As nascent messenger RNA (mRNA) is transcribed, it becomes associated with a variety of proteins, collectively termed heterogeneous nuclear ribonucleoproteins (hnRNPs) that will accompany the mRNA transcript through its lifecycle. HnRNPs play critical roles in all aspects of mRNA metabolism and function. From the regulation of transcription to splicing, transport, localization, and stability, these proteins are essential for proper regulation of gene expression. HnRNP A1, a member of the hnRNP A/B family, is one of the best studied hnRNPs (Bekenstein and Soreq 2013; Jean-Philippe et al. 2013). Here, we review the structure and diverse molecular functions of hnRNP A1, with particular emphasis on its involvement in human diseases.
Historically, the presence of cytoplasmic granules and fibrils in basophilic cells has been initially demonstrated using electron microscopy studies (Porter 1954; Sjostrand and Hanzon 1954; Palade 1955), which revealed structures of 150–200 A in diameter. Using amphibian oocytes, Gall (Gall 1956) demonstrated that these granules and fibrils contain RNA and that they exist in both the nucleus and the cytoplasm, providing one of the first major steps in our understanding of the biology of hnRNPs. Biochemical characterization of nuclear extracts showed that newly synthesized mRNA is an integral part of 30–40 S ribonucleoprotein (RNP) particles in vivo (Samarina et al. 1968). However, the protein composition of these particles remained unknown. In 1977, Beyer et al. (1977) analyzed nuclear extracts of HeLa cells and characterized six “core” hnRNP particles (A1, A2, B1, B2, C1, and C2), with A1/A2 comprising 60% of the protein mass of the particle. Using cross-linking studies in intact cells, van Eekelen et al. (1981) identified other hnRNP proteins, and later studies used immune-purified hnRNPs and two-dimensional gel electrophoresis to further characterize the protein components of hnRNP particles (Pinol-Roma et al. 1988). These and other studies have collectively shown that over 20 proteins (termed hnRNP A-U) are associated with nascent mRNA (pre-mRNA) as well as with fully processed mature mRNA (Dreyfuss et al. 1993).
hnRNP A1: Structure
Human hnRNP A1 is a 320 amino acid protein, the domains of which can be grossly divided by their N- or C-terminal locations. The N-terminus of hnRNP A1 (also called Unwinding/Unfolding Protein, or UP1 domain) spans residues 1–196 and contains two RNA-recognition motifs (RRM1 and RRM2), whereas the C-terminal region contains a glycine-rich domain (GRD, 45% glycine) which harbors an RGG box RNA-binding domain and the M9 nuclear targeting sequence. RRMs are the most prevalent RNA-binding motifs and can be found in many RNA-binding proteins in addition to hnRNPs. They are composed of four beta strands and two alpha helices that make a compact alpha/beta structure, with the antiparallel beta sheets forming the major RNA interacting surface. Conserved aromatic residues are important for interactions with RNA. RRM1 and RRM2 show great sequence similarity, yet they are not functionally equivalent and are not redundant. For example, deletion, duplication, or swaps of these RRM domains demonstrated that they have distinct roles in regulation of alternative splicing (Mayeda et al. 1998). Each RRM contains two conserved sequences, RNP1 and RNP2, that are juxtaposed on the beta 3 and beta 1 strands and make direct contact with the RNA. The crystal structure of UP1 has been solved at 1.1 A resolution, which revealed alternative yet spatially correlated side chain conformations in the two conserved phenylalanine residues of the first RRM (Vitali et al. 2002). The two RRMs interact in solution (Barraud and Allain 2013), suggesting functional relevance to these features.
The GRD is important for interactions with both RNA and proteins, including self-interactions of hnRNP A1 (Cartegni et al. 1996; Fisette et al. 2010; Nadler et al. 1991). Functionally, the GRD is necessary for alternative splicing activity (Mayeda et al. 1994) and is important for cooperative binding of nucleic acids (Nadler et al. 1991). Deletion of the GRD of hnRNP A2, another member of the hnRNP A/B family, results in cytoplasmic accumulation, suggesting that this region may also be involved in nuclear import. The arginine–glycine–glycine (RGG) domain within the larger glycine-rich region is believed to be mostly important for RNA interactions, as has been shown for hnRNP U (Kiledjian and Dreyfuss 1992). The RGG domain contains repeats of an RGG tripeptide that are interspersed with aromatic residues. Interestingly, the arginine residues within the RGG box can be methylated, and this modification was suggested to modulate RNA binding of hnRNP A1 (Kim et al. 1997). Indeed, methylation was shown to reduce the ability of hnRNP A1 to bind single-stranded nucleic acids (Rajpurohit et al. 1994). Thus, both the GRD sequence and its post-translational modifications play pivotal roles in the diverse hnRNPA1 activities.
HnRNP A1 does not contain a “classic” nuclear localization sequence (NLS). Rather, the M9 sequence, also contained within the larger GRD, is critical for nuclear localization of hnRNP A1 (Siomi and Dreyfuss 1995). Strikingly, the M9 sequence is also important for hnRNP A1 export to the cytoplasm (Michael et al. 1995), a feature that other NLSs do not seem to have. The Transportin 1 and Transportin 2 factors that are distantly related to importin beta mediate hnRNP A1 nuclear import (Pollard et al. 1996) (Rebane et al. 2004). Once imported to the nucleus, RanGTP causes Transportin 1 to dissociate from hnRNP A1 (Siomi et al. 1997), releasing hnRNP A1 to bind mRNA. In conclusion, both the N-terminal and C-terminal domains are important for interactions with nucleic acids, whereas interactions with protein partners are likely mostly mediated by the C-terminal domains alone.
Functions of hnRNP A1
hnRNP A1 has been ascribed multiple different functions, mostly related to RNA metabolism but also related to DNA. Here, due to space limitations, we will describe in brief four of these functions: transcription, splicing, miRNA biogenesis, and telomere regulation.
hnRNP A1 functions both as an enhancer and a repressor of gene expression, depending on the context and the specific gene that is regulated. For example, hnRNP A1 binds the APOE promoter and activates APOE gene expression (Campillos et al. 2003). Interestingly, this interaction between hnRNP A1 and the APOE promoter is modulated by the 219 T/G single nucleotide polymorphism (SNP) within the promoter. As this polymorphism is associated with risk of Alzheimer’s disease (AD), myocardial infarction, and early-onset coronary heart disease, this interaction may have important consequences on human health (Campillos et al. 2003). Several instances of gene expression repression by hnRNP A1 have also been reported. These include the human thymidine kinase (Lau et al. 2000), gamma fibrinogen chain (Xia 2005), and vitamin D receptor (Chen et al. 2003). Further examples have been reviewed by Jean-Philippe et al. (2013). Taken together, these reports indicate that hnRNP A1 interacts with promoter regions and regulates transcription in a context-dependent manner.
The term “splicing” refers to the process of intron removal and the joining of exons in two transesterification reactions. Those are catalyzed by the spliceosome – a large macromolecular machine composed of protein factors and five small U RNAs (Wahl et al. 2009). In constitutive splicing, introns are removed and all exons are joined. Supporting its general role in splicing, hnRNP A1 shows preferred binding to the 3′ of introns (Swanson and Dreyfuss 1988). In addition, hnRNP A1 interacts with U2 snRNP, and thus is expected to have important roles in the early steps of spliceosome assembly (Buvoli et al. 1992). More recently, hnRNP A1 was shown to associate with the U2 auxiliary factor (U2AF), and to displace it from transcripts that do not contain the 3′ splice site AG (Tavanez et al. 2012). Thus, hnRNP A1 acts as a proof reader to allow precise splicing reactions.
Research over the past few decades has demonstrated that the vast majority of transcripts in metazoans can also undergo alternative splicing, a process in which, through combinatorial selection of exons and selective retention of introns, the proteomic diversity is massively increased. Several groups of splicing factors which regulate both constitutive and alternative splicing have been described, most notably Serine-Arginine (SR) proteins and hnRNPs. By binding enhancer and silencer mRNA sequences located in introns and exons, these factors enable precise splicing to occur, even though the 3′ and 5′ splice sites, i.e., the cis-acting sequence elements that signal the break points between exons and introns, are inherently short and degenerate, especially in higher eukaryotes. In addition to its general roles in splicing, hnRNP A1 plays more specific roles in the regulation of alternative splicing. A well-studied role of hnRNP A1 is to bind exonic and intronic splicing silencer elements (ESS and ISS, respectively) and antagonize the function of splicing enhancers such as SR proteins, thus promoting the choice of more distal splice sites and exon exclusion (Mayeda and Krainer 1992). Several mechanisms for splicing suppression by hnRNPs have been suggested and include looping out of the intervening RNA, as has been shown for an exon of hnRNP A1 (Blanchette and Chabot 1999). Therefore, hnRNP A1 may sustain the diversity of alternative splicing.
microRNA (miRNA) are short noncoding RNAs that regulate gene expression at the posttranscriptional level. During miRNA biogenesis, pri-miRNA sequences are cleaved in the nucleus by the microprocessor complex. Following their export to the cytoplasm, further processing by DICER generates mature miRNA. Examples of regulation of miRNA biogenesis by hnRNP A1 include the transition of pri-miR 18a to pre-miR 18a, as knockdown of hnRNP A1 results in reduced levels of pre-miR 18a and increased levels of a reporter harboring miR 18a binding sites (Guil and Caceres 2007). HnRNP A1 also regulates the production of let-7a, a highly conserved miRNA that belongs to the let-7 family, which was the first miRNA to be reported. In this case, hnRNP A1 operates as a negative regulator of the processing of let-7a pri-miRNA. This process involves antagonistic functions of hnRNP A1 and KSRP, a protein which promotes miRNA maturation (Michlewski and Caceres 2010). Additionally, hnRNP A2/B1, and possibly also hnRNP A1 regulate miRNA sorting into exosomes in a SUMOylation-dependent mechanism (Villarroya-Beltri et al. 2013). Thus, hnRNP A1 may have far-reaching implications in miRNA biology and its associated functions in human health.
Burd and Dreyfuss identified AUUUA as a “winner” hnRNP target consensus sequence (Burd and Dreyfuss 1994). Interestingly, in addition to 5′ and 3′ splice sites, this sequence is also found in telomere sequences. Indeed, hnRNP A1 binds telomeric sequences (Ding et al. 1999), and has been found to regulate telomere elongation (LaBranche et al. 1998), such that telomeres of cells deficient in hnRNP A1 are shorter than those in control cells. HnRNP A1 protects telomeric sequences from degradation to maintain long 3′ overhangs (Dallaire et al. 2000). Interestingly, hnRNP A1 binds both telomeric DNA sequences and the RNA component of human telomerase (hTR) (Fiset and Chabot 2001). Also, hnRNP A1 can bind G quadruplex structures found in telomeres and unwind them (Enokizono et al. 2003). Importantly, this function is related to the ability of hnRNP A1 to stimulate telomere elongation (Zhang et al. 2006). A detailed thermodynamic characterization of hnRNP A1 binding and unfolding of G-quadruplex structures (Hudson et al. 2014) describes how binding is coupled to and drives these unfolding events.
DNA protein kinase (DNA-PK) is an important regulator of telomere integrity. Among other substrates, DNA-PK phosphorylates hnRNP A1, and this reaction is stimulated by DNA as well as by the telomerase RNA component (Ting et al. 2009). Vaccinia-related kinase 1 (VRK1) is another identified hnRNP A1 kinase, and hnRNP A1 phosphorylation by VRK1 is required for telomere maintenance (Choi et al. 2012). Importantly, hnRNP A1, together with protection of telomeres 1 (POT1) and the telomeric repeat-containing RNA (TERRA), displaces replication protein A (RPA) from telomeres after DNA replication, promoting telomere capping and integrity (Flynn et al. 2011). In conclusion, hnRNP A1 plays multiple roles both in RNA metabolism at large and in the regulation of telomeres’ length. It is therefore expected that mutations or other mechanisms leading to impaired hnRNP A1 function would be detrimental to human health, as is explored in more detail below.
hnRNP A1: Roles in Human Disease
Interrogations of the specific roles of genes and proteins in human disease often initiate with the extreme cases of cancer and neurodegenerative disease, both because of the importance of these two disease types to human health and well-being and since they represent the two ends of a scale reflecting cellular development and senescence. HnRNP A1 is no exception, and a lot of effort has been invested in exploring its involvement in cancer and neurodegenerative diseases.
Early reports demonstrated that hnRNP A1 levels are higher in transformed and proliferating cells than in differentiated tissues, suggesting that hnRNP A1 may be important for cancer progression (Biamonti et al. 1993). Indeed, the oncoprotein BCR/ABL stabilizes hnRNP A1 by preventing its degradation, and an hnRNP A1 mutant that is deficient in its nuclear export suppresses colony formation in transformed cultured cells and reduces tumorigenic potential in-vivo (Iervolino et al. 2002). Multiple other studies report increased hnRNP A1 levels in different cancers. These include lung cancer (Pino et al. 2003), colon cancer (Ushigome et al. 2005), cervical cancer (Fay et al. 2009), hepatocellular carcinoma (Zhou et al. 2013), oral squamous cell carcinoma (Yu et al. 2015), among others. In the context of cervical cancer, it is noteworthy that hnRNP A1 binds the human papillomavirus type 16 late regulatory element and is upregulated during virus-infected epithelial cell differentiation (Cheunim et al. 2008). Importantly, and suggesting a more general role in cancer, the oncogenic transcription factor c-myc upregulates hnRNP A1, which results in inclusion of exon 10 of the pyruvate kinase transcript. The PKM variant of pyruvate kinase, which includes exon 10 is expressed in embryos and is reexpressed in multiple cancers, promoting aerobic glycolysis. Therefore, the c-myc/hnRNP A1/pyruvate kinase pathway may underlie a general pathway that regulates alternative splicing patterns that are required for tumor cell proliferation (David et al. 2010). Because hnRNP A1 upregulates myc levels in an IL-6-dependent pathway (Shi et al. 2008), it is tempting to speculate that myc/hnRNP A1 interactions form a positive feedback loop that increases the levels of these proteins in cancer. Upregulation of hnRNP A1 hence emerges as a common theme in multiple cancers, opening new venues for the development of novel therapeutics.
Various neurodegenerative diseases are accompanied by reduced hnRNP A1 levels, and recent evidence suggests critical functional roles for hnRNP A1 in neurons. Autoantibodies against hnRNP A1 develop both in patients with multiple sclerosis (MS) and in those with T-lymphotropic virus type 1 (HTLV-1)-associated myelopathy/tropical spastic paraparesis (HAM/TSP), a rare disease that can be indistinguishable from MS (Levin et al. 2002). In brain sections, hnRNP A1 antibodies caused inhibition of neural firing (Levin et al. 2002). In-vivo, injection of hnRNP A1 antibodies in mice with experimental autoimmune encephalomyelitis (EAE), a widely used model for MS, worsened disease progression and caused neurodegeneration (Douglas et al. 2016). Taken together, these reports indicate that autoimmune reactions to hnRNP A1 are damaging to brain functioning.
Supporting a role for hnRNP A1 in disease-related aberrant RNA metabolism, hnRNP A1 interacts with TDP-43 (Buratti et al. 2005), the major component of ubiquitinated inclusions found in ALS, as well as in FTD-TDP (Neumann et al. 2006). Importantly, mutations in hnRNP A1 were recently found in two families with multisystem proteinopathy, a rare complex phenotype that associates with frontotemporal dementia (FTD), Paget disease of bone (PDB), inclusion body myopathy (IBM), and amyotrophic lateral sclerosis (ALS) (Kim et al. 2013). In the same study, hnRNP A1 mutation was found in one familial case of ALS. Following, a missense mutation in hnRNP A1 was found in two more families with inclusion body myopathy (Izumi et al. 2015). The identified mutations enhance the propensity of hnRNP A1 to form fibrils, exacerbate its recruitment to stress granules and drive cytoplasmic inclusions of hnRNP A1 in animal models (Kim et al. 2013). Importantly, however, the occurrence of mutations in hnRNP A1 in ALS is likely rare; several follow-up studies failed to detect them (Calini et al. 2013; Le Ber et al. 2014). Thus, the association of hnRNP A1 mutations with ALS remains controversial at this time (Li and Wu 2016).
Nonmutated hnRNP A1 may be implicated in ALS and FTD due to its sequestration by repeat-associated RNA. Recently, a GGGGCC hexanucleotide repeat expansion in the C9ORF72 gene has been demonstrated to be the most common genetic cause of familial ALS and FTD-TDP (DeJesus-Hernandez et al. 2011; Renton et al. 2011). Interestingly, RNA transcribed from these hexanucleotide repeats form RNA foci that are found in patients as well as in animal and cellular models. It has been suggested that toxicity of these repeats may arise at least in part by sequestration of RNA-binding proteins. Indeed, hnRNP A1 is one of the proteins that are bound by both sense and antisense GGGGCC transcripts (Cooper-Knock et al. 2015). HnRNP A1 has also been implicated in other repeat-associated diseases, as a null allele of the Drosophila hnRNP A1 orthologue Hrb87F increases the toxicity of poly glutamine (Poly-Q) expansions (Sengupta and Lakhotia 2006). HnRNP A1 may thus be sequestered by various types of disease-related RNA repeats, and its loss of function leads to exacerbation of disease progression.
hnRNP A1 loss, without noticeable aggregation, has been observed in postmortem human cortical samples from AD patients (Berson et al. 2012). Correspondingly, lentiviral-mediated knockdown of hnRNP A1 causes dendritic loss and reduced synaptic size in cultured neurons without induction of apoptosis. In-vivo, hnRNP A1 loss drives learning and memory deficits in mice, further suggesting that its loss in AD may contribute to impaired neuronal function (Berson et al. 2012). Interestingly, hnRNP A1 regulates the alternative splicing of the amyloid precursor protein (APP), the precursor for amyloid-beta (Aβ) (Donev et al. 2007). Aβ is the main component of amyloid plaques that are found in the AD brain and is highly toxic to neurons both as insoluble aggregates as well as soluble oligomers. Overexpression of hnRNP A1 results in reduced levels of amyloid-beta (Donev et al. 2007), further suggesting that low hnRNP A1 levels are also harmful in the context of AD. Of note, impaired cholinergic signaling in vivo dramatically reduces hnRNP A1 levels (Berson et al. 2012). As cholinergic neurons are a particularly sensitive neuronal population in AD, it is conceivable that a negative feedback loop exists, in which Aβ accumulation impairs cholinergic signaling, which in turn results in reduced hnRNP A1 levels, thus further promoting Aβ accumulation. Together, these studies demonstrate critical roles for hnRNP A1 in the maintenance of the nervous system, with its loss of function being causally involved in a number of different human neurodegenerative diseases. In addition, toxic gain of function mechanisms may exist in cases where hnRNP A1 forms cytoplasmic aggregates.
Our survey demonstrated that hnRNP A1 exemplifies the functional complexity of hnRNP protein families, which range from transcription to splicing, transport, and RNA stability, as well as other functions that are related to DNA, such as regulation of telomeres. It is therefore not surprising that impairments in hnRNP A1 carry grave consequences in the context of human health, with implications to common diseases such as cancer and various neurodegenerative diseases. As a highly abundant nuclear protein, hnRNP A1 was one of the first hnRNPs to be identified. Later structural and biochemical studies laid the groundwork for our current understanding of RNA-binding proteins, their domains, and their modes of interaction with nucleic acids. Notably, hnRNP A1 levels seem to be exquisitely regulated, with both upregulation and downregulation leading to potential damaging effects. Therefore, any future attempt to manipulate hnRNP A1 levels as a therapeutic strategy should be taken with care. As we better understand the mechanisms that control hnRNP A1 expression and function, as well as its interacting partners, we may begin to develop specific therapeutics to address its role in diverse human diseases.
The authors are grateful to all of our group members who contributed to the interest in hnRNP A1 and to the work involved. This study was supported by the Israeli Ministry of Science Aging program (to H.S) and by the NIH (F32-NS084667 to A.B.). We apologize to any colleagues whose work could not be covered due to space limitations.
- Buratti E, Brindisi A, Giombi M, Tisminetzky S, Ayala YM, Baralle FE. TDP-43 binds heterogeneous nuclear ribonucleoprotein A/B through its C-terminal tail: an important region for the inhibition of cystic fibrosis transmembrane conductance regulator exon 9 splicing. J Biol Chem. 2005;280:37572–84. doi: 10.1074/jbc.M505557200.PubMedCrossRefGoogle Scholar
- Cheunim T, Zhang J, Milligan SG, McPhillips MG, Graham SV. The alternative splicing factor hnRNP A1 is up-regulated during virus-infected epithelial cell differentiation and binds the human papillomavirus type 16 late regulatory element. Virus Res. 2008;131:189–98. doi: 10.1016/j.virusres.2007.09.006.PubMedPubMedCentralCrossRefGoogle Scholar
- Douglas JN, Gardner LA, Salapa HE, Lalor SJ, Lee S, Segal BM, et al. Antibodies to the RNA-binding protein hnRNP A1 contribute to neurodegeneration in a model of central nervous system autoimmune inflammatory disease. J Neuroinflammation. 2016;13:178. doi: 10.1186/s12974-016-0647-y.PubMedPubMedCentralCrossRefGoogle Scholar
- Kim S, Merrill BM, Rajpurohit R, Kumar A, Stone KL, Papov VV, et al. Identification of N(G)-methylarginine residues in human heterogeneous RNP protein A1: Phe/Gly-Gly-Gly-Arg-Gly-Gly-Gly/Phe is a preferred recognition motif. Biochemistry. 1997;36:5185–92. doi: 10.1021/bi9625509.PubMedCrossRefGoogle Scholar
- Le Ber I, Van Bortel I, Nicolas G, Bouya-Ahmed K, Camuzat A, Wallon D, et al. hnRNPA2B1 and hnRNPA1 mutations are rare in patients with "multisystem proteinopathy" and frontotemporal lobar degeneration phenotypes. Neurobiol Aging. 2014;35:934 e5–6. doi: 10.1016/j.neurobiolaging.2013.09.016.CrossRefGoogle Scholar
- Nadler SG, Merrill BM, Roberts WJ, Keating KM, Lisbin MJ, Barnett SF, et al. Interactions of the A1 heterogeneous nuclear ribonucleoprotein and its proteolytic derivative, UP1, with RNA and DNA: evidence for multiple RNA binding domains and salt-dependent binding mode transitions. Biochemistry. 1991;30:2968–76.PubMedCrossRefGoogle Scholar
- Shi Y, Frost PJ, Hoang BQ, Benavides A, Sharma S, Gera JF, et al. IL-6-induced stimulation of c-myc translation in multiple myeloma cells is mediated by myc internal ribosome entry site function and the RNA-binding protein, hnRNP A1. Cancer Res. 2008;68:10215–22. doi: 10.1158/0008-5472.CAN-08-1066.PubMedPubMedCentralCrossRefGoogle Scholar
- Ting NS, Pohorelic B, Yu Y, Lees-Miller SP, Beattie TL. The human telomerase RNA component, hTR, activates the DNA-dependent protein kinase to phosphorylate heterogeneous nuclear ribonucleoprotein A1. Nucleic Acids Res. 2009;37:6105–15. doi: 10.1093/nar/gkp636.PubMedPubMedCentralCrossRefGoogle Scholar
- Villarroya-Beltri C, Gutierrez-Vazquez C, Sanchez-Cabo F, Perez-Hernandez D, Vazquez J, Martin-Cofreces N, et al. Sumoylated hnRNPA2B1 controls the sorting of miRNAs into exosomes through binding to specific motifs. Nat Commun. 2013;4:2980. doi: 10.1038/ncomms3980.PubMedPubMedCentralCrossRefGoogle Scholar