a single c. elegans puf protein binds rna in multiple modes · asinglec. elegans puf protein binds...

10.1261/rna.1545309Access the most recent version at doi: 2009 15: 1090-1099 originally published online April 15, 2009RNA

Yvonne Yiling Koh, Laura Opperman, Craig Stumpf, et al.

PUF protein binds RNA in multiple modesC. elegansA single

References

http://rnajournal.cshlp.org/content/15/6/1090.full.html#related-urlsArticle cited in:

http://rnajournal.cshlp.org/content/15/6/1090.full.html#ref-list-1This article cites 40 articles, 13 of which can be accessed free at:

serviceEmail alerting

click heretop right corner of the article orReceive free email alerts when new articles cite this article - sign up in the box at the

http://rnajournal.cshlp.org/subscriptions go to: RNATo subscribe to

Copyright © 2009 RNA Society

Cold Spring Harbor Laboratory Press on October 31, 2011 - Published by rnajournal.cshlp.orgDownloaded from

http://rnajournal.cshlp.org/lookup/doi/10.1261/rna.1545309

http://rnajournal.cshlp.org/content/15/6/1090.full.html#ref-list-1

http://rnajournal.cshlp.org/content/15/6/1090.full.html#related-urls

http://rnajournal.cshlp.org/cgi/alerts/ctalert?alertType=citedby&addAlert=cited_by&saveAlert=no&cited_by_criteria_resid=rna;15/6/1090&return_type=article&return_url=http://rnajournal.cshlp.org/content/15/6/1090.full.pdf

http://rnajournal.cshlp.org/subscriptions

http://rnajournal.cshlp.org/

http://www.cshlpress.com

A single C. elegans PUF protein binds RNA

in multiple modes

YVONNE YILING KOH,1,2 LAURA OPPERMAN,1,4 CRAIG STUMPF,1 ARPITA MANDAN,1,5 SUNDUZ KELES,3

and MARVIN WICKENS1

1Department of Biochemistry, University of Wisconsin–Madison, Madison, Wisconsin 53706, USA2Microbiology Doctoral Training Program, University of Wisconsin–Madison, Madison, Wisconsin 53706, USA3Departments of Statistics and Biostatistics and Medical Informatics, University of Wisconsin–Madison, Madison, Wisconsin 53706, USA

ABSTRACT

PUF proteins specifically bind mRNAs to regulate their stability and translation. Here we focus on the RNA-binding specificityof a C. elegans PUF protein, PUF-11. Our findings reveal that PUF-11 binds RNA in multiple modes, in which the protein canaccommodate variable spacings between two distinct recognition elements. We propose a structural model in which flexibilityin the central region of the protein enables the protein to adopt at least two distinct structures, one of which results in baseflipping.

Keywords: 39UTRs; RNA decay; RNA–protein interactions; translational regulation

INTRODUCTION

RNA–protein interactions control every step in an mRNA’slife. The specificities of regulatory proteins for RNAdetermine which mRNAs are controlled and how. The 39

untranslated region (39UTR) is a key repository forelements that govern stability, translation, and localization(Wickens et al. 2002). Several families of RNA-bindingproteins have been identified that interact with suchelements. Here we focus on the PUF family of RNA-binding proteins, chosen because of their biological impor-tance and the striking modularity of their interactions withRNA.

PUF proteins are highly conserved RNA-binding pro-teins that bind to specific RNA sequences. Members of thePUF family include Drosophila melanogaster Pumilio andCaenorhabditis elegans fem-3 binding factor (FBF) (Zamoreet al. 1997; Zhang et al. 1997). PUF proteins elicitrepression or decay of the mRNAs to which they bind(Ahringer and Kimble 1991; Wreden et al. 1997; Wharton

et al. 1998; Olivas and Parker 2000; Chagnovich andLehmann 2001; Goldstrohm et al. 2006, 2007) (for review,see Wickens et al. 2002; Goldstrohm and Wickens 2008).Each PUF appears to interact with multiple mRNAs, from40 to 220 in yeast (Gerber et al. 2004) to more than 1000 inhumans (Galgano et al. 2008; Morris et al. 2008). Multiplenatural targets of C. elegans PUF proteins have beenidentified by genetics, interaction tests, immunocytochem-istry, and association in extracts (Zhang et al. 1997;Crittenden et al. 2002; Lamont et al. 2004; Thompsonet al. 2005; Lee et al. 2006, 2007) (for review, see Kimbleand Crittenden 2007). While the biological roles arediverse, the maintenance of diverse stem cells appears tobe a common function (Wickens et al. 2002; Morrison andKimble 2006; Kimble and Crittenden 2007), as may be itsroles in the nervous system (Schweers et al. 2002; Dubnauet al. 2003; Ye et al. 2004; Chen et al. 2008).

PUF proteins have a reiterated repeat structure. Virtuallyall PUF proteins possess eight repeats, each of whichcomprises a three-helix domain of z40 amino acids(Zamore et al. 1997; Edwards et al. 2001; Wang et al.2001). These domains lie on one another to form a ladderof eight a-helices, termed RNA recognition helices, whichbind RNA. In the crystal structure of human Pumilio1bound to RNA (Wang et al. 2002), specific contacts areformed between the eight consecutive recognition helicesand nucleotides, yielding a one-helix/one-base bindingmode (Fig. 1A). Each recognition helix uses several amino

Present addresses: 4Department of Genetics, University of Wisconsin–Madison, Madison, WI 53706, USA; 5Department of Chemistry, IndianInstitute of Technology, Kalyanpur, India 208016.

Reprint requests to: Marvin Wickens, Department of Biochemistry,University of Wisconsin–Madison, Madison, WI 53706, USA; e-mail:[email protected]; fax: (608) 262-9108.

Article published online ahead of print. Article and publication date are athttp://www.rnajournal.org/cgi/doi/10.1261/rna.1545309.

1090 RNA (2009), 15:1090–1099. Published by Cold Spring Harbor Laboratory Press. Copyright � 2009 RNA Society.




acid side chains to contact the cognate base via hydrogenbonds, van der Waals forces, and stacking interactions(Wang et al. 2002).

The residues that form base-specific contacts in humanPumilio1 are highly conserved among all PUF proteins, yetnot all PUF proteins bind the same sites (Gerber et al. 2004;Bernstein et al. 2005; Opperman et al. 2005; Seay et al.2006; Stumpf et al. 2008b). The binding sites of C. elegansPUF-8 and FBF comprise a UGU and more 39 AU element(Opperman et al. 2005). In the PUF-8 binding site, thesetwo elements are spaced by 2 nucleotides (nt), and in FBF,by three. This distinction enables the two PUF proteins toefficiently bind only their own sites, yet use nearly identicalsets of atomic contacts. The structure of FBF bound toRNA indicates that the additional base is turned awayfrom the protein, or ‘‘flipped’’ (Y. Wang, L. Opperman, M.Wickens, and T. Hall, in prep.). Flipped bases appear to bea common principle used to achieve PUF specificity, as theyare seen in yeast Puf4p complexes with natural target RNAsand in human Pumilio1 bound to noncognate RNA (Guptaet al. 2008; Miller et al. 2008). In both FBF and yeast Puf4p,the identity of the additional base is unimportant for pro-tein binding (Miller et al. 2008; Y. Wang, L. Opperman, M.Wickens, and T. Hall, in prep.). Indeed, flexibility innucleotide identity at a particular position in a PUF recog-nition site may be good prima facie evidence of flipping.

C. elegans possess 10 PUF proteins, which cluster intofour groups based on sequence similarity: these are termed

the FBF, PUF-8/-9, PUF-5/-6/-7, and PUF-3/-11 groups(Fig. 1B; Stumpf et al. 2008b). Each of the three groupsanalyzed thus far has a distinctive RNA-binding specificity,as indicated in Figure 1; each recognizes a UGU sequence atthe 59 end of the core site, as is typical of all validated PUFprotein targets. Importantly, the biological functions of thedifferent groups of worm PUF proteins are distinct, butoverlap (e.g., Zhang et al. 1997; Lamont et al. 2004;Bachorik and Kimble 2005; Lublin and Evans 2007; Noldeet al. 2007). These findings imply that the different groupsof proteins control distinct sets of mRNAs. Similarly, yeastPuf3p binds 220 mRNAs, yet few of these bind the otherfour canonical PUF proteins of that organism (Gerber et al.2004).

In this study, we sought to determine the RNA-bindingspecificity of C. elegans PUF-11. Our data reveal surprisingflexibility in PUF-11’s sequence specificity, and suggestthat this protein binds to RNA in multiple modes. Wesuggest that Puf-11–RNA complexes exist in at least twodifferent conformations, which enable recognition of dif-ferent groups of mRNAs. We propose a structural modelfor the interaction.

RESULTS

PUF-11 binds three classes of RNA sequences

To characterize the RNA-binding specificity of PUF-11, weperformed a selection experiment using the yeast three-hybrid system (Fig. 2A; Hook et al. 2005; Stumpf et al.2008a). We constructed a library of plasmids that encodedhybrid RNAs. The RNAs possessed a UGU trinucleotide,flanked by three randomized nucleotides upstream (59) andseven randomized nucleotides downstream (39). Thelibrary was introduced into yeast together with a plasmidencoding a PUF-11/AD fusion protein. RNAs that boundPUF-11 activated expression of the reporter gene, HIS3,and so were identified by growth on selective media (Fig.2B). Secondary screens confirmed that the RNAs obtainedalso activated a second reporter gene, LacZ, and that theyrequired the PUF-11/AD protein to do so (data notshown). A total of 66 unique RNA sequences were obtained(Fig. 2B).

The set of sequences fit three position weight matrices(PWMs), and so yielded three classes of RNA sequences(Fig. 3; see Materials and Methods). Class I comprised 27sequences, Class II, 25 sequences, and Class III, 16sequences. We determined the relative binding affinitiesof all the RNAs to PUF-11 by assaying for b-galactosidaseactivity using the three-hybrid system. The magnitude ofb-galactosidase activity in the system correlates with theaffinity of the protein–RNA interaction (Hook et al. 2005).The range of b-galactosidase activities was similar in allthree classes (Fig. 3). In particular, the high-affinity sites ofeach class bound with similar affinities.

FIGURE 1. PUF proteins bind distinct sites. (A) Hs PUM1; structureof human Pumilio1–RNA complex (PDB accession code 1M8Y)(Wang et al. 2002). Eight RNA-recognition helices (purple instructure, red in diagram) contribute side chains to contact theRNA bases (blue). Hydrogen-bonding interactions are indicated byred lines in the structure; stacking interactions are not shown. (B)Four clusters of C. elegans PUF proteins (Stumpf et al. 2008b). PUFproteins are grouped based on amino acid similarity; known bindingsites are indicated for each cluster.

A PUF protein with two binding modes

www.rnajournal.org 1091




The three RNA classes differ in sequence. Each classpossessed an AU dinucleotide downstream of the UGUtrinucleotide (Fig. 3). The following properties were diag-nostic of each class.

d Class I: The AU dinucleotide was located in positions +6and +7. An adenosine was highly preferred in position+5.

d Class II: The AU dinucleotide was located in positions +7and +8. The identities of bases between the UGU and AUelements were more degenerate than in Classes I and III.Class II RNAs were enriched for C at the -1 position;Classes I and III were not.

d Class III: The AU dinucleotide was located in positions+7 and +8. An adenosine was highly preferred in position+5, as in Class I.

Taken together, these data suggest distinct modes of PUF-11 binding, which differ in the spacing of the UGU and AUelements, and in preferences at the �1 and +5 positions.

To compare affinities for the three classes quantitatively,we analyzed PUF-11 binding in vitro. We selected one RNAfrom each class that closely matched the consensus and

had bound PUF-11 with high affinity in the three-hybridassay. These three RNAs were used in an electrophoreticmobility-shift assay together with purified recombinant

FIGURE 3. PUF-11 binds to three classes of RNAs. b-galactosidaseactivity as determined using strains possessing each of the indicatedRNAs. b-galactosidase activity reflects the strength of PUF–RNAinteractions (Hook et al. 2005). Values are relative to RNA I-1. Thesequence in the relevant region of each bound RNA is indicated andcorresponds to the portion that had been randomized. RNAs arenamed as X-Y, where X indicates the class, and Y indicates relativeaffinity for PUF-11 within the class. RNAs I-1, II-1, and III-1 wereincluded when testing each class of RNAs, to provide standards. Theconsensus sequence for each class was derived using WebLogosoftware and displays information content and frequency of nucleo-tide occurrence at each position.

FIGURE 2. Identification of PUF-11 binding sites. (A) Yeast three-hybrid selection used to identify PUF-11 binding sites. The random-ized region of the RNA library is illustrated using N to indicate arandomized base. Interaction between PUF-11 and RNA X will triggerexpression of reporter genes HIS3 and LacZ. (B) Summary of three-hybrid selection experiment. A total of 666 positive clones wereobtained. Of these, 283 clones were recovered from Escherichia coliand sequenced. Sixty-six unique sequences were present among the283 clones, and these unique sequences yielded three classes of RNAs.

Koh et al.

1092 RNA, Vol. 15, No. 6




PUF-11 fused to glutathione S-transferase (GST) (Fig. 4A).GST–PUF-11 bound RNA I-1 with an apparent Kd of 43 6

3 nM; RNA II-1, 62 6 4 nM; and RNA III-1, 54 6 5 nM(Fig. 4B). We conclude that the avidities of these RNAs forGST–PUF-11 were similar, corroborating the three-hybriddata.

Different specificities

To compare the RNA-binding specificity of PUF-11 to thatof other C. elegans PUF proteins, we tested PUF-11 bindingto the known binding sites of other PUF proteins, and thebinding of other PUF proteins to PUF-11 binding sites. Forthis purpose, we used two RNAs from each class thatbound with high and low affinities. Using the three-hybridassay, we determined that PUF-11 bound the two sequencesfrom each class as expected (Fig. 5). PUF-11 did not bindthe PUF-8 site (NRE) and bound the PUF-5 site (5BE) onlyweakly. (Weak binding to the 5BE was due to an adven-titious site in the vector sequence.) Conversely, PUF-5bound its cognate site (the 5BE), but none of the PUF-11sites (Fig. 5). PUF-8 bound its cognate site, the NRE; it alsobound the Class I RNA, I-26. The interaction of PUF-8with this RNA was expected, as the RNA possessed two keydeterminants of PUF-8 binding: an AU dinucleotide inpositions +6 and +7 and an A at +4 (Opperman et al.2005). Conversely, PUF-11 did not bind the NRE because

the NRE has a C rather than an A at +5. We conclude thatPUF-11’s specificity is distinct from those of PUF-5 andPUF-8.

Different modes of RNA binding

Class I and Class III RNAs generally possessed an A atposition +5. To test the role of this position in greaterdepth, we analyzed base substitutions at this position ineach representative RNA using the three-hybrid assay (Fig.6A). In RNA I-1, an adenosine was absolutely required for

binding: any substitution reduced bind-ing to background levels. However, inRNA II-1, no such requirement wasobserved: any base was tolerable. RNAIII-1 was intermediate: either purinesupported binding, while both pyrimi-dines did not. To quantify the effects ofthe +5 base identity, we analyzed bind-ing of GST–PUF-11 in vitro to RNAs ofeach class that possessed either A or C atthis position (Fig. 6B). The data con-firm the three-hybrid results: In RNA I-1 and RNA III-1, a +5A was stronglypreferred, but in RNA II-1, A or Cbound equally well. We infer thatPUF-11 may recognize the base at posi-tion +5 of Classes I and III, but notClass II, again suggesting differentmodes of RNA binding.

Class II RNAs generally possessed a Cat position �1. To test the role of thisnucleotide systematically, we analyzedbase substitutions at this position ineach representative RNA (Fig. 7). Inboth three-hybrid (Fig. 7A) and gel-shift assays (Fig. 7B), a C at position�1 was important for PUF-11 bindingto RNA I-1 and RNA II-1, but not for

FIGURE 4. PUF-11 binds high-affinity sites of each RNA class with comparable affinities invitro. (A) RNAs from each class that bound PUF-11 with high affinity in the three-hybridassay. The sequences of the synthetic RNA oligonucleotides used for EMSAs are shown. Thecore binding element of each RNA is highlighted. Numbering begins with +1 at the first U ofthe core UGU trinucleotide. (B) EMSAs with purified recombinant GST–PUF-11. Proteinconcentrations are indicated above the gel. Two complexes of different electrophoreticmobilities were observed: The complex migrates slower than free RNA. The fraction ofRNA bound was plotted against protein concentration to obtain apparent Kd values for eachRNA.

FIGURE 5. PUF proteins with different specificities. Three-hybridanalysis of different C. elegans PUF proteins. Relative binding of RNAsto PUF-11 was normalized to RNA I-1; binding to PUF-5 wasnormalized to 5BE; and binding to PUF-8 was normalized to NRE.






binding to RNA III-1. In I-1 and II-1, both pyrimidinespermitted binding, but C was preferred (Fig. 7A). Thepreference for a �1C in Class I RNAs was surprising, sincethat preference was not observed in the three-hybridselection experiments (Fig. 2); this likely reflects thestringency of the selection, which permitted detection ofweaker Class I binding sites. Taken together, our resultssuggest that PUF-11 recognizes the base at the �1 positionin Class I and Class II RNAs, but not in Class III.

Our results support the following conclusions: Theidentities of bases at �1 and +5 are important determinantsof binding, and the requirements vary among classes ofbinding sites. These findings support the hypothesis thatPUF-11 can bind RNA in multiple modes.

Alignment of RNA and protein

The presence of UGU and AU elements in all three RNAclasses suggest that PUF-11 forms base-specific contactswith this element. To align specific PUF-11 repeats with thedifferent classes of RNAs, we analyzed compensatory

mutations between the protein and the RNAs (Figs. 8–10).The base specificity of PUF repeats can be switched ratio-nally, by changing the identities of specific amino acid sidechains that make edge-on contacts with the bases (Wanget al. 2002). This manipulation makes it possible to assignspecific helices to the bases they recognize. In several PUFproteins, the three conserved residues of the RNA-recogni-tion helix in Repeat 7 specify a guanosine (Wang et al. 2002;Miller et al. 2008; Y. Wang, L. Opperman, M. Wickens, andT. Hall, in prep.; D. Zhu, C. Stumpf, M. Wickens, and T.Hall, in prep.). To construct PUF-11 mutant proteins, wesubstituted these three residues (serine, histidine, and glu-tamate) at analogous positions in other repeats, and testedthose mutant proteins against RNAs containing mutations atspecific bases. In this fashion, we altered the base specificitiesof repeats 2, 3, and 4 in PUF-11, and aligned their RNA-recognition helices with different RNAs.

Class I RNAs (Fig. 8)

Class I RNAs possess an AU in positions 6 and 7, as doRNAs bound by human Pumilio1 (Fig. 8A). Thus, we

FIGURE 6. Mutational analysis of position +5 in PUF-11 bindingsites. (A) Base substitutions at position +5 of three RNAs analyzedusing the three-hybrid assay. The core binding element of each RNA isshown; UGU is shaded and the mutated base is underlined.b-galactosidase values are relative to the respective wild-type RNAs.(B) Base substitutions at position +5 of three RNAs analyzed in vitro.Synthetic RNAs with the indicated base substitution at position +5were analyzed in EMSAs with purified recombinant GST–PUF-11. Krel

is the ratio of the Kd relative to the respective wild-type RNAs.

FIGURE 7. Mutational analysis of position -1 in PUF-11 bindingsites. (A) Base substitutions at position �1 of three RNAs analyzedusing the three-hybrid assay. The core binding element of each RNA isshown; UGU is shaded and the mutated base is underlined.b-galactosidase values are relative to the respective wild-type RNAs.(B) Base substitutions at position �1 of three RNAs analyzed in vitro.Synthetic RNAs with the indicated base substitution at position �1were analyzed in EMSAs with purified recombinant GST–PUF-11. Krel

is the ratio of the Kd relative to the respective wild-type RNAs.

Koh et al.

1094 RNA, Vol. 15, No. 6




predicted that PUF-11 bound RNA in an uninterruptedone-helix/one-base fashion, as does Pumilio (Fig. 8A). Toalign Repeat 4 with RNA I-1, we mutated its Asn-His-Glnto Ser-His-Glu, and then tested RNA base specificity atpositions 4, 5, or 6 of RNA I-1 (Fig. 8B). If PUF-11 doesbind in the same manner as human Pumilio1, then theRepeat 4 mutant protein should bind an RNA with a G atnucleotide 5 (rather than an A, as in wild type). We testedthe mutant protein with wild-type RNA I-1 and nine otherRNAs, each containing a single-base substitution at posi-tions 4, 5, or 6 (Fig. 8B). Wild-type PUF-11 bound RNAswith an A, U, or G at position 4, and required A at bothpositions 5 and 6. The Repeat 4 mutant protein, however,required a G at nucleotide 5 and did not bind wild-type orother mutant RNAs. Thus, switching Asn-His-Gln to Ser-His-Glu changed the specificity from A to G at nucleotide5. Based on these data, we conclude that Repeat 4 of PUF-11 recognizes nucleotide 5 of RNA I-1.

We used the same strategy to align Repeat 3 with RNAI-1 (Fig. 8C). We predicted that switching Cys-Arg-Gln toSer-His-Glu would change the specificity for nucleotide 6from A to G. Indeed, the Repeat 3 mutant protein boundwhen nucleotide 6 was a G, which the wild-type protein didnot bind. Both proteins still accepted A at that position.Thus, we positioned Repeat 3 opposite of nucleotide 6 inRNA I-1. Surprisingly, the mutant protein bound the RNA

when nucleotide 7 was a G, which thewild-type protein did not. Either Repeat3 toggles between nucleotides 6 and 7 orthe mutation in the stacking amino acidalso changes identity in the adjacentbase. Indeed, arginine stacks poorly onguanosine, and that may be relieved bythe His substitution (Morozova et al.2006).

To align repeat 2 with RNA I-1, wechanged Asn-Phe-Gln to Ser-His-Glu(Fig. 8D). The repeat 2 mutant proteinbound when nucleotide 7 was a G,which the wild-type protein did not.Both proteins still accepted A at thatposition. Thus, we were able to positionrepeat 2 opposite of nucleotide 7 inRNA I-1.

Based on these compensatory muta-tions, we positioned repeats 2, 3, and 4opposite nucleotides 7, 6, and 5, respec-tively (Fig. 8A). We conclude that PUF-11 bound RNA I-1 in a continuous one-helix/one-base manner as does humanPumilio1.

Class II RNAs (Fig. 9)

In Class II RNAs, the AU dinucleotidelies at positions 7 and 8, 1 nt further from the UGU than inClass I. Nonetheless, we hypothesized that Repeats 2 and 3might still recognize the AU element (Fig. 9A). To alignRepeat 3 with RNA II-1, we changed Cys-Arg-Gln to Ser-His-Glu (Fig. 9B) and predicted that it would now bind Grather than A. Indeed, that was the case: The Repeat 3mutant protein preferred G (and accepted A), while thewild type-protein preferred A. Thus, we positioned repeat 3opposite nucleotide 7 in RNA II-1.

To align Repeat 2 with RNA II-1, we changed Asn-Phe-Gln to Ser-His-Glu (Fig. 9C). We predicted that thespecificity for nucleotide 8 would change from U to G.Wild-type protein preferred U over the other bases atnucleotide 8, but the Repeat 2 mutant protein preferred G(and A). We conclude that Repeat 2 binds nucleotide 8 inRNA II-1. Together, these results indicate that PUF-11accommodates an additional base in RNA II-1 as comparedwith RNA I-1. This mode of binding was distinct from thatin RNA I-1, where eight repeats contact 8 nt.

Class III RNAs (Fig. 10)

The spacing in Class III RNAs is the same as in Class II (Fig.10A). To align Repeat 2 with RNA III-1, we changed Asn-Phe-Gln to Ser-His-Glu (Fig. 10B). We predicted thespecificity for nucleotide 8 would change from U to G,

FIGURE 8. Alignment of PUF-11 repeats and RNA I-1. (A) A model illustrating how eachPUF-11 repeat (R1–R8, shaded) could interact with eight bases (A8–U1, unshaded) in the corebinding element of RNA I-1. Black lines indicate hydrogen bonds; stacking interactions andvan der Waals forces are not shown. Triangles indicate conserved interactions among PUFproteins; asterisks and a gray box indicate mapped interactions between PUF-11 and RNA I-1based on B–D. (B–D) Binding of Repeats 4, 3, and 2 mutant proteins to RNA I-1. Bindingscheme is shown on the left, with compensatory protein and RNA mutations unshaded.Binding of wild-type and mutant proteins to RNAs using the three-hybrid assay is shown onthe right. The new interactions that map the base to the a-helix are indicated by arrows.






and, indeed, this was the case. Thus, Repeat 2 lies oppositenucleotide 8 in RNA III-1.

Summary

The compensatory mutant strategy enabled us to alignrepeats with RNA in all three classes. Invariably, PUF-11Repeats 2 and 3 recognized the AU dinucleotide, but atdifferent nucleotide locations relative to UGU. PUF-11bound RNA I-1 in an uninterrupted one-helix/one-basemanner, but accommodated an additional nucleotide inRNA II-1 and RNA III-1. We conclude that PUF-11employs distinct RNA-binding modes with different RNAs.

DISCUSSION

PUF-11 binds to RNA in at least two modes, defined bydifferences in the spacing of two recognition elements,UGU and AU, and preferences for nucleotides at �1 and+5. A cytosine at �1 and an adenosine at +5 are preferredin certain contexts. The distance between the UGU and AUelements can either be 2 or 3 nt; the difference in spacinghas little effect on affinity. Compensatory mutants reveal

that the same helices recognize AU, andpresumably the UGU, in each mode.

We propose a structural model ofPUF-11 interactions with RNA. Thismodel focuses on flexibility in the centralregion of the protein and base flipping(Fig. 11), and draws on the knownstructures of human Pumilio1 andC. elegans FBF. In the ‘‘curved and in’’conformation, the central region ofPUF-11 is curved as in human Pumilio1,resulting in a one-to-one alignment ofbases and a-helices, and no base flipping(Fig. 11, left). In the ‘‘flat and flipped’’conformation, PUF-11 is flattened in theregion opposite nucleotides +5 and +6,resulting in base flipping (Fig. 11, right).These two states are analogous to thoseseen in complexes of human Pumilio1with cognate RNA, which is ‘‘curved andin’’ (Wang et al. 2002), and of FBF-2,which is ‘‘flat and flipped’’ (Y. Wang,L. Opperman, M. Wickens, and T. Hall,in prep.). Similarly, human Pumilio1 canbind noncognate RNA in a manner thatinvolves base flipping (Gupta et al.2008). It is striking that PUF-11 bindswith comparable affinity in either mode,implying that base-flipping does notresult in a dramatic energy cost, despitethe presumed loss of specific contactswith the protein. Base-to-base stacking

likely compensates.In FBF and yeast Puf4p, the identities of flipped bases are

flexible, while the identities of bases that make protein-specific contacts are constrained (Miller et al. 2008;Y. Wang, L. Opperman, M. Wickens, and T. Hall, inprep.). In Class I RNAs, the highly conserved bases betweenUGU and AU suggest that these nucleotides form specificcontacts with PUF-11 in a continuous one-helix/one-basemanner (as in human Pumilio1). In Classes II and III,however, the degeneracy of bases between the two motifssuggests that one or more nucleotides turn away from theprotein. UGU invariably is recognized by Repeats 6–8 ofPUF-11, and AU by Repeats 2 and 3.

In one simple view, interaction with one element, saythe UGU, would enable a ‘‘search’’ for additional recogni-tion elements nearby. Upon that second interaction, withthe AU dinucleotide, bases that do not interact with theprotein are left stacked on one another. It is unclearwhether these two protein conformations exist in theapo-protein, or are induced upon binding the RNA. Ineither case, the flattened protein may be able to accommo-date an even greater number of additional bases betweenthe UGU and AU.

FIGURE 9. Alignment of PUF-11 repeats and RNA II-1. (A) A model illustrating how eachPUF-11 repeat (R1–R8, shaded) could interact with nine bases (A9–U1, unshaded) in the corebinding element of RNA II-1. Black lines indicate hydrogen bonds; stacking interactions andvan der Waals forces are not shown. Triangles indicate conserved interactions among PUFproteins; asterisks and a gray box indicate mapped interactions between PUF-11 and RNA II-1based on B and C. (B,C) Binding of Repeats 3 and 2 mutant proteins to RNA II-1. Bindingscheme shown on left, with compensatory protein and RNA mutations unshaded. Binding ofwild-type and mutant proteins to RNAs using the three-hybrid assay is shown on the right. Thenew interactions that map the base to the a-helix are indicated by arrows.

Koh et al.

1096 RNA, Vol. 15, No. 6




In Classes II and III RNAs, UGU and AU are spacedidentically; the distinction between these classes may lie inprecisely which central bases contact the protein. In Class IIRNAs, the tolerance of PUF-11 for all bases at position +5indicates that base may flip; in Class III RNAs, thepreference for A at position +5 suggests that it contactsPUF-11, which implies that +4 may flip. In both instances,the contact is likely to be with the same binding pocket ofRepeat 5. Indeed, this is likely to be the same site to whichthe adenosine (or guanosine) at +4 binds in Class I RNAs.

The distinctive features of complexes with Class I, II, andIII RNAs may enable them to be used differentially in vivo.For example, proteins that interact with the PUF-RNAcomplexes in vivo, such as CPEB-1, NANOS-3, and GLD-2/GLD-3, could readily discriminate one class from another.Protein partners might also restrict flexibility of the apo-protein, driving its interaction with specific mRNAs.

The �1 position presents another key facet of specificity,in which PUF-11 strongly prefers a cytosine in Classes I andII RNAs. In this respect, PUF-11 echoes yeast Puf3p; abinding pocket at the C terminus of the Puf3p RNA-binding domain specifically recognizes a C at the �2position (D. Zhu, C. Stumpf, M. Wickens, and T. Hall, inprep.). We suggest that PUF-11 possesses an analogousbinding pocket for a cytosine. The identity of the �1position was unimportant in binding to Class III RNAs;this may be due to the use of the cytosines at either �2 or�3 in the place of �1. Alternatively, the precise interactionsin positions +4 to +6 could affect the pocket thatrecognizes the �1C.

Our data demonstrate that a single PUF protein recog-nizes different sites using distinct modes of RNA binding.This greatly diversifies the possibilities for regulation, andthe potential effects of coregulators. Structural analysis

will reveal the precise nature of thecomplexes on the three classes of RNAand will open the door to mechanisticstudies of how those different complexesform and are used for regulation in vivo.

MATERIALS AND METHODS

Constructs for yeast three-hybridassay

The RNA-binding region of PUF-11 (Gen-Bank accession no. NM_171363), compris-ing amino acids 50–505, was cloned into anactivation domain vector, pGADT7 (Stumpfet al. 2008b). For specificity testing ofdifferent PUFs (Fig. 5), full-length PUF-5(NM_063413) and residues 127–519 ofPUF-8 (NM_063122) were cloned into pAC-TII activation vectors. The PUF-11 mutantsused in the compensatory experiments (Figs.8–10) were created by site-directed muta-

genesis. Amino acid changes were as follows: for PUF-11 Repeat 4mutant, N272S and Q276E; Repeat 3 mutant, C235S, R236H, andQ239E; Repeat 2 mutant, N199S, F200H, and Q203E. Therandomized RNA library was constructed by cloning DNAoligonucleotides (CCGGCTAGCNNNTGTNNNNNNNAATTTAATAAAGCATG) into the XmaI and SphI sites of p3HR2 (Stumpfet al. 2008b). All other RNA constructs used in the three-hybridassay were created in the same fashion.

Yeast three-hybrid assay

Three-hybrid assays were performed in the YBZ-1 yeast strain asdescribed previously (Hook et al. 2005). Selection experiments

FIGURE 10. Alignment of PUF-11 repeats and RNA III-1. (A) A model illustrating how eachPUF-11 repeat (R1–R8, shaded) could interact with nine bases (A9–U1, unshaded) in the corebinding element of RNA III-1. Black lines indicate hydrogen bonds; stacking interactions andvan der Waals forces are not shown. Triangles indicate conserved interactions among PUFproteins; asterisks and a gray box indicate mapped interactions between PUF-11 and RNA III-1based on B. (B) Binding of Repeat 2 mutant protein to RNA III-1. Binding scheme is shown onthe left, with compensatory protein and RNA mutations unshaded. Binding of wild-type andmutant proteins to RNAs using the three-hybrid assay is shown on the right. The newinteractions that map the base to the a-helix are indicated by arrows.

FIGURE 11. Model of PUF-11 interactions with RNA. The protein isdepicted as a series of discs corresponding to PUF repeats, arranged intwo groups of four. Interactions of bases (ellipse) with bindingpockets in each repeat also are shown. Alternate conformations inthe central region of the protein alter the overall curvature, as hasbeen observed with FBF (Y. Wang, L. Opperman, M. Wickens, and T.Hall, in prep.), and affect base flipping. In one conformation, theprotein is curved and the bases are ‘‘in,’’ contacting the protein (left,‘‘Curved and In’’). In the other conformation, the protein is flat andthe central bases flip (right, ‘‘Flat and Flipped’’). The alteration ofstructure may involve different helical packing in that region, ratherthan a distinct ‘‘hinge’’ region. See the text for details.






were carried out using medium lacking histidine and in thepresence of 5 mM 3-aminotriazole. A total of 283 out of 666positive interactors were recovered in Escherichia coli and reintro-duced into naı̈ve YBZ-1 to confirm the interaction and specificity.Three-hybrid assays were conducted with the following modifi-cations: 100 mL of a saturated culture was diluted into 4-mLselective medium and allowed to grow for 2.5 h to reach an OD660

of 0.1–0.2. b-galactosidase activity was normalized to an OD660 of0.1. b-galactosidase values are represented as an average of threerepetitions, with standard deviation shown in each figure.

Protein construct and purification

PUF-11 (amino acids 50–505) was cloned into pGEX6P1 (Amer-sham) as GST fusion proteins and purified as described previously(Hook et al. 2005). Protein was eluted from the glutathione-sepharose resin (Amersham) without cleavage and dialysis. Elutionbuffer consists of 50 mM reduced glutathione (Sigma), 1X PBS,0.2% Tween-20, 150 mM NaCl, 0.1% b-mercaptoethanol, and 30%glycerol (pH 8.0). Protein purity was measured by SDS gelelectrophoresis and concentration determined by a Bradford assay.

Electrophoretic mobility-shift assay

A total of 100 fmol of RNA oligonucleotides (Dharmacon), 32P 59

end-labeled using T4 kinase, were combined with a range ofprotein concentrations in 10 mM HEPES (pH 7.4), 1 mM EDTA,20 mM NaCl, 1 mM DTT, and 0.002% Tween-20 for 30 min atroom temperature. Loading dye (5 mL of 6% glycerol and 0.06%Bromophenol Blue) was added to each 10-mL reaction beforeloading 5 mL on a pre-run nondenaturing 5% polyacrylamide gel(BioRad). Gels were resolved in 1X TBE at 100 V for 30 min at4°C. Gels were then exposed to storage phosphor screens for 1 hand scanned using a Typhoon 9410 Workstation (GE Healthcare).The fraction of bound RNA relative to total RNA in the reactionwas determined using ImageQuant (Amersham). The apparentKd, concentration of protein at which half-maximal bindingoccurs, was determined using GraphPad Prism 4. The Kd valuesand standard errors reported here were from four independentexperiments.

Computational analysis of sequences

We used a mixture of position weight matrices (PWMs) (Barashet al. 2003) to characterize whether these sets of sequences are bestdescribed by multiple RNA sequence classes. This probabilisticmodel assumed that each sequence originated from one of the KRNA sequence classes and model fitting involved inferring theclass of each sequence as well as choosing the best K. Weconsidered values between 1 and 6 for K and multiple modelselection criteria yielded K = 3 as the optimal number.

ACKNOWLEDGMENTS

We appreciate the advice and comments of members of theWickens laboratory and the help of Laura Vanderploeg inpreparing the figures. Y.K. is supported by the A*STAR NationalScience Scholarship from Singapore, S.K. by NIH grantHG003747, and M.W. by NIH grants 144-MK47 and 144-QS25.A.M. was supported by a Khorana Summer Fellowship during herparticipation in these studies.

Received January 7, 2009; accepted March 2, 2009.

REFERENCES

Ahringer, J. and Kimble, J. 1991. Control of the sperm-oocyte switchin Caenorhabditis elegans hermaphrodites by the fem-3 39 untrans-lated region. Nature 349: 346–348.

Bachorik, J.L. and Kimble, J. 2005. Redundant control of theCaenorhabditis elegans sperm/oocyte switch by PUF-8 and FBF-1, two distinct PUF RNA-binding proteins. Proc. Natl. Acad. Sci.102: 10893–10897.

Barash, Y., Elidan, G., Friedman, N., and Kaplan, T. 2003. Modelingdependencies in protein-DNA binding sites. Proceedings of theSeventh Annual International Conference on Research in Computa-tional Molecular Biology. ACM Press, pp 28–37. New York.

Bernstein, D., Hook, B., Hajarnavis, A., Opperman, L., andWickens, M. 2005. Binding specificity and mRNA targets of a C.elegans PUF protein, FBF-1. RNA 11: 447–458.

Chagnovich, D. and Lehmann, R. 2001. Poly(A)-independent regula-tion of maternal hunchback translation in the Drosophila embryo.Proc. Natl. Acad. Sci. 98: 11359–11364.

Chen, G., Li, W., Zhang, Q.S., Regulski, M., Sinha, N., Barditch, J.,Tully, T., Krainer, A.R., Zhang, M.Q., and Dubnau, J. 2008.Identification of synaptic targets of Drosophila pumilio. PLoSComput. Biol. 4: e1000026. doi: 10.1371/journal.pcbi.1000026.

Crittenden, S.L., Bernstein, D.S., Bachorik, J.L., Thompson, B.E.,Gallegos, M., Petcherski, A.G., Moulder, G., Barstead, R.,Wickens, M., and Kimble, J. 2002. A conserved RNA-bindingprotein controls germline stem cells in Caenorhabditis elegans.Nature 417: 660–663.

Dubnau, J., Chiang, A.S., Grady, L., Barditch, J., Gossweiler, S.,McNeil, J., Smith, P., Buldoc, F., Scott, R., Certa, U., et al. 2003.The staufen/pumilio pathway is involved in Drosophila long-termmemory. Curr. Biol. 13: 286–296.

Edwards, T.A., Pyle, S.E., Wharton, R.P., and Aggarwal, A.K. 2001.Structure of pumilio reveals similarity between RNA and peptidebinding motifs. Cell 105: 281–289.

Galgano, A., Forrer, M., Jaskiewicz, L., Kanitz, A., Zavolan, M., andGerber, A.P. 2008. Comparative analysis of mRNA targets forhuman PUF-family proteins suggests extensive interaction withthe miRNA regulatory system. PLoS One 3: e3164. doi: 10.1371/journal.pone.0003164.

Gerber, A.P., Herschlag, D., and Brown, P.O. 2004. Extensiveassociation of functionally and cytotopically related mRNAs withPuf family RNA-binding proteins in yeast. PLoS Biol. 2: E79. doi:10.1371/journal.pbio.0020079.

Goldstrohm, A.C. and Wickens, M. 2008. Multifunctional deadenylasecomplexes diversify mRNA control. Nat. Rev. Mol. Cell Biol. 9:337–344.

Goldstrohm, A.C., Hook, B.A., Seay, D.J., and Wickens, M. 2006. PUFproteins bind Pop2p to regulate messenger RNAs. Nat. Struct. Mol.Biol. 13: 533–539.

Goldstrohm, A.C., Seay, D.J., Hook, B.A., and Wickens, M. 2007. PUFprotein-mediated deadenylation is catalyzed by Ccr4p. J. Biol.Chem. 282: 109–114.

Gupta, Y.K., Nair, D.T., Wharton, R.P., and Aggarwal, A.K. 2008.Structures of human Pumilio with noncognate RNAs revealmolecular mechanisms for binding promiscuity. Structure 16:549–557.

Hook, B., Bernstein, D., Zhang, B., and Wickens, M. 2005. RNA–protein interactions in the yeast three-hybrid system: Affinity,sensitivity, and enhanced library screening. RNA 11: 227–233.

Kimble, J. and Crittenden, S.L. 2007. Controls of germline stem cells,entry into meiosis, and the sperm/oocyte decision in Caenorhab-ditis elegans. Annu. Rev. Cell Dev. Biol. 23: 405–433.

Lamont, L.B., Crittenden, S.L., Bernstein, D., Wickens, M., andKimble, J. 2004. FBF-1 and FBF-2 regulate the size of the mitoticregion in the C. elegans germline. Dev. Cell 7: 697–707.

Koh et al.

1098 RNA, Vol. 15, No. 6




Lee, M.H., Hook, B., Lamont, L.B., Wickens, M., and Kimble, J. 2006.LIP-1 phosphatase controls the extent of germline proliferation inCaenorhabditis elegans. EMBO J. 25: 88–96.

Lee, M.H., Hook, B., Pan, G., Kershner, A.M., Merritt, C., Seydoux, G.,Thomson, J.A., Wickens, M., and Kimble, J. 2007. Conservedregulation of MAP kinase expression by PUF RNA-binding proteins. PLoS Genet. 3: e233. doi: 10.1371/journal.pgen.0030233.

Lublin, A.L. and Evans, T.C. 2007. The RNA-binding proteins PUF-5,PUF-6, and PUF-7 reveal multiple systems for maternalmRNA regulation during C. elegans oogenesis. Dev. Biol. 303:635–649.

Miller, M.T., Higgin, J.J., and Hall, T.M. 2008. Basis of altered RNA-binding specificity by PUF proteins revealed by crystal structuresof yeast Puf4p. Nat. Struct. Mol. Biol. 15: 397–402.

Morozova, N., Allers, J., Myers, J., and Shamoo, Y. 2006. Protein-RNAinteractions: Exploring binding patterns with a three-dimensionalsuperposition analysis of high resolution structures. Bioinformatics22: 2746–2752.

Morris, A.R., Mukherjee, N., and Keene, J.D. 2008. Ribonomicanalysis of human Pum1 reveals cis-trans conservation acrossspecies despite evolution of diverse mRNA target sets. Mol. Cell.Biol. 28: 4093–4103.

Morrison, S.J. and Kimble, J. 2006. Asymmetric and symmetric stem-cell divisions in development and cancer. Nature 441: 1068–1074.

Nolde, M.J., Saka, N., Reinert, K.L., and Slack, F.J. 2007. TheCaenorhabditis elegans pumilio homolog, puf-9, is required forthe 39UTR-mediated repression of the let-7 microRNA target gene,hbl-1. Dev. Biol. 305: 551–563.

Olivas, W. and Parker, R. 2000. The Puf3 protein is a transcript-specific regulator of mRNA degradation in yeast. EMBO J. 19:6602–6611.

Opperman, L., Hook, B., DeFino, M., Bernstein, D.S., and Wickens, M.2005. A single spacer nucleotide determines the specificities oftwo mRNA regulatory proteins. Nat. Struct. Mol. Biol. 12: 945–951.

Schweers, B.A., Walters, K.J., and Stern, M. 2002. The Drosophilamelanogaster translational repressor pumilio regulates neuronalexcitability. Genetics 161: 1177–1185.

Seay, D., Hook, B., Evans, K., and Wickens, M. 2006. A three-hybridscreen identifies mRNAs controlled by a regulatory protein. RNA12: 1594–1600.

Stumpf, C., Opperman, L., and Wickens, M. 2008a. Analysis of RNA-protein interactions in a yeast three-hybid system. In Methods inenzymology (eds. L.E. Maquat and M. Kiledjian). Vo. 449, pp. 295–315. Academic, Burlington, VT.

Stumpf, C.R., Kimble, J., and Wickens, M. 2008b. A Caenorhabditiselegans PUF protein family with distinct RNA binding specificity.RNA 14: 1550–1557.

Thompson, B.E., Bernstein, D.S., Bachorik, J.L., Petcherski, A.G.,Wickens, M., and Kimble, J. 2005. Dose-dependent control ofproliferation and sperm specification by FOG-1/CPEB. Develop-ment 132: 3471–3481.

Wang, X., Zamore, P.D., and Hall, T.M. 2001. Crystal structure of aPumilio homology domain. Mol. Cell 7: 855–865.

Wang, X., McLachlan, J., Zamore, P.D., and Hall, T.M. 2002. Modularrecognition of RNA by a human pumilio-homology domain. Cell110: 501–512.

Wharton, R.P., Sonoda, J., Lee, T., Patterson, M., and Murata, Y.1998. The Pumilio RNA-binding domain is also a translationalregulator. Mol. Cell 1: 863–872.

Wickens, M., Bernstein, D.S., Kimble, J., and Parker, R. 2002. A PUFfamily portrait: 39UTR regulation as a way of life. Trends Genet.18: 150–157.

Wreden, C., Verrotti, A.C., Schisa, J.A., Lieberfarb, M.E., andStrickland, S. 1997. Nanos and pumilio establish embryonicpolarity in Drosophila by promoting posterior deadenylation ofhunchback mRNA. Development 124: 3015–3023.

Ye, B., Petritsch, C., Clark, I.E., Gavis, E.R., Jan, L.Y., and Jan, Y.N.2004. Nanos and Pumilio are essential for dendrite morphogenesisin Drosophila peripheral neurons. Curr. Biol. 14: 314–321.

Zamore, P.D., Williamson, J.R., and Lehmann, R. 1997. The pumilioprotein binds RNA through a conserved domain that defines anew class of RNA-binding proteins. RNA 3: 1421–1433.

Zhang, B., Gallegos, M., Puoti, A., Durkin, E., Fields, S., Kimble, J.,and Wickens, M.P. 1997. A conserved RNA-binding protein thatregulates sexual fates in the C. elegans hermaphrodite germ line.Nature 390: 477–484.






a single c. elegans puf protein binds rna in multiple modes · asinglec. elegans puf protein binds...

Documents