The four Zn fingers of MBNL1 provide a flexible platform for recognition of its RNA binding elements
© Cass et al; licensee BioMed Central Ltd. 2011
Received: 11 January 2011
Accepted: 6 May 2011
Published: 6 May 2011
Muscleblind-like 1 (MBNL1) is an alternative splicing factor containing four CCCH Zinc fingers (ZnFs). The sequestration of MBNL1 by expanded CUG and CCUG repeats is a major component in causing myotonic dystrophy. In addition to binding the structured expanded CUG and CCUG repeats; previous results suggested that MBNL1 binds single-stranded RNAs containing GC dinucleotides.
We performed a systematic analysis of MBNL1 binding to single-stranded RNAs. These studies revealed that a single GC dinucleotide in poly-uridine is sufficient for MBNL1 binding and that a second GC dinucleotide confers higher affinity MBNL1 binding. However additional GC dinucleotides do not enhance RNA binding. We also showed that the RNA sequences adjacent to the GC dinucleotides play an important role in MBNL1 binding with the following preference: uridines >cytidines >adenosines >guanosines. For high affinity binding by MBNL1, the distance between the two GC dinucleotides can vary from 1 to 17 nucleotides.
These results suggest that MBNL1 is highly flexible and able to adopt different conformations to recognize RNAs with varying sequence configurations. Although MBNL1 contains four ZnFs, only two ZnF - GC dinucleotide interactions are necessary for high affinity binding.
Alternative pre-mRNA splicing significantly increases genome diversity with recent measurements suggesting that greater than 90% of genes undergo alternative splicing [1–3]. The different protein isoforms generated through alternative splicing can alter function or cellular localization, or it may provide a mechanism to regulate the levels of the protein by leading to non-productive splicing and RNA turnover (reviewed in [4, 5]). Regulated alternative splicing is dependent upon the alternative splicing factors present in specific cell types or at different developmental stages. Many of these alternative splicing factors appear to regulate hundreds of exons and they function by recognizing specific RNA sequence elements within or near the regulated exons (reviewed in ).
Muscleblind-like 1 (MBNL1) is an alternative splicing factor that is associated with the disease myotonic dystrophy (DM). Patients with DM have a CTG or CCTG repeat expansion in an untranslated region of their genome. For DM type 1 (DM1), the CTG repeats are found in the 3' UTR of the DMPK gene, and for DM type 2 (DM2), the CCTG repeats are found within intron 1 of the CCHC-type zinc finger (ZnF), nucleic acid binding protein (CNBP) gene [7, 8]. These CTG and CCTG repeats, when transcribed, form CUG/CCUG RNA stem loops that bind and sequester MBNL1 into nuclear foci resulting in a loss of normal MBNL1 function in the cell (reviewed in ). In studying the molecular basis of DM and the role MBNL1 plays, it was discovered that MBNL1 regulates the alternative splicing of a host of developmentally regulated transcripts [10–13]. In the cells of DM patients, adult splicing patterns revert back to embryonic splicing patterns and these incorrect splicing events are thought to cause many of the symptoms of the disease (reviewed in ).
Several MBNL1 binding sites have been identified, and a few of these have been characterized [10, 14–17]. Complicating the analysis of MBNL1 binding to these sites and the CUG and CCUG repeats has been differentiating between the effects of RNA structure on MBNL1 binding and sequence specific RNA binding. MBNL1 interacts with RNA through its four CCCH ZnFs, although it is possible that other regions of the protein interact with RNA as well or play an important role in regulating splicing . The ZnFs of MBNL1 fold into two compact domains ZnFs1/2 and ZnFs3/4 [19, 20]. A crystal structure of ZnFs3/4 in complex with a 6 mer RNA (CGCUGU) revealed that these ZnFs bind single-stranded RNA through specific recognition of the Watson-Crick face of the GC dinucleotide . Each ZnF (3 and 4) interacts with RNA in a similar manner and the structure of the ZnFs domain in complex with RNA lead Teplova and Patel to propose that MBNL1 binds RNA in a looped conformation . This model predicts that for each ZnF to bind a GC dinucleotide a significant RNA linker must separate the GC dinucleotides because the RNA binding surfaces of ZnFs3/4 (or 1/2) are on opposing sides of the domain . Binding to RNAs containing three different RNA linkers of 5, 10 and 15 nucleotides between the GC dinucleotides to ZnFs3/4 of MBNL1 supported this model . However these experiments were only performed with the isolated ZnFs3/4 and it was unclear if the spacer would be important in the context of the four ZnFs of MBNL1. To determine if the spacing between GC dinucleotides was important for MBNL1 binding and to determine if the sequence of the RNA linker between the GC dinucleotides affected RNA binding, a series of single-stranded RNAs varying linker distances and different linker sequences were studied with MBNL1.
Results and discussion
To better understand the RNA binding specificity of human MBNL1, we used a version of the protein that contains amino acids 2-260. Previous studies have shown that MBNL1 2-260, which contains all four ZnFs, binds RNA as well as the corresponding full-length protein . This version of the protein was used due to its ease in expression and purification in a bacterial system. This region of the protein has been shown to be sufficient for splicing regulation .
Determination of the number of GC dinucleotides required for high affinity binding by MBNL1
To determine if these RNAs are single-stranded, a thermal denaturation assay was used to monitor the structure of these RNAs. For all RNAs tested with one to four GC dinucleotides, the melting point was below 40°C (Figure 1G), while the (CUG)4 RNA which was previously shown to be structured and bind MBNL1  has a melting point of 60.6 +/- 0.6°C (Figure 1G). These thermal denaturation results indicate that these model RNAs are primarily single-stranded and can be used to test MBNL1's specificity in the absence of RNA structure. A possibility we cannot discount is that when MBNL1 binds RNA it could stabilize weak RNA structures; therefore we cannot completely rule out the role of RNA structure. Further experiments would be needed to explore this possibility.
As expected, when both GC dinucleotides are mutated to any other sequence or their polarity is reversed (2CC, 2GG and 2CG), MBNL1 binding is eliminated at the concentrations of protein tested (Figure 1B). Although we had found previously that MBNL1 prefers uridines over the other three nucleotides , these results showed that MBNL1 does not bind to long stretches of uridines alone with appreciable affinity.
Surprisingly, the presence of a single GC dinucleotide was sufficient for high affinity binding by MBNL1 (Figure 1C). The 1GC RNA bound with a Kd of 45 nM while the 2GC RNA bound with a Kd of 11 nM. These results indicate that MBNL1 requires only one GC dinucleotide for one of its four ZnFs to achieve significant RNA binding. Although the 1GC RNA bound with relatively high affinity, the complex appeared to dissociate during the gel shift assay based on the observed smearing (Figure 1C). To confirm the gel shift assay results, a competition assay was used. In this assay, MBNL1 was pre-bound to a radioactive 2GC RNA to form a stable complex. Then increasing amounts of cold competitive RNA were added (Figure 1D). The competition results showed a larger difference between the 1GC and 2GC RNAs compared to the binding assays (15-fold in the competition assay versus 4-fold in the gel shift assay); however these results are consistent with the gel shift assay and further demonstrate that MBNL1 only requires one GC dinucleotide for nanomolar affinity binding by MBNL1 to RNA. The differences in relative affinities between the 2GC and 1GC RNAs in the gel shift and competition assays may be due to the instability of the 1GC-MBNL complex making quantification of this complex in the gel shift assay difficult to measure accurately.
To determine if more than two GC dinucleotides increased RNA binding by MBNL1, a longer RNA (29 nucleotides, Figure 1E) containing up to four GC dinucleotides was used for binding studies. The oligonucleotides with two, three, or four dinucleotides all behaved in a similar manner (Figure 1F). MBNL1 formed a single complex with all RNAs and the Kds were all within error of each other (2.5 to 4.0 nM). A RNA containing one GC dinucleotide (1GC#2) was also tested in the context of the longer RNA and it showed an approximately 5-fold increase in Kd with lower complex stability (smearing) as seen with the shorter 1GC RNA (Figure 1, compare 1GC and 1GC#2). These results indicate that adding more than two GC dinucleotides did not enhance MBNL1 binding to RNA in this context.
The observation that RNAs containing three or four GC dinucleotides didn't bind MBNL1 better than RNAs containing just two GC dinucleotides was surprising. It suggests that only two of the four ZnFs require GC motifs for high affinity binding while the other two ZnFs may bind other sequences such as a poly-uridine or are involved in an intramolecular interaction that precludes RNA binding. This intramolecular interaction is supported by the MBNL1 ZnFs3/4 crystal structure which formed a ZnF4 - ZnF4 interaction with a 1320 Å2 interface that would preclude ZnF4 from binding RNA . Thus we might predict that ZnF2 and ZnF4 are involved in a higher order structure while ZnF1 and ZnF3 are involved in RNA binding. Although it is possible that the two ZnFs involved in RNA binding could be any of the four ZnFs, as none of the ZnFs alone have been directly tested for RNA binding. In comparing this model with the known endogenous MBNL1 targets, it can be seen that many of these binding sites contain only two YGCY motifs suggesting that a mode in which MBNL1 uses only two of its ZnFs to bind YGCY motifs could be a common type of interaction for MBNL1 [10, 14–16].
An alternative explanation for the lack of increased MBNL1 binding to more than two GC motifs is that the RNA we used may not allow all four GC dinucleotides to bind all four ZnFs. However, in RNA 3GC#3 the three GC dinucleotides are spaced by nine and twelve nucleotides, and the model proposed by Teplova and Patel describing MBNL1 binding to RNAs with multiple GC dinucleotides suggests that this type of spacing would allow for binding by three of the four ZnFs . Our data described later also suggests that an RNA containing two GC dinucleotides with only a single nucleotide spacing does not result in a dramatic decrease in binding. This supports the idea that the spacing of the GC dinucleotides was not a problem in these RNA constructs.
The role of nucleotides adjacent to the GC dinucleotides in MBNL1 RNA binding
Replacing the pyrimidines with purines between the GC motifs significantly reduced MBNL1 binding. The presence of adenosines adjacent to the GC dinucleotide weakened binding to MBNL1 by 5-fold while guanosines in these positions reduced binding by 40-fold (Figure 2B, RNAs AGCA and GGCG). Placing two UGCU motifs into a polyA RNA resulted in more than a 10-fold reduction in binding compared to the 2GC RNA (Figure 2C, A4 RNA), indicating that positions outside of the UGCU motifs also affect MBNL1's ability to bind RNA.
Our characterization of MBNL1 binding specificity is consistent with our previously proposed MBNL1 binding site motif of YGCY . However, there seems to be an additional requirement of a uridine rich context as seen by the 10-fold decrease in binding between a cytidine rich and a uridine rich environment. This YGCY and U-rich requirement is not explained by the crystal structure of ZnFs3/4 in complex with RNA. In this structure ZnF3 only interacts with the GC dinucleotide and ZnF4 makes one base specific contact with the downstream uridine, therefore recognizing a GCU motif . One explanation for this apparent discrepancy between our results and the crystal structure is that important protein-RNA contacts were not captured in the crystal structure. It is possible that ZnFs1/2 makes different RNA contacts than ZnFs3/4 or amino acids outside of the crystallized domains interact with RNA. Further studies are required to determine which regions of the protein are responsible for these additional RNA sequence requirements.
The effects of the separation between GC dinucleotides in MBNL1 RNA binding
The proposed model explains how MBNL1 may recognize a wide range of single-stranded RNA targets, but it does not address how MBNL1 interacts with structured RNAs such as the CUG/CCUG repeats or stem-loop targets found in its pre-mRNA targets. It is possible that MBNL1 binds the YGCY motifs in the structured RNAs when these nucleotides are accessible (GC dinucleotides flipped out so the Watson-Crick face of the GC dinucleotides can be bound by the ZnFs domains) in these RNA targets. The flexibility of the linker between the ZnFs domains would allow MNBL1 to easily reach two sites on different parts of a structured RNA. Alternatively, MBNL1 may recognize its structured RNA targets using a different mode of recognition. In the future biochemical approaches using techniques in addition to the gel shift assay and structural studies will be necessary to determine if MBNL1 use one or more modes of binding to recognize single-stranded and structured RNA substrates.
This study suggests that the ZnFs domains of MBNL1 are linked together by a highly flexible linker (Figure 4), which allows MBNL1 to recognize a wide variety of RNA targets. Although MBNL1 contains four ZnFs, it appears that only two ZnFs binding GC motifs are necessary for high affinity RNA binding.
Materials and methods
The MBNL1 protein used in these studies contains amino acids 2-260 and was described previously . Purification of the protein and removal of the GST tag was described previously  except that for our studies the protein was stored at -80°C rather than at -20°C.
RNA synthesis, labeling, and purification
All RNAs were synthesized by Integrated DNA Technologies and 5' end labeled with [γ-32P]ATP. RNAs were then purified on a 10% denaturing polyacrylamide gel (19:1) and eluted into TE Buffer (10 mM Tris, 1 mM EDTA). RNA was then ethanol precipitated with glycogen and resuspended in TE Buffer. Finally a Bio-spin 6 column (Bio-Rad) equilibrated in TE Buffer was used.
RNA thermal denaturation assay
1.5 μM RNA was incubated at 95°C for two minutes in 100 mM NaCl, 21 mM Tris 7.5, and 5 mM MgCl2. RNA was then incubated on ice for 30 minutes. The thermal denaturation analysis was performed on a Carey UV/VIS Spectrophotometer. The absorbance of the RNA was monitored at 260 nm while the temperature was increased from 15-90°C. Melting temperatures were determined by calculating the derivative of the absorbance versus temperature using the Cary program "Thermal". All RNAs were tested three times and Tm values compared. The Tm of all the RNAs used in these studies were measured at below 40°C while the (CUG)4 RNA was measured with a Tm of 60.6 +/- 0.6°C.
Gel mobility shift assay
RNA was initially heated to 95 °C for two minutes and then placed directly onto ice. RNA was then mixed with MBNL1 at concentrations indicated in the figures. The final reaction conditions were 100 mM NaCl, 21 mM Tris pH 7.5, 0.1 mM EDTA, 2 mg/mL BSA, 0.1 mg/mL Heparin, 0.5 mM BME, 5 mM DTT, 0.02% Triton X-100, 5 mM MgCl2, 10% glycerol, 0.02% bromophenol blue, and 0.02 nM RNA. The samples were incubated at room temperature for 30 minutes and then loaded onto a pre-chilled 6% polyacrylamide (19:1) gel running at 50 volts. Once the samples were loaded the voltage was increased to 150 volts and the gel was run for 30 minutes. After running, the gel was dried and exposed overnight.
Quantitation of the binding curve was described previously . The data was fit using the equation where m1 = a fitting factor, m0 = nM protein concentration, and m2 = nM Kd.
Competition assays were run under the same conditions as the gel shift assays except that the protein and radioactive RNA were incubated for 20 minutes and then the competitive RNA was added to the reaction and allowed to equilibrate for 30 minutes.
Quantitation of the competition assays was done using Kaleidagraph. The data was fit to the equation where m2 = competitive RNA Kd, m1 = Kd to radiolabeled RNA, Pt = total protein added in the reaction, and m0 = amount of competitive RNA. When the competitive RNA and the radiolabeled RNA were the same, the equation was simplified to , assuming m2 and m1 are the same.
Modeling of MBNL1 binding RNA
PDB files 3D2N and 3D2S were used to create Figure 4. Each PDB file was opened in separate Pymol windows. 3d2S contained ZnFs3/4 bound to RNA. The structure originally contained a ZnFs3/4 dimer bound to RNA. The extra ZnFs3/4 was removed and the RNA bound to ZnF4 was removed for the figure. In the other window the ZnFs1/2 structure (3D2N) was oriented to place the hypothesized cytidine binding pocket towards the unbound cytidine from 3D2S. Figures were then made of each of these windows and were merged in Photoshop. The amino acid linker was then added by linking the C-terminal amino acid from ZnFs1/2 with the N-terminal amino acid from ZnFs3/4.
Acknowledgements and funding
We thank members of the Berglund lab for experimental advice and critical reading of the manuscript. This research was supported by Grants AR053903 and AR053903-03S1 from NIH to JAB.
- Wang ET, Sandberg R, Luo S, Khrebtukova I, Zhang L, Mayr C, Kingsmore SF, Schroth GP, Burge CB: Alternative isoform regulation in human tissue transcriptomes. Nature. 2008, 456: 470-476. 10.1038/nature07509PubMed CentralView ArticlePubMedGoogle Scholar
- Pan Q, Shai O, Lee LJ, Frey BJ, Blencowe BJ: Deep surveying of alternative splicing complexity in the human transcriptome by high-throughput sequencing. Nat Genet. 2008, 40: 1413-1415. 10.1038/ng.259View ArticlePubMedGoogle Scholar
- Sultan M, Schulz MH, Richard H, Magen A, Klingenhoff A, Scherf M, Seifert M, Borodina T, Soldatov A, Parkhomchuk D: A global view of gene activity and alternative splicing by deep sequencing of the human transcriptome. Science. 2008, 321: 956-960. 10.1126/science.1160342View ArticlePubMedGoogle Scholar
- Graveley BR: Alternative splicing: increasing diversity in the proteomic world. Trends Genet. 2001, 17: 100-107. 10.1016/S0168-9525(00)02176-4View ArticlePubMedGoogle Scholar
- Hartmann B, Valcarcel J: Decrypting the genome's alternative messages. Curr Opin Cell Biol. 2009, 21: 377-386. 10.1016/j.ceb.2009.02.006View ArticlePubMedGoogle Scholar
- House AE, Lynch KW: Regulation of alternative splicing: more than just the ABCs. J Biol Chem. 2008, 283: 1217-1221.View ArticlePubMedGoogle Scholar
- Brook JD, McCurrach ME, Harley HG, Buckler AJ, Church D, Aburatani H, Hunter K, Stanton VP, Thirion JP, Hudson T: Molecular basis of myotonic dystrophy: expansion of a trinucleotide (CTG) repeat at the 3' end of a transcript encoding a protein kinase family member. Cell. 1992, 69: 385-PubMedGoogle Scholar
- Liquori CL, Ricker K, Moseley ML, Jacobsen JF, Kress W, Naylor SL, Day JW, Ranum LP: Myotonic dystrophy type 2 caused by a CCTG expansion in intron 1 of ZNF9. Science. 2001, 293: 864-867. 10.1126/science.1062125View ArticlePubMedGoogle Scholar
- Lee JE, Cooper TA: Pathogenic mechanisms of myotonic dystrophy. Biochem Soc Trans. 2009, 37: 1281-1286. 10.1042/BST0371281View ArticlePubMedGoogle Scholar
- Ho TH, Charlet BN, Poulos MG, Singh G, Swanson MS, Cooper TA: Muscleblind proteins regulate alternative splicing. EMBO J. 2004, 23: 3103-3112. 10.1038/sj.emboj.7600300PubMed CentralView ArticlePubMedGoogle Scholar
- Kanadia RN, Shin J, Yuan Y, Beattie SG, Wheeler TM, Thornton CA, Swanson MS: Reversal of RNA missplicing and myotonia after muscleblind overexpression in a mouse poly(CUG) model for myotonic dystrophy. Proc Natl Acad Sci USA. 2006, 103: 11748-11753. 10.1073/pnas.0604970103PubMed CentralView ArticlePubMedGoogle Scholar
- Kanadia RN, Johnstone KA, Mankodi A, Lungu C, Thornton CA, Esson D, Timmers AM, Hauswirth WW, Swanson MS: A muscleblind knockout model for myotonic dystrophy. Science. 2003, 302: 1978-1980. 10.1126/science.1088583View ArticlePubMedGoogle Scholar
- Dansithong W, Paul S, Comai L, Reddy S: MBNL1 is the primary determinant of focus formation and aberrant insulin receptor splicing in DM1. J Biol Chem. 2005, 280: 5773-5780.View ArticlePubMedGoogle Scholar
- Warf MB, Berglund JA: MBNL binds similar RNA structures in the CUG repeats of myotonic dystrophy and its pre-mRNA substrate cardiac troponin T. RNA. 2007, 13: 2238-2251. 10.1261/rna.610607PubMed CentralView ArticlePubMedGoogle Scholar
- Yuan Y, Compton SA, Sobczak K, Stenberg MG, Thornton CA, Griffith JD, Swanson MS: Muscleblind-like 1 interacts with RNA hairpins in splicing target and pathogenic RNAs. Nucleic Acids Res. 2007, 35: 5474-5486. 10.1093/nar/gkm601PubMed CentralView ArticlePubMedGoogle Scholar
- Sen S, Talukdar I, Liu Y, Tam J, Reddy S, Webster NJ: Muscleblind-like 1 (Mbnl1) promotes insulin receptor exon 11 inclusion via binding to a downstream evolutionarily conserved intronic enhancer. J Biol Chem. 2010, 285: 25426-25437. 10.1074/jbc.M109.095224PubMed CentralView ArticlePubMedGoogle Scholar
- Du H, Cline MS, Osborne RJ, Tuttle DL, Clark TA, Donohue JP, Hall MP, Shiue L, Swanson MS, Thornton CA, Ares M: Aberrant alternative splicing and extracellular matrix gene expression in mouse models of myotonic dystrophy. Nat Struct Mol Biol. 2010, 17: 187-193. 10.1038/nsmb.1720PubMed CentralView ArticlePubMedGoogle Scholar
- Grammatikakis I, Goo YH, Echeverria GV, Cooper TA: Identification of MBNL1 and MBNL3 domains required for splicing activation and repression. Nucleic Acids Res. 2010Google Scholar
- He F, Dang W, Abe C, Tsuda K, Inoue M, Watanabe S, Kobayashi N, Kigawa T, Matsuda T, Yabuki T: Solution structure of the RNA binding domain in the human muscleblind-like protein 2. Protein Sci. 2009, 18: 80-91.PubMed CentralView ArticlePubMedGoogle Scholar
- Teplova M, Patel DJ: Structural insights into RNA recognition by the alternative-splicing regulator muscleblind-like MBNL1. Nat Struct Mol Biol. 2008, 15: 1343-1351. 10.1038/nsmb.1519View ArticlePubMedGoogle Scholar
- Goers ES, Purcell J, Voelker RB, Gates DP, Berglund JA: MBNL1 binds GC motifs embedded in pyrimidines to regulate alternative splicing. Nucleic Acids Res. 2010, 38: 2467-2484. 10.1093/nar/gkp1209PubMed CentralView ArticlePubMedGoogle Scholar