcoma has shown that the presence of APBs is associated with increased mortality, further underscoring the prognostic value of APBs [44].

Second, ALT telomere length is highly heterogeneous [20]. In human ALT cells, the telomere length ranges from extremely short (<1 kb) to unusually long (>20 kb) [45, 46]. Though recent sequencing results show that normal human cells also exhibit telomere length heterogeneity, the range (~2 to 10 kb) is much smaller than that reported in ALT cancer cells [47]. The origin of ALT telomere length is not fully understood yet, but it has been shown to depend on APB-mediated localization of the BTR complex to telomeres [48].

Third, there are elevated levels of telomere sister chromatid exchange (T-SCE) [49]. Sister-chromatid exchange is a process that occurs during DNA replication, where two sister chromatids undergo breakage and rejoining through homologous recombination (HR) resulting in the physical exchange of regions between the parental strands in the duplicated chromosomes [50]. In ALT cells, T-SCEs occur much more frequently than in telomerase-positive cells, indicating a reliance on recombination-based telomere maintenance [51]. Studies have indicated that T-SCEs alone are unlikely to be the primary telomere maintenance mechanism in ALT-positive cells [49, 52]. Unequal T-SCEs can cause some daughter cells to inherit shorter telomeres, resulting in telomere loss, while others acquire longer telomeres, enhancing their proliferative capacity [53]. The combined activity of T-SCEs and homologous recombination likely contributes to the highly heterogeneous telomere lengths characteristic of ALT cells [51, 52, 54].

Lastly, there are abundant extrachromosomal telomere DNA repeats (ECTRs) [55, 56]. ECTRs include telomeric circles (T-circles), which are the double-stranded circular DNAs composed of telomeric repeats, C-circles, which are the partially single-stranded C-rich telomeric DNA circles, and linear telomeric fragments and they are thought to arise from the recombination and replication events in the ALT pathway [56–60]. The level of ECTRs is often used as an indicator of ALT activity, but it remains unknown whether ECTRs are merely byproducts of ALT or they actively contribute to ALT telomere elongation.

1.3. DNA synthesis mechanisms in ALT

Although ALT uses HDR to maintain telomeres, they primarily rely on noncanonical HDR called break-induced replication (BIR) rather than the classical HR mediated by RAD51 recombinase. In BIR, 3′ single-stranded DNA overhangs invade a homologous sequence, initiating replication that copies the invaded sequence to the distal end of the chromosome in a POLD3/4-dependent manner [18, 61–63]. In budding yeast mutants lacking functional telomerase, two distinct ALT pathway types were discovered [25, 64, 65]. Type I involves RAD51- and RAD52-dependent amplification of repetitive subtelomeric sequences, while type II is a RAD52-dependent but Rad51-independent mechanism for maintaining telomeres by expanding telomeric repeats [62]. While both RAD51 and RAD52 have been shown to play crucial roles in human ALT cells, the process is thought to resemble a RAD51-independent mechanism, more akin to the type II pathway observed in yeast [65–67].

BIR activity in human ALT cells peaks during the G2 and early M phases [42, 57, 68], enabling conservative DNA replication at telomeres without relying on conventional S-phase replication. BIR in G2 phase was demonstrated by the ALT telomere DNA synthesis in APBs (ATSA) assay, where EdU labeling was used to detect nascent DNA synthesis in APBs in G2 arrested cells [67]. The ATSA phenotype is observed exclusively in ALT-positive cells but not in ALT-negative cells [67]. RAD52 has been identified as a key factor for ATSA, facilitating displacement-loop (D-loop) formation, which are structures formed by the invasion of ssDNA ends into double-stranded DNA, and likely aiding in the annealing of the 3’-single-stranded DNA (ssDNA) telomeric overhang with a homologous template [67, 69, 70]. RAD52 depletion significantly reduces ATSA, whereas RAD51 depletion has no effect on ATSA [67]. A RAD52-independent but RAD51AP1/TERRA-dependent ALT pathway has also been observed during G2 and is responsible for the formation of c-circles; this pathway, although RAD52-independent, nevertheless requires BLM and POLD3/4, suggesting it still undergoes the BIR mediation [67, 71, 72].

In early mitosis, ALT telomere DNA synthesis continues through mitotic DNA synthesis (MiDAS). While G2-phase telomere synthesis is unique to ALT-positive cells, MiDAS functions as a broader mechanism for addressing telomeric replication stress in both ALT-positive and ALT-negative cells, with notably greater activity in ALT-positive cells [68, 73, 74]. Similar to G2-phase BIR, MiDAS requires a PCNA–POLD3-based replisome for DNA synthesis [69]. However, unlike G2-phase BIR, ALT-associated MiDAS relies on SLX4 [69]. In addition, MiDAS in ALT appears to be primarily RAD52-dependent. In contrast, RAD51 depletion does not reduce MiDAS activity but results in fragile telomeres and telomere dysfunction-induced foci (TIF) [68].

Though MiDAS was previously considered not significantly involved in ALT telomere lengthening [75, 76], a recent study suggests otherwise. Using cell cycle synchronization and single-molecule telomere analysis (SMAT) on stretched DNA fibers [75, 77], this study revealed that G2 BIR is non-productive, coinciding with an increase in ECTRs. In contrast, productive synthesis occurs during the G2-to-M transition, facilitated by translesion DNA polymerase-mediated MiDAS [75].

1.4. ALT promoters

Since ALT is generated by replication-induced DNA damage, proteins that generate replication stress and enhance HDR are found to play an important role in promoting ALT activation.

The first one is the DNA helicase BLM. BLM is an important genome stabilizer for normal cell maintenance, regulating DNA replication, recombination, and both homologous and non-homologous double-strand break repair pathways [78]. BLM functions in conjunction with TOP3A and RMI1/2 to form the BLM-TOP3A-RMI1/2 (BTR) complex [15, 48, 79]. The BTR complex is essential for ALT activation, enabling telomere clustering, supporting the APB formation, MiDAS at telomeres, and the subsequent increase in C-circle levels [67, 80–83]. Anchoring the BTR complex to telomeres is sufficient to induce ALT activity [48]. More importantly, in ALT-negative cells, inducing the APB formation together with BLM expression can trigger ALT phenotypes, highlighting its indispensable role in the ALT pathway [81]. BLM plays multiple roles in ALT. It helps process recombination intermediates during strand invasion, which is crucial for initiating telomere synthesis through POLD3/4 [82]. It also helps dissolve Holliday junctions while preventing exchanges of telomeric sequences between sister chromatids and, therefore, preserving the original orientation of the telomeric DNA strands [80, 82]. A recent study has revealed that BLM uses its helicase function to specifically target the nick between Okazaki fragments on the lagging strand and generate 5’ C-rich telomeric flaps from strand displacement DNA synthesis by the polδ and PCNA replisome [84]. These 5’ C-rich flaps can trigger replication-associated DNA damage responses to initiate ALT.

TERRA levels are significantly upregulated in ALT-positive cancers [37, 85, 86]. In addition to increasing replication stress at telomeres, TERRA R-loops have been shown to contribute to the activation of the RAD52-independent ALT pathway [71]. These TERRA R-loops increase the presence of DNA G-quadruplexes (G4s) at telomeres, which promote the formation of telomeric D-loops, that are required for ALT BIR [71]. Further research has shown that RAD51AP1 directly binds to TERRA and mediates TERRA R-loop homeostasis through a chromatin-directed mechanism that suppresses TERRA accumulation. This regulation prevents transcription-replication collisions (TRCs) at ALT telomeres [72].

Orphan nuclear receptors (NRs), such as COUP-TF1 and COUP-TF2, bind more abundantly to telomeres in ALT cells via variant TCAGGG telomeric repeats compared to normal or telomerase-positive cells [87, 88]. Recent studies have shown that the tethering of orphan NRs to telomeres initiates the formation of APBs, dependent on the zinc finger protein ZNF827, and promotes active ALT telomere DNA synthesis. Additionally, NRs work synergistically with ATRX-DAXX deficiency to activate the ALT pathway [89].

In addition to Rad52 and Rad51AP1 mentioned above, another repair factor that is important for ALT is the DNA translocase RAD54 [90]. In ALT-positive cells, RAD54 is recruited to telomeres in response to DNA damage, where it enhances DNA synthesis by facilitating D-loop formation and extension. Depletion of RAD54 reduces telomeric DNA synthesis, lowers C-circle levels, and increases unresolved recombination intermediates.

1.5. ALT repressors

There are also repressors for the ALT pathways. Some repressors must be mutated for ALT to be activated, while others are in place to prevent excess telomere damage so ALT can be productive.

The ATRX/DAXX complex is the most well-known ALT repressor. ATRX/DAXX is enriched at telomeric repeats, where it regulates the deposition of the histone variant H3.3 to these regions of the genome [35, 38]. Genome sequencing has shown that mutations in the ATRX/DAXX complex and the histone variant H3.3 are frequently found in ALT-positive cancers [22, 24, 91], and ATRX acts as a cellular ALT repressor [29, 92, 93]. ATRX deficiency promotes ALT by disrupting the regulation of TERRA, leading to TERRA upregulation and R-loop induced replication stress [85, 94]. In supporting the suppressing role of ATRX, expressing ATRX in ALT cells reduces ALT phenotypes [95]. A recent study demonstrated that endogenous telomerase activity cannot compensate for telomere dysfunction associated with ATRX loss. As a result, cells must adapt the ALT pathway for telomere maintenance to achieve immortality, indicating that ATRX depletion alone is sufficient to drive ALT activation [96].

SLX4 acts as a suppressor of ALT by regulating the resolution of recombination intermediates at telomeres. In contrast to the BTR complex, which supports non-crossover dissolution and productive telomere extension, the SLX4 complex with SLX1 and ERCC4 facilitates both crossover and non-crossover resolutions that can terminate telomere extension prematurely [82]. The SLX4 interacting protein (SLX4IP) further promotes SLX4’s function by antagonizing BLM’s dissolution capacity [97]. Loss of SLX4 leads to increased ALT phenotypes, including APB formation, C-circle accumulation, and telomere extension, while overexpression of SLX4 reduces these markers [82]. However, disrupting both SLX4IP and SLX4 led to catastrophic levels of telomere instability, resulting from the unregulated dissolution of telomeric homologous recombination intermediates by BLM [97]. Hence, the regulatory balance between SLX4, SLX4IP, and BLM is crucial for maintaining ALT activity in an efficient and productive manner [76]. On the other hand, SLX4 is essential for ALT-associated MiDAS, as mentioned above.

Similar to SLX4, the ATP-dependent DNA-annealing helicase SMARCAL1 counteracts replication stress by promoting the reversal of stalled replication forks, thereby suppressing ALT [98–100]. Additionally, SMARCAL1, along with RPA2, Rad51, and SLX4, forms a machinery known as replication fork regression, which rescues collapsed replication forks [57].

Other important ALT repressors are FANCM and FANCD2. FANCM and FANCD2 are two key proteins in the Fanconi anemia (FA) pathway. The primary role of the FA pathway is to remove DNA interstrand crosslinks (ICLs) to maintain genome stability [101, 102]. It has also been shown to interact with various other repair processes, such as homologous recombination, nucleotide excision repair, and translesion synthesis, to address a diverse array of DNA lesions [102]. FANCD2 counteracts BLM-dependent telomere extension in ALT-positive cells by promoting the intramolecular resolution of stalled replication forks [103]. FANCM, an ATPase and DNA translocase, suppresses ALT activity in a manner akin to SMARCAL1 by remodeling stalled replication forks and promoting fork reversal to alleviate replication stress at telomeres [100, 104]. Additionally, FANCM regulates TERRA levels and suppresses R-loop formation, which helps minimize replication stress and prevent fork stalling [105]. FANCM also interacts with BLM in the BTR complex, influencing their branch migration activity, perhaps counteracts BLM’s function as an ALT promoter [105, 106]. Both loss of FANCD2 and FANCM in ALT-positive cells leads to ALT hyperactivity, evidenced by increased extrachromosomal telomeric DNA formation and enhanced telomeric DNA synthesis [103, 105, 106].

Lastly, BRCA2 plays a vital role in maintaining telomere stability by dynamically interacting with telomeric G-quadruplexes, facilitating the remodeling of stalled replication forks, and restarting telomere replication, particularly through RAD51 loading [107]. When BRCA2 is absent, this dynamic interaction is disrupted, resulting in replication stress that produces TERRA RNA, which forms R-loops at telomeres. The accumulated R-loops trigger telomere clustering, contributing to APB formation and ALT activation [108].

2. Biomolecular condensates in DNA repair

2.1. Biomolecular condensates

The discovery of the first membraneless compartment, the nucleolus, dates to the 1830s [109]. Since then, similar compartments have been identified in the nucleus, cytoplasm, and on membranes of nearly all eukaryotic cells [110, 111]. These structures, now termed biomolecular condensates to emphasize the concentrating of biomolecules within [110], are vital for various cellular functions ranging from signaling pathways, gene expression, and endocytosis to cell fate determination. Recent research has found that dysfunction in biomolecular condensates is linked to a wide range of pathological and physiological processes across various biological scales, making them promising targets for therapy [112–114].

2.2. Formation of biomolecular condensates.

Unlike traditional organelles enclosed by membranes, condensates form when their components, such as proteins and nucleic acids, condense into a phase with distinct chemical and physical properties. Many are proposed to form via liquid-liquid phase separation (LLPS) in which molecules separate from their surroundings to form a new liquid phase when biomolecule concentrations exceed the saturation concentration [115]. These liquid concentrates can coalesce and have a dynamic exchange of components within and with the surroundings. Liquid condensates may undergo an aging process and turn into condensates with more viscoelastic properties or even solids [116]. Others have been shown to form through percolation-coupled phase separation, resulting in condensates that are viscoelastic network fluids. These condensates exhibit sequence-, chemistry-, and structure-specific distributions of clusters, which can assemble even at concentrations significantly below the saturation threshold for phase separation [117].

Despite differences in detailed mechanisms, biomolecular condensate formation relies on multivalent interactions between its components [118] (Fig. 2A). These interactions include cation-anion, dipole-dipole, cation-π, π-π, and hydrophobic forces among proteins, DNAs, and RNAs [119]. Intrinsically disordered regions (IDRs), which are flexible protein segments lacking a defined structure, are found to be overrepresented in condensates due to their ability to enable dynamic multivalent interactions [120, 121]. Protein modular domains, which are repetitive regions within proteins, also promote condensate formation by enabling multiple binding events [118, 122–124].

Figure 2. Biomolecular condensates formation and function.

A. Biomolecular condensate formation occurs through phase separation, driven by multivalent interactions between its components. Condensate assembly, dynamics, and stability are regulated by multiple factors, including salt concentration, pH levels, ATP levels, thermodynamic conditions, and post-translational modifications. B-D. Biomolecular condensates have various functions. B Biomolecular condensates influence biochemical reactions. (a) Concentration: Enzymes and substrates are co-localized within the same condensate, enhancing their proximity and significantly accelerating the reaction rate. (b) Selectively and Exclusion: Non-specific regulators or inhibitors are excluded from the reaction environment, optimizing conditions for greater reaction efficiency. (c) Sequestration: Key molecules are sequestered away from active sites or substrates, providing precise control over the reaction rate. C. The condensation process or the coalescence of liquid condensates can generate forces to organize cellular structures, such as facilitating DNA clustering. D. Phase separation can be used to sense cellular environmental changes and respond to cell stress by forming condensates to facilitate specific reactions.

These multivalent interactions can be regulated by post-translational modifications (PTMs), such as phosphorylation, to control condensate assembly, disassembly, and dynamics. For example, DYRK3 kinase functions as a “dissolvase” that prevents phase-separated organelles, such as stress granules and splicing speckles, from condensing in the mitotic cytoplasm. DYRK3 activity dissolves these organelles through phosphorylation, enabling their reformation at appropriate stages in the cell cycle [125]. Likewise, mitotic phosphorylation of Ki-67 increases charge blockiness, a parameter that quantifies the number of charge clusters (charge blocks) along a polypeptide chain, thereby promoting phase separation and the formation of the chromosome periphery, while mitotic phosphorylation of NPM1 reduces charge blockiness, and phase separation, resulting in nucleolar dissolution [126]. Notably, protein phosphorylation regulates the phase separation of IDRs by modulating charge blockiness rather than targeting specific sites.

Environmental factors such as salt concentration, pH level, ATP level, and temperature, play a role in condensate formation [119, 127–129] (Fig. 2A). For example, proteins can undergo two types of temperature-dependent phase transitions: lower critical solution temperature (LCST) and upper critical solution temperature (UCST) [130]. In LCST phase transitions, phase separation occurs as temperature rises. LCST phase behavior is typically seen in sequences rich in hydrophobic amino acids, like elastin-like polypeptides, which demonstrate this “inverse” temperature dependence: they remain mixed at lower temperatures and phase-separate when heated. In UCST phase transitions, proteins remain mixed at higher temperatures and phase-separate upon cooling. UCST behavior is often observed in polar sequences enriched with residues like serine, asparagine, and glutamine, where enthalpic interactions drive the separation at lower temperatures. P granules in Caenorhabditis elegans have been shown to dissolve with increasing temperature and recondense when cooled [131]. This temperature response can be captured in an in vivo phase diagram, suggesting that, despite the cell’s active processes, P granule formation and dissolution are governed by local thermal equilibrium, highlighting a thermodynamic basis for regulating cellular condensates.

2.3. Functions of biomolecular condensates

Biomolecular condensates are involved in functions at the molecular, cellular, and tissue scales. At the molecular level, biomolecular condensates are implicated in modulating processes like transcription, RNA processing and translation [132, 133]. At the cellular level, condensates are closely involved in processes such as molecular transport and signal transduction [134–136]. At the tissue level, condensates play a role in the immune response, participating in both innate and adaptive immune pathways [137, 138]. They are involved in key signaling cascades, such as the cGAS-STING pathway [138–140], T cell receptor (TCR) and B cell receptor (BCR) signaling pathways [141–143].

Mechanistically, biomolecular condensates perform their functions in several ways (Fig. 2B). First, by bringing molecules into close proximity, condensates can accelerate reaction rates by concentrating enzymes and substrates [144], thereby enhancing reactions like RNA splicing and RNA degradation [145, 146]. Beyond recruitment, condensates can also enhance activity by selectively excluding negative regulators, thereby removing inhibitory influences [141, 144]. Consequently, biomolecular condensates can efficiently localize to specific regions within the cell, controlling the subcellular distribution of specific molecules and allowing them to influence various cellular processes through positional effects [112, 144]. Conversely, they can reduce reaction activity by sequestering key molecules away from their active sites or substrates, effectively controlling the availability of these components and modulating cellular processes [144, 147, 148].

In addition, the phase separation process or the coalescence of condensates can also generate forces to organize cellular structures (Fig. 2C). Chromatin, for example, can undergo phase separation through histone tail interactions to form dense liquid-like droplets, regulated by factors like linker histone H1, nucleosome spacing, and histone acetylation [149]. Histone phase separation provides a physical basis for chromatin organization, allowing the genome to be compact yet accessible for cellular processes. Moreover, heterochromatin protein 1 alpha (HP1α), known for its role in gene silencing, can undergo phase separation, particularly when phosphorylated at its N-terminal extension [150]. This phase separation appears vital for creating compacted chromatin regions, enabling gene silencing by sequestering chromatin within phase-separated droplets. CasDrop, a CRISPR-Cas9-based optogenetic technology to induce condensate at targeted genomic loci, reveals that condensates can mechanically pull together targeted genomic loci through surface tension-driven coalescence while excluding non-targeted regions, effectively acting as mechano-active filters within the genome [151]. Lastly, biomolecular condensates also contribute to the organization of membrane-bound organelles, such as the Golgi apparatus, where the Golgi matrix protein GM130 has been observed to form liquid-like droplets in cells, suggesting it may contribute to Golgi organization through phase separation, contributing to the structural integrity and flexibility of the Golgi apparatus [152].

Phase separation can also act as a sensory mechanism to detect and respond to environmental changes, particularly temperature shifts [153] (Fig. 2D). For instance, the poly(A)-binding protein (Pab1) in yeast undergoes phase separation in response to mild temperature increases, specifically around the organism’s heat-shock temperature [154]. This phase separation is highly sensitive, responding to only a few degrees of temperature change, allowing the cell to react quickly to thermal stress [154]. Pab1’s sensitivity to temperature is further enhanced by its response to pH changes, as heat stress often coincides with a drop in pH [155]. This dual sensitivity allows Pab1 to act as an effective sensor for thermal stress and may help the cell adapt rapidly by forming stress granules during stressful conditions [153, 154].

2.4. Biomolecular condensate in DNA damage repair.

Biomolecular condensates are implicated in efficient DNA damage repair, particularly in response to DSBs, which are among the most severe forms of DNA damage [156–159]. 53BP1 is the first DNA repair protein identified to undergo phase separation [158]. 53BP1 phase separation is promoted by 53BP1’s oligomerization domain and DNA damage-induced transcription of long noncoding RNAs (ncRNAs) [160–162]. 53BP1 is a scaffold protein engaging with factors including the DSB sensor MRE11/RAD50/NBS1 (MRN) complex [163], DNA repair protein BRCA1 and RIF1 [164], and the damage response transducers ATM and ATR to favor non-homologous end joining (NHEJ) pathway over the HR [162, 165, 166]. The formation of 53BP1 condensates helps concentrate these factors to reinforce the DNA repair pathway choice. Notably, the tumor suppressor and cell cycle arrest regulator p53 are co-assembled within the 53BP1 condensate [161], which plays an important role in its activation [167, 168]. Therefore, 53BP1 condensates not only concentrate repair factors to DNA damage site but also selectively recruit specific client proteins (Fig. 2B).

Poly(ADP-ribose) polymerase 1 (PARP1) is one of the earliest responders to DNA damage and plays a critical role in recruiting DNA repair proteins through its poly(ADP-ribosyl)ation (PARylation) activity. PARP1 forms co-condensates with DNA depending on its three zinc finger domains to hold broken DNA ends together and make it enzymatically active for PAR synthesis [169, 170]. PARP1 will be released from DNA through PARylation, which recruits stabilizing proteins like FUS to initiate DNA repair. This orchestration ensures that repair factors are precisely localized and activated for a rapid response [171]. PARP1 condensates significantly promote DNA single-strand break ligation following PARylation [169]. Therefore, PARP1 condensates can facilitate DNA repair by generating forces to bring DNA ends together and concentrating repair factors (Fig. 2B and C).

In addition to facilitating repair, condensates play a critical role in regulating DNA damage response through modulating post-translational modifications (PTMs). A notable example is SLX4-driven condensates that concentrate factors in the SUMO-RNF4 pathway to promote protein SUMOylation and ubiquitination and enable chromatin extraction and targeted protein modifications [172]. In addition, TopBP1 has been reported to self-assemble into micrometer-sized condensates, activating a key component of the DNA damage response, the checkpoint kinase ATR [173]. Notably, the condensation of TopBP1 is directly linked to the phosphorylation and activation of the effector kinase CHK1 [173, 174]. Together, these condensates orchestrate various PTMs in DNA repair by creating different chemical environments (Fig. 2B).

3. Bimolecular Condensates at ALT telomeres

ALT is based on HDR but uses BIR, which differs from classic HR in many ways. Consequentially, condensates involved in the ALT pathway share some similarities with condensates in normal DNA repair, but many unique condensates are also present in the ALT pathway. Below, we summarize ALT-related condensates and discuss their relation to normal repair condensates.

3.1. ALT-associated PML bodies (APBs)

PML bodies localize to telomeres to form APBs only in ALT cells, though PML bodies are found to be associated with other genomic loci occasionally or dynamically in non-ALT cells [175]. Given the essential role of APBs in ALT, many studies have focused on understanding how APBs are formed [176]. Early work shows that many factors contribute to APB formation, leading to the hypothesis that APBs form via multiple pathways [177]. A key step among the many pathways is SUMOylation, the post-translational modification process of covalently adding the small ubiquitin-like modifiers (SUMOs) to target proteins [178]. First, the SUMOylation of telomeric proteins is required for APB formation [25, 39, 40, 81, 179]. Second, SUMOylation of PML is essential for APB formation, similar to the formation of normal PML bodies [180] . Lastly, the recruitment of repair proteins such as RAD51AP1 and BLM to ALT telomeres is mediated by SUMOylation [72, 181].

Recent work has linked SUMOylation-mediated phase separation at telomeres to APB formation [40] (Fig. 3). This was demonstrated by the induction of APBs by controlling SUMO phase separation at telomeres with a chemical dimerization system [40, 182]. Targeting SUMO or SIM (the SUMO interaction motif), a short stretch of hydrophobic amino acids with an acidic region that none-covalently binds to SUMO [183, 184], to telomeres has resulted in the formation of distinct, bright, and round foci for both SUMO and telomere, with a high degree of colocalization between SUMO and telomere foci. These foci also exhibited fusion behaviors characteristic of phase-separated condensates. The SUMO-mediated phase separation results in APB formation and telomere clustering. The phase separation is mediated by SUMO-SIM interactions as targeting SUMO and SIM mutants that lack the ability to interact with each other failed to induce phase separation and APB formation. Supporting the phase separation model, ALT operates as a self-perpetuating process within APBs, where replication stress triggers a SUMO-dependent loop to recruit DNA damage response proteins for telomere extension [181]. In addition, it has been shown that ALT telomere-localized synthetic polySUMO/polySIM condensates, which mimic APBs, along with co-expression of BLM, can activate ALT-like phenotypes and lead to MiDAS [81].

Figure 3. Condensates at ALT telomeres.

ALT relies on several condensates, including APBs, SUMO condensates, TERRA-LSD1 condensates, and RAD52-RPA condensates, to facilitate telomere elongation through BIR. These condensates might function at different stages of the ALT process and likely cooperate to achieve optimal telomere maintenance. The model is predicted as, TERRA-LSD1 condensates promote R-loop formation, triggering replication stress that initiates ALT. The collapsed replication fork is resected by BLM, generating ssDNA overhangs that are coated by RPA. On these telomeric overhangs, RPA-RAD52 co-condensates facilitate SUMO enrichment at telomeres, which promotes SUMO-mediated phase separation between DNA repair factors and PML via SUMO-SIM interactions. These SUMO condensates, with or without PML, can cluster telomeres and concentrate repair factors to facilitate BIR.

Interestingly, targeting SIM to telomeres has led to APB formation and telomere clustering but no telomere DNA synthesis, while targeting SUMO has induced all three phenotypes [40, 182]. The differences were attributed to the enrichment of DNA repair factors after targeting SUMO but not SIM, indicating that APB’s role in promoting telomere DNA synthesis depends on its chemical composition. However, APB’s role in promoting telomere clustering relies solely on the liquid-like properties [40]. This was shown by creating non-APB condensates at ALT telomeres with disordered proteins, which can cluster telomeres fine but fail to induce telomere DNA synthesis.

What remains unsolved for the APB phase separation model is why PML bodies only uniquely localize to ALT telomeres but not to other DNA damage sites where protein SUMOylation also occurs [185]. One possible reason is that telomere-binding proteins can serve as substrates for DNA damage-induced SUMOylation to increase the SUMO level to enable PML body nucleation at ALT telomeres.

3.2. SUMO condensates

In normal DNA damage events, SUMO has been shown to facilitate physical interactions between DNA repair proteins, acting like a “molecular glue” that stabilizes repair protein complexes at the damage site [186]. This begs the question of whether SUMO also acts as a glue to promote DNA repair factor interactions at ALT telomeres without APBs. In an ALT cancer cell line depleted of PML, thus lacking APBs, we show that targeting SUMO to telomere also leads to signatures of phase separation, telomere clustering, and telomere DNA synthesis, just like in APB-containing control cells [182] (Fig. 3). SUMO targeting also concentrates proteins such as RAD52, RAD51AP1, RPA, and BLM at telomeres without PML. This SUMOylation-dependent concentration of DNA repair factors is driven by SUMO-SIM interactions, which enable their co-phase separation at telomeres. Additionally, RAD52 itself can phase-separate, agreeing with what has been demonstrated in budding yeast [182, 187]. More importantly, RAD52 condensates can enrich BLM to facilitate ALT in a SUMO-dependent manner, with or without PML. This work demonstrates that fundamentally, it is SUMO-mediated phase separation that organizes repair factors to sustain ALT activity. The importance of APBs lies in the PML body’s ability to enrich SUMO on telomeres.

As discussed above, SLX4 forms nuclear condensates that compartmentalize the SUMO/RNF4 pathway and activate SUMO- and ubiquitin-dependent processes in DNA repair [172]. While SLX4 primarily acts as a suppressor of ALT activity by opposing BLM’s function [82], it has also been shown that SLX4 is required for ALT-associated MiDAS [69]. How SLX4 functions in relation to the SUMO condensates/APBs in the ALT pathway remains to be investigated.

3.3. Shelterin complex

The Shelterin complex, particularly its components TRF1 and TRF2, exhibits phase separation characteristics, forming condensates with telomeric DNA repeats [188, 189]. While TRF1 and TRF2-driven condensates contribute to telomere stability, it remains unclear whether these condensates are, to some extent, linked to ALT function. Live telomere imaging shows that without inducing protein phase separation, ALT telomeres are not able to cluster with themselves, and as long as the liquid-like condensates are formed, ALT telomeres can be clustered [40]. This suggests that either TRF1 and TRF2 do not form liquid condensates or that their ability to coalesce with each other is inhibited at ALT telomeres. However, as mentioned above, SUMOylation of shelterin proteins, mediated by ligase MMS21, is required for APB formation [190]. The ability of TRF1 and TRF2 to undergo phase separation may help increase their ability to enhance SUMO level at ALT telomeres for APB formation.

3.4. TERRA condensates

In addition to form R-loops, TERRA interacts with various proteins, including DNA repair factors like FUS and BRCA1, helping to facilitate telomere DNA replication and protect stressed or shortened telomeres [191–194]. TERRA’s interaction with proteins also plays a role in ALT. TERRA interacting protein endonuclease XPF is recruited by telomeric R-loops for DNA damage response, thereby triggering BIR for telomere replication and driving ALT activity [195]. Recently, we found lysine-specific demethylase 1A (LSD1), is recruited to ALT telomeres in a TERRA-dependent manner by interacting with TERRA’s G4s, which drives phase separation and leads to the formation of TERRA-LSD1 condensates [196]. These condensates specifically promote ALT activity by facilitating R-loop formation at telomeres. They also enrich R-loop-promoting proteins, such as RAD51AP1, to enhance homology-directed DNA synthesis and telomere elongation. As a result, TERRA-LSD1 condensates contribute to the replication stress and facilitate the recruitment of essential repair factors (Fig. 3).

3.5. RPA condensates

Replication protein A (RPA) stabilizes transiently formed ssDNA, protecting it from nucleolytic degradation and interacting with various DNA repair factors to provide a physical platform for the DNA damage response [197, 198]. ATR activation on RPA-coated ssDNA triggers cell cycle checkpoints, stabilizes stalled replication forks, facilitates DNA repair, and generally primes the cell for optimal genome integrity maintenance [199]. In ALT cells, RPA has also been shown to coat ssDNA generated by BLM-DNA2 mediated long-range resection in ALT activation [81, 200]. RPA protects telomeric ssDNA not only during S/G2 phases but also at post-MiDAS sites in the G1 phase in ALT-positive cells [200].

Although RPA has extremely high affinity for ssDNA, it can undergo phase separation into liquid droplets through incorporating ssDNA into dynamic condensates [201], where an abundance of free RPA facilitates rapid exchange of RPA on ssDNA [202]. It is worth noting that the phase separation effect of RPA is triggered by sub-stoichiometric amounts of ssDNA rather than by RPA itself or double-stranded DNA [201]. This dynamic condensation process allows RPA to form a concentrated reservoir around ssDNA, facilitating rapid exchange with free RPA and handover to downstream effectors [201]. Consequently, it maintains a balance between ssDNA stabilization and support for subsequent repair activities. As shown by a phase-separation-impaired but ssDNA-binding-proficient mutant of RPA increases markers of replication stress-induced telomere fragility, however, displaying altered ALT activity characterized by reduced telomere clustering, impaired RAD52 recruitment, and eventually heightened telomere loss [201] (Fig. 3).

4. Concluding marks

Biomolecular condensates play important roles in organizing cellular processes, and their involvement in the ALT pathway highlights their significance in cancer biology. ALT relies on PML, SUMO, TERRA and DNA repair factor mediated condensates to facilitate telomere elongation through DNA damage response and HDR. Although these condensates play distinct roles in ALT activation, how they collaborate to support different aspects of ALT telomere elongation remains to be investigated. It is possible these condensates represent temporal changes of the repair condensates at ALT telomeres. Supporting this, TERRA, RPA, SUMO and various DNA repair factors are all detected to co-localized with APBs. TERRA-LSD1 promotes R-loop formation, inducing replication stress that activates ALT. RPA condensates protect ssDNA overhangs generated during ALT BIR, transferring them to downstream effectors like RAD52, which help form SUMO condensates and APBs. APBs then cluster telomeres and concentrate repair factors, facilitating homologous recombination at telomeric DNA (Fig. 3).

Other condensates found in normal DNA repair might play a role in ALT as well. 53BP1 localizes to ALT telomeres in response to replication or nuclease-induced DNA damage [203, 204]. 53BP1 condensates might help cluster damaged telomeric DNA ends within APBs and collaborate with p53 to support ALT cell viability. PARylation has been shown to regulate ALT via the chromatin-assembly factor HIRA [205]. PARP1 condensates might facilitate this function by concentrating necessary proteins to ensure a chromatin environment that is adaptable to ALT. How these repair condensates collaborate with existing ALT-specific condensates in ALT telomere maintenance remains to be investigated.

In summary, the unique formation, dynamics, and interaction of various condensates help ALT-positive cancer cells to bypass telomerase and maintain chromosomal stability, thereby sustaining uncontrolled proliferation. Future research to unravel the molecular dynamics and regulatory mechanisms of ALT condensates, including how modifications in condensate-associated proteins and phase separation processes affect ALT cell survival would be important. Given the potential to selectively target ALT-positive cancers, developing drugs that disrupt biomolecular condensate formation or function could offer a highly specific and effective treatment approach. Targeting these phase-separated condensates may enable precision therapies that exploit ALT-dependent tumors’ unique vulnerabilities, ultimately expanding the arsenal of anti-cancer treatments.