Nucleic Acids Research, 1995, Vol. 23, No. 24 4963-4970

Identification of a 67 kDa protein that bindsspecifically to the pre-rRNA primary processing sitein a higher plantManuel Echeverria* and Sylvie Lahmy

Laboratoire de Physiologie et Biologie Moleculaire Vegetale, Universite de Perpignan, URA CNRS 565, Avenuede Villeneuve, 66860 Perpignan Cedex, France

Received October 9, 1995; Revised and Accepted November 16, 1995

ABSTRACT

In radish pre-rRNA primary processing cleavageoccurs at a UUUUCGCGC element (motif P) mapped inthe 5'-external transcribed spacer (Delcasso-Tremou-saygue et al., 1988). Significantly, motif P is part of acluster of homologous elements including threeUUUUCCGG elements (motifs A123) and a singleUUUUGCCCC element (motif B). Here we used theEMSA to identify in radish extracts an RNA-bindingactivity, NF C, that specifically interacts with thepre-rRNA A123BP sequence. Using different RNAprobes and competitors we show that NF C recognisesa 38 base RNA sequence including the 3'-end of motifA3 and motifs B and R NF C binds to poly U, but not topoly A, poly C or poly G. Therefore we used poly (U)Sepharose chromatography as a final step to obtainpure NF C fractions. These, analysed by SDS-PAGE,revealed two major polypeptides of 67 and 60 kDa.According to UV cross-linking analysis the 67 kDapolypeptide corresponds to NF C activity, while the 60kDa species is a proteolysed form of this protein. Wealso showed that NF C is enriched in nuclear extracts.Based on its stringent RNA substrate specificity andits nuclear localisation we propose that NF C isinvolved in pre-rRNA primary processing in plants.

INTRODUCTION

Nuclear genes coding for the 25S, 5.8S and 18S ribosomal RNAs(rRNAs) are organised as head-to-tail tandem repeats of thecoding sequences separated by internal spacers and a largeintergenic spacer (IGS). Transcription of this unit by RNApolymerase I (RNA pol I) gives a primary rRNA precursor(pre-rRNA) comprising both therRNA and the transcribed spacersequences. The pre-rRNA is subsequently processed in thenucleolus to generate the mature rRNAs. Pre-rRNA processinginvolves specific nucleolytic cleavages which remove the inter-genic spacers, base modification and ribosome assembly (forreviews see 1,2).

In all eukaryotes the first pre-rRNA processing eventcorresponds to an endonucleolytic cleavage on the 5'-externaltranscribed spacer (5'-ETS), several hundred nucleotides up-stream from the 5'-end of the 18S coding sequence. Recent dataobtained by transient expression of cloned rDNA in Xenopuslaevis show that the first pre-rRNA cleavage occurs concomitant-ly with rDNA transcription, generating the typical Miller spreadof eukaryotic rRNA transcription units characterised by terminalballs at the 5'-end of the nascent pre-rRNA (3). These terminalballs, observed in all eukaryotes including plants (4), are largeribonucleoprotein particles (RNP) involved in the 5'-ETS proces-

sing (3). In vertebrates the pre-rRNA primary cleavage has beenreproduced in vitro using crude extracts from mouse and X. laeviscells (5-7). The pre-rRNA processing activity detected in vitrodepends on U3 small nucleolarRNA (snoRNA) and is correlatedwith the formation ofRNP on the 5'-ETS detected by electropho-retic mobility shift assay (EMSA). Critical sequences directingpre-RNA primary processing and RNP formation are restricted toa few nucleotides encompassing the cleavage site which are

relatively conserved in vertebrates (6,8). The RNP complexesinclude various polypeptides, some of which are homologousbetween mammals and X.laevis (6).

In plants the identification of cis- and trans-acting factorscontrolling pre-rRNA processing has been limited by thedifficulties in obtaining cell-free systems capable of specificrDNA transcription and/or pre-rRNA processing. The completesequence of the 5'-ETS has been reported for many plantsrevealing a great heterogeneity of size and sequence betweenspecies (9). In a few plant species the transcription start site (TIS)and pre-rRNA primary processing site have been localised bymapping the 5'-ends of the major pre-rRNAs intermediates usingS 1 nuclease and primer extension analysis. Using this approachin radish, a crucifer which has the most simple IGS structureamong plants (see 9), our laboratory mapped the pre-rRNAprimary processing site to a single rDNA motif (site P) located onthe 5'-ETS (Fig. IA; 10). Intriguingly, this motif is found just a

few bases downstream from four homologous motifs correspon-ding to A123 and B (Fig. IA). Computer analysis predicting RNAfolding suggests that the A123BP clustered motifs induceimportant secondary structures on the 5'-end of the nascentpre-rRNA (11; data not shown).

* To whom correspondence should be addressed

4964 Nucleic Acids Research, 1995, Vol. 23, No. 24

A+1

TIS- H Ir-'-fl-f

AXBPr

740

I I B

rBP (71 at)

At A} AM% Ml I

rA123 (6lnt)

MI

HS200

r-UiSI41

SimI

DEAE SepbseI

LU.

PoW lU S

US3W

F4ure 1. (A) Structure ofthe radish rDNA ES regionencompassing the pre-rRNA pinmary processing site (datafromrefs 10,1 1). Angled arrows indicate major5'-endsofpre-rRNAs interediates. +1 indicates the TIS. + 186 indicates the pre-rRNA primary processing site. +740 indicates the 5'-end of 18Scoding sequence indicatedby stippling. Boxes A123, B and P show the five homologous motifs. NF B indicates the sequence-specific DNA-binding factor previously identified and its cognatesite. (B) Strucure of fte probes rA123BP, rBPand rA123 synthesized in vitro. The length ofeach probe is given in brackets. Heavy lines indicate pre-rRNA sequences.Thin lines are 7 and 4 bases of vector sequence at the 5'- and 3'-ends of each probe respectively. A detailed sequence is given only for rA123BP between motifs A1and P. Broken lines show the corresponding 5'- and 3'-extremities of probes rA'23 and rBP. Open boxes indicate motifs A123, B and P. (C) Diagram of the procedurefor protein fractionation of radish whole cell extracts.

We recently identified in radish a nuclear factor, NF B, thatbinds to a double-stranded DNA sequence including site B(Fig. lA; 11). The specificity ofNF B is very strict as it does notbind to the homologous A123 or P adjacent motifs. In additionNF B does not bind to single-stranded substrates thus excludingthe possibility dtat it is a pre-rRNA binding factor (11). Wesuggested that NP B controls the elongation rate ofRNA pol I in thisregion of the 5'-ETS by coupling rDNA trarnription to pre-rRNAprimry processing (11), which in vivo occur concomitanty (3).Taken together these data suggest that the A123BP clustered

motif could have a regulatory role in pre-rRNA processing. Weproposed that upon transcription the A123BP sequences induceRNA secondary stuctures which would be recognised bytrans-acting factor(s) involved in the pre-rRNA primary cleavagein crucifers (11). This prompted us to search in radish extracts forprotein factor(s) recognising this pre-rRNA sequence in vitro.Using the EMSA here we report the identification in radishextracts of a protein which specifically binds to the pre-rRNAsequence encompassing both the primary processing site P andthe homologous site B. We purified this protein by affinitychromatography and determined its polypeptide composition byUV cross-linking photolabelling and gel filtration chromatogra-phy. We also present evidence suggesting that nuclear fractionsare enriched in this factor. According to its stringent RNA-bindingspecificity and its physical properties this factor is the first exampleof sequence-specific interaction between a nuclear protein and the

5'-end of pre-rRNA in plants. We suggest that it could be a novelcomponent of the pre-rRNA processing machinery.

MATERIALS AND METHODS

Plant material

Radish seeds (Raphanus sativus cv. rond rose A bout blanc,National) were purchased fomn Vilmorin. Three-day-old seedlingsgrown in daylight conditions were used for all protein extrtions.

RNA probes and oligonucleotides

The rDNA sequences transcribed in vitro to generate the

pre-rRNA probes were obtained by PCR amplification of the IGSsequence from plasmid pD12-3 (10) using a pair of oligonucleo-tides containing EcoRI and HindI: restriction sites at theirextremities. The radish IGS sequences amplified were: A123BP(103-205), BP (147-205) and A123 (108-156) (nuceotides are

numbered relative to the transcription initiation site). The

amplified DNA fragments were then inserted into the EcoRIIHindIH site of pGEM-3Zf(+) (Promega), generating plasmidspGA123BP, pGBPandpGA123 respectively. These plasmids werethen linearised by HindmI and used as templates for in vitrotranscription by T7 RNA polymerase to generate RNA probesrAI23BP, rBP and rA123. The anti-sense RNA probes asrAI23BPor asrBP were obtained from transcription of EcoRl linearisedpGA123BP or pGB with SP6 RNA polymerase. In vitrotranscription was performed according to conditions specified byPromega, using [cr-32P]CTP as a labelled precursor.

The double-stranded rBP/DNA probe was prepared by mixing50 000 c.p.m. of probe rBP with 100 ng of complementary DNA

C

BrV23BP (114 at)

A! a2 i B P

I

-

p II

pop-nm

Nucleic Acids Research, 1995, Vol. 23, No. 24 4965

oligonucleotide sequence (corresponding to anti-sense oli r38) inbuffer 50 mM Tris-HCl pH 8, 10 mM MgCl2 heating 5 min at65°C and slowly decreasing the temperature.

Oligonucleotides (Eurogentec) used for competition were:Oli r49: CCGGUGGACGAUUUUGCCCCUGAUAUGAAAUUUUCGCGCU-UUGACGGACOli r38: CCGGUGGACGAUUUUGCCCCUGAUAUGAAAUUUJUCGCGOli r35: UGGACGAUUUUGCCCCUGAUAUGAAAUUUUCGCGCOli r28: CCGGUGGACGAUUUUGCCCCUGAUAUGAAA

Preparation of radish seedling extracts

The fractionation procedure is shown schematically in Figure IC.Whole cell extracts from radish seedlings were prepared aspreviously described (11). The RNA-binding activity correspon-ding to NF C was followed by EMSA with probe rBP. Proteinswere measured by Bradford (12). All buffers contained 1 mMEDTA, 1 mM EGTA, 10 ,ug/ml pepstatin A, 0.5 ,ug/ml leupeptin,1 mM PMSF, 1 mM benzamide. All steps were carried out at 0°C.Initially 100 g of fresh tissue were homogenised in a Waringblender in 800 ml of buffer 1 (50 mM Tris-HCl pH 8.5, 10 mMMgCl2, 10% sucrose, 10mM 2-mercaptoethanol, 20% glycerol).The homogenate was filtered through Miracloth, 1/10 vol 3.8 M(NH4)2SO4 was added. After 1 h of stirring the homogenate wascentrifuged for 1 h at 38 000 r.p.m. in a Ti 45 rotor (Beckman).The supernatant was precipitated by addition of 0.33 g(NH4)2SO4/ml extract. After 1 h of stirring precipitated proteinswere recovered by centrifugation. The pellet was resuspended in5 ml buffer I1/00 (50 mM Tris-HCl pH 8, 6 mM MgCl2, 5 mMDTT, 15% glycerol and 100mM KCI) and dialysed for 3 h against150 ml of the same buffer. The dialysed extract, corresponding toS100 fraction (Fig. IC), was then loaded on a 15 ml column ofDEAE-Sepharose CL6B (Pharmacia) equilibrated with buffer11/100. The NF C activity was eluted with buffer 11300mM KCI.The peak of protein fractions were pooled, diluted with buffer IIto a final concentration of 200 mM KCI, and loaded on a 10 mlcolumn of Heparin-Sepharose CL6B (Pharmacia) equilibratedwith buffer I1/200mM KCI. The column was washed with buffer11/300 mM KCI. NF C was eluted with buffer 11/600 mM KCI.The fractions containing NF C activity were pooled, whichcorresponds to HS600 fraction (Fig. IC). Typically 100 g ofseedlings yielded 10 mg of proteins in this fraction.For final purification HS 600 fractions from three different

preparations (corresponding to 300 g of tissue) were pooled anddialysed 2 h against buffer 11/100mM KCI. The dialysed pool wasloaded on a 2 ml column of poly(U) Sepharose 4B (Pharmacia)equilibrated with buffer 11/100 mM KCI. All NF C activity wasretained. The column was washed with buffer II/300mM KCI andNF C activity was eluted with buffer 111800 mM KCI. Fractionscontaining the peak of NF C activity were pooled, representingthe US800 fraction, and stored at -80°C. For long storage 0.1mg/ml ofbovine serum albumin (BSA) was added to this fraction.Nuclear extracts from radish seedlings were prepared starting

from 40 g of fresh tissue according to Green et al. (13).

EMSA analysis

RNA-binding reactions were performed in 15 gl of buffer (50mM Tris-HCl pH 8,6mM MgCl2, 5mM DTE, 15% glycerol and100mM KCI) containing 2 ,ug yeast tRNA, 10 U RNasin, and the

competitor was thoroughly extracted with phenol-chloroform.Reactions were started by adding 10 000-20 000 c.p.m. of theRNA probe. Incubation was carried out for 20 min at 0°C. Afterincubation the reaction was directly loaded on 5% polyacryl-amide gel (29/1 acrylamide/bisacrylamide, 20 x 16 x 0.75 cm).The gel was pre-run in TBE buffer (45 mM Tris-HCl pH 8.3,45 mM boric acid, 0.5 mM EDTA) and electrophoresis wascarried out in the same buffer for 1 h at 15 mA. After migrationthe gel was dried and exposed overnight with autoradiographicfilm.

SDS-PAGE and UV cross-linking analysis

UV cross-linking analysis was done according to McEwan (14).For UV cross-linking RNA binding assays were done as usualusing 15 ,ul US800 or 5 gl of DE100 fractions (Fig. IC), 100 000c.p.m. of 32P-labelled probe rBP or asrBP and 200 ng yeast tRNA.RNA-binding assays was carried out at 0°C for 20 min. Afterincubation the reaction mixture was transfered to a Parafilmsheet, placed in a Stratalinker (Stratagene) and irradiated for 5min with a 254 nm UV lamp. After irradiation droplets weretransfered back into Eppendorf tubes and 1 jig RNase A and 1 jgRNase T1 were added. Tubes were then incubated for 30 min at37°C. The RNase reaction was stopped by addition of 1 volloading buffer.

Separation of proteins by SDS-polyacrylamide gel electro-phoresis (SDS-PAGE) was performed on a 12.5% acrylamide gel(15). Protein samples were diluted with 1 vol loading buffer (4%SDS, 0.125 M Tris-HCl pH 6.8, 10% glycerol, 0.2 M DTT and0.2% bromophenol blue), and heated to 95°C before loading.Electrophoresis was done on a Protean II apparatus (BioRad).After running gels were stained using the Silver Staining Plus Kit(BioRad), dried and exposed for autoradiography.

Proteins from US800 fraction were concentrated by acetoneprecipitation. Nine volumes of cold acetone were added to 150 ,ulof this fraction. The sample was kept for 1 h at -20°C,centrifuged, and the protein pellet was resuspended in 20 jlloading buffer.

Western blot

Western blots were carried out as described by Domon et al. (16).Proteins from nuclear and S100 fractions (6 ,ul each) wereresolved by SDS-PAGE (15) on a mini-Protean II cell (BioRad)and transfered onto a nitrocellulose membrane (Schleicher &Schuell). Membranes were then immersed in 5% gelatin inTBS-T buffer (20 mM Tris pH 7.6, 137 mM NaCl, 0.1% Tween20). Membranes were incubated overnight at room temperaturewith anti-serum directed against Brassica napus C24 protein(BnC24) (17), Arabidopsis thaliana NADPH thioredoxin reduc-tase (NTR) (18) and the large subunit of the tobacco rubisco(rbcL) (19), using a 1/20 000 (anti-BnC24 and anti-NTR) or

1/40 000 (anti-rbcL) dilution in TBS-T containing 0.1% non-fatmilk. Membranes were then washed in TBS-T buffer andincubated 2 h at room temperature with goat anti-rabbit antibody-conjugated horseradish peroxidase (1/20 000 dilution). Forvisualisation of protein bands blots were washed with TBS-T,incubated with ECL (enhanced chemiluminiscence) reagentsolutions (Amersham) and exposed to X-ray films.Anti-BnC24 and anti-NTR serum were provided by J. SaLez-

Va'squez and Y Meyer (our laboratory). Anti-rbcL serum was

indicated protein fractions. Yeast tRNA (Boehringer) used as kindly provided by J. Fleck (CNRS-Strasbourg).

4966 Nucleic Acids Research, 1995, Vol. 23, No. 24

RESULTS

Structure of the primary pre-rRNA processing siteregion of radish

The structural organisation of the 5'-ETS region including thepre-rRNA primary processing site ofradish is shown in Figure lA(data from 10,11). The pre-rRNA primary processing site wasmapped 186 bases downstream from the TIS, in aUUUUCGCGC element (motif P in Fig. lA). This motif,characterised by four Us followed by alternating CG residues ispreceded by three UUUUCCGG elements (motifs A123) and asingle UUUUGCCCC element (motif B). Motif B differs frommotifs A123 and P because it has a single G separating the twopyrimidine stretches. The double-stranded DNA form of motifBis specifically recognised by a nuclear factor NF B previouslycharacterised in radish extracts (11). This cluster of motifs isfound nowhere else in the radish rDNA unit.

Probe

Id ol'HS6901

-A 12-3IP iBP rA. '-BP rAX12-t 21 4 h 4 ".: 4 h 2 4 6

pC-_ _

41

d L I~ --

V anel i 4 ` i, ".I-*I1 3. 14 15 1h 1 7

Identification of a specific pre-rRNA-binding activity inradish seedling extracts

To identify radish factors that specifically bind to the pre-rRNAsequences we used the EMSA to reveal stable interactionsbetweenRNA probes containing the A123BP sequences (Fig. IB)and fraction HS600, a partially purified protein fraction fromseedlings (Fig. IC). Fractionation of the extracts was necessaryto reduce probe degradation and proteolysis which occur in crudefractions. To reduce non-specific interactions between proteinsand RNA probes the binding assays were done in the presence ofa large excess of yeast tRNA. Using these conditions incubationof probe rAI23BP with increasing amount of HS 600 fractionresults in the proportional formation of a major complex C ofreduced electrophoretic mobility (Fig. 2A, lanes 2-5). Thiscomplex reveals a specific protein/rA123BP interaction as noretarded complex is detected with probe asrAI23BP, correspon-ding to the anti-sense pre-rRNA sequence (lanes 10-13) orwithout protein (lane 1). Pre-incubation of the HS600 fraction at50°C or with proteinase K eliminates the rAI23BP bindingactivity (not shown). We called this specific rAI23BP-bindingactivity present in the radish HS600 fraction Nuclear Factor C(NF C).We tested optimal conditions for NF C activity. The binding of

NF C to rAl23BP requires Mg2+ with an optimum of 5-10 mMand is resistant to 400mM KCI (results not shown). Treatment ofthe HS600 extracts with microccocal nuclease had no effect on Ccomplex formation (result not shown), which suggests that thebinding ofNF C to the probe does not require an additional RNAcomponent (5,7).

Finally we must say that in most preparations a second complex'pc' appears migrating slightly faster then complex C, and theratio of C to pc varies between preparations (see Figs 2-4). Thisis due to the extreme sensitivity of NF C to the high proteolyticactivity of plant extracts: all C complex is transformed in pccomplex by incubating crude protein extracts at room tempera-ture or by omitting protease inhibitors from extraction buffers(results not shown). This suggests that pc is a proteolysed formof C complexThe recognition of probe rAI23BP by NF C is specific. To

determine whether this RNA-binding activity recognises thewhole AI23BP sequence we did the EMSA with separated rA123and rBP probes (Fig. IB). A retarded complex C is detected only

P'robe rKP Ss (IJil(itIIS64mI 2 2 4 6

1)(

I Ss>m

anc

Figure 2. (A) Specific binding of a radish protein fraction to probes rAI23BPand rBP detected by EMSA. Binding reactions were perfonned with theindicated amount ofHS600 fraction in the presence of 2 1ig oftRNA and 2 fmolofprobe (10 00-20 000 c.p.m. depending on theC content ofeach probe). Thestructure of the probes is shown in Figure 1 except for the asrA123BP. This isa 121 basesRNA corresponding to the anti-sense A123BP sequence synthesizedin vitro (see Materials and Methods). F, free probe; C, RNA/protein specificcomplex; pc, proteolysed form ofcomplex C; d, degraded probe. (B) NFC doesnot bind to a double-stranded RNA/DNA substrate. The EMSA was carried outwith the indicated amount of HS600 fraction and 10 000 c.p.m. ofsingle-stranded rBP or double-stranded rBP/DNA probe (see Materials andMethods). Fss and Fds show the migration ofthe corresponding single-strandedand double-stranded free probes.

with probe rBP (Fig. 2A, lanes 6-9). No complex is detected withprobe rAI23 (Fig. 2A, lanes 14-17). Complex C detected withrAI23BP and rBP responds in a similar manner to specificoligonucleotides competitors confirming that both probes arerecognised by the same protein fraction (not shown). The strongersignal of retarded rA123BP complex (lanes 2-5) compared withthe signal detected with the rBP probe (lanes 6-9) does not simplyreflect the 2-fold higher specific activity of the longer probe. Thisindicates that NF C has a greater affinity for the rAI23BP than forthe rBP probe and suggests that additional bases in the longerprobe increase the RNA-binding affinity of NF C.A very important result is the lack of binding of HS600 protein

fraction to probe rA123 (Fig. 2A, lanes 14-17), which containsthree UUUUCCGG motifs (Fig. IB). This clearly shows that thebinding ofNF C to rBP and rA123BP implicates a sequence-spe-cific interaction and is not simply due to recognition of theU-stretches present in these probes. In addition this indicates that

4

F im-

Nucleic Acids Research, 1995, Vol. 23, No. 24 4967

competitor

3 B P_- UUCCGUGACGAUUUUGCCCCUGAUAUGAAACwCCGCuuurBP

II uIuay-rB

2

3 r

4

55 _

6 a=L

I I

M.

1, me- #-- oli r49

-- oli r38

r--- oli r38 i

pmoles0 1 2 4

los&-A....

mB lotit .

oli r35 wtn"oli r28 N _s

poly - U A C G

ng _0o a

F I

Lane 1 2 3 4 5 6 7 8 9 10 11 12

Figure 4. NF C binds to poly U. EMSA was carried out with 2 jl of HS600fraction and 10000 c.p.m. ofprobe rBP in the presence of the indicated amountof ribonucleotide homopolymers. (-) no protein in the binding assay.

Figure 3. Competing NF C binding to rBP with specific oligonucleotides.Thestructure of the oligonucleotide competitors is shown. rBP corresponds to theunlabelled rBP probe sequence synthesized in vitro. Heavy lines and thin linesindicate pre-rRNA and vector sequences respectively (as in Fig. IB). Oli r38,oli r38mB, oli r35 and oli r28 are RNA oligonucleotides.Vertical broken linesindicate the 5'- and 3'-end of each competitor. The two U to G mutations on olir38mB are indicated. The RNA-binding assays were performed with 2 p1 ofUS800 fraction and 2 fmol of probe rBP. The indicated amounts of competitorswere added prior to addition of the probe.

RNA-binding activities are also distinguished by their relativeaffinity for ribonucleotide homopolymers (20). The binding ofNF C to rBP was displaced by an excess of poly U (Fig. 4, lanes3-6) but not by poly A (lanes 7 and 8), poly C (lanes 9 and 10)or poly G (lanes 11 and 12). So NF C binds to poly U but has littleaffinity for poly A, poly C or poly G.

Purification of NF C

the motifs recognised by this RNA-binding activity are restrictedto the rBP sequence.Next we tested the relative affinity ofNF C for single-stranded

versus double-stranded specific substrates. A double-strandedrBP/DNA probe was made by annealing the rBP probe to aDNAoligonucleotide corresponding to the anti-sense sequence (seeMaterials and Methods). This anti-sense oligonucleotide is notrecognised by NF C (not shown). NF C, which binds tosingle-stranded probe rBP (Fig. 2B, lane 1), does not bind to thedouble-stranded probe rBP/DNA (Fig. 2B, lanes 2-4). Neitherdoes it binds to dsDNA substrate (not shown). Thus NF C bindsonly to a single-stranded RNA or DNA substrate. This result isimportant because it excludes the possibility that NF C binds tothe rDNA.To further delimit the sequence recognised by NF C we

performed competition studies using oligonucleotides derivedfrom the rBP sequence (Fig. 3). First we showed that theunlabelled rBP competitor (synthesized with the T7 RNApolymerase from linearised plasmid pGBP/HindIll) displaces NFC binding to the probe rBP when present in excess in the bindingassay (line 1). Oli r49, which lacks 27 bases from the 5'- and3'-ends of probe rBP, displaced NF C binding to the rBP probe aswell as the unlabelled rBP competitor (compare lines 1 and 2). Olir49 can be further deleted at its 3' end to keep just site P: theresulting oli r38 still efficiently competes for the binding of NFC to rBP (line 3). A further reduction of oh r38 sequence hasdrastic effects on NF C binding: oli r35, lacking the CCGG basesfrom motif A3 (line 5) or oli r28 lacking the P site (line 6) havelittle capacity to compete for NF C binding. Finally mutating theB site also affects the binding ofNF C: oli r38 mB, in which twoUs of the B motif were replaced by two Gs has a lower capacityto displace the C complex (line 4) when compared to oli r38.These data show that the minimal sequence efficiently recognisedby NF C is oli r38 which includes the 3' end of motif A3 and theB and P motifs.

NF C binds to poly U (Fig. 4). Therefore HS600 fractionscontaining NF C activity were further fractionated by poly (U)Sepharose chromatography (Fig. IC). All NF C activity wasretained on the column equilibrated with a 100 mM KCl bufferand subsequently washed with a 300 mM KCl buffer. NF C wasstep eluted with a 800 mM KCl buffer. Peak fractions with NF Cactivity detected by EMSA were pooled, corresponding to theUS800 fraction. NF C was here separated from NF B, a sequencespecific rDNA-binding activity previously described (11), whicheluted in the 300 mM KCl wash (not shown).To identify NF C the pooled US800 proteins were first

separated by SDS-PAGE and visualised by silver staining. TheUS800 NF C fraction is highly enriched in two polypeptides of67 and 60 kDa (Fig. 5A, lane 1). To confirm a correlation of theseproteins withNFC activity we usedUV cross-linking photolabel-ling to visualise the polypeptides directly interacting with proberBP (14). Two polypeptides were specifically photolabelledwhich comigrate with the 67 and 60 kDa species detected bysilver staining (Fig. 5A, lanes 3 and 7). Preincubation of theUS800 fraction at 50°C (lane 5), with proteinase K (lane 6) oraddition of an excess of specific competitor oli r49 to the bindingassay (lane 4) reduces labelling of these bands. No polypeptidewas labelled in the absence ofUV irradiation (not shown), whenUV cross-linking was done without the protein fraction (lane 2),with a DE100 protein extract (Fig. IC) which is devoid of NF Cactivity (lane 8) or upon incubation of NF C activity with ananti-sense asrBP probe (lane 9). These results confirm that thelabelling of 67 and 60 kDa species by UV cross-linking can beattributed to NF C.The 67/60 kDa ratio ofUV photolabelling perfectly correlates

with the C/pc complex ratio detected by the EMSA in thispreparation as well as in all different preparations ofNF C tested(not shown). Thus the 67 and 60 kDa proteins are respectivelyrelated to C and pc complex formation. Considering that pc is aproteolysed form of the C complex this suggests that the 60 kDa

I

4968 Nucleic Acids Research, 1995, Vol. 23, No. 24

I AN

+ + + + + 1)1: +cp 5;Do PK as

.-protein h h

uty .nd 6 6 6

Ni-E. sI 0

.i 6 3

protein sIoo N.F.-6 6

94-

433-

( _-A

vc o- ONr-r*s~~~~~~~~~~~~~~~~~~~~~~ g :i -fi.P.,, u

30 -

20-

B

0.

0 .1ccI

0.01-;1}04- L

0 s10

E.iS;4.^6

WU

V _it

J1all

_. _.........

k Dai

Figur 5. (A) Identification of NF C by SDS-PAGE and UV cross-linkinganalysis. Proteins of the US800 fractions were separated by SDS-PAGE andvisualised eiherby silver saining ofthe gel or byUV cross-linlkng photolabelling,as indicated. For silver staining visuatisaton 150 p1 of US800 pool were

concentate by acetone precipitabon prior to loadng on the gel. For UVcross-lning analysis the binding assays were perfonned with probe rBP and 15p1 ofUS800 fraction indicated by (+). Control reactions were done without protein(-)orwith 5 p1ofDElOOfiaction (DE, see Fig. 1) which isdevoidofNFC activity.Odter controls were binding reactions performed with an excess of oh r49 (cp),using a US800 faction pre-eated 5 min at 50°C (500), or treated with ProteinaseK (PK). An additional control (as) is a binding assay perfonned with 15 p1 ofUS800 fraction and the anti-sense rBP sequence probe. The insert shows the C/pcratio deteced with EMSA made with 1 and 2 of US800 pool fraction. (B)Detenination of the size of native NF C activity by Sephadex-G-100chromatography. The column was loaded with US800 fraction and the elution ofNFC activity was deeumined by the EMSA witi the rBP probe (see Materials andMethods). The afow indiates the Kav of the peak of NF C activity (C and pccomplexes). Filled cirles indicates the Kav ofproteis used forcolumn calibration:(1) ovalbumin (43 000 kDa), (2) bovine serum albumin (67 000 kDa), (3) humantrasfemn (81 000 kDa) and (4) rabbit aldolase (158 000 kDa). Kav = Ve - VJVt- VO, where Ve is the elution volume of the protein, Vt and VO are the total and theexclusion volume of the column respectively (31).

species derives from limited proteolysis of the 67 kDa species,without affecting its RNA-binding specificity.To confn-m the previous data and determine whether the 67 kDa

species binds to rBP probe as a monomer or a dimer we evaluatedthe size of native NF C by gel filtration chromatography. When

I

1-h1c. ___-

NIVR -*P -w

lint 24-W-

Figure 6. (A) EMSA analysis for NF C activity in the nuclear fraction. NF Cactivity was assayed with the indicated amount of protein fraction and the rBPprobe. (nd) stands for non-detectable protein by Bradford assay. Oli r38 (lane3) or oli r35 (lane 4) competitors (100 ng each, shown in Fig. 3) were added tothe extracts prior to addition of the probe. (B) Western blot analysis to evaluatethe relative amount of the rbcL, NTR and BnC24 proteins in the S100 and thenuclear fractions. The arrows indicate the bands corresponding to each of theseproteins. N.E. stands for nuclear extract. The original SO00 fraction is shownin Figure IC.

the US800 fraction was passed on a Sephadex-G-100 column theNF C activity generating both C and pc complexes eluted as asingle peak corresponding to a molecular mass of -80 kDa (Fig.SB). This is higher then the 67 kDa estimated by SDS-PAGE forNF C but clearly differs from the -130 kDa expected for a dimer.This suggests that the 67 kDa species binds to the rBP probe asa monomer forming the complex C detected by the EMSA.

NF C is a nuclear activity

Pre-rRNA processing occurs in the nucleolus (1,2). Therefore wesearched for NF C in nuclear extracts (13) prepared from radishseedlings. TheEMSA shows a strong NFC activity in the nuclearfraction which binds to the rBP probe forming a complex (Fig.6A, lane 2) that comigrates with complex C detected with pureNF C fraction (lane 1). The C complex detected with nuclearextracts is displaced by an excess of oli r38 (lane 3) but not by olir35 (lane 4) as shown forNFC (Fig. 3). The strong smear detectedin theseEMSA, as well as the low amount ofpc detected (Fig. 6A,lanes 2-8) is due to a very rapid degradation of complex Coccurring in crude fractions (not shown). We confirmed thepresence of the 67 kDa protein corresponding to NF C in thenuclear extracts by UV cross-linking analysis (result not shown).To confirm an enrichment of NF C in the nuclear fraction we

compared the amount ofNF C activity detected by EMSA to theamount of proteins ofknown subcellular localisation detected byWestern blot in the nuclear and S100 total cellular fractions. NFC activity is comparable in the nuclear (Fig. 6A, lanes 5 and 6) andS100 fraction (lanes 7 and 8). A similar result is obtained forBnC24, a nuclear protein from crucifers (17,21) that is alsodetected at the same level in both fractions (Fig. 6B, lanes 1 and2). The relative distribution ofNF C and BnC24 clearly contrastwith the distribution of NADPH-thioredoxin reductase (NTR), a

cytosohic protein (18) which is only found in the S100 extract(Fig. 6B, lane 1) and is undetected in the nuclear fraction (Fig. 6B,

1+1kDa

I.11,

p::1 WW-0 ,wNNW

Lane 1 2 3 4 5 6 8 9

Nucleic Acids Research, 1995, Vol. 23, No. 24 4969

lane 2). The distribution of NF C also clearly differs from thepartition of the large subunit of rubisco (rbcL), a chloroplasticprotein (19). Most of rbcL is in the S100 fraction (Fig. 6B, lane1), with only a very minor percentage detected in the nuclearextracts (Fig. 6B, lane 2). The minor rbcL signal contaminatingthe nuclear extract can be explained because rbcL is the mostabundant protein in plant cells (19). We confirmed by EMSA theabsence of NF C activity in protein extracts prepared frompurified radish chloroplasts (not shown).These results indicate that NF C activity is enriched in the

nuclear fraction and suggests that it is a nuclear protein.

DISCUSSION

We have identified and purified a nuclear protein, NF C, thatbinds to the pre-rRNA sequence encompassing the primaryprocessing site in radish. The most striking characteristic of thisRNA-binding activity is its strict substrate specificity: NFC bindsonly to probes rA123BP or rBP (Fig. 2A). It does not bind to anyother sequence such as tRNAs, which is in excess in the bindingassays or anti-sense pre-rRNA sequences (Figs 2A and 5). Mostimportant the lack of binding to rA123 (Fig. 2A) shows thatNF C can discriminate between the very similar rBP and rAI23sequences. This demonstrates that the interaction of NF C withrAI23BP or rBP is sequence-specific and is not simply due torecognition of the U stretches in these probes. Based on thecompetition studies we delimited the minimal RNA sequencerecognised by NF C to 38 bases, extending from the 3' end CCGGof motif A3 to the P site, including the B motif (Fig. 3). Astructural analysis using the Mulfold program suggests that this38 bases RNA has a high potential to form a stable stem-loopstructure which is conserved in all probes recognised by NF C(not shown). Elimination of the CCGG 3' end from the A3 motifor mutation of two Us of B motif into Gs (oli r35 and oli r38mBrespectively) would drastically alter the secondary structure ofthese oligonucleotides compared with oli r38 (not shown). Whenwe assayed these competitors we showed that NF C has a loweraffinity for oli r35 and oli r28mB in comparison to oli r38 (Fig.3). This could indicate that the secondary structure of the RNAmay also be an important recognition element for the binding ofNF C (20).Another important result is that NF C binds only to single-

stranded probes, and not to double-stranded substrate (Fig. 2B).So NF C cannot bind to the rDNA, thus differing from NF B, aradish factor which binds to double-stranded rDNA but not to theRNA sequence (11). Indeed NF B and NF C activities wereseparated on the poly (U) Sepharose chromatography (notshown).Another major point concerns the protein structure of NF C.

Based on UV cross-linking analysis and gel filtration chromatog-raphy NF C corresponds to a 67 kDa protein that binds as amonomer to the probe to form complex C which is detected byEMSA (Fig. 5). The 60 kDa protein would derive from aproteolysis of the 67 kDa form occurring in the first steps ofprotein extraction (not shown) and is correlated to pc complexformation (Fig. 5). Indeed, because of the extreme lability ofNFC we cannot exclude the possibility that the 67 kDa species isalready a proteolysed form of a major precursor. This situation,where a limited proteolysis of a precursor originally undetectedby the EMSA, generates a heterogenous population of sequence-

Based on its RNA substrate specificity and its molecular massNF C would differ from nucleolar proteins with RNA-bindingactivity identified in plants. In onion cells the plant homologue offibrillarin, a -38 kDa protein involved in pre-rRNA processing inmammals and yeast (2,23) has been identified as a 37 kDa proteinthat cross-reacts with antibodies directed against human fibrilla-rin (24). In maize a cDNA coding for a 16 kDa proteincharacterised by an RNA-binding domain and a GAR nucleolarmotif has been cloned (25). This protein binds to poly G and polyU in vitro (25) and has been immunolocalised in the nucleolus(26). The substrate and role of this protein are unknown.

In plants specific RNA-binding has been shown only forchloroplast proteins involved in the 3'-end processing of thechloroplast RNAs (27, and references therein). Specific bindingto nuclear-encoded RNAs has been reported only for a 50 kDaprotein from bean which binds to the 3'-end of an mRNA codingfor a proline-rich polypeptide (28). Poly(A) binding proteins havealso been identified in plants but no specific binding has beenshown for these activities (29,30). To our knowledge NF C is thefirst example in plants of a nuclear protein that binds specificallyto the 5'-end of a nuclear-encoded RNA.A main question is the function of NF C. It is tempting to

propose that NF C is involved in pre-rRNA primary processingin radish. The main argument is that NF C strictly binds to theradish pre-rRNA sequence including the primary processing sitemapped in vivo (10). It does not bind to the non-coding rDNAstrand or to the rDNA IGS (Fig. 2). In agreement with thisproposal we found that NF C activity is enriched in nuclearextracts and could be a nuclear protein (Fig. 6). Additionalsupport to this hypothesis is that homologous BP clusters almostidentical to those in radish are found in other crucifer IGSs,always located in the 5'-ETS (not shown). Though primarypre-rRNA processing sites have not been reported for thesespecies, this suggests a conservation of putative NF C bindingsites associated with pre-rRNA processing sites.

In vertebrates the pre-rRNA primary cleavage correlates withthe formation of a large RNP complex at the pre-rRNA primaryprocessing site in vivo (3) and in vitro (5-7). These RNPcomplexes, detected by EMSA, include at least six proteins, someof which are homologous between mouse and X.laevis (6). Theindividual roles of these polypeptides are not known though onecould correspond to a specific pre-rRNA endonuclease (7). Inplants terminal balls identified as pre-rRNA processing com-plexes (3) are also detected at the 5'-end of the nascent pre-rRNA(4). NF C could be one component of this plant pre-rRNAprocessing complex. Various roles should then be considered.NF C could be directly involved in the pre-rRNA primarycleavage, as shown for a purified murine endoribonuclease whichforms a stable complex with the mouse pre-rRNA and cleaves itat the primary processing site (7). Alternatively the binding ofNFC to the pre-rRNA could serve to direct the factors involved in thecleavage to the processing site. Another possibility is that NF Ccould modulate the secondary structure of the pre-rRNA toactivate (or inhibit) pre-rRNA processing.

In conclusion we present here the first evidence obtained inplants for a nuclear protein, NF C, that binds specifically to apre-rRNA sequence which includes the primary processing site.These results, in agreement with our previous data (11), suggestthat the A123BP cluster could be an important element involvedin pre-rRNA processing. We are now currently studying the effectof NF C binding on the pre-rRNA at the primary and secondaryspecific DNA binding factors has already been described (22).

4970 Nucleic Acids Research, 1995, Vol. 23, No. 24

structural level in vitro. In addition as NF C has been purified weintend to clone the corresponding cDNA in order to elucidate theidentity of NF C.

ACKNOWLEDGEMENTS

We are much indebted to Dr M. Delseny for his constant supportand critical reading of the manuscript. We are also grateful to DrJ. J. Toulme for discussing our results. We are also grateful to DrJ. Fleck for providing the tobacco antiserum against rbcL. We alsowish to thank Prof. P. Penon for advice on the elaboration of themanuscript and Dr G. Hull for correcting the English. Also we are

grateful to G. Jung and C. Salvat for their contribution to thiswork.

REFERENCES

1 Sollner-Webb, B. and Mougey, E.B. (1991) Trends Biochem. Sci. 16,58-62.

2 M61ese, T. and Xue, Z. (1995) Curr: Opin. Cell Biol. 7, 319-324.3 Mougey, E.B., O'Reilly, M., Osheim, Y, Miller, O.L., Beyer, A. and

Sollner-Webb, B. (1993) Genes Dev. 7, 1609-1619.4 Greimer, R. and Deltour, R. (1984) Bio. Cell. 50, 237-246.5 Kass, S. and Sollner-Webb, B. (1990) MoL Cell. Biol. 10, 4920-4931.6 Mougey, E.B., Pape, L. and Sollner-Webb, B. (1993) Mol. Cell. Biol. 13,

5990-5998.7 Eichler, D.C., Liberatore, J.A. and Shumard, C.M. (1993) Nucleic Acids

Res. 21, 5775-5781.8 Craig, N., Kass, S. and Sollner-Webb, B. (1991) MoL Cell. Biol. 11,

458-467.9 Hemleben, V. and Zentgraf, U. (1994) In L. Nover (ed.) Plant Promoters

and Transcription Factors. Springer-Verlag, Berlin Heidelberg, pp 3-24.

10 Delcasso-Tremousaygue, D., Grellet, F., Panabikres, F., Ananiev, E. andDelseny, M. (1988) Eur. J. Biochem. 172, 767-776.

11 Echeverria, M., Penon, P. and Delseny, M. (1994) Mol. Gen. Genet. 243,442-452.

12 Bradford, M.M. (1976) Anal. Biochem. 72, 248-254.13 Green, P., Kay, S.A., Lam, E. and Chua, N.H. (1989) In vitro DNA

footprintng. Plant Mol. Biol. Manual BJJ, 1-22.14 McEwan, I.J. (1991) In J.P. Jost and H.P. Saluz (eds) A Laboratory Guide

to In Vitro Studies ofProtein-DNA Interactions, Birkhauser Verlag, Basel,pp 71-78.

15 Laemli, U. (1970) Nature 227, 680-685.16 Domon, J.M., Dumas, B., Lain6, E., Meyer, Y., David, A. and David, H.

(1995) Plant Physiol. 108, 141-148.17 Saez-Vasquez, J., Raynal, M., Mezza-Basso, L. and Delseny, M. (1993)

Plant Mol. Biol. 23, 1211-1221.18 Jacquot, J.P., Rivera-Madrid, R., Marinho, P., Kollarova, M., Le Mar6chal,

P., Miginiac-Maslow, M. and Meyer, Y. (1994) J. Mol. Biol. 235,1357-1363.

19 McIntosh, L., Poulsen, C. and Bogorad, L. (1980) Nature 288, 556-560.20 Burd, C.G. and Dreyfuss G. (1994) Science 265, 615-621.21 Saez-Vasquez, J. (1995) Ph D thesis, Universite de Perpignan.22 Smid, A., Finsterer, M. and Grummt, I. (1992) J. Mol. Biol. 227, 635-647.23 Tollervey, D., Lehtonen, H., Jansen, R., Kern, H. and Hurt, E.C. (1993)

Cell 72, 443-457.24 Cerdido, A. and Medina, F.J. (1995) Chromosoma, 103, 625-634.25 Ludevid, M.D., Freire, M.A., Gomez, J., Burd, Ch.G., Albericio, F., Giralt,

E., Dreyfuss, G. and Pages, M. (1992) Plant J. 2, 999-1003.26 Alba, M.M., Culiafiez-Macia, A., Goday, A., Freire, M.A., Nadal, B. and

Pages, M. (1994) Plant J. 6, 825-834.27 Nickelsen, J. and Link, G. (1993) Plant J. 3, 537-544.28 Zhang, S. and Mehdy, M.C. (1994) Plant Cel 6, 135-145.29 Yang, J. and Hunt, A.G. (1992) Plant Physiol. 98, 1115-1120.30 Belostotsky, D.A. and Meagher, R.B. (1993) Proc. Natl Acad. Sci. USA 90,

6676-6690.31 Stellwagen, E. (1990) In Deutscher M.P. (ed.) Guide to Protein

Purification, Methods in Enzymology, Academic Press, San Diego, 182,317-328.