In the eukaryotic cell nucleus, chromatin forms 3-D structures at multiple levels, including domains, loops, A and B compartments (relating to active and inactive chromatin, respectively), topologically associating domains (TADs) and territories (13-16).

In early research, SAF-A was found to form large aggregates in vitro and induce DNA loops to gather around them (17) (Fig. 2A). When SAF-A is depleted in mouse hepatocytes, lamina-associated domains increase, compartments switch, TAD boundary strengthens, and chromatin loop intensities decrease, demonstrating that SAF-A plays a vital role in maintaining the higher-order chromatin organization (18).

Further studies demonstrated that SAF-A regulates chromatin structure via RNA. Jiao et al (19) found that SAF-A binds to histone acetyltransferases p300 in cancer cells (19). Facilitated by heparanase (HPSE) enhancer RNA, the SAF-A-p300 complex is enriched on the super-enhancer, consequently leading to chromatin looping between the HPSE promoter and the super-enhancer. Consistently interacting with chromatin-associated RNAs, SAF-A forms oligomerization that induces de-compaction of large-scale human interphase chromatin structure (8) (Fig. 2A). In this way, SAF-A maintains genomic stability (8). When human monocyte THP-1 cells are infected by vesicular stomatitis virus (VSV), SAF-A interacts with viral infection-induced RNAs, mediating the openness and activation of antiviral immune genes (20). Puvvula and Moon (10) treated cancer cells with cell-penetrating peptides derived from SAF-A, and found that the SAP-derived peptide rather than the RGG-derived one promotes chromatin compaction in HCT116 (colorectal), T47D (breast) and UMUC3 (bladder) cancer cells. Based on these observations, RNA is proposed to be essential for SAF-A-induced chromatin de-compaction.

Besides RNA, DNA was also reported to be necessary for SAF-A to regulate chromatin structure. In the study of Kolpa et al (21), similar to the C280 SAF-A deletion mutant (only containing RGG domain), the expression of ∆RGG (containing DNA-binding domain) and G29A mutants (in lack of DNA-binding ability) was able to release C0T-1 RNA from chromatin, resulting in chromatin condensation. Despite this, their regulatory mechanism is different. ∆RGG SAF-A mutant mainly replaced endogenous SAF-A from chromatin, while the G29A mutant still combined with C0T-1 RNA and could not bind to chromatin (21). Therefore, to some degree, SAF-A may play a role in bridging C0T-1 RNA and chromatin to regulate chromatin architecture. Depending on the SAP domain, SAF-A was also found to combine with MARs and the pericentromere tandem repeats in chromocenters (6,22). Therefore, it was hypothesized that SAF-A tethers MARs to the chromocenter to organize nuclear architecture (1).

In eukaryotic cells, transcription of protein-coding genes, including initiation, elongation and termination phases, contains numerous regulatory proteins affecting either the RNA polymerase II (Pol II) machinery or chromatin structure (23-26).

Previous studies have indicated that SAF-A is involved in initial transcriptional regulation. Actin in the nucleus regulates transcription by remodeling chromatin, regulating transcription factor (TF) location, or binding to RNA polymerase (27-30). Through extracellular vesicles derived from embryonic stem cells, SAF-A is transferred into human coronary artery endothelial cells to combine with actin. The SAF-A-actin complex leads to enhanced RNA Pol II phosphorylation and its level on vascular endothelial growth factor (VEGF) promoter, upregulating VEGF expression (4) (Fig. 2B). When the SAF-A-actin complex is disrupted by H19 [a long non-coding (lnc) RNA], the phosphorylation of the Pol II C-terminal domain (CTD) is inhibited, and consequently, Pol II-mediated transcription is prohibited (31). Additionally, SAF-A was found to associate with elements within the promoter regions. In Sertoli cells, SAF-A has been identified to bind directly to the promoter regions of Sox8 and Sox9, thereby enhancing their expression (32) (Fig. 2B). Similarly, IL21-anti-sense RNA 1 interacting with SAF-A binds to the IL21 promoter, which is essential for regulating IL21 transcription (33). Research on embryonic stem cells also revealed that not only does SAF-A bind to the Oct4 proximal promoter, but it also interacts with endogenous Pol II. And depletion of SAF-A impairs Oct4 expression (34).

In previous studies, SAF-A appeared to have two-way adjusting effects on transcription elongation mediated by Pol II. According to the study of Obrdlik et al (35), SAF-A, actin and p300/CBP-associated factor (PCAF) were associated with the phosphorylated Pol II CTD; and the actin-SAF-A interaction assisted Pol II transcription elongation depending on PCAF (Fig. 2B). On the contrary, in another study, SAF-A inhibited Pol II elongation. Through the middle domain, SAF-A was sufficient to combine with Pol II and repressed TF IIH-mediated Pol II CTD phosphorylation, which inhibited Pol II elongation (36) (Fig. 2B).

SAF-A has also been implicated in various aspects of RNA metabolism, including normal and alternative pre-mRNA splicing and RNA transporting (Fig. 2C).

Ye et al (37) inactivated SAF-A expression in murine hearts and found that compared with wild type, SAF-A mutant in hearts results in extensive intron retention and cassette skipping. In contrast to the wild-type samples, mouse cortices with SAF-A mutations showed 850 differentially spliced genes, and the gene splicing included exon skipping, an alternative 3' or 5' splice site and intron retention (38).

Then, by what mechanism does SAF-A regulate pre-mRNA splicing? Xiao et al (39) found that SAF-A binds to all the small nuclear RNAs essential for splicing major and minor intron classes and regulates U2 small nuclear ribonucleoprotein maturation and Cajal body morphology. However, Vu et al (40) demonstrated the different splicing regulatory mechanisms of SAF-A. Competed with hnRNP L (a splicing repressor) to combine with an RNA cis-element in the exon 3 of caspase-9, SAF-A promotes the exon 3, 4, 5, 6 cassette inclusion in the mature caspase-9 mRNA, determining the preferential expression of caspase-9a rather than caspase-9b.

It has been identified that SAF-A is responsible for transporting and stabilizing mRNA via directly binding to it. Zhao et al (41) found that SAF-A binds to tumor necrosis factor α (TNF-α), IL-6 and IL-1β mRNAs and that downregulation of SAF-A expression in macrophages induces the decreased half-life of these cytokine mRNAs. Besides, Toll-like receptor signaling in macrophages leads to SAF-A translocation from nuclear to cytoplasm (41). During this translocation, SAF-A is proposed to stabilize pro-inflammatory cytokine mRNAs (41). In agreement with the aforementioned study, SAF-A was found to bind to the 3' untranslated region of TNF α mRNA to stabilize it and specifically enhance TNF α expression (42). However, other studies suggested that SAF-A stabilizes mRNA depending on other molecules. Pan et al (43) found that interaction between LIM domains-containing 1 antisense RNA 1 (LIMD1-AS1) and SAF-A is essential for stabilizing LIMD1 mRNA in non-small cell lung cancer. Furthermore, Lu et al (44) reported that following lipopolysaccharide stimulation in human intestinal epithelial cells, functional intergenic repeating RNA element cooperating with SAF-A may regulate the stabilization of vascular cell adhesion protein 1 and IL12p40 mRNAs through targeting the AU-rich elements of these mRNAs.

In cells, ionizing radiation (IR) and reactive oxygen species may result in clustered oxidized bases and DSBs that are also induced by V(D)J recombination (45-47). Oxidized base lesions in the human genome are repaired via DNA glycosylase (DG) repair, also known as base excision repair (BER). Repair enzymes, including Nei endonuclease VIII-like (NEIL) and certain non-repair proteins, such as SAF-A, initiate this process (48,49). In previous studies, SAF-A was demonstrated to interact directly with NEIL1 and stimulate NEIL1-mediated repair of oxidative base damage (48,49). In addition to NEIL1, SAF-A combines with NEIL2, enhancing NEIL2-initiated transcribed gene-specific repair of oxidized bases (50) (Fig. 2D).

However, BER of bi-stranded oxidized bases can lead to additional DSBs, causing loss of DNA fragments. This may be avoided if DSBs are repaired through non-homologous end joining (NHEJ) mediated by Ku, DNA-dependent protein kinase catalytic subunit (DNA-PKcs), X-ray repair cross-complementing protein 4 (XRCC4), DNA ligase IV, and cernunnos-XRCC4-like factor complex, preceding BER (51-53). In response to irradiation-induced DSBs, SAF-A is rapidly recruited to DNA damage sites (54) (Fig. 2D). Besides, DSBs cause phosphorylated SAF-A at Ser59, which is exclusively dependent on DNA-PK (55,56) (Fig. 2D). In NHEJ-deficient cells treated with IR or clastogenic drugs, the extent and duration of SAF-A phosphorylation at Ser59 increase (56), which suggests that phosphorylated SAF-A may be a biomarker for testing the capacity of the cells to repair DSBs by NHEJ. A previous study showed that SAF-A phosphorylation at Ser59 is related to transient NEIL1 release from chromatin and BER prevention. Then, due to dephosphorylation, SAF-A reactivates BER by relieving DG inhibition (57). These results suggested that DSB repair by NHEJ or BER is partly determined by whether SAF-A is phosphorylated. However, a recent study found that increased SAF-A stabilizes or recruits NHEJ factors via liquid-liquid phase separation, including Ku80, 53BP1, DNA-PKcs and the shieldin complex at DSBs during antibody class-switch recombination (58) (Fig. 2D). The finding supplies SAF-A may balance between BER and NHEL through complex mechanisms.

During mitosis, kinetochore-microtubule (MT) attachment and chromosome congression at the spindle equator is essential to accurate chromosome segregation (59,60). Interestingly, SAF-A binds to not only Aurora-A and targeting protein for Xklp2 (TPX2) at spindle MTs, but also nucleolin at the outer kinetochore, whereby SAF-A stabilizes kinetochore-MT attachment (61) (Fig. 2E). Via non-coding RNA, SAF-A interacts with centromere protein W (CENP-W, an inner centromere component crucial for forming a functional kinetochore complex). Notably, SAF-A-CENP-W interaction increases their stability, and they co-localize at the MT-kinetochore interface during mitosis (62) (Fig. 2E). Furthermore, SAF-A depletion in cells leads to delayed mitosis, unsuccessful chromosome alignment, and spindle assembly (61).

In addition, SAF-A is dynamically phosphorylated during mitosis, and in human cells, phosphorylation at serine 2 (S2), S3, S4, S59, S66 and S270 is related to mitosis (63-65). Douglas et al (66) found SAF-A is phosphorylated at S59 by polo-like kinase 1 instead of DNA-PK and is dephosphorylated by protein phosphatase 2A. Mutations of SAF-A S59 in cells lead to aberrant mitoses, including polylobed nuclei, delayed passage, misaligned and lagging chromosomes (66). However, the regulatory mechanism of phosphorylated SAF-A on mitosis remains elusive.

Previous studies revealed that SAF-A mainly promotes proliferation (67-71), aggressiveness (19) and migration (71) of tumors (Table I). SAF-A is overexpressed in hepatocellular carcinoma or acute myeloid leukemia, and the proliferation of these tumors is inhibited in vitro and in vivo due to SAF-A downregulation (68-70). Further studies demonstrated that SAF-A promotes tumor proliferation through combining with lncRNAs, such as promoter of CDKN1A antisense DNA damage activated RNA (PANDA) in esophageal squamous cell carcinoma and hepatocyte nuclear factor 4 alpha-AS1 in neuroblastoma (67,69). Jiao et al (19) also used several cancer cell lines and found that SAF-A interaction with HPSE enhancer RNA enhances HPSE expression and activity, which promotes tumorigenesis and aggressiveness. In the research of Han et al (71), SAF-A was found to interact with dead box helicase 5 (DDX5) to promote the proliferation and migration of triple-negative breast cancer. On the one hand, the SAF-A-DDX5 complex leads to alternative splicing (AS) of minichromosome maintenance protein 10 pre-mRNA and subsequent activation of Wnt/β-catenin signaling. On the other hand, the complex enhances the expression of LIM domain only protein 4 by locating in its transcriptional start sites, activating PI3K-Akt-mTOR signaling (71). Regulating AS, SAF-A is involved in the pathogenesis of gastric cancer and renal clear cell carcinoma (72,73). Zhang et al (74) found that SAF-A also binds to P-element-induced wimpy testis-interacting RNA-1742 (piRNA-1742) to stabilize USP8 mRNA and promote the progression of renal cell carcinoma. Through interaction with a family with sequence similarity 171 B (an independent prognostic factor for bladder cancer progression), SAF-A regulates the CCL2 pathway to facilitate M2 macrophage polarization in the tumor microenvironment, thereby promoting bladder cancer development (75).

Further studies have demonstrated a correlation between SAF-A and resistance to chemotherapy. Shi et al (76) discovered that the knockout of SAF-A in human T24 cells enhances bladder cancer susceptibility to cisplatin via inhibition of cell proliferation and impairment of cellular invasion and migration capabilities. In the research of Wang et al (77), the knockdown of SAF-A increases selinexor sensitivities of multiple myeloma (MM) cells in vitro and in mouse models and MM patients with relatively low SAF-A expression response to selinexor.

However, a few studies demonstrated that SAF-A might suppress tumor cell proliferation and enhance sensitivities to chemotherapy. Pan et al (43) reported that LIMD1 inhibits proliferation and promotes apoptosis in non-small cell lung cancer cells, and SAF-A interacts with lncRNA LIMD1-AS1 to stabilize LIMD1 mRNA. Puvvula and Moon (10) dealt with various tumor cells with SAF-A RGG- or SAP-derived peptides. Depending on different mechanisms, these peptides suppress breast, bladder, colorectal and prostate cancer cell proliferation. The SAF-A RGG-derived peptides mainly alter SAF-A splicing and binding targets, while the SAP-derived regulates global epigenetic marks to induce DNA damage reaction and cell death (10). Li et al (78) also identified that lncRNA SFTA1 could augment cisplatin sensitivity in treating lung squamous cell carcinoma by upregulating SAF-A.

The contradictory effects of SAF-A on tumors may result from different tumor types or test methods. Therefore, more animal and clinical experiments are needed to identify its functions further.

Previous research revealed that SAF-A plays dual roles in antiviral response. Through the N-terminal fragment (aa1-86), SAF-A targets the 3' long terminal repeat of human immunodeficiency virus type 1 mRNA and blocks viral replication in cells (79). Furtherly, in both in vitro and in vivo animal experiments, SAF-A has been reported to be associated with the initiation of innate immunity against viruses, including severe fever with thrombocytopenia syndrome virus, VSV, and herpes simplex virus type 1 by recognizing nuclear or cytoplasmic viral RNA (12,20,80).

However, Gupta et al (81) found that SAF-A is associated with the leader RNA of VSV and co-localizes with the virus in the cytoplasm of infected cells, implying that SAF-A may contribute to the life cycle of VSV and its pathogenesis. In recent studies, SAF-A was found to negatively regulate innate immune responses against infectious bursal disease viruses and porcine epidemic diarrhea virus and induce immune escape of these viruses (82,83).

Therefore, the diverse roles of SAF-A played in these experiments may be due to different viruses or hosts, and more animal experiments are needed to determine the roles of SAF-A in antiviral immunity.

Caliebe et al (84) first revealed that patients with a chromosome deletion 0.440 Mb in region 1q44 bearing SAF-A suffer from seizures and speech delay. Similarly, Depienne et al (85) reported that 1q43q44 containing SAF-A microdeletion leads to intellectual disability (ID) and epilepsy. Employing molecular genetic testing, a series of pathogenic SAF-A mutants have been identified to be associated with developmental delay and neurodevelopmental disorder showing severe ID with delay of speech and language, early-onset seizures, and autistic features as well (85-90). Despite this, the regulatory mechanism of SAF-A on neurodevelopment remains elusive. In a recent study, deletion of SAF-A in cultured neural progenitors and murine brains resulted in varied gene expression and AS, apoptosis of neural cells and abnormalities in neuronal migration, suggesting that SAF-A is essential for the development of cortex (38).