Diverse cellular functions of barrier-to-autointegration factor and its roles in disease

Barrier-to-autointegration factor (BAF; encoded by BANF1) is a small, ubiquitous and highly conserved 89 amino acid (aa) metazoan protein (Zheng et al., 2000; Harris and Engelman, 2000). Although its localization can be variable, being dependent on cell type, age and cell cycle stage, BAF is found throughout the cytoplasm, nucleoplasm and enriched on the nuclear envelope (NE) (Haraguchi et al., 2007). BAF has numerous binding partners, including DNA, histones and proteins of the NE (Shimi et al., 2004; Haraguchi et al., 2008; Molitor and Traktman, 2014). Originally discovered as a host protein that is hijacked by retroviruses to prevent suicidal autointegration (Lee and Craigie, 1998), BAF was subsequently found to have anti-viral functions in the cytoplasm and nucleus (Wiebe and Traktman, 2007; Ibrahim et al., 2013; Jamin et al., 2014b). Beyond these innate immune functions, BAF is required for multiple fundamental cellular processes, including post-mitotic nuclear envelope reformation (Haraguchi et al., 2001; Gorjánácz et al., 2007; Haraguchi et al., 2008; Asencio et al., 2012; Samwer et al., 2017), repair of damage to the barrier function of the NE (Halfmann et al., 2019; Young et al., 2020), transcriptional regulation (Margalit et al., 2007; Wang et al., 2002; Huang et al., 2011; Cox et al., 2011) and the DNA damage response (Montes de Oca et al., 2009; Bolderson et al., 2019). Collectively, the fundamental cellular role of BAF is to protect genome integrity and ensure successful completion of mitosis. The latter involves at least two distinct functions for BAF – post-mitotic reconstitution of the nuclear envelope around chromosomes (Haraguchi et al., 2001 2008; Samwer et al., 2017) and promotion of inter-chromosome cohesion to form a single nucleus (Samwer et al., 2017). Loss of BAF in Caenorhabditis elegans (Zheng et al., 2000) and Drosophila melanogaster (Furukawa et al., 2003) during embryogenesis is lethal, likely due to defects in post-mitotic nuclear reformation. Not surprisingly, altered BAF expression is linked to both a rare genetic progeric disorder (Puente et al., 2011; Cabanillas et al., 2011) and some forms of cancer (Lai et al., 2010; Li et al., 2017, 2018; Zhang, 2020). Despite being a small dynamic protein with no known enzymatic properties, BAF has diverse cellular roles that rely on its ability to bind DNA and its numerous protein interactions. Here, we review our current understanding of the roles of BAF in key cellular processes, and the mechanisms by which it accomplishes those functions through self-association, numerous binding partners and posttranslational regulation, as well as the role of BAF in human disease.

By way of a helix-hairpin-helix DNA-binding domain, BAF binds to double-stranded DNA (dsDNA, hereafter just denoted DNA) in a sequence-independent manner (Zheng et al., 2000; Harris and Engelman, 2000; Umland et al., 2000) (Fig. 1). As BAF forms stable homodimers, and eventually oligomers, it is capable of stably binding DNA with a dissociation constant of ∼10 pM (Samwer et al., 2017); this allows it to ‘bridge’ and condense intra- and/or inter-molecular DNA strands (Zheng et al., 2000) of both foreign (Lee and Craigie, 1998; Wiebe and Traktman, 2007; Kobayashi et al., 2015) and genomic DNA (Samwer et al., 2017). This ability to bind, condense and functionally crosslink DNA is key to most of the fundamental cellular roles for BAF.

For such a small protein, BAF has numerous reported interactions with other proteins, with one BAF interactome including over 70 potential partners (Montes de Oca et al., 2009). However, the most characterized protein interactions of BAF are those of the LAP2-Emerin-MAN1 (LEM)-domain proteins (Cai et al., 2001, 2007; Lee et al., 2001; Shumaker et al., 2001; Mansharamani and Wilson, 2005) (for more information on LEM-domain proteins, see Box 1). The LEM domain is a conserved globular domain of ∼40 aa that mediates binding to a BAF dimer. Upon BAF homodimerization, a binding cleft is created for the LEM domain (Cai et al., 2001), and several LEM-domain proteins that are part of the inner nuclear membrane (INM) proteins rely, at least in part, on their association with nuclear BAF for their recruitment to and retention at the INM (Haraguchi et al., 2008; Holaska et al., 2003; Montes de Oca et al., 2009).

A-type lamins are also well-characterized BAF interactors (Holaska et al., 2003; Montes de Oca et al., 2009; Capanni et al., 2010, 2012); these lamins are type-V intermediate filament proteins that help form the nuclear lamina in many metazoan cells and also partially rely on BAF for envelope localization (Haraguchi et al., 2001, 2008). Although some LEM-domain proteins directly associate with A-type lamins (Wagner and Krohne, 2007; Berk et al., 2014; Pałka et al., 2018), BAF dimers can act as a bridge between LEM-domain proteins and A-type lamins by binding to both the LEM domain of a protein and the immunoglobulin (Ig)-like fold of A-type lamins, potentially stabilizing the LEM–lamin connection (Samson et al., 2018) (Fig. 1). This interaction suggests that BAF, A-type lamins and nuclear LEM-domain proteins function synergistically. Moreover, the high affinity of BAF for the core histones H3 and H4 and linker histone H1.1 may contribute to its enrichment at the NE due to association with peripheral heterochromatin (Montes de Oca et al., 2005, 2011). Other BAF interactors include various transcriptional regulators (Wang et al., 2002; Holaska et al., 2003; Montes de Oca et al., 2009; Huang et al., 2011) and DNA damage repair proteins (Montes de Oca et al., 2009; Bolderson et al., 2019) (see Table 1 for a more extensive list).

BAF has also been reported to form heterodimers with BAF-like (BAF-L; encoded by BANF2), which is 40% identical and 53% similar to BAF, but is incapable of binding DNA (Tifft et al., 2006). BAF-L has been speculated to regulate the binding of BAF to DNA or other binding partners (Tifft et al., 2006), perhaps in a tissue-specific manner, as it is expressed most abundantly in the testis and pancreas (Tifft et al., 2006; Elkhatib et al., 2017).

Despite these associations, a GFP-tagged BAF has an extremely high mobility, with fluorescence recovery after photobleaching (FRAP) recovery halftimes (t1/2) reported as 0.26 s in human cells (Shimi et al., 2004) and 2.24 s in C. elegans (Margalit et al., 2007; Bar et al., 2014), likely due to posttranslational regulation modifications (Shimi et al., 2004). Curiously, regardless of its small size and high mobility, GFP–BAF does not passively diffuse through nuclear pores (Shimi et al., 2004). Instead, GFP–BAF exhibits an unexpected subcellular compartmentalization with populations predominantly, if not exclusively, retained in either the cytoplasm or nucleoplasm. However, within the nucleus there is rapid exchange of GFP–BAF between the nucleoplasm and NE (Shimi et al., 2004; Haraguchi et al., 2008; Molitor and Traktman, 2014; Birendra et al., 2017). Owing to the high mobility observed with GFP-tagged BAF, it is likely that BAF transiently interacts with many of its binding partners in a ‘touch-and-go’ manner (Shimi et al., 2004), although any behavior of BAF fused to GFP, a fluorescent protein that is approximately three times its size, should be cautiously interpreted.

BAF–DNA complexes are stable in vitro (Skoko et al., 2009); however, in cells, the ability of BAF to bind to DNA is regulated by N-terminal phosphorylation (Nichols et al., 2006). The serine-threonine (Ser-Thr) vaccinia-related kinases (VRKs), which have homology with vaccinia virus B1 kinase (Nichols and Traktman, 2004), play roles in phosphorylation events that regulate cell cycle progression (Kang et al., 2007; Molitor and Traktman, 2014), cell signaling (Kang and Kim, 2008; Klerkx et al., 2009), DNA-damage responses (Sanz-García et al., 2012; Salzano et al., 2015) and apoptosis (Monsalve et al., 2013). The VRK kinases also phosphorylate BAF (Nichols et al., 2006; Birendra et al., 2017; Park et al., 2015) at residues Thr-2, Thr-3 and/or Ser-4, decreasing its affinity for DNA and, possibly, LEM-domain proteins (Nichols et al., 2006). The Ser-Thr protein phosphatases PP2A (Gorjánácz, 2013) and PP4 (Zhuang et al., 2014) dephosphorylate BAF.

Upon its entry into the nucleoplasm, BAF can be phosphorylated by the nucleoplasmic VRK1 (Nichols et al., 2006) and to a lesser extent by the NE transmembrane-anchored VRK2A, the predominant splice isoform of VRK2 (Birendra et al., 2017). Knockdown of VRK1 has been shown to increase the levels of nuclear BAF, in addition to reducing its mobility in the nuclear compartment, mostly likely through an enhanced affinity to DNA (Molitor and Traktman, 2014), whereas VRK1 overexpression leads to a decrease in the nuclear BAF (Nichols et al., 2006). VRK3, which was previously thought to be an inactive pseudokinase due to mutations in three conserved amino acids in the catalytic kinase domain (Scheeff et al., 2009), has more recently been reported to be capable of phosphorylating Ser-4 on BAF (Park et al., 2015). Ectopic expression of phosphomimetic GFP–BAF exhibits a similar subcellular localization to the wild-type protein, whereas an unphosphorylatable BAF mutant is predominantly nucleoplasmic (Jamin et al., 2014b; Halfmann et al., 2019). These results indicate that N-terminal phosphorylation of BAF regulates its DNA-binding affinity, other BAF-binding interactions and subcellular localization. Interestingly, the reduced expression of VRK1 (Vinograd-Byk et al., 2018) or complete loss of VRK3 (Kang et al., 2017) in mice leads to central nervous system defects. In humans, VRK1 mutations lead to the neurodegenerative disorder pontocerebellar hypoplasia type 1A (PCH1A; 607596), characterized by early-onset spinal muscular atrophy with death in late childhood (Renbaum et al., 2009; Najmabadi et al., 2011), and variants in VRK2 are strongly associated with multiple psychiatric and neurological disorders, including schizophrenia, major depressive disorder and genetic generalized epilepsy (reviewed in Li and Yue, 2018). Direct links between BAF and neurological disorders remain unreported, indicating that these phenotypes may be unrelated to VRK-mediated BAF regulation.

BAF was originally discovered and named for its role in binding to cytosolic viral DNA (Lee and Craigie, 1998; Wiebe and Traktman, 2007; Ibrahim et al., 2013; Chen and Engelman, 1998), after it had been purified from the cytosol of NIH3T3 cells as a key component of retroviral pre-integration complexes (PICs) (Lee and Craigie, 1998; Cai et al., 1998). After a retrovirus enters a cell and during the early phases of retroviral infection, the retrovirus reverse transcribes the viral RNA genome. While in transition to the nucleus, the newly transcribed viral DNA is reorganized into a PIC, containing both viral- and host-encoded proteins (Greene and Peterlin, 2002). Moloney murine leukemia virus (MoMLV) PICs from infected NIH3T3 cells were depleted of BAF by salt extraction and shown to suffer ‘suicidal’ autointegration and loss of host integration (Lee and Craigie, 1998). Upon BAF reconstitution, the salt-stripped MoMLV PICs regained host integration capability (Lee and Craigie, 1998). The recruitment of BAF (‘barrier-to autointegration factor’) to PICs compacted and ‘cross-bridged’ the viral DNA protecting it from ‘suicidal’ autointegration and thus increasing viral DNA integration into the host genome (Fig. 2).

This hijacked pro-viral role for BAF appears to contradict its antiviral function, which is likely its original cellular function in the cytoplasm. This can be seen most clearly in BAF-mediated inhibition of poxviruses, such as vaccina virus, where BAF binds cytoplasmic viral DNA to inhibit viral genome replication (Wiebe and Traktman, 2007) and transcription (Ibrahim et al., 2013), likely through the condensation and isolation of the viral DNA from other DNA-binding proteins (Fig. 2). To combat this antiviral role of BAF, vaccinia viruses utilize protein kinase B1, a conserved Ser-Thr protein kinase essential for the viral life cycle (Rempel and Traktman, 1992), which has homology with the metazoan VRKs that phosphorylate BAF (Lee and Craigie, 1998; Nichols et al., 2006; Birendra et al., 2017). The viral B1 kinase phosphorylates BAF to prevent binding with viral DNA (Ibrahim et al., 2013; Jamin et al., 2014b) (Fig. 2). Accordingly, B1 mutagenesis inhibits BAF phosphorylation, increases BAF colocalization with viral DNA and decreases viral production (Ibrahim et al., 2013). The protein phosphatase PP2A appears to counteract B1-induced BAF phosphorylation in infected cells (Jamin et al., 2014b).

There is also a reported antiviral role for BAF in the nucleus. The DNA virus Herpes Simplex Virus type 1 (HSV-1) forms cytoplasmic PICs for nuclear translocation, where it replicates and transcribes its genes (for a review, see Full and Ensser, 2019). Upon HSV-1 infection, cytosolic BAF is dephosphorylated and rapidly localizes to the nucleus, suggesting an ability to sense the nuclear viral DNA (Jamin et al., 2014a). Nuclear BAF has also been shown to interact with an HSV DNA-binding protein (Oh et al., 2015). Although overexpression of wild-type BAF has a minor impact on viral yield, expression of a non-phosphorylatable and nuclear-restricted BAF mutant reduces viral DNA replication and viral protein expression (Jamin et al., 2014a), suggesting that non-phosphorylated nuclear BAF inhibits viral DNA replication and transcription. These studies further support the idea that the antiviral functions of BAF are compartmentally regulated by phosphorylation.

BAF rapidly binds to any cytoplasmic DNA, including experimentally introduced DNA. For instance, in HeLa cells transfected with DNA-coated polystyrene beads, BAF binds to the exposed DNA and assembles an NE-like membrane, potentially enabling the DNA-coated beads to avoid LC3 recruitment and thus autophagic engulfment by recruitment of LEM-domain protein-containing membranes (Kobayashi et al., 2015) (Fig. 2). Furthermore, loss of BAF in mouse microglial cells results in increased levels of cytosolic double-stranded DNA and an enhanced innate immune response (Ma et al., 2020), indicating that BAF may promote degradation of cytosolic DNA and thus suppress innate immune responses elicited by cytosolic DNA. Collectively, this body of evidence suggests that BAF likely modulates innate immune responses that are triggered by cytosolic DNA of foreign or endogenous origin via uncharacterized mechanisms.

The NE is a specialized extension of the endoplasmic reticulum (ER) that separates the nucleoplasm from the cytoplasm during interphase. The NE protects and organizes the genome, and establishes cellular compartmentalization enabling regulation of cell signaling (for more information on the composition of the nuclear envelope, see Box 2). BAF plays essential roles in both mitotic nuclear disassembly and post-mitotic nuclear reassembly.

In order to initiate NE breakdown, BAF must be phosphorylated by VRKs in early mitosis, triggering its release from chromatin and LEM-domain proteins (Gorjánácz et al., 2007; Molitor and Traktman, 2014) (Fig. 3). As the NE disassembles during early mitosis, transmembrane NE proteins disperse into the ER, while soluble NE proteins, such as the lamins, disperse into the cytoplasm (Ungricht and Kutay, 2017). Disruption of BAF phosphorylation during early mitosis stalls the process of nuclear disassembly (Gorjánácz et al., 2007; Molitor and Traktman, 2014). Near the end of mitosis, BAF is rapidly dephosphorylated by PP2A (Asencio et al., 2012; Gorjánácz, 2013) and PP4 (Zhuang et al., 2014). Ankle2 is required to recruit PP2A to BAF for its dephosphorylation (Asencio et al., 2012; Snyers et al., 2018). Absence of PP2A activity (Ahn et al., 2019) or Ankle2 (Asencio et al., 2012; Snyers et al., 2018) causes BAF to remain hyperphosphorylated during mitosis and prevents proper NE reformation. The dephosphorylation of BAF by PP2A and PP4, as well as inhibition of VRK1, triggers binding of BAF to the surface of mitotic chromosomes, which facilitates NE reformation around decondensing chromatin in late anaphase and early telophase (Gorjánácz, 2013).

Depletion of BAF in mitotic cells leads to NE membrane invagination, multinucleation and formation of micronuclei upon mitotic exit (Gorjánácz et al., 2007; Zhuang et al., 2014; Samwer et al., 2017). In both C. elegans and D. melanogaster, loss of BAF causes mitotic phenotypes, including anaphase chromosome bridges and dysmorphic nuclei, and results in embryonic lethality (Zheng et al., 2000; Furukawa et al., 2003; Gorjánácz et al., 2007; Margalit et al., 2005a). These mitotic phenotypes indicate that BAF is integral to the proper reformation of a single nucleus in each daughter cell following cell division. Indeed, post-mitotic recruitment of the NE around chromosomes to form a single nucleus is mediated by BAF interacting with and recruiting LEM-domain proteins and A-type lamins, as well as binding to DNA and compacting chromosomes at the central region of the assembling nuclear rim (called the ‘core’ region) during telophase (Haraguchi et al., 2001, 2008; Samwer et al., 2017) (Fig. 3). Starting in late anaphase, INM proteins, including several transmembrane LEM-domain proteins, emerge from the ER and begin to re-associate with the chromatin (Ulbert et al., 2006; Anderson et al., 2009; Haraguchi et al., 2008). Although LEM-domain proteins can interact directly with chromatin at this core region, non-phosphorylated BAF localized at the core has been suggested to be crucial for the recruitment of these transmembrane LEM-domain proteins and, therefore, the de novo NE formation around this area (Haraguchi et al., 2001; Gorjánácz et al., 2007; Haraguchi et al., 2008; Asencio et al., 2012) (Fig. 3). Interestingly, expression of a BAF mutant unable to bind to LEM-domain proteins in BAF-depleted HeLa cells rescues the multinucleation phenotype, whereas expression of a BAF mutant that is unable to dimerize does not (Samwer et al., 2017). These findings suggest that the role of BAF in nuclear reassembly is more complex than simply recruiting the newly forming NE membrane.

In addition to the core region, BAF accumulation has been observed at late-segregating acentric chromosome fragments and along the tethers that connect acentric fragments to newly forming daughter nuclei (Warecki et al., 2020). BAF enrichment on acentric fragments begins to dissipate only following reincorporation into the primary nucleus (Warecki et al., 2020). The ability of BAF to dimerize while bound to chromosomes drives chromatin condensation and the reincorporation of most of the lagging chromosomal fragments (Warecki et al., 2020). Moreover, in C. elegans, the LEM-domain protein LEM-3 (Ankle1 in mammalian cells) contains an enzymatically active endonuclease domain in its C-terminus (Brachner et al., 2012) and is essential in maintaining genomic integrity during cell division (Dittrich et al., 2012) by ensuring proper resolution of chromatin bridges (Hong et al., 2018). There is evidence suggesting that Aurora B kinase is responsible for phosphorylation and possibly recruitment of LEM-3 to the midbody of the chromatin bridges (Hong et al., 2018). However, through interactions with the LEM-domain of LEM-3, BAF may recruit LEM-3 to the chromatin bridges for their initial resolution, which could facilitate BAF-mediated chromatin condensation into the respective newly forming nuclei (Warecki et al., 2020; Samwer et al., 2017), as is the case for the acentric fragments. Collectively, this suggests that although BAF is partially responsible for recruitment of LEM-domain proteins and A-type lamins, and thus, reconstitution of the NE, the crosslinking of surface DNA between neighboring chromosomes by BAF oligomers ensures the formation of a single nucleus (Warecki et al., 2020; Samwer et al., 2017) (Fig. 3).

Perhaps the most fundamental role for the NE is to form a well-regulated barrier that separates the nuclear and cytoplasmic constituents during interphase. However, upon exposure to mechanical forces (Denais et al., 2016; Hatch and Hetzer, 2016; Zhang et al., 2019b; Halfmann et al., 2019) or intrinsic loss of nucleoskeletal integrity (Chen et al., 2018; Yang et al., 2017), the nuclear envelope can be structurally compromised, leading to nuclear rupture and the loss of nucleocytoplasmic compartmentalization. Under normal conditions, nuclear ruptures are repaired, and, although disputed, evidence suggests proteins from the endosomal sorting complexes required for transport (ESCRT)-III machinery are responsible (Willan et al., 2019). Efficient repair stops the unrestricted exchange of cellular constituents between the cytosol and nucleus, which can lead to DNA damage (Raab et al., 2016; Xia et al., 2018), chromosome rearrangements (Maciejowski et al., 2015; Zhang et al., 2015) and activation of innate immune signaling pathways (Mackenzie et al., 2017), as well as mislocalization of cytoplasmic and nucleoplasmic cellular components (Gupta et al., 2010; De Vos et al., 2011). Nuclear rupture has also been implicated as a potential pathogenic mechanism in several diseases, including laminopathies (De Vos et al., 2011), cancer (Denais et al., 2016) and autoimmunity (Mackenzie et al., 2017).

BAF has been shown to localize to sites of nuclear rupture (Young et al., 2020; Denais et al., 2016; Halfmann et al., 2019), and it is predominantly non-phosphorylated cytoplasmic BAF that is primed to bind to cytosolic DNA during these ruptures (Halfmann et al., 2019) (Fig. 2), in a manner similar to what occurs during some viral infections (Jamin et al., 2014b). Further evidence, using DNA-binding deficient mutants of BAF that do not normally enrich at the rupture site, supports the notion that its DNA-binding affinity is the determining factor for recruitment to ruptures (Halfmann et al., 2019). Furthermore, our studies also indicate that BAF is required to recruit transmembrane LEM-domain proteins and membrane repair proteins, as well as the membranes themselves to sites of rupture for efficient repair (Halfmann et al., 2019) (Fig. 2). As nuclear ruptures repair, BAF dissociates from the rupture site and gradually diffuses into the nucleoplasm and along the NE (Halfmann et al., 2019). This diffusion of BAF is likely facilitated by VRK1-mediated phosphorylation in the nucleoplasm and VRK2A-mediated phosphorylation along the NE (Nichols et al., 2006; Molitor and Traktman, 2014; Birendra et al., 2017; Blanco et al., 2006). Following rupture, the population of cytoplasmic BAF appears to be largely redistributed to the nucleoplasm (Halfmann et al., 2019). Future studies are needed to ascertain how cytoplasmic localization of BAF is restored.

Although nuclear rupture has been implicated in activation of innate immunity (Mackenzie et al., 2017), this correlation has only been seen in the context of micronuclei ruptures and not those of the primary nucleus (Gentili et al., 2019). It has been suggested that BAF may protect the cell from a basal immune response resulting from endogenous cytosolic self-DNA of nuclear and mitochondrial origin (Ma et al., 2020). Collectively, current evidence suggests that BAF is a first-responder to nuclear ruptures to ensure the repair of the rupture; however, in some cases it may also help ensure that an innate immune response is not triggered by nuclear rupture.

Reports indicate BAF may positively and negatively regulate expression of endogenous genes, although the underlying mechanisms remain unclear. In C. elegans, BAF was found to repress the expression of eff-1, which encodes a protein involved in somatic cell fusion, by directly binding to the eff-1 promoter (Margalit et al., 2007). However, BAF binds to DNA in a sequence-independent manner (Zheng et al., 2000; Harris and Engelman, 2000; Umland et al., 2000), making it reasonable to speculate that BAF coordinates with other gene regulatory factors to ‘target’ specific genes. Furthermore, BAF may indirectly act as a transcriptional regulator for several genes through its associations with DNA, thereby bringing BAF into close proximity of other regulators, which it can interact with. For example, BAF indirectly associates with the murine transcription factor cone-rod homeobox factor (Crx) to repress the Crx-dependent promotor (Wang et al., 2002; Huang et al., 2011). Other transcriptional regulators that associate with BAF as determined by affinity purification include p15 (also known as SUB1 and PC4), NonO, Requiem (also known as DPF2) and LEDGF (also known as PSIP1) (Montes de Oca et al., 2009), and they may contribute to a BAF-dependent regulation of gene expression. Further evidence for a role of BAF in indirectly regulating gene expression comes from a report that mouse embryonic stem cells (ESCs) depleted of BAF display reduced expression of the ESC pluripotency-associated transcription markers Sox2, Oct4 and Nanog, as well as an increase in mesoderm and trophectoderm markers (Cox et al., 2011). Additionally, both mouse and human ESCs exhibit decreased cell survival following BAF depletion (Cox et al., 2011), indicating that BAF is vital for maintaining the viability of pluripotent cells early in development, perhaps through its requirement for maintaining viability of all dividing cells. In human skin, BAF expression is predominantly nuclear in the proliferative basal layers of epidermis, but is shifted to the cytoplasm in the upper epidermal keratinocytes at the surface of the skin (Takama et al., 2013). Although the mechanisms mediating BAF redistribution are unclear, the shift in BAF localization along with its reported role in maintaining cells in a pluripotent state suggests that a predominant nuclear BAF population is critical in pluripotent cells, likely functioning in genomic regulation.

BAF also interacts with proteins that indirectly regulate transcription, such as A-type lamins (Bermeo et al., 2015; Zhang et al., 2019a; Flint Brodsly et al., 2019; Margalit et al., 2005a), LEM-domain proteins (Holaska et al., 2003; Margalit et al., 2005a) and histones (Montes de Oca et al., 2005, 2011). In these instances, DNA could be initially ‘targeted’ by lamin-anchored LEM-domain proteins, before coming in close contact with BAF (Segura-Totten and Wilson, 2004). Conversely, these BAF–protein interactions could be mediated by nonspecific binding of BAF to DNA that enables BAF-mediated recruitment of regulatory proteins. It is also important to note that BAF competes with the conserved transcription repressor germ cell-less (GCL; also known as GMCL1) to bind to the LEM-domain protein emerin in vitro (Holaska et al., 2003), suggesting that BAF could also have regulatory roles at a protein–protein level, rather than only a DNA–protein level. Histones may also facilitate specific BAF–DNA interactions, as BAF binds to histones H1.1, H3 and H4 (Montes de Oca et al., 2005, 2011). Furthermore, BAF overexpression alters epigenetic histone modifications, increasing histone hypermethylation and decreasing histone acetylation, potentially decreasing the expression of genes throughout the genome (Montes de Oca et al., 2011; Oh et al., 2015). Thus, although the molecular mechanisms remain unclear, BAF might also be an important epigenetic regulator, perhaps by influencing histone markers utilized to regulate transcription.

Several BAF-interacting proteins are involved in DNA damage responses (Montes de Oca et al., 2009; Moser et al., 2020), and both BAF and the LEM-domain protein emerin interact with the DNA damage response proteins of the CUL4–DDB–ROC1 complex, specifically DDB1, DDB2, CUL4A and PARP1 (Montes de Oca et al., 2009; Bolderson et al., 2019). This complex is required for the repair of DNA single-strand breaks (Pines et al., 2012). Poly(ADP-ribosyl)ation of DDB2 by PARP1 is important in regulating the stability of the complex, as well as its chromosome retention time (Pines et al., 2012). Upon UV exposure, the association of BAF or emerin with DDB1, DDB2 and CUL4A increases over time (Montes de Oca et al., 2009). Additionally, BAF colocalizes with PARP1 upon induction of single-strand breaks through oxidative damage with H₂O₂. PARP1 activity is negatively regulated by the binding of BAF to the NAD⁺-binding domain of PARP1 (Bolderson et al., 2019). In fact, BAF is the only identified PARP1 regulator that inhibits PARP1 activity by binding to this domain. Another DNA repair protein that may be influenced by BAF is the LEM-domain protein Ankle1, which has endonuclease activity. Accumulation of Ankle1 in the nucleus causes DNA cleavage and activation of DNA damage response pathways in a LEM-domain-dependent manner (Brachner et al., 2012). Although BAF has not been shown to directly interact with the Ankle1 LEM domain, it is plausible that BAF may facilitate DNA targeting and retention of Ankle1. Regardless, it appears that BAF functions to regulate DNA damage repair through its association with various repair factors, perhaps by helping target repair factors to damage and/or regulating their activity.

In D. melanogaster, homozygous deletion of baf results in lethality at the larval-pupal stage with evidence of mitotic defects (Furukawa et al., 2003). Survival until this developmental stage is likely enabled by maternally derived BAF mRNA and protein (Furukawa et al., 2003). In contrast, RNAi knockdown of C. elegans BAF-1 leads to early embryonic death with a profound mitotic defect (Zheng et al., 2000; Margalit et al., 2005b), perhaps because the RNAi inhibits new BAF production from maternally derived mRNA. For embryos that escape lethality, likely due to incomplete RNAi-mediated knockdown, there were defects in the positioning of gonads and distal tip cells (Margalit et al., 2005b). In baf-1-null C. elegans, maternally derived mRNA and protein are thought to enable development to early adulthood with an absence of notable mitotic defects (Margalit et al., 2007). However, a number of cell-type-specific roles of BAF are evident in this model, such as a considerably smaller body size, deterioration of muscle cell integrity, premature fusion of epidermal seam cells, failure to form vulva, defective gamete formation and mislocalized distal tip cells (Margalit et al., 2007). Collectively, these results suggest that BAF is critical for the process of cell division, but that it also likely has a post-mitotic role in the differentiation, regulation and maintenance of cells in developing and adult tissues.

With the variety of roles for BAF in multiple cellular processes, it is not surprising that BAF is increasingly being associated with human disease. An autosomal recessive mutation of BANF1 (c.34G>A [p.A12T]) in two separate unrelated individuals has been reported to cause an atypical progeroid syndrome called Néstor–Guillermo progeria syndrome (NGPS) (Puente et al., 2011; Cabanillas et al., 2011). NGPS patients share clinical features, including failure to thrive, lipoatrophy, osteoporosis, pseudosenile facial appearance and normal cognition, with the more characterized Hutchinson–Gilford progeria syndrome (HGPS), which is predominantly caused by mutations in LMNA (Puente et al., 2011; Cabanillas et al., 2011). However, NGPS is described as a chronic progeria due to the longer life span of patients and milder symptoms (Puente et al., 2011; Cabanillas et al., 2011). Furthermore, NGPS patients do not display signs of coronary dysfunction, atherosclerosis or metabolic complications, all hallmarks of HGPS. At the cellular level, NGPS patients exhibit nuclear abnormalities, such as blebbing and aberrations, consistent with those described in other progeric laminopathies (Puente et al., 2011). Interestingly, heterozygous individuals display normal phenotypes (Cabanillas et al., 2011), and ectopic expression of GFP–BAF in NGPS patient fibroblasts rescues the nuclear abnormalities (Puente et al., 2011), suggesting that one BANF1 allele is sufficient for proper functionality, although the A12T NGPS mutation does not appear to be a complete loss of function.

The Ala12 residue mutated in NGPS is highly evolutionarily conserved (Puente et al., 2011), indicating it may play an important role in BAF structure and/or function. Structural analyses indicate that the A12T mutation does not alter the secondary or tertiary structure of BAF (Paquet et al., 2014), nor is the mutation in a region that would inhibit BAF dimerization (Samson et al., 2018). Additionally, localization of the A12T mutant to the nuclear envelope is unaltered compared to wild-type (WT) protein (Paquet et al., 2014). The change from alanine to a bulky threonine residue could potentially interfere with the binding partners of BAF. Indeed, in an in vitro DNA-binding assay, the A12T mutant bound DNA less efficiently than its WT counterpart (Paquet et al., 2014). In addition, the location of the mutation may alter the association with the Ig-like fold of A-type lamins (Loi et al., 2016; Samson et al., 2018). Furthermore, accumulation of unrepaired DNA damage and increased cellular senescence are phenotypic hallmarks of HGPS (Kubben and Misteli, 2017; Romero-Bueno et al., 2019), and BAF has been shown to interact with DNA repair proteins (Montes de Oca et al., 2009; Bolderson et al., 2019). This suggests that the NGPS mutation could result in abnormal DNA repair and accumulation of DNA damage as seen in HGPS (Romero-Bueno et al., 2019). In fact, the A12T mutant binds more strongly to and inhibits the DNA repair mechanism mediated by PARP1, and NGPS patient fibroblasts exhibited defective DNA repair of induced oxidative lesions (Bolderson et al., 2019). Importantly, depletion of BAF in NGPS patient cells was sufficient to restore repair of oxidative lesions (Bolderson et al., 2019). Clearly, there are several possible mechanisms by which the mutation in BAF could lead to NGPS.

Recently, BAF expression levels have been implicated in certain cancer prognoses. Increased expression in breast (Lai et al., 2010), esophageal (Li et al., 2017) and gastric cancer (Li et al., 2018) are all associated with poor prognosis in patients. In triple-negative breast cancer, BANF1 mRNA overexpression positively correlated with proliferative and metastatic markers (Zhang, 2020). These recent studies potentiate the use of BAF expression levels as a novel marker for cancer diagnoses and prognoses and provoke questions as to the possible role(s) of elevated BAF in cancer.

Although BAF is a small non-enzymatic protein, its well-regulated coordination with numerous binding partners and ability to bind and condense DNA make it critical for several key cellular processes. During interphase, BAF is responsible for recognizing and binding cytosolic DNA, repairing nuclear ruptures and genomic regulation (Fig. 2). Following mitosis, BAF both condenses chromatin and recruits newly forming nuclear membranes around the chromatin to enable the formation of a single nucleus with a functional NE (Fig. 3).

Because of these diverse roles and the phenotypes that arise when its expression is altered, the original evolutionary purpose of BAF remains unclear. Did BAF evolve as a sensor for intrinsic immunity or for genomic integrity? Perhaps BAF is a consequence of multicellularity in cells lacking a protective cell wall to enable the repair in nuclei that were increasingly susceptible to nuclear rupture from extrinsic forces? Or maybe BAF was necessary for efficient reformation of nuclei following open mitosis? Whatever the original evolutionary need for this unique protein, further characterization of BAF and its regulators is likely to inform on critical cellular processes and reveal cellular mechanisms of disease.

We thank Charles Halfmann and Kelsey Scott for their discussions and feedback on the manuscript.