Splicing site recognition by synergy of three domains in the splicing factor RBM10

Abstract

The splicing factor RBM10 and the close homologues RBM5 and RBM6 govern the splicing of oncogenes such as Fas, NUMB and Bcl-X. The molecular architecture of these proteins includes zinc fingers (ZnF) and RNA recognition motifs (RRM). Three of these domains in RBM10, which constitute the RNA-binding part of this splicing factor, were found to individually bind RNAs with μM affinities. It was thus of interest to further investigate the structural basis of the well-documented high affinity RNA-binding by RMB10. Here, we investigated RNA-binding by combinations of two or three of these domains, and we thus discovered that a polypeptide containing RRM1, ZnF1 and RRM2 connected by their natural linkers recognizes specific sequences of the Fas exon 6 mRNA with 20 nM affinity. NMR structures of the RBM10 domains RRM1, ZnF1, and the V354del variant of RRM2 further confirmed that the interactions with RNA are driven by canonical RNA recognition elements. The well-known high-fidelity RNA splice site recognition by RBM10, and probably by RBM5 and RBM6, can thus be largely rationalized by a cooperative action of RRM and ZnF domains.

INTRODUCTION

The splicing factor RBM10 modulates the cellular isoform rates of multiple apoptotic genes, such as Fas¹, NUMB² and Bcl-x,^{3, 4} and has been linked to the onset of multiple types of cancer.^{5, 6} In their physiological functions, RBM5, RBM10 and the more distant homologue RBM6 mediate exon skipping as well as inclusion events, and participate in the processing of multiple oncogenes. While in some alternative splicing events the involvement of the three RBM proteins favors the same gene product, the individual splicing factors may also lead to the production of different isoforms.² For example, in the alternative splicing of Fas, RBM5 and RBM10 behave similarly, by promoting exon 6 skipping and thus increasing the yield of an antiapoptotic isoform. In contrast, in the processing of NUMB, RBM10 induces exon 9 inclusion, while RBM5 increases skipping.^{1, 2, 4, 7–9}

RBM10, RBM5 and RBM6 share similar domain architectures, which include an N-terminal RNA recognition region consisting of a zinc finger (ZnF1) flanked by two RRMs (RRM1 and RRM2),^10–12 and a C-terminal protein-interacting region containing a second zinc finger (ZnF2), an OCRE sequence motif^{13, 14} and a G-patch motif (Fig. 1A). The RNA recognition domains in RBM5 and RBM10 have about 60% sequence identity, recognize similar splice sites, and together with the OCRE domain^15–17 govern Fas isoform ratios. In this manuscript we report on recognition of the exon 6 in Fas by cooperative action of the RRM1, ZnF1 and RRM2 domains of RBM10, and present NMR structures of the individual RBM10 RNA-binding domains.

Figure 1. RBM5/6/10 domain architectures and recognition of exon 6 from Fas by RBM10 fragments.

(A) Schematic representation of the domain architectures of the three proteins. The three globular domains studied here are highlighted in color. At the bottom, the RRM1–ZnF1 and RRM1–ZnF1–RRM2 polypeptide fragments of RBM10 used for RNA binding assays are indicated by thick colored lines and indication of the chain ends. (B) Protein–RNA association curves with a 22-nucleotide RNA sequence from exon 6 of Fas, UAAUUGUUUGGGGUAAGUUCUU. The data were obtained using nitrocellulose binding assays with the RBM10 polypeptide fragments indicated in the figure. The same color code is used as in panel (A). Average values from three independent experiments are shown as geometric symbols, and the standard deviations are represented as vertical bars. K_D values obtained with Hill-equation fitting are indicated in the lower right.

RESULTS

The biological functions of RBM5, RMB6 and RBM10 indicate that these splicing factors bind to mRNA with high affinity.^{2, 4, 7, 9, 17–19} Here, we set out to investigate the structural basis of these implicated interactions. Working with RBM10, we measured the binding affinities for RNA fragments from exon 6 of Fas of the individual RRMs, the zinc finger, and of combinations of two and three of these globular domains. We also determined NMR structures of the individual domains, which may support future studies of the general mechanisms of action by this class of splicing factors.

Multidomain RNA recognition by the splicing factor RBM10

In order to investigate the impact of synergies between multiple RNA binding domains in RBM10 during Fas recognition, we prepared RBM polypeptide fragments of variable lengths and evaluated their affinities for the 22-nucleotide sequence UAAUUGUUUGGGGUAAGUUCUU found in exon 6 of Fas (Fig. 1B). High affinity was observed for a three-domain construct containing RRM1, ZnF1 and RRM2 connected by the natural linkers in RBM10 (RRM1–ZnF1–RRM2), with K_D = 20 nM, as compared to 2.5 μM and 5.5 μM, respectively, for the individual domains RRM1 and RRM2, and 845 nM for ZnF1 (Fig. 1B). The K_D value for ZnF1 is similar to the affinites observed for other members of the ZRANB2 family.^{11, 20} A construct of residues 128–250 containing RRM1 and ZnF1 (RRM1–ZnF1) was found to bind with intermediate affinity (K_D = 412 nM; Fig. 1B). Overall, the data of Fig. 1B show that synergies among three RBM10 domains can afford high-affinity recognition of a sequence Fas mRNA.

NMR Structures of the RBM10 domains RRM1, ZnF1 and RRM2[V354del].

For the NMR structure determinations we followed the J-UNIO protocol with non-uniform sampling of the 3D heteronuclear-resolved [¹H,¹H]-NOESY data sets^21–23, high-quality structures were thus obtained, as indicated by the statistics presented in Table 1.

Table 1.

Input for the structure calculations and characterization of bundles of 20 energy-minimized CYANA conformers representing the NMR structures of the RBM10 domains RRM1, ZnF1 and RRM2[V354del].

Quantitya	RRM1	ZnF1	RRM2[V354del]
NOE upper distance limits	1960	420	1693
Intraresidual	457	135	442
Short range	540	109	556
Medium range	376	77	278
Long range	587	99	417
Dihedral angle constraints	385	179	336
Residual target function value (Å2)	1.66 ± 0.30	0.34 ± 0.05	1.44 ± 0.24
Residual NOE violations
Number ≥ 0.1 Å	5 ± 2	1 ± 1	6 ± 1
Maximum (Å)	0.14	0.12	0.13
Residual dihedral angle violations
Number ≥ 2.5°	1 ± 1	1 ± 1	0 ± 0
Maximum (°)	1.56	1.78	0.87
Amber energies (kcal/mol)
Total	-3462 ± 105	-1117 ± 87	- 3801 ± 57
Van der Waals	-279 ± 18	-208 ± 15	- 223 ± 18
Electrostatic	- 3994 ± 92	- 1519 ± 112	- 3291 ± 68
RMSD from ideal geometry
Bond lengths (Å)	0.0091	0.0081 6	0.0078
Bond angles (°)	1.48	1.31	1.44
RMSD to the mean co-ordinatesb (Å)b
bb	0.46 ± 0.06	0.41 ± 0.07	0.68 ± 0.08
ha	0.93 ± 0.09	1.01 ± 0.12	1.04 ± 0.09
Ramachandran plot statistics (%)c c
Most favoured regions	83.5	85.3	74.5
Additional allowed regions	13.9	13.3	21.3
Generously allowed regions	2.6	0.9	3.1
Disallowed regions	0	0.5	1.1

^aExcept for the top six entries, which describe the input generated in the final cycle of the ATNOS/CANDID/CYANA calculation,^36–38 the last entries refer to the 20 best CYANA conformers after energy minimization with OPALp (see text). Where applicable, the average value for the bundle of 20 conformers and the standard deviation are given.

^bbb indicates the backbone atoms N, C^α, and C’; ha stands for all heavy atoms.

^cAs determined by PROCHECK.

RRM1 shows a variation of the canonical RRM architecture, with a four-stranded antiparallel β-sheet and two α-helices (Fig. 2A).^{24, 25} The helices α1 and α2 contain the residues 141–153 and 181–191, and β-strands are formed by the polypeptide segments 128–132 (β1), 157–162 (β2), 174–178 (β3), 195–198 (β3’) and 201–205 (β4). The unique feature of RRM1 is that the linker between α2 and β4 forms an additional strand, β3’, which is not part of the canonical RRM β-sheet.^{10, 24, 25}

Figure 2. NMR structures of RBM10 domains.

The RBM10 domains RRM1(A), ZnF1(B) and RRM2[V354del] (C) are represented by bundles of 20 conformers (left) and ribbon presentations of the conformer closest to the mean coordinates (right). The chain ends are identified by N and C.

The ZnF1 domain adopts a ZRANB2-like structure (Fig. 2B)^{11, 20} containing an α-turn with residues 240–244 and four short β-strands of residues 315–318, 324–328, 336–338 and 340–344. Zn²⁺ is coordinated by four cysteines in positions 319, 322, 333 and 336, which are located within or near the four β-strands.

The RRM2[V354del] structure contains the helices α1 and α2 with residues 310–324 and 351–359, and the four β-strands are formed by the polypeptide segments 301–305 (β1), 329–334 (β2), 344–349 (β3) and 378– 381 (β4) (Fig. 2C). RRM2[V254del] thus adopts a canonical RRM structure, and therefore it did not come as a surprise that this deletion did not have a significant effect on RNA binding (Fig. 3). In view of the near-identity of these domain structures with those for which detailed studies of RNA-complexation have been reported^10,24,25, we hypothesize that they also have similar patterns of contacts with RNAs.

Figure 3. Association curves of RRM2 and RRM2[V354del] with the 22-nucleotide RNA fragment UAAUUGUUUGGGGUAAGUUCUU from exon 6 of Fas.

Same measurement details and presentation as in Fig. 1(B).

DISCUSSION

The function of RBM10 in the regulation of Fas alternative splicing has been shown to be based on both, specific recognition of oligonucleotide motifs in the mRNA and interactions with supplementary splicing factors^1,10. While the C-terminal region of RBM10 is involved in protein recruitment, the N-terminal region with the three RNA binding domains RRM1, ZnF1 and RRM2 ensures recognition and binding of specific splice sites on the mRNA, which is the focus of the present work. Previous studies with tandem RRM constructs showed enhanced affinities when compared with respect to the individual RRMs.^24–29 RBM10 now provided an opportunity to investigate possible cooperativity between RRMs and a zinc-finger domain, which exhibits a different structure and a different mode of RNA recognition. Introduction of non-RRM domains into splicing factors obviously carries the promise to diversify RNA recognition, which has found extensive use in nature (Fig. 1A)

Individually, the RRMs and ZnF1 in RBM10 have nearly identical structural properties to corresponding domains in other splicing factors,^{10, 12, 24–26, 30–33} with the sole exception that RBM10-RRM1 includes an additional, non-canonical regular secondary structure element, β3’ (Fig. 2A). We now showed that recognition of exon 6 in Fas by the combination of the three RNA binding domains of RBM10 results in low nM binding affinities, although the individual domains have weak affinities for RNA binding (Fig. 1B).

Complementarity of the biological functions of the splicing factors RBM5 and RBM10 has previously been extensively investigated.^{2, 4, 7, 17–19, 34, 35} The high sequence homology and the close similarity of the three-dimensional structures of their RNA binding domains, in as far as they are available,^{11, 12} suggest a homologous RNA recognition mode, which coincides with evidence that the two splicing factors may target closely related mRNA motifs.^{2, 32, 35}

Abbreviations:

APSY

automated projection spectroscopy

ASCAN

software for automated side-chain resonance assignment

ATNOS

software for automated NMR peak picking

CANDID

software for automated NOE assignment

CYANA

software for NMR structure calculation

EDTA

ethylenediaminetetraacetic acid

HSQC

heteronuclear single-quantum coherence spectroscopy

J-UNIO

protocol for automated determination of NMR structures of proteins

MATCH

software used for backbone NMR chemical shift assignments

NOE

nuclear Overhauser effect

NOESY

nuclear Overhauser effect spectroscopy

PDB

protein data bank

RMSD

root-mean-square deviation

TEV

tobacco etch virus