
==== Front
J Biol Chem
J Biol Chem
The Journal of Biological Chemistry
0021-9258
1083-351X
American Society for Biochemistry and Molecular Biology

S0021-9258(24)02141-0
10.1016/j.jbc.2024.107640
107640
Research Article
Structural basis for RNA recognition by the C-terminal RRM domain of human RBM45
Chen Xi 12
Wei Qinghao 12
Yang Zhongmei 12
Chen Xiaolei 12
Guo Shuoxuan 12
Jiang Meiyu 12
Wang Mingzhu wangmzh@ahu.edu.cn
123∗
1 Institutes of Physical Science and Information Technology, Anhui University, Hefei, Anhui, China
2 School of Life Sciences, Anhui University, Hefei, Anhui, China
3 Key Laboratory of Human Microenvironment and Precision Medicine of Anhui Higher Education Institutes, Anhui University, Hefei, Anhui, China
∗ For correspondence: Mingzhu Wang wangmzh@ahu.edu.cn
08 8 2024
9 2024
08 8 2024
300 9 1076408 2 2024
8 7 2024
© 2024 The Authors
2024
https://creativecommons.org/licenses/by/4.0/ This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
RBM45 is an RNA-binding protein with roles in neural development by regulating RNA splicing. Its dysfunction and aggregation are associated with neurodegenerative diseases such as amyotrophic lateral sclerosis (ALS) and frontotemporal lobar dementia (FTLD). RBM45 harbors three RRM domains that potentially bind RNA. While the recognitions of RNA by its N-terminal tandem RRM domains (RRM1 and RRM2) have been well understood, the RNA-binding property of its C-terminal RRM (RRM3) remains unclear. In this work, we identified that the RRM3 of the RBM45 sequence specifically binds RNA with a GACG sequence, similar but not identical to those recognized by the RRM1 and RRM2. Further, we determined the crystal structure of RBM45RRM3 in complex with a GACG sequence-containing single-stranded DNA. Our structural results, together with the RNA-binding assays of mutants at key amino acid residues, revealed the molecular mechanism by which RBM45RRM3 recognizes an RNA sequence. Our finding on the RNA-binding property of the individual RRM module of RBM45 provides the foundation for unraveling the RNA-binding characteristics of full-length RBM45 and for understanding the biological functions of RBM45.

Keywords

neurodegenerative disease
amyotrophic lateral sclerosis (ALS)
RNA binding protein
RNA modification
crystal structure
RNA–protein interaction
Abbreviations

ALS amyotrophic lateral sclerosis

FP fluorescence polarization

FTLD frontotemporal lobar dementia

FUS fused in sarcoma

HOA homo-oligomer assembly

IPTG isopropyl-β-D-thiogalactoside

KD dissociation constant

NLS nuclear localization sequence

PDB Protein Data Bank

PEG polyethylene glycol

RBM45 RNA-binding motif protein 45

RBP RNA-binding protein

RMSD root-mean-square deviation

RRM RNA-recognition motif

ssDNA single-stranded DNA

TDP-43 TAR DNA-Binding Protein 43

Reviewed by members of the JBC Editorial Board. Edited by Ronald Wek
==== Body
pmcRNA-binding motif protein 45 (RBM45) is an RNA-binding protein (RBP) involved in neural development (1, 2). It is highly expressed in the fetal brain, and the expression level gradually decreases with development (1). Normally, the C-terminal nuclear localization sequence (NLS) of RBM45 makes it predominantly localized to the nucleus, but RBM45 can shuttle between the nucleus and cytoplasm (3, 4). RBM45 was reported to bind to thousands of RNAs, mostly in intronic regions of pre-mRNA, regulating RNA splicing (5). Abnormalities in RBM45 have been observed to be associated with certain neurodegenerative diseases. Mutations in RBM45 have been found linked to amyotrophic lateral sclerosis (ALS) and frontotemporal lobar dementia (FTLD) (6). In the neurons and glia cells of ALS and FTLD patients, RBM45 was found to aggregate in the cytoplasmic inclusions and co-localize with TDP-43 (TAR DNA-Binding Protein 43) (3, 7), a well-understood ALS-linked RBP (8). These results suggested that aggregation and mislocalization of RBM45 should play a role in the development of neurodegenerative diseases such as ALS and FTLD, although the mechanism remains unclear.

Characterizing the RBM45-recognizing RNAs is essential for understanding the biological function of RBM45 as well as its aggregation in cells from patients with neurodegenerative diseases. RBM45 consists of three RNA-recognition motif (RRM) domains, two at the N-terminus (RRM1 and RRM2) and one at the C-terminus (RRM3), and a pseudo-RRM domain in the middle (Fig. 1A). While the pseudo-RRM domain reportedly mediates self-oligomerization of RBM45, therefore named the homo-oligomer assembly (HOA) domain (3), and interacts with FUS (fused in sarcoma) (9), another well-known ALS-linked RBP (10, 11), the three RRM domains are responsible for binding RNA. Containing multiple RNA-binding modules is common in RBPs (12, 13). They recognize the same sequence or different sequences, binding RNA independently or cooperatively. Previous studies demonstrated that all three RRMs of RBM45 have RNA-binding ability (5, 14, 15). RBM45 prefers to bind RNAs containing GAC or bipartite GAC sequences in vivo (5). Two independent RBP recognition sequence screening studies have identified the full-length RBM45 bound GACGAC (16) or ACGC (17) sequence RNA in vitro. Our previous study has identified that both N-terminal RRM domains (RRM1 and RRM2) of RBM45 recognize GAC sequences (14), however, the RNA recognition property of the C-terminal RRM (RRM3) remains unclear.Figure 1 Structure of RBM45RRM3. A, the domain structure of RBM45. B, sequence alignment of three RRM domains of RBM45. The identical residues are highlighted in red background; the similar residues are highlighted in red letters. Residues interacting with G1, A2, and C3 through their side chains are indicated by red pentagrams; residues interacting with G4 through their side chains are indicated by red blank pentagrams. C, two views of the overall structure of the RBM45RRM3. The cartoon structure is colored in rainbow, with the N-terminus in blue and the C-terminus in red. D, structural comparison of RRM3 with RRM1 (PDB code 7CSX) and RRM2 (PDB code 7CSX) of RBM45. RRM3, RRM1, and RRM2 are colored in green, gray, and wheat, respectively. The distinct N-terminal loop of RRM3 is colored in blue. The β2–β3 loop and the α2–β4 loop are indicated by red dashed circles.

In this work, we identified that the C-terminal RRM domain of human RBM45 (RBM45RRM3) sequence specifically binds GACG sequence RNA and solved the crystal structures of RBM45RRM3 in apo form and in complex with a single-stranded DNA that mimicked the recognized RNA. Our structural result, coupled with the results of our mutation experiments, elucidated the mechanism of RNA recognition by RBM45RRM3.

Results

Crystal structure of the C-terminal RRM of RBM45

The core region (residues 392–464) of RRM3 of RBM45 shares ∼27% and ∼19% of sequence identities with RRM1 and RRM2, respectively, significantly lower than that between RRM1 and RRM2, which is about 32% (Fig. 1B). To determine the structure of the C-terminal RRM (RRM3) of RBM45, we tried a series of different RBM45 fragments containing RRM3 for crystallization, and finally obtained crystals of a fragment from residue 370 to the C-terminus (hereafter named RBM45RRM3, Fig. 1A) and resolved its crystal structure at 2.4-Å resolution. There are four RBM45RRM3 molecules per asymmetry unit, containing residues from Gln370 to Arg466 or Glu467. The last nine residues at the C-terminus, including the NLS, are disordered in all four molecules. Except for a few amino acid residues at the very N-terminus (residues 370–373), there is no significant difference between the four chains (Fig. S1), so only one chain will be described hereafter. RBM45RRM3 bears the canonical RRM fold, which is characterized by a βαββαβ topology. The four β-strands form an antiparallel β–sheet in the order of β4–β1–β3–β2, while the two α-helices are located on the same side of the sheet (Fig. 1C). The exposed side of the β-sheet is predominantly positively charged, indicating a potential RNA binding surface (Fig. S2). The overall structure of RBM45RRM3 can be well aligned with the AlphaFold-predicted model (18) with a Cα root-mean-square deviation (RMSD) of 0.42 Å (for 86 residues). The major differences occur in the N-terminal loop, C-terminal loop, and some other loop regions (Fig. S3), and these regions also show diversity among the four molecules in the asymmetric unit (Fig. S1).

The overall fold of RBM45RRM3 is similar to RRM1 and RRM2 of RBM45 (PDB code 7CSX), with Cα RMSDs of 0.86 Å (for 61 residues) and 1.06 Å (for 46 residues), respectively. However, they are significantly different in some loop regions. The β2–β3 loop of RRM3 is much shorter than those of RRM1 and RRM2, whereas the α2–β4 loop of RRM3 is a little shorter than that of RRM1 and significantly longer than that of RRM2. The most significant difference is that RRM3 has a distinctive N-terminal loop surrounding α1 and touching α2 as well as the α2–β4 loop (Fig. 1D). However, this pre-RRM loop, which is absent in RRM1 and RRM2, as well as in other structurally known RRMs, is not involved in the predicted RNA binding surface (Figs. 1D and S2).

The C-terminal RRM of RBM45 recognizes GACG RNA

Previous studies have shown that both the RRM1 and RRM2 domains of RBM45 recognized GAC sequence RNA (14). To identify whether RBM45RRM3 recognizes the same sequence, we analyzed the binding of RBM45RRM3 with GAC-contain RNA using fluorescence polarization (FP) assay. RBM45RRM3 bound a 5′-FAM labeled 6-mer RNA, 5′-GGACGG-3′, with a much higher affinity (dissociation constant (KD) of 2.34 ± 0.38 μM) than that for a non-specific sequence (polyU) 6-mer RNA (with KD of 199.4 ± 113.5 μM) (Fig. 2A), suggesting that RBM45RRM3 binds RNA sequence-specifically and likely also prefers the GAC sequence. We also analyzed the binding affinity of RBM45RRM3 to single-stranded DNA (ssDNA). The FP results showed that RBM45RRM3 bound 5′-GGACGG-3′ ssDNA 6-mer with a KD of 8.77 ± 0.61 μM (Fig. S4), suggesting that RBM45RRM3 can also bind ssDNA, although the affinity is much weaker than that for RNA.Figure 2 The RNA-binding property of RBM45RRM3. A, the fluorescence polarization (FP) measurements of the binding affinities of RBM45RRM3 to the GGACGG sequence and polyU 6-mer RNA. B, the FP measurements of the binding affinities of RBM45RRM3 for RNA 6-mers substituted at each site. For brevity, the results of only one representative substitution for each site are shown here, and the full FP results of substitutions are shown in Fig. S5. The substituted bases are highlighted in red. The data shown here are the averages of three replicates. The error bars indicate the standard deviations of three replicates.

To identify the precise recognition sequence of RBM45RRM3, we substituted each residue of the 6-mer RNA with other nucleotides and analyzed the changes in binding affinities using the FP method. The substitutions of the second guanine, the adenine, the cytosine, or the third guanine of the 6-mer RNA resulted in ∼4-fold, 25–30-fold, 26–39-fold, or 6.5–14-fold reductions of the RBM45RRM3-binding affinity, respectively, whereas the substitutions in the first and last guanines did not significantly change the RBM45RRM3-binding affinity (Figs. 2B and S5). It should be noted that since the assays for some RNA variants failed to reach saturation of binding, the calculation of KD values for these assays should be considerably inaccurate. Nevertheless, the FP results still clearly showed that these sequence variants significantly reduced the RRM45RRM3-binding affinity of RNA. These results demonstrated that RBM45RRM3 recognizes the GACG sequence, which is slightly different from the GAC sequence recognized by RRM1 and RRM2 of RBM45.

As RBM45RRM3 was recently reported as an N6-methyladenosine (m6A) reader, we also analyzed its binding ability for m6A-containing RNA. The FP result showed that RBM45RRM3 bound 5′-FAM labeled GG(m6A)CGG 6-mer RNA with a KD of 4.62 ± 0.55 μM (Fig. S6), slightly weaker than that for unmethylated RNA, which implied that RBM45RRM3 does not exhibit specific affinity for m6A modified RNA.

Crystal structure of RBM45RRM3 in complex with ssDNA

To explore the RNA-recognition mechanism of RBM45RRM3, we attempted to co-crystallize RBM45RRM3 with RNA or ssDNA containing GACG sequence and finally solved the crystal structure of RBM45RRM3 in complex with an ssDNA 7-mer, 5′-GACGCAG-3′, at a resolution of 1.65 Å. There are also four RBM45RRM3 molecules per asymmetric unit, however, the space group and the packing pattern of the complex crystal are totally different from those of the protein-alone crystal. Clear electron densities of ssDNA can be observed at the presumed RNA binding site, allowing unambiguous ssDNA model-building (Fig. 3, A and B). The final model includes the first six nucleotides of the ssDNA 7-mer, hereafter named from G1 to A6. The structures of the four protein chains, as well as their bound ssDNAs, are very similar (Fig. S7), thus we will describe only one RBM45RRM3 chain and its bound ssDNA hereafter. The ssDNA binds to the positively charged β-sheet surface of RBM45 with its 5′ end bound to the β4 side and its 3′ end bound to the β2 side of the β-sheet, consistent with the conventional RNA binding pattern of RRMs (19). There was no direct interaction observed between the pre-RRM loop and ssDNA (Fig. 3A). Superimposition with the protein-alone structure shows that the C-terminal loop of RBM45RRM3 in the complex structure is in a different direction (Fig. 3C). This loop is directly involved in the interaction with ssDNA, suggesting that the conformational change is most likely a result of ssDNA binding. The ssDNA/RNA-binding-induced conformational changes of the C-terminal loop have also been found in other RRMs, such as RRM1 and RRM2 of RBM45 (14). The most N-terminus of RBM45RRM3 also exhibits different conformations in the two structures (Fig. 3C). However, this region does not interact with the ssDNA, and it also shows conformational variety in the protein-alone structure (Fig. S1), suggesting that this conformational inconsistency is more likely due to crystal packing. The Cα RMSD between the RBM45RRM3 structure in the complex and the AlphaFold-predicted model is about 0.30 Å (for 74 residues), comparable to that of the protein-alone structure. Interestingly, in those regions that directly interact with ssDNA, such as the C-terminal loop and the β2–β3 loop, the AlphaFold-predicted model is more similar to the complex structure, whereas in regions that do not contact ssDNA, such as the most N-terminal loop, the AlphaFold-predicted model is more similar to the protein-alone structure (Fig. S8).Figure 3 Structure of RBM45RRM3in complex with ssDNA. A, two views of the overall structure of RBM45RRM3–ssDNA complex. RBM45RRM3 is presented as a green cartoon; ssDNA is shown as magenta sticks. The simulated annealing map of ssDNA (Fo–Fc map contoured at 3.0 σ) is shown as gray meshes. B, the surface electrostatic potential of RBM45RRM3. Positively charged and negatively charged areas are indicated in blue and red, respectively. ssDNA is presented as magenta sticks. C, structural comparison of RBM45RRM3 in the complex and protein-alone structures. The complex structure and the protein-alone structure are presented as green and gray cartoons, respectively. The N-terminus and C-terminus of RBM45RRM3 in the complex structure are highlighted in blue and red, respectively. D, comparison of RBM45RRM3–ssDNA structure with RRM1–ssDNA and RRM2–ssDNA structures (PDB code 7CSZ). RRM3, RRM1, and RRM2 are presented as green, gray, and wheat cartoons, respectively; the RRM3-, RRM1-, and RRM2-bound ssDNA are presented as magenta, gray, and wheat sticks, respectively.

Compared with the structure of the RBM45RRM1-2–ssDNA complex (PDB code 7CSZ), which also contained the GACG motif, the binding patterns of A2 and C3 are remarkably consistent with the corresponding bases bound by RRM1 and RRM2. Three RRMs of RBM45 bind G1 (or the corresponding guanine) at similar sites, but the orientations of the purine rings of the three guanines are different. The purine rings of guanines bound by RRM3 or RRM1 are approximately parallel to the β-sheet, whereas the purine ring of RRM2-bound guanine is almost perpendicular to the β-sheet. The corresponding guanines of G4 bind at quite different positions in RRM1 and RRM2, and the binding of G4 in RRM3 is closer to that in RRM1 (Fig. 3D).

Structural details of nucleic acids recognition

The structure of RBM45RRM3–ssDNA complex presents the details of how RRM3 of RBM45 recognizes each base of ssDNA. Specifically, G1 binds to a hydrophobic patch formed by the side chains of Phe395 and Val397 in β1 and Met460 in β4; a γ-methyl of Val397 side chain contacts the purine ring of G1 via the CH–π interaction; the side-chain amino group of Lys458 in β4 forms hydrogen bonds with N7 and O6 of G1 (Fig. 4, A and B). The purine ring of A2 π–π stacks with the aromatic side chain of Phe395; the N1 and N3 of A2 form hydrogen bonds with the main-chain amino group of Asp463 in the C-terminal loop and the side-chain hydroxyl group of Tyr431 in β3, respectively; the N6 forms a hydrogen bond with the main-chain carbonyl group of Leu461 via a water molecule (Fig. 4, A and B). The hydroxyl group of Tyr431 contacts the pyrimidine ring of C3 through OH–π interaction; two side-chain amino groups of Arg393 in β1 form hydrogen bonds with O2 and N3 of C3; the main-chain carbonyl group of Asp463 forms a hydrogen bond with N4 of C3. The aromatic side chain of Tyr431 does not stack with C3 but instead stacks with the guanidine group of Arg393, stabilizing its conformation (Fig. 4, A and C). For G4, its purine ring π–π stacks with the side chain of Tyr422, whereas its N1 and N2 form two hydrogen bonds with the side-chain carboxyl group of Glu420 in β2 and its O6 forms a hydrogen bond with the side-chain amino group of Lys433 in β3. In addition to interactions with the base moieties of ssDNA, RBM45RRM3 also contacts the backbone of ssDNA. The ribose moieties of G1 to G4 bind to a hydrophobic patch formed by side chains of Val429, Phe395, and Val424, whereas the phosphate of A2 forms hydrogen bonds with the main-chain amino of N428 via a water molecule (Fig. 4, A–C). The densities of C5 and A6 are weak (Fig. 3A), suggesting that the structures of C5 and A6 are largely the result of crystal packing. C5 interacts weakly with RBM45RRM3, except that the phosphate of C5 forms two hydrogen bonds with the main-chain amino group and the side-chain hydroxyl group of Ser425 (Fig. 4, A and C), whereas A6 does not directly contact the protein (Fig. 3, A and B). It should be noted that introducing 2′-hydroxyl groups to ssDNA in the complex structure does not result in any observable steric hindrance, which suggests that our complex structure might also represent the binding of RNA. Furthermore, the fact that RBM45RRM3 binds RNA with significantly stronger affinity than DNA suggests that the 2′-hydroxyl groups may introduce additional interactions, although these interactions are hard to extrapolate from our protein-ssDNA complex structure. Interestingly, the side chains of several key amino acids, including Phe395, Arg393, and Tyr431, exhibited distinct conformations in the complex structure compared with those in the protein-alone structure (Fig. S9), implying that the side chains of these residues may undergo conformational changes during RBM45RRM3 bind RNA.Figure 4 Structural details of nucleic acids recognition. A, schematic diagram of RRM45RRM3-ssDNA interaction. The bases, deoxyriboses, and phosphates are represented as rectangles, pentagons, and circles, respectively. The recognized bases are highlighted in yellow background. B, interactions of RBM45RRM3 with G1 and A2. C, interactions of RBM45RRM3 with C3 and G4. Protein chains are presented as green cartoons; nucleotide residues are presented as magenta sticks; amino acid residues that interacted with ssDNA are presented as green sticks; hydrogen bonds are indicated by magenta dashed lines.

Binding assays of the interaction surface

To validate the ssDNA/RNA-binding mechanism revealed by our structural results, we generated single mutants of several key amino acid residues and measured their RNA-binding affinities using the FP method. Single mutations of two aromatic residues, Phe395 and Tyr422, which stack with the purine rings of A2 and G4, respectively, to alanine reduced the RNA-binding affinity by ∼10- and ∼25-fold, respectively (Fig. 5A), suggesting that they both play important roles in RNA-binding. Mutating Tyr431 to phenylalanine or alanine reduced the RNA-binding affinity by 1.8- or ∼50-fold, respectively (Fig. 5A), suggesting that in addition to the direct interaction of its side-chain hydroxyl group with the adenine and cytosine bases, the stabilization of Arg393 side-chain conformation may be more important for RNA binding. To verify the importance of Arg393, which forms two hydrogen bonds with the base of cytosine, in RNA binding, we attempted to generate the R393A mutant, but it was not expressed well in E. coli. However, we successfully obtained the R393Q mutant protein, which showed an approximately 10-fold decrease in RNA-binding affinity, proving that Arg393 is critical for RNA binding (Fig. 5B). For Lys458, which forms hydrogen bonds with the first guanine of the GACG sequence, the K458A mutation reduced RNA-binding affinity by ∼11-fold. For Glu420 and Lys433, which form hydrogen bonds with the last guanine of the GACG sequence, the E420A and K433A mutations reduced the RNA-binding affinities by ∼5- and 12-fold, respectively (Fig. 5B). These results indicate that these amino acid residues that form hydrogen bonds with bases also play crucial roles in RNA binding. We would like to caution that the binding assays for some mutants did not reach saturation, therefore, the KD calculations for these assays might be inaccurate. However, our FP results still clearly showed that these mutations significantly reduced the RNA-binding affinity of RBM45RRM3.Figure 5 RNA-binding affinities of RBM45RRM3mutants. A, FP measurements of RNA-binding affinities of three key aromatic amino acid mutants. B, FP measurements of RNA-binding affinities of several other key residue mutants. The data shown here are the averages of three replicates. The error bars indicate the standard deviations of three replicates.

Discussion

In this study, we verified that the RRM3 domain of RBM45 binds RNA in a sequence-specific manner like its RRM1 and RRM2 domains, and identified that RBM45RRM3 recognizes RNA sequence GACG, which is slightly different from the GAC sequence recognized by RRM1 and RRM2. Our structural and biochemical results showed that three aromatic residues, Phe395 in β1, Tyr431 in β3, and Tyr422 in β2, play crucial roles in RNA binding. Two of them, Phe395 and Tyr422, stack with A2 and G4, respectively, whereas Tyr431 does not stack with any base but contacts the pyrimidine ring of C3 through OH–π interaction and stacks with Arg393, another key amino acid residue for RNA binding. Stackings of these aromatic side chains with the bases provide the essential driving force to bind RNA. Single mutations in these aromatic residues significantly weakened the RNA-binding affinity. Several other amino acid residues form hydrogen bonds with bases through their side-chain or main-chain groups, further promoting the binding to RNA and supposedly providing sequence-specificity (20). Among them, for those residues that interact with bases through side chains, including Arg393, Glu420, Lys433, and Lys458, their single mutations also significantly reduced the RNA-binding affinities. These residues, as well as the three aromatic residues described above, are conserved in RBM45s from Drosophila to humans (Fig. S10), suggesting that the mechanism of RNA recognition by RBM45RRM3 should be conserved across species. Mutation in one of these residues, Lys458, has been found linked to ALS and FTLD (6), suggesting that the RNA-binding ability/specificity of RBM45 is likely to play an important role in the development of these neurodegenerative diseases.

Our biochemical results showed that RBM45RRM3 exhibited different specificities for different nucleotides among the recognized GACG sequences. RBM45RRM3 is highly selective for the central dinucleotide of the GACG sequence, adenine and cytosine, as substituting any of them with any other nucleotides resulted in a more than 25-fold decrease in affinity. In contrast, substituting the first guanine in the GACG sequence with other nucleotides only reduces the affinity by ∼4-fold. RBM45RRM3 is moderately selective for the latter guanine, with a 6.5–14-fold higher affinity for guanine than for other nucleotides. Our structural result revealed that the adenine and cytosine are bound to RBM45RRM3 through interactions between their bases and the conserved aromatic residues in RNP2 and RNP1, respectively, which is the classical mechanism for nucleotide binding by RRM domains (19). In addition, two bases each form three hydrogen bonds with the main-chain or side-chain atoms of RBM45RRM3. The base of the last guanine stacks with the aromatic side chain of Tyr422, a non-classical binding site in β2, and forms three hydrogen bonds with side-chain atoms of RBM45RRM3. In contrast, the first guanine base interacts less with the protein; it does not stack with any aromatic residues and only forms two hydrogen bonds with the protein through the side chain amino group of Lys458.

RBM45RRM3 shares a portion of its recognition sequence, GAC, with RRM1 and RRM2. However, the mechanisms of the three RRMs recognizing these three nucleotides are not exactly the same. The recognition of the core dinucleotide, the adenine and the cytosine, are conserved among the three RRMs of RBM45. The two nucleotides recognized by RBM45RRM3 can be superimposed well with those bound by RRM1 and RRM2. The purine rings of adenines all stack with the conserved phenylalanine in RNP2, and each purine ring forms two hydrogen bonds with main-chain groups in the C-terminal loop. A minor difference is that the bases of RRM3-and RRM2-bound adenine each form an additional hydrogen bond with the side-chain hydroxyl group of the conserved tyrosine in RNP1, whereas there is no such hydrogen bond in RRM1-bound adenine, as the corresponding aromatic residue in RRM1 is phenylalanine (Figs. 1B and 6B). For the cytosine, the pyrimidine rings all form two hydrogen bonds with the side chain of the arginine (Arg393/Arg27/Arg122) in RNP2 and one hydrogen bond with a main-chain carbonyl group in the C-terminal loops, and all interact with the side chain of the conserved aromatic residues (Y431/F70/Y165) in RNP1 (Fig. 6C). In contrast, the three RRMs interact with the first guanine quite differently. Comparatively, the binding of RRM3 is more similar to that of RRM1. However, the base of the RRM1-bound guanine forms two additional hydrogen bonds with the side-chain carboxyl group of Asp114 in the C-terminal loop. The purine ring of RRM2-bound guanine exhibits a different orientation. It is sandwiched between the side chains of Met126 and Ile188 and forms two hydrogen bonds with the main-chain carbonyls of Phe124 and Arg186 (Fig. 6A). The observed interaction of RRM3 with the base of this guanine is less than that of either RRM1 or RRM2, consistent with the fact that RRM3 is relatively less selective for the first guanine of its recognized sequence. Although RRM1 does not sequence-specifically recognize the last guanine of GACG as RRM3 does, binding of this guanine is also observed in the RRM1–ssDNA structure. Interestingly, the details of the two RRMs binding this guanine are similar. The RRM1-bound guanine also stacks with the side chain of an aromatic residue (Trp55) in α2 and forms two hydrogen bonds with an acidic residue (Asp53) in α2 as the RRM3-bound guanine does. The RRM3-bound guanine forms an additional hydrogen bond with the side chain of a lysine residue (Lys433) in α3, whereas the RRM1-bound guanine interacts with the side chain of an arginine residue (Arg27) in α1 through a water molecule (Fig. 6D). However, although RRM1 binds this guanine in a manner similar to that of RRM3, biochemical assays showed that its binding to this guanine is sequence nonspecific (14). One possible explanation might be that stacking with tryptophan and hydrogen bonding mediated by water may confer more flexibility to the binding site of RRM1, thus allowing it to bind other nucleotides, whereas this binding site in RRM3 is less flexible and only suitable for binding guanine.Figure 6 Structural comparison of the ssDNA binding detail of three RRMs of RBM45. A–C, the structural comparisons of G1, A2, and C3 binding of three RRMs, respectively. D, the structural comparison of G4 binding of RRM3 and RRM1. The key amino residues in RRM3, RRM1 (PDB code 7CSZ), and RRM2 (PDB code 7CSZ) are presented as green, gray, and wheat sticks, respectively; the RRM3-, RRM1-, and RRM2-bound ssDNA are presented as magenta, gray, and wheat sticks, respectively; the hydrogen bonds in RRM3–ssDNA, RRM1–ssDNA, and RRM3–ssDNA complexes are indicated by magenta, gray, and wheat dashed lines, respectively.

RBM45 is reportedly an m6A reader that recognizes m6A through its RRM3 domain (5). However, our binding assays showed that RBM45RRM3 did not preferentially target m6A. Our structural data also did not support that RBM45RRM3 prefers to recognize m6A. Our complex structure indicated that the N6 methyl group of m6A in the syn conformation, which is the major conformation of m6A (21, 22), neither introduces additional interactions nor leads to steric hindrance (Fig. S11), suggesting that RBM45RRM3 is unlikely to be selective for the N6 methylation of adenine. There are two possible reasons for the inconsistency of our results with the previously reported results: one is that the HOA domain, which is not included in our biochemical and structural studies, may affect the binding of RRM3 to m6A; the other is that N6 methylation of adenine may destabilize the Watson-Crick base pairing, affecting the secondary structure of RNA and predisposing the N6 methylated GACG motifs to adopt single-stranded conformation (23, 24), which would be more favorable for RBM45 binding. Either reason would make full-length RBM45 prefer to bind N6-methylated RNAs in vivo, although a single RRM domain does not exhibit any preference for N6-methylation modification of short RNA fragments in vitro.

RBM45 contains three RRM domains, which all have the ability to bind RNA. Their recognition sequences are similar, but not identical. Containing multiple RRM domains is common in RBPs. The RNA binding specificity and affinity of individual RRM domains are usually limited, and the coordination of multiple RRMs will help to improve the RNA-binding specificity and/or affinity (12, 25). Although for most multiple-RRM RBPs, the RNA sequences recognized by their individual RRM domains are not well characterized to date, available studies with clear biochemical and/or structural evidence have shown that different RRMs of the same RBP may either recognize similar sequences or very different sequences. For example, all three RRMs of HuR bind polyU sequences (26, 27, 28), and its RRM3 can also bind UUA (28); the RRM1, RRM3, and RRM4 of PTB all recognize UCU, while its RRM2 recognizes CU (29); RRM1 and RRM2 of TDP-43 recognize GGUG and UG, respectively (30); and RRM1 and RRM2 of hnRNP A1 recognize UAG(G) and (U)AG (the sequences in parentheses denote the variations in the results of different studies), respectively (31, 32, 33). The recognition sequences of the different RRMs in these RBPs are identical or similar and share a core sequence, UU for HuR, CU for PTB, UG for TDP-43, and (U)AG for hnRNP A1. This is also the case for RBM45; although the sequences recognized by the three RRMs are not identical, they share a core trinucleotide, GAC. In contrast, for some RBPs, the sequences recognized by their different RRMs are quite different. For example, RRM1 of RBM5 recognizes GG (34), while its RRM2 binds CU or GA (35); RRM2 and RRM3 of Prp24 recognize (G)AG(A) and AAC, respectively, while its RRM1 and RRM4 do not bind RNA (36, 37, 38). While the N-terminal tandem RRM domains (RRM1 and RRM2) interact with each other to form two antiparallel RNA-binding sites both binding GAC sequences (14), the C-terminal RRM domain (RRM3) is predicted not to interact tightly with any other domains (18), resulting in a separate binding site recognizing GACG sequence. The 3-nucleotide or 4-nucleotide recognition sequence implies a limited sequence specificity. Whereas multiple RRM domains with contiguous RNA-binding sites improve both the sequence-specificity and affinity of RBPs, multiple RRM domains that do not form contiguous binding sites, as in the case of RBM45, can only improve the RNA-binding affinity and capacity (19, 20, 39). RBM45 most likely prefers to bind RNAs containing multiple GAC or GACG motifs. RBM45 was identified to recognize GAC or bipartite GAC sequences RNAs in vivo (5), consistent with the expectation based on our biochemical and structural results for individual RRM domains. Transcripts of two genes associated with neurodevelopment, NTRK2 and WNT3, were recently reported as targets of RBM45. RBM45 was found to bind to the introns of these transcripts, introns 14 and 16 of NTRK2 pre-mRNA, and intron 2 of WNT3 pre-mRNA (5), all of which contain a large number of GAC and GACG motifs. These in vivo results support our hypothesis that RBM45 tends to bind RNAs containing multiple GAC or GACG motifs. However, because of the short sequence lengths, multiple GAC and GACG sequences are expected to present in most RNAs with thousands or more nucleotides, which is common in introns. Obviously, containing multiple GAC or GACG sequences is not enough to make it a target of RBM45. Our present and previous (14) biochemical and structural results showed that all RRMs of RBM45 bind RNA in extended conformation; therefore, whether an RNA molecule can be a target of RBM45 depends not only on its sequence but also on its secondary structure. As the shared recognition sequence of the three RRM domains of RBM45, GAC matches the sequence feature of the most common m6A modification site "DRACH" or "RRACH" motif (40, 41, 42, 43, 44), coupled with the fact that m6A-modified RNA tends to adopt single-stranded conformation (23), no matter whether or not the C-terminal RRM domain of RBM45 is a real m6A reader, RBM45 would likely prefer to bind RNAs containing the m6A modification, as demonstrated in a recently published work (5). In this manner, RBM45 could potentially mediate the regulation of RNA splicing and/or post-splicing events by m6A modification.

Experimental procedures

Cloning, protein expression, and purification

The DNA encoding the RRM3 domain of human RBM45 (RBM45RRM3, residues 370–476) was obtained using the Polymerase Chain Reaction (PCR) method from a codon-optimized full-length RBM45 gene synthesized by Sangon Biotech (Shanghai China) and cloned into pET-22b (+) vector with a C-terminal His-tag using restriction enzymes NdeI and XhoI. The recombinant plasmid was transformed into Escherichia. coli BL21 (DE3) cells and culture in LB medium containing 50 ug/ml ampicillin at 37 °C until OD600 reaches 0.6 to 0.8. Protein expression was induced by adding 0.3 mM isopropyl-β-D-thiogalactoside (IPTG). Cells were further incubated at 16 °C for 20 h and then collected by centrifugation. For protein purification, cells were suspended in a lysis buffer containing 20 mM Tris HCl, pH7.5, and 200 mM NaCl, and lysed by sonication. After centrifuging at 12,000 rpm for 60 min, the supernatant was loaded to a Ni-Chelating SFF column (Cytiva) and eluted with lysis buffer containing 200 mM imidazole. The eluate was desalted and loaded to an SP SFF column (Cytiva) equilibrated with a buffer containing 20 mM Tris HCl, pH7.5, and 50 mM NaCl, and eluted by a NaCl gradient. The purified protein was concentrated by centrifugal ultrafiltration for crystallization and biochemical experiments. The mutants of RBM45RRM3 were expressed and purified by the same method using plasmids generated by site-directed mutagenesis PCR. The primers used for molecular cloning are listed in Table S1.

RNA and DNA oligonucleotides

The 5′-FAM labeled RNA and single-stranded DNA (ssDNA) used for fluorescence polarization assays and the ssDNA used for co-crystallization were purchased from General Biosystems.

Crystallization

Crystallizations of RBM45RRM3 and RBM45RRM3–ssDNA complex were performed by sitting-drop vapor diffusing method. For the crystallization of RBM45RRM3, the protein was concentrated to 16.8 mg/ml in a buffer containing 20 mM Tris HCl, pH 7.5, and 180 mM NaCl. The beat crystals were obtained in a solution containing 20% (w/v) polyethylene glycol (PEG) 3350, 0.1 M Sodium citrate, pH 5.5, and 18% (v/v) 2-propanol at 16 °C. The crystal was soaked in a reservoir solution with 10% (v/v) glycerol added for several seconds and then flash-frozen in liquid nitrogen before data collection. For co-crystallization with ssDNA, the protein was mixed with a 7-nt 5′-GACGCAG-3′ ssDNA at a molar ratio of 1:1.2 and incubated on ice for 30 min before crystallization. The best crystals were obtained in a solution containing 0.1 M Sodium succinate, pH 5.3, and 8% (w/v) PEG 2000 at 16 °C. The crystal was soaked in a reservoir solution with 20% (v/v) glycerol for several seconds and flash-frozen in liquid nitrogen before data collection.

Data collection and structure determination

All the X-ray diffraction data were collected at beamline BL18U of the Shanghai Synchrotron Radiation Facility (SSRF) at a wavelength of 0.97915 Å, using a Pilatus 6M detector, and processed with the HKL3000 package (45). The crystals of RBM45RRM3 and RBM45RRM3–ssDNA belonged to the C2221 space group and P21 space group, respectively. The structures were solved by molecular replacement method with PHASER (46) in the CCP4 suite (47), using a model predicted by AlphaFold2 (18) as the search model. The models were then refined using REFMAC5 (48) and rebuilt with COOT (49). Clear electron density of DNA was observed in the map of RBM45RRM3–ssDNA structure, allowing unambiguous DNA model building with COOT. The final models were refined with PHENIX (50). The statistics of data collection and structure refinement are shown in Table 1.Table 1 X-ray data collection and structure refinement statistics

Structures	RBM45RRM3	RBM45RRM3‒ssDNA	
Data collection			
 Wavelength (Å)	0.97915	0.97915	
 Space group	C2221	P21	
 Cell dimensions			
 a, b, c (Å)	71.26, 74.60, 195.84	42.40, 83.77, 77.76	
 α, β, γ (°)	90, 90, 90	90, 96.87, 90	
 Resolution (Å)	50.00–2.40 (2.49–2.40)	50.00–1.65 (1.71–1.65)	
 Rmerge	0.148 (1.357)	0.121 (0.974)	
 I/σI	16.9 (2.0)	12.4 (2.7)	
 CC1/2	0.996 (0.714)	0.998 (0.841)	
 Completeness (%)	100.0 (100.0)	99.8 (99.8)	
 Redundancy	9.7 (10.2)	5.0 (5.1)	
 Total/Unique reflections	200,735 (20,694)	327,484 (64,857)	
Refinement			
 Resolution (Å)	35.49–2.41	38.97–1.65	
 No. reflections	20,388	64,589	
 Rwork/Rfree	0.183/0.221	0.158/0.207	
 No. atoms			
 Protein	3069	3235	
 DNA	/	450	
 Ligand	2	18	
 Water	219	636	
 B-factors (Å2)			
 Protein	39.3	18.8	
 DNA	/	31.2	
 Ligand	58.4	57.2	
 Water	42.6	36.7	
 R.m.s. deviations			
 Bond lengths (Å)	0.004	0.004	
 Bond angles (°)	0.694	0.741	
 MolProbity score	1.27	1.01	
 Ramachandran plot			
 Favored	97.9	99.0	
 Allowed	2.1	1.0	
 Outliers	0	0	
∗Values in parentheses are for the highest-resolution shell.

Fluorescence polarization analysis

The fluorescence polarization (FP) assay was used to analyze the RNA/ssDNA-binding affinities of RBM45RRM3 and its mutants. The FAM-labeled RNA or ssDNA at a final concentration of 100 nM was incubated with increasing concentrations of protein in binding buffer containing 10 mM Tris-HCl, pH 7.5, 200 mM NaCl, 1 mM DTT, 1 mM EDTA, and 5% (v/v) glycerol in dark at 25 °C for 30 min. The 535 nm fluorescence polarizations were measured by the SpectraMax Paradigm Multi-Mode detection platform (Molecular Devices) with an excitation wavelength of 485 nm. Three parallel measurements were performed for each sample. The data were fitted to the one-site model using the GraphPad software.

Data availability

The structures of RBM45RRM3 and RBM45RRM3–ssDNA have been deposited into the Protein Data Bank (PDB) with accession numbers 8WQ3 and 8WQ5, respectively.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Supporting information

Supplementary data

Acknowledgments

We thank the staff at beamlines BL18U and BL19U1 of the National Facility for Protein Science in Shanghai (NFPS) at the Shanghai Synchrotron Radiation Facility (SSRF) for assistance during X-ray data collection. We thank the staff at the core facility of the Institute of Health Sciences & Technology, Anhui University, for providing technical support.

Author contributions

Q. W., X. C., Z. Y., X. C., M. J. and S. G. investigation; Z. Y., X. C., and M. W. methodology. X. C. and M. W. data curation, X. C. and M. W. validation; X. C. visualization; X. C. and M. W. writing–original draft; M. W. conceptualization; M. W. funding acquisition; M. W. project administration; M. W. supervision; M. W. visualization; M. W. writing–review & editing.

Funding and additional information

This work was supported by the 10.13039/501100003995 Natural Science Foundation of Anhui Province (2108085MC75 to M. W.).
==== Refs
References

1 Tamada H. Sakashita E. Shimazaki K. Ueno E. Hamamoto T. Kagawa Y. cDNA cloning and characterization of Drb1, a new member of RRM-type neural RNA-binding protein Biochem. Biophys. Res. Commun. 297 2002 96 104 12220514
2 Cooper-Knock J. Robins H. Niedermoser I. Wyles M. Heath P.R. Higginbottom A. Targeted genetic screen in amyotrophic lateral sclerosis reveals novel genetic variants with synergistic effect on clinical phenotype Front. Mol. Neurosci. 10 2017 370 29170628
3 Li Y. Collins M. Geiser R. Bakkar N. Riascos D. Bowser R. RBM45 homo-oligomerization mediates association with ALS-linked proteins and stress granules Sci. Rep. 5 2015 14262
4 Mashiko T. Sakashita E. Kasashima K. Tominaga K. Kuroiwa K. Nozaki Y. Developmentally regulated RNA-binding protein 1 (Drb1)/RNA-binding motif protein 45 (RBM45), a nuclear-cytoplasmic trafficking protein, forms TAR DNA-binding protein 43 (TDP-43)-mediated cytoplasmic aggregates J. Biol. Chem. 291 2016 14996 15007 27226551
5 Choi S.H. Flamand M.N. Liu B. Zhu H. Hu M. Wang M. RBM45 is an m(6)A-binding protein that affects neuronal differentiation and the splicing of a subset of mRNAs Cell Rep. 40 2022 111293
6 van der Zee J. Dillen L. Baradaran-Heravi Y. Gossye H. Kocoglu C. Cuyt I. Family-based exome sequencing identifies RBM45 as a possible candidate gene for frontotemporal dementia and amyotrophic lateral sclerosis Neurobiol. Dis. 156 2021 105421
7 Collins M. Riascos D. Kovalik T. An J. Krupa K. Krupa K. The RNA-binding motif 45 (RBM45) protein accumulates in inclusion bodies in amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration with TDP-43 inclusions (FTLD-TDP) patients Acta Neuropathol. 124 2012 717 732 22993125
8 Neumann M. Sampathu D.M. Kwong L.K. Truax A.C. Micsenyi M.C. Chou T.T. Ubiquitinated TDP-43 in frontotemporal lobar degeneration and amyotrophic lateral sclerosis Science 314 2006 130 133 17023659
9 Gong J. Huang M. Wang F. Ma X. Liu H. Tu Y. RBM45 competes with HDAC1 for binding to FUS in response to DNA damage Nucleic Acids Res. 45 2017 12862 12876 29140459
10 Kwiatkowski T.J. Jr. Bosco D.A. Leclerc A.L. Tamrazian E. Vanderburg C.R. Russ C. Mutations in the FUS/TLS gene on chromosome 16 cause familial amyotrophic lateral sclerosis Science 323 2009 1205 1208 19251627
11 Vance C. Rogelj B. Hortobagyi T. De Vos K.J. Nishimura A.L. Sreedharan J. Mutations in FUS, an RNA processing protein, cause familial amyotrophic lateral sclerosis type 6 Science 323 2009 1208 1211 19251628
12 Lunde B.M. Moore C. Varani G. RNA-binding proteins: modular design for efficient function Nat. Rev. Mol. Cell Biol. 8 2007 479 490 17473849
13 Corley M. Burns M.C. Yeo G.W. How RNA-binding proteins interact with RNA: molecules and mechanisms Mol. Cell 78 2020 9 29 32243832
14 Chen X. Yang Z. Wang W. Qian K. Liu M. Wang J. Structural basis for RNA recognition by the N-terminal tandem RRM domains of human RBM45 Nucleic Acids Res. 49 2021 2946 2958 33577684
15 Wang J. Ganaie S.S. Cheng F. Xu P. Ning K. Wang X. RNA binding motif protein RBM45 regulates expression of the 11-Kilodalton protein of parvovirus B19 through binding to novel intron splicing enhancers mBio 11 2020 e00192-20
16 Ray D. Kazan H. Cook K.B. Weirauch M.T. Najafabadi H.S. Li X. A compendium of RNA-binding motifs for decoding gene regulation Nature 499 2013 172 177 23846655
17 Dominguez D. Freese P. Alexis M.S. Su A. Hochman M. Palden T. Sequence, structure, and context preferences of human RNA binding proteins Mol. Cell 70 2018 854 867 29883606
18 Tunyasuvunakool K. Adler J. Wu Z. Green T. Zielinski M. Zidek A. Highly accurate protein structure prediction for the human proteome Nature 596 2021 590 596 34293799
19 Maris C. Dominguez C. Allain F.H. The RNA recognition motif, a plastic RNA-binding platform to regulate post-transcriptional gene expression FEBS J. 272 2005 2118 2131 15853797
20 Auweter S.D. Oberstrass F.C. Allain F.H. Sequence-specific binding of single-stranded RNA: is there a code for recognition? Nucleic Acids Res. 34 2006 4943 4959 16982642
21 Engel J.D. von Hippel P.H. Effects of methylation on the stability of nucleic acid conformations: studies at the monomer level Biochemistry 13 1974 4143 4158 4606508
22 Liu B. Shi H. Rangadurai A. Nussbaumer F. Chu C.C. Erharter K.A. A quantitative model predicts how m(6)A reshapes the kinetic landscape of nucleic acid hybridization and conformational transitions Nat. Commun. 12 2021 5201 34465779
23 Roost C. Lynch S.R. Batista P.J. Qu K. Chang H.Y. Kool E.T. Structure and thermodynamics of N6-methyladenosine in RNA: a spring-loaded base modification J. Am. Chem. Soc. 137 2015 2107 2115 25611135
24 Liu N. Dai Q. Zheng G. He C. Parisien M. Pan T. N(6)-methyladenosine-dependent RNA structural switches regulate RNA-protein interactions Nature 518 2015 560 564 25719671
25 Afroz T. Cienikova Z. Cléry A. Allain F.H.T. One, two, three, four! How multiple RRMs read the genome sequence Methods Enzymol. 558 2015 235 278 26068744
26 Wang H. Zeng F. Liu Q. Liu H. Liu Z. Niu L. The structure of the ARE-binding domains of Hu antigen R (HuR) undergoes conformational changes during RNA binding Acta Crystallogr. D Biol. Crystallogr. 69 2013 373 380 23519412
27 Pabis M. Popowicz G.M. Stehle R. Fernández-Ramos D. Asami S. Warner L. HuR biological function involves RRM3-mediated dimerization and RNA binding by all three RRMs Nucleic Acids Res. 47 2019 1011 1029 30418581
28 Ripin N. Boudet J. Duszczyk M.M. Hinniger A. Faller M. Krepl M. Molecular basis for AU-rich element recognition and dimerization by the HuR C-terminal RRM Proc. Nat. Acad. Sci. U. S. A. 116 2019 2935 2944
29 Oberstrass F.C. Auweter S.D. Erat M. Hargous Y. Henning A. Wenter P. Structure of PTB bound to RNA: specific binding and implications for splicing regulation Science 309 2005 2054 2057 16179478
30 Lukavsky P.J. Daujotyte D. Tollervey J.R. Ule J. Stuani C. Buratti E. Molecular basis of UG-rich RNA recognition by the human splicing factor TDP-43 Nat. Struct. Mol. Biol. 20 2013 1443 1449 24240615
31 Ding J. Hayashi M.K. Zhang Y. Manche L. Krainer A.R. Xu R.M. Crystal structure of the two-RRM domain of hnRNP A1 (UP1) complexed with single-stranded telomeric DNA Genes Dev. 13 1999 1102 1115 10323862
32 Kooshapur H. Choudhury N.R. Simon B. Muhlbauer M. Jussupow A. Fernandez N. Structural basis for terminal loop recognition and stimulation of pri-miRNA-18a processing by hnRNP A1 Nat. Commun. 9 2018 2479 29946118
33 Beusch I. Barraud P. Moursy A. Clery A. Allain F.H. Tandem hnRNP A1 RNA recognition motifs act in concert to repress the splicing of survival motor neuron exon 7 Elife 6 2017 e25736 28650318
34 Soni K. Jagtap P.K.A. Martinez-Lumbreras S. Bonnal S. Geerlof A. Stehle R. Structural basis for specific RNA recognition by the alternative splicing factor RBM5 Nat. Commun. 14 2023 4233 37454201
35 Song Z. Wu P. Ji P. Zhang J. Gong Q. Wu J. Solution structure of the second RRM domain of RBM5 and its unusual binding characters for different RNA targets Biochemistry 51 2012 6667 6678 22839758
36 Martin-Tumasz S. Reiter N.J. Brow D.A. Butcher S.E. Structure and functional implications of a complex containing a segment of U6 RNA bound by a domain of Prp24 RNA 16 2010 792 804 20181740
37 Montemayor E.J. Curran E.C. Liao H.H. Andrews K.L. Treba C.N. Butcher S.E. Core structure of the U6 small nuclear ribonucleoprotein at 1.7-A resolution Nat. Struct. Mol. Biol. 21 2014 544 551 24837192
38 Montemayor E.J. Didychuk A.L. Yake A.D. Sidhu G.K. Brow D.A. Butcher S.E. Architecture of the U6 snRNP reveals specific recognition of 3'-end processed U6 snRNA Nat. Commun. 9 2018 1749 29717126
39 Clery A. Blatter M. Allain F.H. RNA recognition motifs: boring? Not quite Curr. Opin. Struct. Biol. 18 2008 290 298 18515081
40 Wei C.M. Moss B. Nucleotide sequences at the N6-methyladenosine sites of HeLa cell messenger ribonucleic acid Biochemistry 16 1977 1672 1676 856255
41 Meyer K.D. Saletore Y. Zumbo P. Elemento O. Mason C.E. Jaffrey S.R. Comprehensive analysis of mRNA methylation reveals enrichment in 3' UTRs and near stop codons Cell 149 2012 1635 1646 22608085
42 Dominissini D. Moshitch-Moshkovitz S. Schwartz S. Salmon-Divon M. Ungar L. Osenberg S. Topology of the human and mouse m6A RNA methylomes revealed by m6A-seq Nature 485 2012 201 206 22575960
43 Jenjaroenpun P. Wongsurawat T. Wadley T.D. Wassenaar T.M. Liu J. Dai Q. Decoding the epitranscriptional landscape from native RNA sequences Nucleic Acids Res. 49 2021 e7 32710622
44 Kortel N. Ruckle C. Zhou Y. Busch A. Hoch-Kraft P. Sutandy F.X.R. Deep and accurate detection of m6A RNA modifications using miCLIP2 and m6Aboost machine learning Nucleic Acids Res. 49 2021 e92 34157120
45 Minor W. Cymborowski M. Otwinowski Z. Chruszcz M. HKL-3000: the integration of data reduction and structure solution--from diffraction images to an initial model in minutes Acta Crystallogr. D Biol. Crystallogr. 62 2006 859 866 16855301
46 McCoy A.J. Grosse-Kunstleve R.W. Adams P.D. Winn M.D. Storoni L.C. Read R.J. Phaser crystallographic software J. Appl. Crystallogr. 40 2007 658 674 19461840
47 Winn M.D. Ballard C.C. Cowtan K.D. Dodson E.J. Emsley P. Evans P.R. Overview of the CCP4 suite and current developments Acta Crystallogr. D Biol. Crystallogr. 67 2011 235 242 21460441
48 Murshudov G.N. Skubak P. Lebedev A.A. Pannu N.S. Steiner R.A. Nicholls R.A. REFMAC5 for the refinement of macromolecular crystal structures Acta Crystallogr. D Biol. Crystallogr. 67 2011 355 367 21460454
49 Emsley P. Cowtan K. Coot: model-building tools for molecular graphics Acta Crystallogr. D Biol. Crystallogr. 60 2004 2126 2132 15572765
50 Liebschner D. Afonine P.V. Baker M.L. Bunkoczi G. Chen V.B. Croll T.I. Macromolecular structure determination using X-rays, neutrons and electrons: recent developments in Phenix Acta Crystallogr. D Struct. Biol. 75 2019 861 877 31588918
