Kinetic Control of Condensate Function: How the Formation Dynamics of APBs Influence ALT Cancer Telomere Length Heterogeneity

ABSTRACT

Telomerase‐free cancer cells use a recombination‐based alternative lengthening of telomeres (ALT) pathway to maintain telomere length. ALT cancer cells exhibit extreme heterogeneity in telomere length, which requires the formation of a specialized nuclear condensate, the ALT telomere‐associated PML bodies (APBs). Recent findings on APB assembly reveal that APBs form through two mechanistically and kinetically distinct pathways. These pathways modulate the strength of ALT activity. In this essay, I discuss how APB formation dynamics may contribute to telomere‐length variability in ALT cells and consider the broader implications of these findings for other chromatin‐associated nuclear condensates.

The dynamics of PML body formation at telomeres in cancer cells using the alternative lengthening of telomeres (ALT) pathway for telomere elongation correlate with different degrees of ALT activity and might contribute to telomere length heterogeneity in ALT cancer cells.

1. ALT Cancer and Telomere‐Length Heterogeneity

Continued proliferation is a hallmark of cancer cells [1]. To sustain this proliferation, cancer cells actively maintain telomere length to counteract shortening during cell division and the resulting replicative senescence. Most cancers reactivate telomerase, a reverse transcriptase silenced in somatic cells, to extend telomeres. A subset of tumors instead employs a recombination‐based mechanism known as alternative lengthening of telomeres (ALT). A defining feature of ALT‐positive cells is extreme variability in telomere length, both between chromosome ends within the same cell and across clonal populations, with individual telomeres ranging from very short (<2–3 kb) to exceptionally long (>30–50 kb) [2]. This heterogeneity contrasts sharply with telomere lengths maintained by telomerase, which are typically constrained to a narrower range of approximately 5–15 kb.

ALT originates from replication‐stress‐induced DNA damage at telomeres, followed by a homology‐directed DNA repair mechanism resembling break‐induced replication (BIR) [3]. BIR uses a simplified replisome dominated by polymerase δ and a dynamic displacement loop (D‐loop) that migrates along the donor template for conservative DNA synthesis [4]. This process is accompanied by frequent template switching and extensive DNA engagement, giving it a stochastic recombination nature. In addition, recombination templates can range from other telomeres to extrachromosomal telomeric repeats (ECTRs), including C‐circles and linear telomeric DNA of various lengths. Unequal use of these templates contributes to highly variable elongation tract lengths [5]. Importantly, not all newly synthesized telomere DNA is incorporated into telomeres; some is excised as ECTRs [6]. While details of the process remain elusive, this would further contribute to differences in the resulting telomere length. In addition, unlike telomerase, which preferentially extends shorter telomeres and thus homogenizes telomere length, ALT has not been shown to have such a preference [7]. Furthermore, the excess replication stress at ALT telomeres might lead to rapid telomere deletions [8], generating extremely short telomeres. Together, these features could explain why ALT telomeres are inherently more heterogeneous in length.

However, this may not be the whole story. ALT recombination is organized by the unique nuclear condensate known as the ALT‐associated PML nuclear bodies (APBs) [9]. The promyelocytic leukemia (PML) nuclear body is scaffolded by the PML protein and contains many other proteins [10]. PML bodies are present in normal cells, mainly in the nucleoplasm. Only in ALT cells is a subset of PML bodies localized to the telomere, forming APBs. APBs thus have many unique components, including telomeric DNA, telomere RNA (TERRA), shelterin components, and, more importantly, DNA repair factors. Concentrating recombination machinery near telomeres has been shown to be a major function of APBs in ALT [9]. As condensates with liquid‐like properties, APBs can coalesce, clustering telomeres in proximity [11]. Furthermore, APBs promote the generation of ECTRs and help retain them near the telomere [12]. ALT cells with the PML protein deleted, and thus lacking APBs, largely reduce telomere‐length heterogeneity [12]. More importantly, there is a baseline level of telomere recombination in these cells [13], demonstrating that APBs are not required for telomere recombination and highlighting their contribution to recombination efficiency and, thus, to the variance in telomere extension in ALT cells.

Recent studies of APB assembly suggest that APB formation dynamics are also important regulators of ALT activity (Figure 1). Below, I will summarize the current understanding of APB assembly, propose how APB formation dynamics contribute to ALT telomere‐length heterogeneity, and discuss the implications for the function of other chromatin‐associated condensates.

FIGURE 1.

Model for APB formation dynamics and its impact on ALT telomere‐length heterogeneity. (A) ALT telomeres experiencing replication stress and DNA damage accumulate SUMO through group SUMOylation of proteins at telomeres, including the shelterin complex, DNA repair factors, and chromatin modifiers. (B) Fluctuations in telomere length, TERRA levels, or chromatin state might cause some telomeres to accumulate more SUMO, thereby nucleating APBs by recruiting PML proteins to these telomeres. APB nucleation on some telomeres then depletes SUMO‐related proteins from the surrounding nucleoplasm, preventing APB nucleation on other telomeres. (C) SUMO‐coated telomeres can fuse with pre‐existing PML nuclear bodies (PML‐NBs) to form APBs. APBs may also fuse with SUMO‐coated telomeres or other APBs, clustering telomeres within. Telomeres that are closer to mobile PML‐NBs or nucleated APBs and positioned favorably within the 3D genome organization to have higher telomere mobility and collision rates with PML‐NBs or nucleated APBs will be preferentially clustered within APBs, while others cannot. (D) Differential kinetics of nucleation and fusion generate APBs of variable sizes and telomere clusters that differ in their ability to recruit break‐induced replication (BIR) factors, generate extrachromosomal telomeric repeats (ECTRs), and extend telomeres. These telomeres, with APB nucleation‐prone features amplified in this process, are then biased toward APB formation in the next cell cycle. Over successive cell cycles, these dynamics give rise to the extreme telomere‐length heterogeneity observed in ALT‐positive cancer cells.

2. APB Formation Dynamics Affect ALT Activities

Like PML nuclear bodies, APBs can be explained by phase separation driven by the SUMO–SIM interaction network [11]. SUMO, the small ubiquitin‐like modifier, can be covalently attached to proteins as monomers for SUMO1 or as polymers for SUMO2 and SUMO3 [14]. SUMO interacts non‐covalently with SUMO‐interaction motifs (SIMs) present in many proteins. Multivalent SUMO‐SIM interactions contribute to condensate formation by forming a dynamic SUMO‐SIM network [15]. Many APB components, including the PML protein itself, the shelterin complex, and the recombination factors, are either SUMOylated, contain SIMs, or both [11]. SUMO‐SIM interactions among APB components are required for APB formation. Even in the absence of the PML protein, SUMO‐SIM interactions among DNA repair proteins can form condensates that mimic APB functions, establishing that APBs are, in essence, SUMO‐rich condensates [13]. Notably, although SUMO‐SIM interactions are essential, they are not the only contributors to the interaction valence that drives APB formation. Many APB components, including PML, RPA, TRF1, TRF2, Rad52, and Rad51AP1, have phase‐separation ability that depends on their oligomerization domains, intrinsically disordered regions (IDRs), and/or nucleic acid‐binding domains [16]. These interactions may contribute as much as SUMO‐SIM interactions to the overall interaction valence in APBs. Nevertheless, SUMO‐SIM interactions are needed to promote heterotypic interactions among these proteins and bring them together in the same condensate [13].

The SUMO‐SIM interaction network in APBs originates from DNA damage‐induced SUMOylation at ALT telomeres [17]. Due to the promiscuous nature of SUMOylation [18], in which SUMOylation enzymes modify proteins in proximity non‐discriminately, the DNA damage‐recruited SUMOylation enzymes modify a variety of proteins at ALT telomeres, including shelterin components such as TRF1 and TRF2, as well as DNA damage proteins such as Rad51AP1 and BLM. This group SUMOylation leads to SUMO enrichment at the telomere, which is both necessary and sufficient for APB formation [11, 13, 19]. The necessity is demonstrated by inhibiting SUMOylation through knockdown or chemical inhibition of SUMOylation substrates or enzymes, and the sufficiency is demonstrated by tethering SUMO or SUMO enzymes to telomeres in the absence of DNA damage. These findings reveal that APB formation is telomere‐instructed and occurs through SUMO enrichment at damaged telomeres.

Live imaging shows that SUMO enrichment drives APB assembly through two distinct pathways: telomere‐driven nucleation of PML bodies and telomere fusion with preexisting PML nuclear bodies [19]. In the nucleation pathway, SUMO‐coated telomeres recruit PML proteins via their SIMs to assemble APBs de novo at telomeres. In the fusion pathway, SUMO‐coated telomeres approach existing PML nuclear bodies, which then coalesce to form APBs. Both nucleation‐ and fusion‐mediated APB formation can occur in the same cell, with nucleation occurring faster and fusion more slowly. Both nucleation‐ and fusion‐mediated APB formation require the SUMO–SIM interaction. Nucleation is strongly associated with high SUMO levels at telomeres, whereas fusion occurs at lower telomeric SUMO levels that are insufficient to nucleate APBs. Once formed, APBs can grow by fusing with SUMO‐coated telomeres or other APBs, a process termed APB growth by fusion, in distinction from APB formation by fusion. Enrichment of SIM at telomeres cannot nucleate APBs, but it can still fuse with PML bodies or other APBs. Telomeric SUMO levels thus act as a molecular switch controlling APB biogenesis.

Importantly, these distinct APB assembly pathways correlate with the strength of ALT activity [19]. Nucleated APBs preferentially recruit the DNA helicase BLM, a critical factor for initiating replication‐associated DNA damage as well as resolving repair DNA intermediates at ALT telomeres [20, 21]. These APBs are associated with elevated ALT activities, including increased telomeric DNA synthesis and C‐circle generation. APBs formed via fusion recruit less BLM and show reduced telomeric DNA synthesis and C‐circle generation compared with nucleated APBs, indicating that APB fusion is linked to lower ALT activity. These findings suggest that APB biogenesis can lead to differences in ALT activity and help control ALT telomere length distribution. However, since not all recombination events lead to productive telomere lengthening in ALT, future work linking APB formation dynamics to telomere length beyond the ALT activities mentioned above would be needed. Nevertheless, these results suggest that factors affecting APB formation via either nucleation or fusion could contribute to ALT telomere elongation and telomere length heterogeneity.

3. APB Nucleation as a Gatekeeper of ALT Engagement

Condensate nucleation follows the “winner‐takes‐all” phenomenon, in which nucleation of the first wave of condensates effectively halts further nucleation by acting as molecular sinks that deplete molecules in the surrounding environment [22, 23]. As a result, high nucleation rates yield many small condensates, whereas low nucleation rates generate fewer, larger condensates. Only a subset of telomeres has APBs in each cell, even during the G2 phase of the cell cycle when ALT activity peaks [9], indicating low APB nucleation rates. These low nucleation rates are unlikely to be caused by limited telomere DNA damage, as introducing DNA damage with a nuclease across all telomeres still yields only a subset of telomeres with APBs. Rather, it might be an intrinsic feature of the system, such as the limited availability of DNA damage response machinery, including SUMOylation enzymes, which damaged telomeres compete for. A subset of telomeres that can trigger a more efficient DNA damage response and faster SUMO accumulation after DNA damage can nucleate APBs more quickly. These faster‐nucleating APBs prevent APB nucleation at other telomeres, resulting in APB‐bearing telomeres being effectively elongated while others are not.

Why would some telomeres nucleate APBs faster? For one, condensate nucleation is inherently stochastic. Even small fluctuations in SUMO abundance at telomeres would allow some telomeres to pass the threshold for APB nucleation faster than others. This inherent stochasticity explains why some telomeres have APBs, while others do not, within the same cell. However, this would only induce heterogeneity in telomere extension within a single cell cycle, as stochastic fluctuations would cause other telomeres to form APBs and elongate in the next cell cycle, converging telomere lengths toward a mean over time for a given cell. One possibility to overcome this averaging is that small fluctuations in telomere features are amplified by APB formation, thereby conferring heritability of APB nucleation across cell cycles, thereby maintaining extreme heterogeneity in telomere length.

One such feature could be telomere chromatin state. Normal telomeres are typically packaged into heterochromatin, whereas ALT telomeres are reported to be less compact [24], rendering them recombinogenic. Paradoxically, heterochromatin marks, including H3K9me3 and H3K27me3, are important for APB formation [25, 26, 27]. Intriguingly, artificial induction of heterochromatin at telomeres by tethering HP1a and KRAB is sufficient to promote APB formation [28]. APB formation, in turn, might affect chromatin state at the telomere, given the reported link between PML bodies and heterochromatin remodeling [29]. Changes in chromatin state due to APB formation might be transmitted to the next cell cycle via epigenetic memory, thereby biasing APB nucleation at these telomeres across cell cycles.

Another such feature could be telomere length. Longer telomeres provide a larger binding platform for shelterin components, DNA damage response factors, and chromatin modifiers that can be SUMOylated. As a result, these telomeres might have high local SUMO abundance after DNA damage and thus nucleate APBs. In normal cells, telomeres are not uniform but span a range of lengths [30]. When a normal cell becomes an ALT cancer cell, these naturally long telomeres would nucleate APBs faster, are more likely to undergo recombination, and become even longer. These elongated telomeres then gain a greater advantage over short ones in APB nucleation during the next cell cycle.

TERRA levels at telomeres could also play a role. TERRA is transcribed from the subtelomeric region into the telomeric repeats at a subset of telomeres in human cells and can act in trans to function at all telomeres [31]. Differences in each telomere's ability to produce and recruit TERRA could lead to variation in TERRA levels at telomeres. TERRA has been shown to promote APB formation and ALT [32, 33, 34, 35]. First, it can hybridize with telomeric DNA to form R‐loops at telomeres. These structures are inherently recombinogenic and elicit DNA damage signaling [32], which is known to promote APB nucleation via SUMOylation [11, 19]. In addition, TERRA can bind various proteins, including shelterin components and recombination factors [34, 36, 37, 38, 39], thereby increasing the local molecular density for APB nucleation [34, 36]. Furthermore, TERRA modulates the telomere chromatin state, such as promoting heterochromatin formation [26, 40], which, as mentioned above, drives APB formation. Therefore, telomeres with high TERRA or TERRA R‐loops could be more likely to nucleate APBs. APBs contain factors such as Rad51 and Rad51AP1 that can interact with TERRA or promote TERRA R‐loop formation [41, 42], raising the possibility that APB formation, in turn, enhances TERRA abundance and R‐loop formation at these telomeres. If this feedback loop is true, it would bias APB nucleation across cell cycles.

Therefore, the low APB nucleation rates ensure that only a subset of telomeres form APBs; the stochastic nature of APB nucleation creates heterogeneity in which telomeres gain access to APBs; the positive feedback of APBs that reinforces APB nucleation‐prone telomere features, such as chromatin state, telomere length, and TERRA levels, could promote biased elongation across cell cycles.

4. APB Fusion as an Amplifier of ALT Activity

APB formation via fusion is defined as the fusion of SUMO‐coated telomeres with preexisting PML nuclear bodies. Live imaging shows that this is a low‐efficiency process that occurs slowly after telomere DNA damage, with only a few events per hour [18]. The low efficiency of the fusion process ensures that only a subset of telomeres gains access to PML bodies in this way. This low efficiency can be attributed to the requirement that SUMO‐coated telomeres collide with PML bodies. The collision rates might reflect the mobility of PML bodies more than that of telomeres, as PML bodies are more mobile [43]. PML nuclear body movement is characterized by a corral model, in which PML bodies diffuse within the chromatin corral and hop between the corrals [44]. Therefore, factors that affect PML body size and number, as well as chromatin accessibility and dynamics, would determine the likelihood of a PML body and a SUMO‐coated telomere being in the same chromatin corral and would influence this fusion process. Variability at the single‐cell level in these features would lead to stochastic APB formation via fusion.

Another type of APB fusion occurs when APBs fuse with SUMO‐coated telomeres or with other APBs to grow. These fusion events drive telomere clustering, in which multiple telomeres come into proximity within APBs. These telomeres can serve as repair templates; thus, these APB fusion types nonlinearly amplify ALT activity. These fusion processes require two telomeres to collide in space, making them inefficient. They are strongly shaped by the three‐dimensional organization of the genome, which constrains telomere mobility and influences telomere‐APB or APB‐APB proximity. Factors that alter genome organization—such as changes in chromatin compaction, transcriptional activity, or nuclear lamina integrity—are likely to affect this type of APB fusion behavior. While genome organization is non‐random at the population level, large variance is observed at the single‐cell level [45], which would lead to stochasticity in APB growth via fusion.

One process that might affect both types of APB fusion mentioned above is increased telomere mobility after telomere damage. Increased telomere mobility after TRF2 deficiency‐induced DNA damage is attributed to 53BP1 [46] and the LINC‐microtubule connection [47]. Increased telomere mobility after DNA damage at ALT telomeres has been observed, but the cause of this increase is unknown [48]. It is possible that nuclear actin plays a role, as it increases DNA damage foci mobility in G2 [49], the cell cycle phase in which ALT is active. Thus, factors that regulate 53BP1, the LINC‐microtubule connection, and nuclear actin might influence ALT telomere mobility and, in turn, APB fusion.

Therefore, APB formation via fusing of telomeres with PML bodies enables additional telomeres to access extension after APB nucleation. APB growth via fusing with telomeres and other APBs nonlinearly amplifies ALT activity on these telomeres. The low efficiency of these fusion processes allows only a subset of telomeres to increase ALT activity via APB fusion. The stochastic nature of nuclear organization at the single‐cell level leads to variability in which telomeres participate in APB fusion and are thus preferentially elongated, contributing to telomere length heterogeneity.

5. Kinetic Control of ALT via Coupling Between APB Nucleation and Fusion

While APB nucleation and fusion can be regulated independently, they can occur in the same cell and are linked in several ways. First, because both pathways rely on a limited set of multivalent SUMO‐mediated interactions, rapid nucleation suggests that many telomeres have a strong affinity for SUMO‐related proteins, including the PML protein and other PML body components. This may reduce the number and size of pre‐existing PML bodies as they lose components to newly nucleated APBs, thereby limiting opportunities for SUMO‐coated telomeres to fuse with PML bodies later. Thus, a low nucleation rate of APBs can help boost APB formation via fusion. However, rapid nucleation of APBs produces more APBs, thereby boosting APB fusion rates with telomeres or with each other. On the other hand, low nucleation leads to larger APBs, which might promote telomere clustering since they would make telomeres larger and large telomeres are shown to attract and cluster small telomeres [48]. The heterogeneity in these coupling forces between APB nucleation and fusion, together with differences in the individual regulation layer discussed above, works in concert to bias different telomeres toward APB formation in each cell. Importantly, differential kinetics of nucleation and fusion generate APBs of variable sizes and telomere clusters that differ in their ability to recruit BIR factors, generate ECTRs, and extend telomeres. Telomere features that favor APB formation, such as telomere length and chromatin state, can be enhanced after APB formation and then inherited in the next cell cycle to bias APB formation. Over successive cell cycles, APB formation dynamics give rise to the extreme telomere‐length heterogeneity observed in ALT‐positive cancer cells.

If confirmed, this model may explain how ALT cells can exhibit both robustness and variability: robustness arises from the recombination pathways available at each telomere, while variability emerges from kinetic biases in APB formation, and thus elevated recombination capability, at only a subset of telomeres. Because APB formation is a threshold‐dependent process, ALT output is exquisitely sensitive to modest perturbations. Changes in protein levels, post‐translational modification efficiency, chromatin mobility, or nuclear architecture can shift kinetic parameters enough to alter ALT behavior. This sensitivity may explain why ALT activity varies widely across cell lines, tumor types, and experimental conditions. From a therapeutic perspective, this kinetic susceptibility suggests that targeting APB assembly dynamics—rather than recombination enzymes per se—could attenuate ALT by reducing the probability of APB formation without fully abolishing DNA repair capacity.

6. Broader Implications for Chromatin‐Associated Condensates

The emerging view of APBs as kinetically regulated condensates that can shape functional outcomes in ALT cancer has implications beyond telomere biology. Many nuclear processes, including DNA repair, transcriptional regulation, and heterochromatin formation, are now understood to involve condensates [50]. As with APBs, the functional effects of these condensates may depend on the dynamics of their formation.

First, condensate nucleation can serve as a rate‐limiting control point, determining which genomic loci gain access to concentrated enzymatic activity. Because condensate components are finite within a cell, early‐forming condensates may sequester shared resources, biasing further growth toward already active sites. In ALT, telomeres that nucleate APBs early gain preferential access to recombination machinery, effectively outcompeting other telomeres. This competition introduces functional asymmetry that persists over many cell cycles and becomes encoded in chromosome architecture. Analogous competitive dynamics likely operate in other nuclear processes. For example, for HP1‐mediated heterochromatin formation, early nucleated HP1 condensates can preferentially recruit chromatin modifiers such as SUV39H1/2 and SETDB1, thereby depriving other loci of access. This kinetic filtering mechanism ensures that only selected sites experience intense regulatory activity, thereby biasing future epigenetic trajectories.

In addition, fusion introduces strong nonlinearity into condensate behavior. When two condensates fuse, their functional capacity can increase disproportionately due to cooperative effects, such as higher local enzyme or substrate availability. In ALT cells, APB fusion not only increases the amount of recombination factor near the telomeres but also disproportionately increases the available recombination template. This nonlinearity explains how modest differences in initial nucleation probability or spatial proximity can yield extreme telomere‐length heterogeneity. Similar amplification mechanisms are likely at play in other nuclear processes. In splicing, for example, genomic loci with a high propensity to associate with nuclear speckles can exhibit disproportionately higher splicing efficiency.

Thus, chromatin‐associated condensates can be viewed as kinetic engines that translate molecular fluctuations, spatial organization, and temporal signals into durable genomic outcomes. This implies that ectopic condensate formation in the nucleus—that is, condensates forming at noncanonical genomic locations, at inappropriate times, or with excessive frequency—can profoundly reshape chromatin function. These effects arise not simply because condensates form “in the wrong place,” but also because their formation alters biochemistry at physiological sites. Such examples have been observed in transcription [51], DNA repair [52], and cell division [53]. Understanding condensate kinetics—how quickly structures form, how often they fuse, and how long they persist—is therefore important for understanding how cells use condensates to regulate chromatin functions in both health and disease.

7. Summary

Recent findings suggest that telomere‐length heterogeneity in ALT‐positive cancers is not merely a byproduct of stochasticity in the DNA recombination pathway but is actively shaped by APB formation dynamics. Multiple APB assembly pathways with distinct temporal profiles provide a mechanistic basis for variable ALT activity at individual telomeres. Understanding the kinetic regulation of APBs may reveal new vulnerabilities in ALT‐dependent cancers and offer broader insights into the functions of other chromatin‐associated condensates.

Author Contributions

H. Zhang drafted and edited the manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

No new data were generated or analyzed in this manuscript.