hnRNP A/B Proteins: An Encyclopedic Assessment of Their Roles in Homeostasis and Disease

Figures

• 

  Figure 1. Human hnRNP A/B proteins are closely related and share similar alternative exonic structures. Protein sequences from the full-length isoform for each hnRNP A/B protein family member were aligned using ClustalW [26]. (A) Percent identity matrix for the full-length isoforms for each hnRNP A/B protein family member: A0 (NCBI Protein accession number NP_006796.1); A1-B (NP_112420.1); B1 (NP_112533.1); A3-1 (NP_001317178.1). Boxes are shaded red to indicate degree of sequence identity. (B) Percent identity matrix for the most common isoforms for each hnRNP A/B protein family member, coloured as in (A). A0 (NP_006796.1); A1-A (NP_002127.1); A2 (NP_002128.1); A3-2 (NP_001317176.1). (C) Sequence alignment for the full-length isoforms of each A/B family protein, from (A). Black and grey highlights indicate exons excluded in alternative isoforms as depicted in Figure 3 (e.g., amino acids 251–302 in A1). Note that the first exon of hnRNP A3 is replaced with an exon encoding just “ME” in isoforms A3-2 and A3-4.

• 

  Figure 2. Updated analysis of mammalian hnRNP A/B protein sequences demonstrates their evolutionary relatedness. Sequences were drawn from the NCBI Orthologs database for each hnRNP A/B family protein, and manually depleted of Low-Quality protein sequences. Different isoforms of each protein within a species were retained. Sequences were then aligned using a Mafft multiple sequence alignment program [27], identical sequences were manually removed, and ambiguous sites were masked using Alistat [28], before generating maximum likelihood trees with VT + F + R4 model in the IQTree program [29]. Original inputs: A0, n = 140; A1, n = 259; A2/B1, n = 428; A3, n = 352. (A) Unrooted phylogram of mammalian hnRNP A/B family proteins, coloured by protein of origin. Scale bar indicates 0.2 amino acid changes per site. (B) Circular tree (polar plot) representation of mammalian hnRNP A/B family proteins with A0 as outgroup. Each dotted line indicates a unique sequence; duplicate sequences have been removed from the plot. Grey circles indicate bootstrap confidence values.

• 

  Figure 3. Domain architecture of hnRNP A/B family proteins and their known isoforms. All regions are to-scale. The total protein for each isoform is in black-and grey, with each shade corresponding to an exon. Red exons are altered in splice variants: for example, note that the A1-A isoform lacks amino acids 251–302 found in the longer A1-B isoform. Green denotes RNA recognition motifs RRM1 (left) and RRM2 (right), and yellow denotes the glycine-rich region, or low-complexity domain (LCD). Within the glycine-rich region, purple denotes RGG (arginine-glycine-glycine) motif-containing region, blue denotes the core prion-like domain (PrLD), and orange denotes the M9 nuclear localization sequence. Note that A0 does not bear an annotated M9 sequence motif. A0 and A3 RGG regions were inferred by the authors, while the A0 and A3 PrLDs were defined using PLAAC prediction algorithms [43].

• 

  Figure 4. Sequence logos for the Hidden Markov Model (HMM) profile of the longest isoforms of each hnRNP A/B protein, generated using the Skylign program [48] through the HMMER implementation available within Skylign.

• 

  Figure 5. hnRNP A1 RRM structures. (A) The secondary structure topology diagram describes the β1α1β2β3α2β4 arrangements of both RRM1 and RRM2 in A1: four antiparallel β-sheets and two α helices as numbered in the figure. (B) A cartoon representation of the 3D structure of A1 RRMs (based on PDB: 1UP1), where RRM1 is shown in cyan and RRM2 in blue. The linker connecting the two RRMs is shown in orange. The key salt bridge interactions (shown as ball-and-stick representations) and hydrophobic contacts (residues shown in line format) at the interface of tandem RRMs are shown. The β-sheets encompassing RNP1 (β3) and RNP2 (β1) motifs in both the RRMs are marked in the structure. (C) A structural ensemble of 20 A1 RRM conformers (PDB code: 2LYV) are shown, and the flexible regions in the structures are marked with arrows.

• 

  Figure 6. 3D structures of DNA- and RNA-bound complexes of A1 RRMs, rotated forward 180° from Figure 5. (A) The structures of RRMs from the free-state (PDB: 1UP1, shown in gray) and the ssDNA-bound complex (PDB: 2UP1, shown in cyan) are overlapped to highlight the structural variations in the protein in response to DNA-binding. The linker loop and the C-terminal segment in RRMs became more ordered upon DNA-binding that led to the formation of short helices shown with arrows. (B) A closer view into the binding mode of 5′ end of ssDNA (shown as green ribbon and nucleotides shown as lines) against the surface of RRM1 (in cyan). The H-bond interactions and π-π between the protein residues and ssDNA are marked. (C) The 3D structure of A1 RRMs (cartoon representation in cyan) in complex with 5-AGU-3′ RNA (shown as green ribbon and nucleotides shown as lines). The key residues engaged in hydrophobic contact with the substrate are shown as stick representations. (D) 2D interaction diagrams showing the interactions between the RRM1-5′-UUAGGUC-3′ complex (left), and RRM2-5′-UCAGUU-3′ complex (right). The RNA molecules are shown as line drawing and key residues are noted. * denotes the residues that…

• 

  Figure 7. 3D structures of hnRNP A2/B1 RRMs. (A) A cartoon representation of the 3D structure of A2/B1 RRMs (based on PDB: 5WWG), with residues that are different from the A1 RRMs coloured in red. The salt-bridge interactions at the interface of tandem RRMs are shown as ball and stick representations, and the consensus in the RNP1 and RNP2 motifs in A1 and A2/B1 RRMs (highlighted in yellow) are shown as text. (B) A closer view into the binding mode of a 10-mer RNA (shown as green ribbon with nucleotides shown as lines) against the surface of RRM1 (in yellow). The H-bond interactions and π-π between the protein residues (in stick representations) and RNA are marked.

• 

  Figure 8. 3D Structures of A1 and A2/B1 LCD fibrils. (A) A cartoon representation of a 3D fibril structure (PDB: 7BX7) formed by multimers of the A1 LCD residues 251–295, with the key residues engaging in different electrostatic and van der Waals interactions are shown in stick representations. (B) A cartoon representation of a 3D fibril structure (PDB: 6WQK) formed by multimers of the A2 LCD residues 263–319 shown in blue. The key aromatic stacking interactions rendered by Y275, Y283, and F309 are shown as sticks. The D290 residue, which is mutated to a valine residue (D290V) in ALS, is rendered in ball-and-stick mode for context.

• 

  Figure 9. Known shared and unique functions of hnRNP A/B family proteins in homeostasis and in disease. Future research will likely alter this understanding.