Mechanisms and genomic implications of break-induced replication

Abstract

DNA double-strand breaks (DSBs) are a severe threat to genome stability, as DSB-repair mechanisms with low fidelity contribute to loss of genome integrity. Break-induced replication (BIR) is a crucial DSB-repair pathway when classical homologous recombination mechanisms fail. BIR is often triggered by stalled or collapsed replication forks, following extensive end resection that generates a single-stranded DNA substrate, which can engage either canonical homology-driven BIR, or microhomology-mediated BIR (mmBIR), which requires shorter sequence homologies than does canonical BIR. BIR is a double-edged sword: it is necessary for DSB repair, but is also culpable for introducing mutations and structural variations that are linked to cancer and genetic disorders. In this Review, we discuss BIR regulation in mammalian cells, and the role of BIR in telomere maintenance and in human disease, as well as in genome engineering. We highlight emerging findings in these areas and advances in technologies that have enabled their discovery and reshape our understanding of this enigmatic repair mechanism.

DNA-double-strand breaks (DSBs) are particularly detrimental to genome integrity, and their timely repair is necessary for cell and organismal viability. Several pathways exist for repairing double-ended DSBs (deDSBs). However, when the second break end is inaccessible or non-existent, the process can shift toward break-induced replication (BIR), which entails unidirectional DNA-repair synthesis over many kilobases. BIR has similarities to long-tract gene conversion, which is thought to occur through an abortive BIR mechanism¹. This scenario typically takes place at damaged replication forks, necessitating homology-directed repair (HDR) (Fig. 1a). These findings build on seminal studies by Meselson and Weigle², as well as Skalka³, which introduced the concept of recombination-dependent DNA replication in bacteriophages. Currently published data support the idea that the majority of BIR takes place during the G2 and M phase of the cell cycle, suggesting that it proceeds after fork damage and the completion of DNA replication^4,5. The first step in BIR, like the other HDR mechanisms, is 5′-to-3′ end resection to create a 3′ single-stranded DNA (ssDNA). This overhang is recognized by replication protein A (RPA) and subsequently displaced by a RAD51 filament. The RAD51-coated 3′ overhang can invade duplex DNA with shared homology from another template before returning to its original chromosome (Fig. 1a).

Fig. 1 ∣. The transition from canonical replication to BIR is responsible for ALT.

a, Schematic of DNA replication during S phase and of BIR. When a replication fork encounters an unresolvable barrier, the replication machinery can collapse and generate a seDSB. At the seDSB, the capture by a single-strand 3′ overhang of a homologous template initiates BIR, which proceeds through unidirectional, conservative DNA synthesis. Unlike canonical, S-phase replication, lagging-strand synthesis during BIR is delayed, resulting in the accumulation of extended ssDNA on the leading strand, which is an inherently mutation-prone intermediate. The extension of BIR is limited by the presence of converging replication forks. b, Diagram of ALT mechanisms in yeast. In yeast, ALT occurs through two pathways. Type I ALT uses BIR, in which a Rad51-coated 3′ end invades donor DNA and generates a tandem of the sub-telomeric Y′ element and telomere repeats (marked as (TG)_n). Type II ALT is a Rad51-independent BIR, in which the 3′ overhang of the TG repeats anneals with the template DNA. The Y′ element is not amplified in this pathway. A recent model also indicates that a type I survivor product can be further elongated by Rad52 and Rad59 to form the hybrid ALT outcome containing both tandem Y′ elements and a long TG repeat sequence¹⁵⁸. c, In human cells, ALT is mediated by the helicase BLM, a component of the BTR complex, which unwinds Okazaki fragments to create a 5′ flap to promote the DNA-damage response that mediates ALT. The endonuclease DNA2 cleaves long 5′ flaps generated by BLM on unligated Okazaki fragments. The translocase FANCM is recruited to BLM-dependent telomere substrates and prevents excessive recombination. Hyperactive ALT occurs in response to FANCM loss, leading to the accumulation of unresolved telomeric recombination intermediates and cell death specifically in ALT-positive cancer cells. PIF1, ATP-dependent DNA helicase PIF1.

BIR is as an evolutionarily conserved DNA-repair mechanism, most studied in the budding yeast model Saccharomyces cerevisiae, in which it was first discovered that BIR could repair broken chromosomes by copying large sections of DNA from a homologous template^5-9. It was later found that BIR can be part of a telomere-maintenance mechanism dubbed alternative lengthening of telomeres (ALT)^10,11. In contrast to RAD51-mediated BIR, which requires long stretches of DNA homology (70–100 bp)^7,12, microhomology-mediated BIR (mmBIR) is thought to be initiated with just 1–3 nucleotides of homology^13-16. mmBIR therefore leads to frequent template switching and intrachromosomal recombination with the genesis of complex genomic rearrangements (CGRs), as observed in cancer^17-20 and some de novo germline syndromes^14,21-23. A similar mechanism termed microhomology/microsatellite-induced replication, described by Payen et al.²⁴, produces segmental duplications in yeast in a Pol32-dependent manner. Both RAD51-dependent BIR and RAD51-independent mmBIR appear to be operative in telomere maintenance by ALT in yeast and human cells^25,26 (Fig. 1b,c). Because BIR synthesis tract length is not currently defined, we settle on an arbitrary definition of 1 kb, which is longer than short-tract synthesis observed during classical homologous recombination^27,28.

Several excellent reviews on BIR have highlighted the history and advances made in the field with a strong emphasis on more extensively studied yeast models^28,29 and related findings in mammalian cells^29-32. Consequently, we aim to avoid redundancy by focusing specifically on mechanistic insights from mammalian cells, and their relevance to human disease and the burgeoning field of genome engineering. We begin by introducing foundational concepts of BIR to provide a basis for established literature on its specific mechanisms and genetic regulation. We then focus on the physiological relevance of BIR in human health and disease. We also touch upon less-studied cases of BIR in organisms other than yeast and mammals (Box 1).

BOX 1: Roles of break induced replication in the evolution of plants and other organisms.

Although much BIR research has been conducted on yeast and mammalian systems, there is growing evidence that BIR has an important role in other organisms. Long-read sequencing enabled the identification of chromosomal structural variants in the Caenorhabditis elegans strain CB4856, formed by BIR¹⁵⁹. Experiments in Drosophila melanogaster utilizing a dicentric chromosome system demonstrated that BIR is a minor, but significant, repair pathway¹⁶⁰. Interestingly, there have also been reports of DNA repair by BIR or suspected BIR mechanisms in plants. Cybrid plant mitochondria undergo HDR at high rates¹⁶¹; in one study of Nicotiana tabacum and Hyoscyamus niger cybrids, which identified 28 recombination events, 9 cases were linked to BIR, making it the most prevalent type of recombination mechanism¹⁶². It is likely that plant genomes, whether due to ploidy, epigenetic plasticity or inherent genetic diversity, can better tolerate genomic rearrangements that are typically detrimental to humans. The latter is a case in which BIR has a net positive effect on organismal fitness and directs its genetic evolution. Plant genomes, although large, contain mostly non-coding DNA that is prone to much higher mutation rates than are coding regions. It has therefore been hypothesized that plant genomes evolved to favor either ‘accurate’ or ‘inaccurate’ DNA-repair mechanisms for coding regions and non-coding regions, respectively¹⁶³.

It is thought that the propensity of BIR and mmBIR to initiate replication at DNA-double-strand breaks without a proper replication fork is responsible for the massive genomic duplications and rearrangements seen in plant genomes^163,164. These observations are mirrored in algae: one study found that nearly all substitutions and indels in the non-coding genome of Chlorosarcinopsis eremi were adjacent to or within short inverted palindromic repeats, making them hotspots for mmBIR¹⁶⁵. In the fungus Pyricularia oryzae, BIR has been shown to take place at the telomeres owing to interstitial telomeric insertions of the MoTeR transposable element, which not only leads to harmful ‘telomere failure’, but also promotes genetic diversity and structural rearrangements that confer fitness advantages, including increased virulence¹⁶⁶. Another beneficial process in the plant kingdom is horizontal gene transfer, which contributes to genomic diversity. Its interplay with BIR and mmBIR highlights an evolutionary strategy of balancing genomic instability with evolutionary adaptations^167,168. One study identified whole-mitochondrial-genome transfers from several moss species to the angiosperm Amborella trichopoda; genomic sequencing revealed microhomology regions in which mmBIR could lead to large structural rearrangements¹⁶⁹. The plant species Lophophytum mirabile also exhibits extensive horizontal gene transfer in its mitochondrial genome, up to 74% of which originates from foreign sources. The high prevalence of microhomology regions suggests this integration was facilitated by mmBIR¹⁷⁰. The frequency of BIR and mmBIR in plant genomes is not coincidental: these are purported to be the primary mechanisms of horizontal gene transfer among plants. Using plant cybrids, one group observed a striking predominance of homologous recombination, with minimal evidence of non-homologous repair pathways¹⁷¹. Additionally, this study revealed the presence of recombination hotspots, often associated with regions of highly repetitive DNA tracts, further supporting the role of BIR and mmBIR in facilitating genomic exchange and structural rearrangement in plant genomes.

The existence of BIR in both the plant and animal kingdoms underscores its fundamental importance, suggesting that it is an evolutionarily conserved mechanism of DNA repair that is vital both for genome integrity and as an evolutionary engine, thereby maintaining a delicate balance between genetic instability and evolutionary resilience.

Comparison between break-induced replication and canonical DNA replication

Genome duplication in S phase entails the formation of an active replisome comprising an origin-recognition complex and the CDC4–MCM–GINS (CMG) helicase, which unwinds duplex DNA at the replication origin to allow leading- and lagging-strand synthesis in a bidirectional, semi-conservative manner³³. Replication origins are defined by DNA sequence and chromatin environment, and licensing of the fully assembled CMG proceeds by an orderly sequence of events. By contrast, BIR occurs as a response to DNA-damage repair. Notably, most studies indicate that the CMG helicase is not essential for BIR, which relies on other helicases to unwind DNA in a 5′–3′ manner^34-36, along with a minimal replisome that is unencumbered by regulatory mechanisms inherent to origin firing during S-phase replication^10,37.

BIR uses a conservative replication mechanism, involving new DNA synthesis on both leading and lagging strands^4,38. This process primarily relies on DNA polymerase δ (Polδ) and Polα, whereas the canonical-replication polymerase Polε is not involved in most reported BIR processes^10,37,39-44. The bias towards Polδ stems from its higher-affinity binding to PCNA through its POLD3 subunit⁴⁵. By contrast, in canonical replication, Polε is tightly regulated through interaction with the CMG helicase^46,47. These features explain how Polδ is favored during BIR, given that PCNA is loaded onto DNA by the clamp-loading replication factor C complex (RFC1–5) at recessed 3′ ends in BIR or 5′ overhangs and can serve as a scaffold for Polδ recruitment. Polδ displays strand-displacement synthesis activity that is thought to facilitate its function in lagging-strand DNA synthesis and to drive bubble migration during BIR^36,48 (Fig. 1).

An important question is how a minimal, break-induced PCNA–Polδ replisome can achieve long-tract DNA synthesis in the absence of the replicative CMG helicase and many other factors that are present at DNA replication forks. In yeast, the 5′–3′ DNA helicase Pif1 fills this void^34-36 (Fig. 1a). Moreover, when Polδ is forced to act as the leading-strand polymerase, as opposed to Polε, it requires both Pol32 and Pif1 (ref. 49). Pif1 itself is essential for BIR and telomerase-independent telomere maintenance^50,51. The Pif1 helicase operates similarly to its yeast ortholog in Drosophila melanogaster under replication stress, and, by extension, BIR⁵². PIF1 has further been implicated in model systems of BIR in human cells but seems to be dispensable for ALT^34,53, suggesting that more than one helicase could fulfill this role in mammalian cells. Indeed, telomere purification following DSB induction shows various damage-responsive DNA helicases, although their contribution to BIR remains unknown⁵³.

The regulation of break-induced replication

BIR can be induced by barriers that block replication, consistent with stalled or damaged forks stimulating the transition from semi-conservative to conservative DNA-repair synthesis³¹ (Fig. 1a). Indeed, BIR can be induced by replication-fork collapse at DNA lesions, including unrepaired DNA nicks that arise from topoisomerase I adducts, to restart the fork^54-58.

Although the mammalian BIR helicase(s) remains enigmatic, additional aspects of the DNA-replication-stress response seem to contribute to long-tract HDR synthesis in BIR. Template switching is a conserved DNA-damage-tolerance mechanism and a feature of BIR in both yeast^59-61 and human cells^53,62. During this process, the invading strand can switch among homologous donor strands. Replication stress due to break-induced replisome encounters with blocking lesions seems to be an important factor in its occurrence. PCNA ubiquitination by the E3 ligase RAD18 serves as a platform for translesion DNA polymerases and for template switching^53,63. There is evidence for both tolerance mechanisms during ALT. The translesion polymerase Polη was found specifically at ALT telomeres and suppressed BIR, ostensibly by preventing replication-fork collapse into DSBs⁶⁴. Recent work has revealed that RAD18 mediates the ubiquitylation of nearly 50% of PCNA at damaged telomeres. Ubiquitin modification of PCNA occurs during BIR, particularly during break-induced replisome encounters with blocking lesions. These events mediated recruitment of the PCNA–ubiquitin-binding nuclease SNM1A to damaged telomeres to execute long-range end-resection and template switching during ALT^53,65. Generation of ssDNA by SNM1A in association with PCNA–ubiquitin during ALT was independent of the nucleases EXO1, MRE11 and DNA2 (refs. 53,66,67), which participate in end-resection at DSBs^68-70. This indicates that BIR during ALT invokes a unique mechanism of end processing to create the ssDNA necessary to initiate recombination. It is currently unclear whether this is a general mechanism of resection during BIR or specific to damaged telomeres. Kaposi’s sarcoma herpesvirus infection induces ALT-like features and BIR at telomeres in cells that do not rely on ALT for telomere maintenance⁷¹. Investigating whether similar resection-dependent template-switching mechanisms are involved in viral genome maintenance would be intriguing.

The attributes that promote BIR efficiency also compromise the fidelity of this repair mechanism. Although the proofreading activity of Polδ suppresses mutagenesis during BIR, high mutation rates have been reported⁷². This could stem, in part, from Polδ having lower fidelity in the context of extension of recombination intermediates at the D-loop⁷³. The modified Polδ with yeast Pol32 or human POLD3 frequently disassociates from the template and increases the risk of strand slippage and template switching errors⁷⁴. Meanwhile, other polymerases such as translesion synthesis polymerase Polζ can conduct error-prone DNA synthesis during BIR⁷⁵. Additionally, the accumulation of long stretches of single-stranded DNA during BIR can be a substrate for the APOBEC3A cytidine deaminase to induce abasic sites and base substitution⁷⁵. Mismatch repair efficiency is also reduced during BIR⁷². Finally, template switching includes extension from multiple DNA templates that contain minimum sequence homology. Although this process can be error-free in principle, it can also result in complex genome rearrangements and amplifications^60,61 (Figs. 2 and 3a,b). For example, yeast translocase Mph1 disrupts BIR progression and promotes template switching and formation of non-crossover products in BIR^36,76,77. The mammalian ortholog of Mph1, FANCM is essential specifically in cells in which ALT is operative (Fig. 1c). FANCM deficiency hyperactivates a lethal form of ALT characterized by excessive telomere recombination and synthesis, and possibly nucleolytic cleavage of telomeres into extrachromosomal circles^78-81. Whether FANCM functions at damaged telomere replication forks or subsequently during BIR is unclear but remains an active area of investigation.

Fig. 2 ∣. Consequences of break-induced replication versus microhomology-mediated break-induced replication.

a, Illustration of the first steps of BIR or mmBIR, beginning with (i) fork collapse during replication, resulting in a single-ended DSB. (ii) The DSB end undergoes 5′–3′ end resection to create a 3′ overhang and (iii) coating of the 3′ overhang with a RAD51 filament before homology search. b, In BIR conditions (that is, the existence of long stretches of homology), low-fidelity DNA synthesis takes place, generating mutations in the newly synthesized DNA strand that were not present in the template strand. c. In mmBIR conditions (short stretches of homology; base pairing indicated as black vertical lines), strand synthesis is limited by frequent collapses or by 3′-end dissociation that cause template switching of the 3′ end. This process creates a DNA structure consisting of complex genomic rearrangements incorporating several non-concurrent sequences.

Fig. 3 ∣. Break-induced replication-driven genomic amplification mechanisms.

a, Collapsed forks at repetitive elements repaired by BIR can use a repair template with a differing number of repeats as the broken strand. The broken strand from chromosome ‘A’ originally contained ‘Repeat 1’ and ‘Repeat 2,’ representing individual repeat elements for a total of two repeats. During BIR-mediated interchromosomal strand invasion, the invading strand uses a non-sister chromatid of chromosome ‘B’ as a template that contains three repeat elements—‘Repeat 1’, ‘Repeat 2’ and ‘Repeat 3’. As a result, the newly synthesized strand is expanded to include all three repeats. Color shading represents sequence orientation: dark, 5′–3′; light, 3′–5′. b, BIR and mmBIR can give rise to complex genomic rearrangements through co-occurring inversion and duplication events when using a sister chromatid as a template. Following fork collapse and 3′-end resection, the broken strand invades the sister chromatid through microhomology to copy locus ‘B’ in the opposite orientation and creates an inversion. This process is followed by fork collapse and re-invasion in the appropriate orientation, starting from locus ‘A’ through ‘B’ and ‘C.’ The resulting DNA product is a duplication of loci A, B and C flanking an inversion of locus ‘B.’ c, CRISPR–Cas9 mediated ‘amplification editing’¹⁴¹ is initiated by targeting of Cas9 nickases to two complementary regions, resulting in BIR-like expansions. The use of a Cas9-fused reverse transcriptase and a pegRNA at the targeted nicks generates 3′ flaps with complementary sequences. These flaps anneal, leading to self-template-driven DNA synthesis that causes unwinding of the template with continued synthesis up to the flap sequence. The resulting product is a duplication ranging from 20 bp to 100 Mbp, depending on the distance between the two generated flaps. This phenomenon has not been observed to occur endogenously in human cells, but its demonstration using genome-engineering tools to invoke endogenous DNA-repair pathways suggests it is a biologically possible variant of BIR.

Although myriad replication-blocking lesions exist, secondary DNA structures cause the transition of stalled replication forks to BIR. Trinucleotide repeats, such as (CAG)n and (CTG)n, can form hairpin structures that stall replication forks and trigger BIR⁸². Similarly, non-B DNA structures, such as AT-rich sequences at common fragile sites, can induce replication stress and DNA breaks^34,83. Additionally, R-loops formed during transcription promote various processes involved in BIR, including end resection and template switching^62,84,85. Most common fragile sites overlap with long transcribed genes⁸⁶. The fragility of these regions can be explained by collisions between replication machinery and either R-loops or transcription machinery.

Recent studies using Cas9 nickases to create persistent nicks on the leading or lagging strands have shed light on this transition^54,56-58,87. Interestingly, nicks on either strand can trigger BIR, although they typically lead to double-ended DSBs repaired through homologous recombination. Such persistent Cas9-induced nicks likely mimic covalent topo I adducts that encounter replication forks^55,88. Additional sources of nicks can activate BIR. For example, ATRX deficiency promotes ALT^89-91 and has been shown to cause persistent nicks on the telomere lagging strand, likely owing to incomplete Okazaki fragment maturation^66,67. BLM helicase was specifically recruited to nicked C-rich strand of telomere DNA, unwinding it to generate long 5′-flap ssDNA, which orchestrates the telomere DNA-damage response. Notably, lagging strand, but not leading strand, Cas9-induced telomere nicks recapitulated this process to activate the ALT mechanism⁶⁶.

It is important to note that BIR may not be the primary pathway for rescuing single-ended DSBs (seDSBs) during replication. Cells lacking essential BIR genes can survive under conditions where nicks are induced^56,57 and BIR at collapsed forks is suppressed by multiple mechanisms. For instance, Mus81-mediated cleavage and replication fork convergence can limit BIR in yeast^54,55. The activation of dormant origins during replication stress may therefore serve to suppress BIR by increasing convergence at stalled forks (Fig. 1a). These S-phase limitations likely explain why a majority of BIR occurs in G2 and mitosis, when converging forks are not available to limit unidirectional synthesis of the break-induced replisome. For example, common fragile sites with origin-poor regions tend to utilize mitotic DNA synthesis to complete replication in mitosis, a RAD51-independent BIR-like mechanism that has similarities to ALT⁸⁶. Recent studies also indicate that lesions such as nicks in the leading- or lagging-strand template are converted into seDSBs and deDSBs, and are repaired mostly by homologous-recombination-mediated gene conversion instead of BIR^54,56-58. These limitations explain how BIR is promoted when only one of the DSB ends contains homology to the donor or when there is a significant distance between the homology sites on the donor so the second end capture is difficult^34,92,93. In agreement, site-specific endonucleases, such as HO and I-SceI, have been widely used in yeast and mammalian cells to establish BIR reporter systems³¹. Similarly, DSBs induced by the telomere-sequence-specific endonuclease TRF1–FokI stimulate POLD3-dependent break-induced telomere synthesis (BITS) and other ALT features with high efficiency^37,94.

Several factors modulate BITS, such as RAD51AP1 (ref. 95), XPF⁹⁶, ATAD5 (ref. 65), and the Fanconi anemia (FA) family of proteins^78-81,97. The telomere repeat-containing RNA (TERRA) forms R-loops at telomeres that are proposed to initiate ALT^84,98-103. Interestingly, TERRA might also regulate ALT through phase separation: it promotes condensate formation in conjunction with lysine-specific demethylase 1A (LSD1)¹⁰⁴. These phenomena could be connected to the predominance of BITS in ALT-associated PML bodies^25,26,105. These large SUMO-containing condensates comprise a hub of DNA-repair proteins that carry out HDR synthesis other than semiconservative DNA replication^26,106,107. The E3 SUMO protein ligase PIAS4 promotes the recruitment of repair proteins to ALT-associated PML bodies in ALT, and condensate formation can be achieved by tethering a SUMO-interacting motif to telomeres. In accordance, SUMO inhibition or PML deficiency strongly reduces ALT, commensurate with telomere shortening^107,108. Given the evidence for clustering damaged replisomes into recombination centers³⁷, an intriguing question is whether BIR is regulated by 3D chromatin organization. Late-replicating domains are typically enriched at the nuclear periphery, and nuclear-actin filaments have a role in relocating stressed replication foci to the nuclear periphery and in regulating the loading of repair machinery^109,110. The SUMO-dependent relocalization of chromatin containing DSBs to the nuclear periphery is crucial for effective DSB repair^111,112. Increased mobility of chromatin has been observed at ALT telomeres following DSBs⁹⁴, suggesting that 3D nuclear organization is a key factor in the BIR process.

Physiological consequences of break-induced replication

BIR is implicated as a driver of several human genetic diseases that are associated with large copy-number variations and chromosomal rearrangements^17,40,59. BIR-repaired DNA is highly mutagenized as a consequence of low-fidelity Polδ-driven replication, with frameshift mutation rates during BIR in yeast being up to 2,800-fold greater than normal replication at an error rate of 1 × 10⁻⁵–1 × 10⁻⁶ (refs. 72,113) (Fig. 2a,b). These structural alterations represent a BIR-driven scar in specific cancers¹¹⁴. In homologous recombination-deficient BRCA1 or BRCA2 (BRCA1/2)-mutated cancers, RAD52-mediated mmBIR¹¹⁵ drives interchromosomal translocations between low-homology sites, leading to extensive structural alterations and large-scale chromosomal duplications^{14,19,116,117} (Fig. 2a,c). The latter is especially consequential for repetitive regions in which high sequence homology renders them susceptible to BIR-mediated expansions and contractions. Centromeric arrays, for instance, can experience expansions of up to ~80-fold¹¹⁸. Similarly, ALT-positive cancers exhibit heterogeneous telomere lengths as a result of stochastic BIR-induced lengthening and shortening^37,38 (Fig. 3a). There is mounting evidence that a chronic replication-stress-associated damage response at ALT telomeres^53,119,120 engages BIR for telomere maintenance⁶⁶. Furthermore, ALT telomeres frequently engage in BIR-mediated template switching. It is unknown which lesion(s) is responsible for ALT stalling, but this process occurs in a BIR-synthesis-dependent manner⁵³. Although template-switching is tolerated at repeat regions, it causes a detrimental accumulation of templated insertions in cancer cells^13,121.

Replication-associated damage can dictate whether BIR promotes cell survival or leads to lethal levels of genome instability. BIR at ALT telomeres provides telomere maintenance in the absence of telomerase; FANCM depletion however induces a hyperactive, lethal ALT phenotype driven by R-loop formation and replication stress^79,81 (Fig. 1b). Thus, whereas a basal DNA damage response is necessary for telomere maintenance, excessive BIR overwhelms cellular tolerance. This phenomenon at ALT telomeres may be mechanistically related to observed increases in long-tract gene conversions at stalled replication forks and synthetic lethal interactions with BRCA1 mutations following FANCM loss¹²². This raises an interesting question: are insights from ALT telomeres generally applicable to BIR across the genome? Understanding what enables ‘sustainable’ BIR in cancer cells to foster evolution and therapy resistance could reveal opportunities to target BIR for conditional cell death through genomic instability. Conversely, it is essential to understand why and how cells avoid choosing BIR as a repair mechanism in certain contexts. Owing to defects in mismatch repair, cancers with microsatellite instability acquire short, repetitive DNA sequences scattered throughout the genome. In particular, large (TA)_n repeats that develop in microsatellite instability-positive cancers form cruciform structures and depend on the Werner (WRN) helicase to maintain genome stability¹²³. WRN depletion is synthetically lethal in these cancer cells owing to the generation of DSBs in the microsatellites by replication-stress-activated structure-specific nucleases^124,125. Despite the extensive homology present in this repetitive DNA, BIR either fails to initiate or is insufficient to overcome replication blocks.

Beyond cancer, BIR is a significant source of genetic instability, even during the earliest stages of embryonic cell division. CGRs can give rise to myriad diseases and developmental disorders, such as those characterized by chromosomal triplications, duplications and inversions (for example, duplication-triplication/inversion-duplication (DUP-TRP/INV-DUP) structures), or by simple deletions and duplications (Fig. 3b and Table 1). Similar structures caused by BIR, such as DUP-TRP/INV-DUP events, have also been documented in yeast⁶⁰. The cause of these CGRs, in both yeast and humans, is supported by models proposing mmBIR-mediated template switching between homologous chromosomes and sister chromatids¹²⁶. These mechanisms have also been proposed to result in an absence of heterozygosity in developing embryos, leading to non-inherited developmental disorders¹²⁶.

Table 1 ∣.

Diseases and syndromes suspected to be caused by BIR or by microhomology-mediated BIR

Disease or syndrome	Chromosomal rearrangement(s)	Locus or gene	Reference
Temple syndrome	DUP-TRP/INV-DUP	Chr. 14	143
Pelizaeus–Merzbacher disease	DUP-TRP/INV-DUP	PLP1	12,114
Lubs X-linked mental retardation syndrome	Duplication	Xq28	144
Syndromic anophthalmia	DUP-TRP/INV-DUP	13q	145
Potocki–Lupski syndrome	Duplication	17p11.2	146,147
MECP2 duplication syndrome	Duplication	MECP2	148,149
CHARGE syndrome	Deletion	CHD7	150
Marfan syndrome	Deletion	FBN1	151
Kallmann syndrome	Deletion	Xp22.31	152
Smith–Magenis syndrome	Deletion	17p11.2	146,153
2q23.1 microdeletion syndrome	Deletion	2q23.1	154
Alveolar capillary dysplasia	Deletion	16q24.1	155
Miller–Dieker syndrome	Deletion	17p13.3	156
Autosomal dominant facioscapulohumeral muscular dystrophy	Deletion	Chr. 4, Chr. 10	157

PLP1, proteolipid protein 1; MECP, methyl CpG binding protein 2; CHD7, chromodomain helicase DNA binding protein 7; FBN1, fibrillin-1.

Repeat expansion disorders highlight the role of the length of DNA sequences and repeat tracts in regulating BIR. For example, in yeast, TG-rich sequences at DSBs inhibit resection of the TG-rich side, while promoting it on the TG-free side¹²⁷. This asymmetry drives homology-driven repair resembling BIR, with a migrating D-loop extending the TG-free end, resulting in non-reciprocal translocations. As previously mentioned, (CAG)_n–(CTG)_n repeats can trigger BIR by forming hairpin structures that stall replication forks in yeast—could similar mechanisms drive their expansion in humans during early embryogenesis? By contrast, BIR-mediated (GAA)_n repeat expansion, which is implicated in Friedrich ataxia, is driven by the formation of DNA–RNA hybrids¹²⁸. Recent evidence shows that single-strand nicks converted into DSBs during replication drive (GAA)_n repeat expansions in a strand-specific manner: leading-strand nicks are repaired through template switching and BIR, and lagging-strand nicks are repaired through homologous recombination^129,130.

An often-overlooked factor influencing BIR and DSB repair is genome spatial organization, which impacts repair kinetics and can manifest as large deletions in disease contexts. Resected single-ended DSBs could extend homology searches over extensive distances (~100–200 kb), thereby driving long-range deletions¹³¹. Chromatin 3D organization can even drive pathogenic repeat expansion: misfolding of topologically associating domains caused by excessive heterochromatinization leads to replication stress and eventually DSBs, which drive (CGG)_n repeats in fragile X syndrome¹³². The interplay between trinucleotide repeat sequence, length, transcriptional activity, 3D organization and strand-specific replication dynamics underscores the complexity of modeling BIR during repeat expansions and necessitates further research into their behavior.

Advancements in long-read sequencing technologies have significantly improved the identification of chromosomal structural variants and their breakpoints. For instance, Oxford Nanopore sequencing has revealed microhomology-flanked breakpoints in copy number variants on chromosome 20 in human pluripotent stem cells, which resulted in tandem duplications of the anti-apoptotic gene BCL2L1¹³³. Similarly, high-fidelity technologies such as PacBio HiFi have enabled the discovery of critical breakpoint junctions underlying DUP-TRP/INV-DUP rearrangements, experimentally validating inverted low-copy repeats as recombinant substrates for BIR^134,135. Existing models to study BIR kinetics in yeast and mammalian cells use techniques such as PCR, high-throughput sequencing and fluorescence reporter assays to elucidate the dynamics of this repair pathway^4,117,136. As technologies continue to advance in resolution and sensitivity, they are likely to uncover more information on breakpoint events in cells that lead to rare diseases or drive cancer growth, offering valuable insights into the fundamental mechanisms underlying BIR and human disease^137-139.

Break-induced replication in the service of genome engineering

Genetic engineering has advanced significantly since the discovery of the CRISPR–Cas9 system. Cas9 endonuclease, guided by an RNA complementary to its DNA target, induces a DSB 3 base pairs downstream of a protospacer adjacent motif¹⁴⁰. Engineered Cas9 variants, such as those with D10A or H840A substitutions, create single-strand nicks on the target or non-target DNA strands, respectively. This approach, in tandem with prime editing guide RNA (pegRNA) and a reverse-transcriptase fusion protein, has enabled ‘amplification editing’ in cells, generating megabase-scale duplications proposed to arise through a BIR-like mechanism (Fig. 3c)¹⁴¹. DNA nicking at targeted loci is followed by generation of complementary 3′ flaps through the reverse-transcriptase template of the pegRNA. Like RAD51-independent BIR, these flaps can anneal to one another and support synthesis of a new DNA strand using each other as templates. The migrating D-loop proceeds to amplify the genomic region until reaching the nick/flap site. The demonstration of this amplification mechanism following the induction of nicks, a common lesion in human cells, prompts speculation about an endogenously occurring editing-like amplification process producing large-scale genomic duplications.

The Cas9 system, with its ability to introduce targeted DSBs, provides a powerful tool for investigating a wide array of DNA-repair responses. The FA pathway directs Cas9-induced DSBs toward template-dependent repair mechanisms, prioritizing single-strand template repair¹⁴². This highlights a potential parallel with BIR, as both single-strand template repair and BIR resolve DSBs through homology-directed repair. FANCD2 is particularly enriched at Cas9-induced DSBs and coordinates repair-pathway choice, raising the possibility that other FA proteins could similarly regulate these decisions at endogenous lesions to bias either towards or away from BIR. These observations also invite speculation that BIR involving single-stranded intermediates can occur and could similarly rely on the FA pathway to guide repair fidelity, linking these pathways to broader mechanisms of genome stability.

Conclusions and future perspectives

BIR is an error-prone DNA repair mechanism with broad implications for genome stability, evolution, and disease. Although extensively studied in yeast, insights from mammalian systems reveal distinct regulatory mechanisms and relevance to human health. The role of BIR in cancer, particularly in maintaining telomeres through the ALT pathway, highlights its dual nature—supporting cell survival while driving genome instability and complex rearrangements. Involvement in diseases such as BRCA1/2-deficient cancers and repeat expansion disorders further underscores its significance in developmental and somatic instability.

Future research should focus on elucidating the molecular machinery underlying BIR in mammalian cells, particularly the helicases and replisome components that enable long-tract synthesis. The relationship between chromatin state, 3D genome organization and replication stress in regulating BIR warrants deeper exploration, as these factors likely determine repair-pathway choice and outcomes. The application of genome engineering tools such as CRISPR–Cas9 offers new avenues to study BIR dynamics and its role in large-scale duplications. Linking BIR to genome editing could uncover new therapeutics targeting BIR-reliant cancers. Furthermore, the evolutionary implications of BIR in plants and other non-model organisms suggest that it has a broad role in shaping genomes to balance genome instability with adaptability.