Various post-transcriptional modifications are required for the maturation of mRNA in eukaryotic cells. These modifications include the addition of 7-methylguanosine (m7G) at the 5′ end, the formation of a polyadenylic acid tail at the 3′ end, and RNA splicing. Alternative splicing allows for the production of different mature mRNAs from a single pre-mRNA molecule. Heterogeneous nuclear ribonucleoproteins (hnRNPs) are a superfamily of RNA-binding proteins that play a key role in regulating the alternative splicing of pre-mRNA [1].

2. Overview of hnRNPs

2.1. Composition of hnRNPs

Heterogeneous nuclear ribonucleoproteins (hnRNPs) are a class of RNA-binding proteins (RBPs). Through immunopurification and two-dimensional gel electrophoresis, 42 kinds of hnRNPs have been identified in Hela cells. These hnRNPs consist of more than 20 major hnRNPs that exhibit relatively higher abundance, along with other hnRNPs ^[2][3][2,3]. The main hnRNPs have a molecular weight ranging from 34 kDa to 120 kDa and are named hnRNP A1 to hnRNP U ^[4][5][6][7][4,5,6,7] based on their molecular weight and structural and functional characteristics. Due to the strong association between hnRNP A1, A2/B1, B2, C1, and C2 and hnRNA, they are referred to as “core proteins” [8].

2.2. hnRNPs: RNA-Binding Domains and Structural Insights

The analysis of cDNA sequences has revealed that hnRNPs possess a modular structure comprising at least one RNA-binding motif and auxiliary domains ^[9][10][9,10] (Figure 1). As recent investigations into the structure of hnRNPs have delved into greater detail, it has been confirmed that hnRNPs consist of three distinct RNA-binding motifs: the RNA recognition motif (RRM) ^[11][12][11,12], the RNA-binding domain consisting of Arg-Gly-Gly repeats (RGG domain) [13], and the K-homology domain (KH domain) [14]. Researchers have conducted comprehensive analyses of the sequences of RNA-binding motifs and the auxiliary domains of hnRNPs, resulting in a well-established understanding of the primary functions associated with each motif. Among these, the RNA recognition motif (RRM) stands out as the most prevalent RNA-binding domain within hnRNPs. Comprising 80–90 amino acid residues, the RRM is characterized by highly conserved RNP1 ([K/R]-G-[F/Y]-[G/A][F/Y]-[I/L/V]-X-[F/Y]) and RNP2 ([I/L/V]-[F/Y]-[I/L/V]-X-N-L) amino acid sequences ^[14][15][14,15]. A typical RRM is composed of a four-stranded anti-parallel β-sheet that folds into two α-helices, forming the characteristic β1α1β2β3α2β4 topology [16]. Functionally, the RRM acts as an RNA binding platform, playing a pivotal role in the specific binding of hnRNPs to pre-mRNA [12]. Studies have revealed that the RNA binding of the RRM often relies on the aromatic amino acid residues within RNP1 (β1) and RNP2 (β3), as well as the first basic amino acid residue of RNP1 [17]. For instance, in hnRNP A1, the RRM domain binds to RNA (UAGGG(A/U)) via aromatic amino acid residues located on the β-sheet [18]. Interestingly, the RRM domain of hnRNP F employs three highly conserved loop structures (loop 1, 3, and 5) to interact with AGGGAU, deviating from the traditional RNP1 and RNP2 binding mode [19]. The β-sheet of hnRNP F’s RRM domain does not engage with RNA [20]. Currently, a diverse array of RRM-RNA complexes have been identified within cells with the ability to predict the RNA segments they recognize [21]. Many hnRNPs also harbor multiple RRM domains, where their role in RNA binding within hnRNPs extends beyond mere summation, intricately entwined with the elaborate conformations they adopt [18]. Only a handful of hnRNPs lack RRM domains, including hnRNP E1-E4, hnRNP J/K, and hnRNP U. RGG also has RNA binding activity [22]. It is characterized by closely spaced arginine–glycine–glycine tripeptide repeat clusters and interspersed aromatic amino acid residues [23]. The RGG box contains many arginines which carry positive charges. These residues can form electrostatic interactions with non-specific RNA to enhance RNA binding. Additionally, they can form a hydrogen bond network with specific RNA to specifically recognize the tertiary structure of RNA, such as stem-loop or convex structures [22]. The KH domain consists of 70 amino acids. Within the KH domain, the conserved GxxG loop plays a crucial role in recognizing and binding RNA bases through hydrogen bonds and hydrophobic interactions ^[24][25][26][24,25,26]. These interactions allow the KH domain to specifically recognize and bind certain RNA sequences. Additionally, KH domains have the ability to recognize longer RNA targets in a tandem manner [14]. Furthermore, certain studies ^[22][27][28][22,27,28] have demonstrated the ability of these RNA-binding motifs to bind to DNA. This suggests that certain hnRNPs may engage in DNA binding to execute additional functions, although this aspect will not be further explored in this study. It is worth noting that hnRNP proteins also contain other domains that mediate important functional specificities. The most common and well-characterized auxiliary domains are the nuclear localization signal (NLS) and M9 domains, both of which are associated with nucleoplasmic shuttling ^[29][30][29,30]. hnRNP A2/B1, C, Q, and R all contain NLS motifs, and NLS binds to the nuclear transfer factor importin α-β heterodimer to facilitate nucleoplasmic shuttling through the nuclear pore complex. Both hnRNP A1 and hnRNP A2/B1 contain the M9 domain, which enables nuclear localization by binding to the transport receptor Trn1 ^[31][32][31,32]. Additionally, hnRNP U also possesses the SAP domain related to DNA binding (Scaffold Attachment Factor A/B Acinus PIAS domain) ^[33][34][33,34] and the ATPase domain to regulate the oligomerization of chromatin-associated RNAs (caRNAs) ^[35][36][35,36]. However, the precise mechanisms of action for these two domains are still unclear. The function of hnRNPs is dependent on their domains and binding affinity to pre-mRNA. In their study, Massimo Caputi et al. [37] discovered that the core binding site of the RRM domain of the hnRNP H family (H, H’, F, 2H9) is GGGA. Additionally, high-affinity binding sequences for hnRNP A1 [38] and C [39] have been identified through selection and amplification from random RNA libraries.

2.3. Functions of hnRNPs

Due to the diverse types and complex structural domains of hnRNPs, they can perform a variety of functions in cells. These functions include processing pre-mRNA into mature mRNA and serving as trans-acting factors that regulate gene expression. hnRNPs also participate in the processing of pre-mRNA alongside other RNPs, playing a crucial role in regulating mRNA transport, localization, translation, and stability (Figure 2). As a result, hnRNPs play a key role in various biological processes within cells. Contemporary research on hnRNPs primarily focuses on their involvement in muscle and neurodegenerative diseases (such as amyotrophic lateral sclerosis) as well as their role in cancer progression ^[40][41][42][40,41,42].

2.3.1. Regulation of Alternative Splicing

The pre-mRNA splicing process in cells relies on the interaction between splicing factors and splice sites. This mainly includes the following steps: (1) The U1 snRNP complex binds to the 5′ splice site ^[43][44][43,44]. (2) U2AF65 binds to the polypyrimidine site (Py-tract) [45]. (3) U2AF35 binds to the AG bases at the 3′ splice site. (4) The U2AF heterodimer binds to the 3′ splice site [46] (Figure 3). Most hnRNPs typically function as inhibitory splicing regulators, and their mechanism is as follows.

• Splice site identification and shelter: hnRNP proteins have the ability to bind to pre-mRNA splice sites, which can impact the formation of spliceosomes. This binding can mask or hinder the recognition of splice sites, leading to differential alternative splicing. Additionally, the binding of hnRNPs can competitively affect the binding of other splicing factors. For instance, hnRNP I’s RRM1 and RRM2 specifically bind to the polypyrimidine sequence of the fourth internal loop of the U1 snRNA stem loop, thereby inactivating the splicing complex A formed by U1 snRNP [47]. Studies have also shown that hnRNP I can target ITSN1 [48], β-tropomyosin [49], Fas [50], and alpha-actinin [51], competitively inhibiting the binding of U2AF65 and preventing the formation of the U2AF heterodimer complex by binding to the polypyrimidine sequence of pre-mRNA. hnRNP A1 interacts with the 3′ splice site of MAPT exon 10 and facilitates the exclusion of exon 10 [52]. Similarly, hnRNP L can bind to pre-mRNA polypyrimidine sequences, competitively inhibiting the binding of U2AF65 to RNA [53]. Furthermore, hnRNP H1/H2 counteracts the activation of the 3′ splice site [54] (Figure 4B).
• Splicing inhibition: hnRNPs can impact RNA structure by interacting with pre-mRNA, resulting in the exclusion of specific exons. This inhibitory effect primarily operates by altering the structure of the splice site. For instance, the RRM3 and RRM4 domains of hnRNP I bind to the polypyrimidine sequence (e.g., CUCUCU) near the pre-mRNA exon, forming an RNA ring structure that hinders the binding of other splicing factors and the formation of splicing complexes ^[51][55][51,55] (Figure 4C).
• Competitive splicing inhibition: In certain cases, hnRNP proteins and other RNA-binding proteins may competitively bind to the same pre-mRNA, which can impact the inclusion or exclusion of specific exons. The majority of competitive splicing occurs between hnRNPs and SRs. For instance, hnRNP H1 can compete with SRSF3 for binding to PRMT5 pre-mRNA, thereby inhibiting the exclusion of PRMT5 exon 3 by SRSF3 [56]. Additionally, hnRNPA1 can competitively bind to the G-rich sequence downstream of β-tropomyosin exon 6B, along with ASF/SF2, leading to the inhibition of exon 6B exclusion. Simultaneously, hnRNP A1 and ASF/SF2 competitively bind to the 5′ splice site of C175G pre-mRNA (C175G is a synthetic 533 nt pre-mRNA sequence that is frequently employed as a standard model for investigating 5′ splice sites). hnRNP A1 competitively inhibits the binding of U1 snRNP to the 5′ splice site of C175G pre-mRNA, while ASF/SF2 enhances the binding of U1 snRNP to the 5′ splice site of C175G pre-mRNA [57] (Figure 4D).

Recent studies have demonstrated that hnRNPs can also function as splicing enhancers, thereby promoting alternative splicing. For instance, SCHAUB M et al. [58] discovered that members of the hnRNP H family play a crucial role in the alternative splicing of HIV-1 tat pre-mRNA by acting as splicing enhancers. Additionally, hnRNP K has been found to bind downstream of the 5′ splice site of IAV pre-mRNA, facilitating the recruitment of U1 snRNP [59]. The underlying reason for these observations may be attributed to the binding site of hnRNPs [60]. The mislocalization of hnRNPs on the opposite side of the 5′ splice site leads to an increased recruitment of U1 snRNP and promotes splice site recognition ^[61][62][61,62].

2.3.2. Regulation of mRNA Stability

hnRNPs play a role in regulating mRNA stability through various mechanisms, including the poly(A) tail, AU-rich elements (AREs), and the 3′UTR. For instance, hnRNP H1 and hnRNP F can enhance the stability of APP mRNA by binding to it through cytoplasmic shuttling mediated by the gly Rich motif, thereby increasing its half-life [63]. Additionally, hnRNP F has been found to regulate the TTP/BRF-mediated degradation of ARE-mRNAs [64]. The RRM2 domain of hnRNP A2/B1 acts as an m6A reader [65], recognizing the N6-methyladenosine (m6A) site on TCF7L2 mRNA to stabilize the poly(A) tail and maintain its mRNA stability. Moreover, hnRNP A2/B1 and hnRNP A1 play a role in the regulation of mRNA deadenylation via the CCR4-NOT deadenylation complex. They achieve this by binding to the UAASUUAU sequence present in the mRNA 3′UTR, thereby influencing the degradation of the transcript [66] (Figure 5B). Interestingly, hnRNP A2/B1 also binds to its own 3′UTR to regulate the ratio of nonsense-mediated RNA decay (NMD) (both for sensitive and insensitive types). When the levels of hnRNP A2/B1 protein increase, it combines with its own pre-mRNA 3′UTR, leading to alternative splicing and the production of a higher proportion of NMD-sensitive mRNAs. These NMD-sensitive mRNAs are subsequently degraded during translation [67]. This auto-regulatory mechanism demonstrates how splicing factors control their own expression levels, providing one possible explanation [68]. As an ARE-binding protein, hnRNP D has the ability to recognize and bind to various ARE-mRNA sequences by targeting uridine residues [69]. Subsequently, it recruits transporters such as eIF4G, PABP, Hsp70, Hsc70, and Hsp27 to form the signal transduction regulatory complex (ASTRC). ASTRC plays a crucial role in initiating the 3′ to 5′ deadenylation-dependent mRNA degradation pathway, thereby promoting mRNA degradation. Among the isoforms of hnRNP D, the specific mechanism of action involves AUF1 interacting with the AU-rich element (ARE) in mRNA, which serves as a recognition site for AUF1. While the mRNA is being translated, AUF1 interacts with the translation initiation factor eIF4G [70], causing AUF1 to be released from the ARE by the ribosome. This enables the ribosome to access the mRNA and initiate translation. Simultaneously, AUF1 forms a complex with the polyadenine nucleic acid-binding protein (PABP), which exposes the polyadenine nucleic acid tail and allows the mRNA to be degraded by nuclease. The p40 isoform is involved in the decay of ARE-mRNA, which regulates lymphokine mRNA stability [71] (Figure 5A). Additionally, studies have demonstrated [72] that the KH3 domain of hnRNP K can bind to the poly(C) site of LAPTM5 3′UTR, thereby enhancing the stability of its transcripts. Different hnRNPs exhibit variations in their regulation of mRNA stability. For instance, hnRNP D, K, I, and Q [73] can all bind to the 3′UTR of mPer3, but they differ in their impact on mPer3 stability. hnRNP K maintains the stability of mPer3, while hnRNP D and Q accelerate its degradation. hnRNP I, on the other hand, does not affect the stability of mPer3. Additionally, hnRNP A1 [74], C [75], U [76], and L [77] can bind to the 3′UTR of certain mRNAs and influence their stability, although the mechanism underlying this remains unclear.

2.3.3. Localization and Transport of mRNAs

RNA molecules transported within cells often contain specific cis-acting elements that are recognized by specific trans-acting factors in the cell. The interaction between these cis-acting elements and trans-acting factors, along with the involvement of molecular motors, enables RNA particles to actively transport on microtubules and actin filaments. The hnRNP AB and hnRNP A2 proteins contain RRM motifs that can bind to the cis-acting element RTS on the 3′UTR of mRNA [78]. This binding facilitates the transfer of mRNA from the nucleus to the cytoplasm. During this process, hnRNP AB may target specific transcripts by stabilizing RNA G4 quadruplexes near the transcript’s RTS. Several transcripts, including MBP, β-actin, Arc, BDNF, CAMKIIα, and Protamine 2, require hnRNP AB for their localization and transport ^{[79][80][81][82][83]}[79,80,81,82,83]. The glycine-rich domain (GYR) of the hnRNP H/F protein interacts with Transportin 1 to facilitate nucleocytoplasmic shuttling and participate in the extranuclear transport of mRNA [84]. The M9 domain of hnRNP A1 is considered the main functional domain for nucleocytoplasmic shuttling, and its nuclear import does not rely on the classic NLS pathway [85].