THE ROLE OF SON IN SPLICING, DEVELOPMENT AND DISEASE

Abstract

SON is a nuclear protein involved in multiple cellular processes including transcription, pre-mRNA splicing and cell cycle regulation. Although SON was discovered 25 years ago, the importance of SON’s function was only realized recently when its roles in nuclear organization and pre-mRNA splicing as well as the influence of these activities in maintaining cellular health were unveiled. Furthermore, SON was implicated to have a key role in stem cells as well as during the onset of various diseases such as cancer, influenza, and hepatitis. Here we review the progress that has been made in studying this multi-functional protein and discuss questions that remain to be answered about SON.

DISCOVERY OF SON: A HISTORICAL PERSPECTIVE

Structure and conservation of SON

The SON gene spans 34,463bp on human chromosome 21 and is composed of 23 exons encoding transcripts of various lengths ranging from ~1.5kb to ~8kb. The protein encoded by SON contains a unique long stretch of repetitive nucleotide sequences, which were initially identified in human and later found to be conserved in SON across multiple species. Discovery of the SON gene came in 1988 when a partial clone called SON3 was isolated from a human embryonic cDNA library and noted as having unusual sequence structure¹. SON3 showed homology to the chicken gallin as well as to human C-MYC and MOS oncogenes¹. Due to this relatedness to DNA binding proteins, the protein encoded by SON gene was named ‘SON DNA binding protein’. Continuation of this work to isolate SON from a human placenta cDNA library demonstrated in 1992 that multiple mRNAs having alternate 5′ ends were produced from the SON gene².

Another group subsequently mapped the genomic location of the human SON gene to a region neighboring GART in the D142H8 YAC clone³. Meanwhile, the mouse homolog of SON was located in a 4-gene cluster with GART, INFAR and CFR2-4⁴. Interestingly, it was shown that gene pairs (CRF2-4, INFAR) and (GART, SON) maintained synteny in human and mouse⁴. SON was later found to be highly conserved, and related sequences are present in in multiple species. SON transcripts of various lengths are expressed differentially in cell lines⁵. Although multiple other SON clones have been documented, including BASS1 (Bax antagonist selected in saccharomyces 1)⁶ and KIAA1019⁷, the full length gene structure of SON was not documented until 2000, when Wynn et al. mapped the SON gene locus and its neighboring genes (GART and DONSON) in both mouse and human⁸. The mouse SON gene is composed of 23 exons encoding a 7215 nt-long ORF⁸. A later study mapped human chromosome 21 transcripts and found that human SON encodes 6 alternative transcripts with the longest of these containing a 7,281 nt-long open reading frame (ORF)⁹.

Determining the structure of the SON gene was met with challenges due to the length of this gene and its repeat-rich nucleotide sequence. The short transcripts initially identified from the SON gene could be attributed to fragmentation of RNA or difficulty in cloning. It will be interesting to use contemporary technology to confirm the existence of only a single full length transcript from the SON locus or to reveal bona-fide shorter transcripts that may have specific roles in human development and/or diseases.

Expression and function of SON

Hints about the potential functions of SON were initially based on its primary sequence and cellular expression. Pioneering studies revealed the ubiquitous but variable expression pattern and putative molecular roles for SON^{5, 10}. Notably, a high expression level in fetal tissue implicated a role of SON in proliferation and/or differentiation of cells during embryonic development⁴. Although SON was initially considered to be a DNA binding protein, the demonstration of its specific localization in nuclear speckles (also called splicing factor compartments) later implicated a role for SON in pre-mRNA processing. Moreover, the similarity of motif structures among SON and serine/arginine-rich (SR) splicing factors further supports that SON plays a role in splicing.

The first piece of experimental evidence that SON acts as a DNA binding protein came from a screening of factors that interact with the X-box consensus sequence (CCTAGCAACAGATGCGT) of a human leukocyte antigen (HLA) Class II promoter¹⁰ where partial SON was cloned from cDNA library as DBP-5. It was noted that the ~7.5 kb ubiquitously expressed DBP-5 transcript encodes a protein containing multiple repetitive motifs¹⁰. Immunostaining of HeLa cells with an antibody raised against DBP-5 revealed its punctate nuclear localization, suggesting a role of DBP-5 in transcription¹⁰. In 1994, Khan et al.⁵ isolated a SON cDNA clone during a screen to identify DNA binding proteins that regulate involucrin expression in keratinocytes⁵. This isolated cDNA clone encoded a partial sequence of SON protein that included a 260 a.a. basic region⁵ containing Ser-Pro-Lys(Arg)-Lys(Arg) repeats found in a variety of DNA binding proteins such as mouse c-Myc and Histone H1¹¹.

Additional motifs later found in both human and mouse SON proteins⁸ pointed toward a role for Son in pre-mRNA processing. Perhaps the most interesting feature noted in the primary sequence of SON is its multiple unique repetitive sequence motifs of unknown function that span nearly 1,200 amino acids. SON also contains an RS domain that is enriched in serine (S)/arginine (R) dipeptides⁸ and typically serves to localize SR protein splicing factors to nuclear speckles¹², that are nucleoplasmic clusters enriched with pre-mRNA splicing complexes^{13, 14}. The presence of an RS domain⁸, the subnuclear localization reported earlier for DBP-5¹⁰, and the confirmed localization of SON in nuclear speckles with pre-mRNA processing factors^{8, 15, 16} all hinted at pre-mRNA splicing functions for SON. Further support for this hypothesis came from other studies that identified a double-stranded RNA binding domain and G-patch domain in SON^{9, 15}, both of which are frequently found in pre-mRNA processing proteins¹⁷.

Taken together, important insights gained from resolving the gene structure of SON included its putative molecular functions in pre-mRNA processing as well as the potential structural organization or scaffolding role of novel repeat motifs not previously found in any other mammalian proteins. Uncovering the functions of these repeats and how they are post-translationally regulated is a current challenge. While SON was initially described as DNA-binding protein, we discuss below how it was later shown that SON has RNA binding ability and is involved in splicing. It is possible that both RNA and DNA binding activities are important for SON’s role in regulating gene expression, since transcription and splicing are coupled and co-regulated¹⁸.

SON REGULATES PRE-MESSENGER RNA SPLICING

SON in nuclear organization

Proteomic analysis of nuclear speckles isolated from mouse liver nuclei identified SON among nearly 200 other proteins^{15, 19}. The majority of factors identified in purified nuclear speckles were pre-mRNA processing proteins, but also included subunits of RNA polymerase II and several transcription factors^{15, 19}. Later, a study that aimed to identify ‘nuclear architecture’ components for nuclear organization and higher order chromatin structure by mass spectrometry, found SON among 502 proteins in insoluble nuclear fractions from HeLa extracts²⁰. Nuclear speckle localization and motif structure of SON described by these studies further suggested that SON may play a role in pre-mRNA splicing. The first comprehensive investigation on the role of SON in nuclear speckles was done by Sharma et al²¹, who assembled the first full-length SON cDNA encoding all predicted motifs, and confirmed the nuclear speckle localization of yellow fluorescent protein (YFP)-SON in HeLa cells²¹. Two antibodies raised against SON showed the co-localization of SON with the splicing factor SRSF1 (previously called SF2/ASF)²¹. In addition, other groups found that SON could directly interact with spliceosome components SRSF1 and PRP8¹⁶, and with SRrp53²². These suggest a direct role of SON in nuclear speckle organization and pre-mRNA splicing. The absence of SON in HeLa cells led to reorganization of many nuclear speckle constituents including the U1 snRNP protein U1-70K, splicing factor SRSF1, exon junction complex (EJC) components, as well as nuclear polyadenylated RNA²¹. Disrupting nuclear speckle organization, either by SON depletion²¹ or by hyperphosphorylation of SR proteins²³ did not alter global transcription. Expression of siRNA-resistant SON rescued nuclear speckle organization in SON-depleted HeLa cells²¹. Expression of various partial SON constructs to rescue depletion of the endogenous protein showed that the unique repetitive domain of SON is essential for its role in organizing splicing factors and polyadenylated RNA in nuclear speckles²¹.

Interestingly, SON’s impact on a nuclear-retained long noncoding RNA called MALAT1 (metastasis-associated lung adenocarcinoma transcript 1; also called NEAT2) was somewhat different than its effect on other nuclear speckle components. MALAT1 is proposed to be a ‘molecular sponge’ that interacts with SR splicing factors and helps to regulate their local availability and activity during alternative splicing of nascent transcripts²⁴. Interestingly, SON repeats are not sufficient for the organization of MALAT1 in nuclear speckles, since SON deletion mutants having the repeats but missing the G-patch and RS domain did not retain MALAT1 in nuclear speckles²⁴. Altogether these data revealed a critical role for SON in organizing not only the protein components but also the RNA components of nuclear speckles, and further implied that SON has a central role in regulating pre-mRNA splicing.

How SON depletion results in dramatic reorganization of pre-mRNA processing factors within mammalian cell nuclei is not known, but it was initially proposed that SON repeats provides a network or scaffold within nuclear speckles to provide a landing pad for RNA processing proteins in these nuclear domains²¹. Additionally, the interactions of SON with EJC proteins reveal a potentially broad interactome for SON as part of mRNP complexes²⁹. Recent evidence that overexpression or knockdown of other nuclear speckle proteins (in particular, certain EJC proteins) leads to nuclear speckle reorganization necessitates a somewhat revised model for SON’s role in nuclear speckle assembly. For example, exogenous expression of EJC factors show similar localization of nuclear speckles as seen in SON-depleted nuclei^{25, 26}. A well-established model for nuclear speckle organization is based on self-organization of components via self-interaction domains²⁷. Self-organizing complexes might include higher-order assembly onto docking sites at extensive repeat motifs in proteins such as SON. This process could scaffold or distribute processing factor complexes throughout speckles via multi-molecular binding. However, intermolecular interactions within pre-mRNA processing complexes could potentially become unbalanced by increased or decreased abundance of individual components, leading to reorganization of pre-mRNA processing factors within nuclear speckles as seen following SON depletion. Absence of such molecular “glue” (i.e. SON) would presumably dislodge self-organized complexes without causing their complete disassembly, and may explain why enhanced distribution of EJC components and other splicing factors is observed at perispeckle regions following SON depletion²¹.

SON in pre-mRNA splicing

SON’s role in pre-mRNA splicing has been demonstrated in multiple studies^28–30. By profiling the expression change after SON knockdown in HeLa cells, microarray analysis by Ahn et al. uncovered 659 genes whose expression was affected²⁸. Examination of downregulated transcripts showed splicing defects at specific introns. Through ultraviolet crosslinking and immunoprecipitation (CLIP), it was demonstrated that SON directly interacts with these transcripts²⁸. Moreover, this study discovered that the RS and G-patch domains are responsible for the splicing activity of SON, whereby the unique repetitive motifs of SON further enhanced the activity of SON in splicing²⁸. This study also demonstrated that SON-regulated introns contain weak or dual-specificity splice sites. Since the absence of SON selectively disrupts the association of splicing factors SC35 with U1-70K and U2AF65, as well as Ser2-phosphorylated RNAP II, SON may regulate co-transcriptional splicing²⁸.

SON is not only involved in constitutive splicing but also plays a role in alternative splicing of exons. The first link between SON and alternative splicing was established by Moore et al³¹, who discovered that SON was a candidate for regulators of Bcl-x and Mcl1 alternative splicing in a genome-wide RNA interference (RNAi) screen. The relationship between SON and alternative splicing was more comprehensively studied later by Sharma et al.²⁹, who showed that SON accumulated at constitutively active β-tropomyosin minigene reporter transcription sites together with other pre-mRNA splicing factors such as SRSF1 and U1-70K further supporting a co-transcriptional role of SON in splicing. In addition, the scaffolding domain of SON is essential for the localization of SON to β-tropomyosin reporter transcripts²⁹. These results imply that although the repeat domains of SON have only a minor role during constitutive co-transcriptional splicing, they may be essential for SON in mediating alternative splicing. Furthermore, a genome-wide exon array study revealed that SON depletion caused defects in all types of alternative splicing²⁹. SON depletion caused exon skipping at transcription loci in situ, but did not affect the recruitment of SRSF1 or U1-70K to the transcription site. One of the most interesting aspects of SON splicing regulation is that it targets only a subset of human transcripts encoding proteins involved in several major biological processes. These proteins participate in a variety of pathways including cell cycle regulation, apoptosis, integrin mediated cell adhesion, smooth muscle contraction, G protein signaling, Wnt mediated signaling, TGF beta signaling, translation factors, glycogen metabolism, and cholesterol biosynthesis. Sharma et al. (2011) validated altered abundance of transcripts that were upregulated (GEMIN5, IL1A, TNFRSF21, CCNG1 and CYP1B1) or downregulated (DNAH2, TNNC1 and CDK5), as well as altered splicing for several transcripts encoding epigenetic regulators HDAC6, SetD8 and ADA whose processing requires SON²⁹.

Another common thread observed by many groups is that SON depletion causes defects in cell survival^{16, 21, 28, 29, 32, 33}. Examination of mitotic cells showed that SON depletion causes spindle defects and mitotic arrest at metaphase²¹. During mitosis, SON colocalizes with pre-mRNA splicing factors in mitotic interchromatin granule clusters (MIGs) (Figure 1), but localization in other mitotic structures such as the mitotic spindle has not been observed. This supports the hypothesis that the mitotic defects resulting from the loss of SON following knockdown result from the impact of SON on splicing of mitotic regulator transcripts, rather than loss of SON from mitotic structures. In HeLa cells, SON is required for proper splicing of a subset of human transcripts that are necessary for maintaining proper cell growth, explaining cell cycle defects seen after SON knockdown in several studies^{16, 21, 28, 29, 32, 33}. Besides transcripts encoding cell cycle proteins, SON also regulates splicing of a subset of transcripts from DNA repair pathways²⁸. Later studies by Lu et al.³⁰ (discussed below) firmly support a model that SON specifically associates with only a subset of human transcripts. While the mechanisms underlying each type of splicing defect remain to be elucidated, these findings taken altogether implicate a role for SON to work in concert with other splicing regulators such as other SR proteins, EJC components, and MALAT1 in coupling transcription with splicing (Figure 2a). The repetitive nature and serine-rich sequence of SON’s repeat motifs is intriguingly analogous to the heptad repeats of RNA polymerase II (RNAPII) carboxy-terminal domain (CTD) that are responsible for loading RNA processing factors onto nascent transcripts^34–36. It is therefore possible that SON regulates pre-mRNA processing at specific splice sites by controlling the availability of factors in the vicinity. We provide a model illustrating how SON could modulate local recruitment of SR proteins to nascent transcripts to affect splicing outcomes (Figure 2b). Work from many groups supports a role for SON as an essential co-transcriptional splicing regulator in both constitutive and alternative splicing. The mechanism or SON-dependent splicing of specific primary transcripts is still unclear. Thus, by continuing to systematically analyze affected splice sites, we will gain a much fuller picture regarding how SON mediates splicing outcomes.

FIGURE 1.

SON localization throughout the cell cycle. SON colocalizes with splicing factor SRSF1 in HeLa cells throughout the cell cycle. Immunofluorescence localization of SON and of SRSF1 show colocalization as well as reorganization of cellular distribution of these splicing factors during mitosis. Specific localization is observed in mitotic interchromatin granules (MIGs; arrow) during mitosis. DNA was stained with DAPI. Panels in order from left to right: interphase, prophase (before nuclear envelope breakdown, prophase (after nuclear envelope breakdown), metaphase, anaphase, telophase, and G1. Bar = 5 μm. Images provided by Keshia Torres-Munoz and Paula Bubulya.

FIGURE 2.

A model for SON as a multifunctional scaffold. (a) SON organizes nuclear speckle components through its repetitive domains while the C-terminus of SON including the G patch is required for its interaction with the nuclear-retained long noncoding RNA MALAT1 (metastasis-associated lung adenocarcinoma transcript 1). Expression and distribution of components impacts the overall subnuclear organization of RNA processing factors. DNA is stained by Hoechst in blue and SON is stained red to mark nuclear speckles. (b) SON and MALAT1 recruit SR proteins to mediate co-transcriptional splicing of sub-optimal splice sites and alternative splicing of exons. Son may remain associated with mature mRNPs²⁹. CTD: c-terminal domain of RNA polymerase II. * represents weak/dual-specificity splice sites.

Importantly, SON not only takes part in pre-mRNA splicing but also plays a part in other steps of mRNA maturation. Knockdown of SON in HeLa cells led to a redistribution of nuclear speckle-localized exon junction complex (EJC)²¹, which marks the position of exon–exon junctions and participate in mRNA export and decay³⁷. Together with other SR proteins, SON was associated with EJC proteins, suggesting a role in downstream mRNA metabolism after splicing³⁸. Furthermore, SON could also directly regulate splicing of transcripts encoding regulators of the nonsense-mediated mRNA decay (NMD) pathway, such as UPF1, SMG1 or DHX34³⁰. The interplay of SON in EJC complex distribution, mRNP regulation and the NMD pathway infers a sophisticated role of SON in the co-regulation between mRNA splicing and post-splicing RNA metabolism.

ROLE OF SON DURING DEVELOPMENT

SON transcripts have been detected in multiple tissues including human embryo¹, placenta², B cell lymphocytes¹⁰, skin⁵, liver⁵, and brain⁷. Studies on SON function during development were done in stem cells isolated from tissues and embryos^{30, 39}. Stem cells are undifferentiated cells with unlimited potential to self-renew and the ability to differentiate into more specialized cells. Pluripotent stem cells, such as embryonic stem cells, can generate cells of all three germ layers, whereas adult stem cells, such as hematopoietic stem cells, can give rise to mature differentiated cells of a specific lineage⁴⁰. Identification and characterization of new regulators in stem cells will be key to deciphering the mechanisms behind the potency of these cells.

SON in embryonic stem cells

A genome-wide RNAi screen revealed SON as a regulator of human embryonic stem cell (hESC) pluripotency and survival⁴¹. SON is also upregulated in both induced pluripotent cells (iPSCs) and hESCs when compared to MRC-5 fibroblasts³⁰. Moreover, hESC differentiation induced by PRDM14 or NFRKB depletion is accompanied by downregulation of SON, suggesting a specific role of SON in the regulation of pluripotency in hESCs³⁰. In addition to pluripotency maintenance, SON is also essential for the acquisition of pluripotency by somatic cells³⁰. RNA-seq analysis revealed that SON deficiency led to an almost exclusive upregulation of intronic expression, suggesting that SON acts as an intron splicing activator³⁰ (Figure 3a). These introns regulated by SON in hESCs have shorter length and weaker splicing strength than non-targeted introns³⁰. Furthermore, the splice sites flanking these SON target introns are GC-enriched³⁰. These suggest the splicing regulation by SON is dependent on intron size, strength of splice sites and nucleotide content of pre-mRNA transcripts.

FIGURE 3.

SON regulates gene expression in stem cells. (a) SON promotes intron removal from hESC maintenance and survival genes³⁰. (b) SON facilitates correct isoform expression of hESC specific genes³⁰. (c) SON may directly or indirectly inhibit expression of miR23a-27a-24-2 cluster to repress GATA2 expression in HSCs³⁹. SON may interact with other factor to directly inhibit transcription of miR23a-27a-24-2 cluster or act indirectly through regulating splicing of other factors responsible for miR23a-27a-24-2 expression.

Splice site strength is positively correlated with intron length⁴². Short introns may undergo different splicing mechanisms than long introns; therefore, the strength of splice sites alone may not be sufficient for the recognition by SON. It was suggested that short introns directly utilize splice donor and acceptor as a splicing entity (intron-based splicing recognition)^{43, 44}, while long introns are excised upon the recognition and pairing of exon splice sites (exon-based splicing recognition)⁴⁵. These two splicing mechanisms cause different effects during splice site mutation. Particularly, mutation of splice sites flanking small introns will cause intron retention whereas mutation of splice sites flanking long introns leads to exon skipping^46–49. These observations may be partially attributed to the distinct sequence composition of small and large introns, as the polypyrimidine tract between 3′ splice site and the branch point are absent in small introns⁴⁷. Since the splicing of short introns is also regulated by SON, it may mainly recognize the splicing signals near splice junction regions and within the intron. SON could be one of the factors responsible for the splicing mechanism differences between short intron and long introns. Beside intron size and splicing strength, signals from splicing enhancers and silencers as well as RNA secondary structure can also affect splice site selection by splicing factors^{50, 51}. A high GC content around splice sites of SON target genes could influence splicing by altering mRNA secondary structure. Furthermore, the presence of other factors interacting with SON, such as transcription factors or splicing factors, may contribute to the selection of splice site as well. It will be intriguing to investigate the SON-splicing mechanism in hESCs in the future.

Interestingly, SON not only regulates the splicing of cell survival genes such as TUBG1, TUBGCP2, AURKB, and FANCG, but also binds to and regulates the splicing of hESC maintenance genes such as OCT4, MED24 and E4F1, and the affected transcripts may be targeted by NMD pathway for degradation³⁰. Besides intron splicing, SON also regulates alternative splicing in hESCs. 940 splicing events in 694 genes were found to be alternatively spliced when SON is depleted and most of these events induce the emergence of minor splice variants with single exon skipping³⁰. Strikingly, these alternative splicing events significantly overlap with SON-regulated introns³⁰, implicating a coordinated regulation of intron splicing and exon skipping. Moreover, SON deficiency induces alternative splicing in hESC specific genes³⁰. This is exemplified by PRDM14 isoform switch after SON depletion. PRDM14 is a pluripotency regulator in mouse⁵² and human⁴¹. In the absence of SON, a previously un-annotated isoform of PRDM14 lacking exon2 was upregulated while the known isoform of PRDM14 was downregulated³⁰. It was further found that this shorter isoform of PRDM14 could not activate the enhancer region or promote pluripotency induction during reprogramming, raising the possibility that altering the splicing of PRDM14 is one of the main events driving hESC differentiation after SON knockdown³⁰ (Figure 3b).

SON in hematopoietic stem cells

Besides hESCs, SON also has an important role in hematopoietic stem cells (HSCs). SON is highly expressed in HSCs and downregulated during hematopoietic differentiation³⁹. Knockdown of SON in these cells causes downregulation of GATA2 protein expression without affecting its mRNA level, suggesting post-transcriptional regulation of GATA2 by SON³⁹. Examination of GATA2 3′-UTR revealed microRNA-27a (miR-27a) and microRNA-24 (miR-24) as potential GATA2 regulators³⁹. Further analysis identified miR27a as the functional miRNA behind GATA2 regulation. It was also found that the promoter of miR-23a-27a-24-2 cluster, which encodes miR-27a, is regulated by SON, indicating a role of SON in transcription repression in addition to its splicing activity³⁹ (Figure 3c). However, it still remains unclear whether transcription regulation by SON is direct versus indirect due to altered abundance and mRNA processing for transcripts encoding other transcription regulators.

ROLE OF SON IN DISEASES

Role of SON in cancer

SON participates in cellular processes important for cancer development such as cell proliferation, apoptosis, as well as DNA damage and repair^{16, 21, 28, 29, 32}, therefore, activity of SON has been linked to tumorigenesis. Early studies found that overexpression of SON fragments could transform mammalian cells to an epithelial-like morphology⁵³ or inhibit cell apoptosis⁶. In leukemia cells, SON interacts with growth arrest inducing protein AML1-ETO instead of its tumorgenic isoform AML1-ETO(C663S)^{32, 54}. In addition, SON translocates from nuclear speckles to the cytoplasm in t(8;21)+ leukemia cells. SON also interacts with tumor suppressor protein von Hippel-Lindau in 786-O renal carcinoma cells³². The expression of SON has been detected in various cancer cell lines such as leukemia-derived, HeLa, and pancreatic cancer cell lines^{16, 21, 28, 29, 32, 33, 54}. SON deficiency inhibited the tumorigenicity of pancreatic tumor cells in vivo³³. While depletion of SON induced significant growth arrest and apoptosis in cancer cell lines, these effects were less pronounced in normal cells^{16, 21, 28, 29, 32, 33}. Taken together, these data implicate SON as a key factor in cell growth that could impact tumorigenesis. Given the specific role of SON in cancer cell survival, SON may serve as a novel target for cancer treatment.

Role of SON in virus infection

Besides cancer development, SON is also involved in pathways regulating virus infection. Transcription repression by SON was not only observed in HSCs but also on viral promoters⁵⁵. SON binds to a negative regulatory element (NRE) of human hepatitis B virus (HBV) via a GA(G/T)AN(C/G)(A/G)CC motif in the NRE, inhibiting transcription of HBV genes and virion production through regulating its core promoter⁵⁵. In addition to its role in HBV infection, SON is a regulator of influenza virus infection as discovered by a genome-wide RNAi screen for host factors critical for influenza virus replication⁵⁶. Reduced influenza viral RNA levels and decreased viral infection were seen following SON knockdown⁵⁶; suggesting that SON is needed for influenza virus replication. However, it still remains unclear if these observations are direct effects or secondary effects of the knockdown due to unsuccessful transcription or splicing of transcripts encoding other cellular proteins needed for viral replication and trafficking. Sorting out primary (direct role in process e.g. mitosis) versus secondary effects (due to misprocessing of other transcripts) under SON knockdown conditions is a serious challenge. Additional knowledge derived from the mapping of direct interfaces between RNA and RNA binding proteins will provide new insights into the rules that govern selective and specific interactions. Techniques such CLIP-seq (cross-linking immunoprecipitation coupled with high-throughput sequencing) will undoubtedly be used more widely to study RNA binding proteins^{57, 58}.

Conclusion

SON is preferentially expressed in undifferentiated stem cells and downregulated during stem cell differentiation^{30, 39}. SON regulates the potency of stem cells through both transcription and splicing control. It remains to be answered whether SON is equally important to all stem cells or only specific types of stem cells. All studies thus far are based on knockdown or overexpression of SON in cell lines, so generation of SON knockout mice will be key for studying its role during development. It will be interesting to learn if SON helps to scaffold the assembly of mature splicing complexes in nuclear speckles. How this would in turn promote the overall stability of RNP complexes, the dynamic exchange of complexes among different nuclear compartments, or the precise mechanisms for SON-mediated splicing are all questions for future study.