4376–4384 Nucleic Acids Research, 1999, Vol. 27, No. 22 © 1999 Oxford University Press

irst,orre.re

meis.itin

hed toetioneseebe

led by

8)inswo,onisofsh-ct

eese

ontÅ

allly

Identification of 23S rRNA nucleotides neighboring theP-loop in the Escherichia coli 50S subunitYuri Bukhtiyarov, Zhanna Druzina and Barry S. Cooperman*

Department of Chemistry, University of Pennsylvania, Philadelphia, PA 19104-6323, USA

Received August 11, 1999; Revised and Accepted September 30, 1999

ABSTRACT

We report the synthesis of a radioactive, photolabile2′′′′-O-methyloligoRNA probe, 2258-53/52(SAz)-48,PHONT 1, and its exploitation in identifying 23S rRNAnucleotides neighboring the so-called ‘P-loop’. Theprobe is complementary to nt 2248–2258 in Escherichiacoli 50S subunits. PHONT 1 contains a p-azidophenacylgroup attached to a phosphorothioate bridge betweenthe nucleotides complementary to the positions2252–2253, such that the photogenerated nitrene ismaximally 17–19 Å from 23S RNA nucleotides G2252and G2253. PHONT 1 binds to the 50S subunit, andphotoincorporates within or immediately adjacent toits target site, as well as into several nucleotidesfalling between G2357 and A2430. The significance ofthese results for the structure of the peptidyl trans-ferase center is considered. The PHONT approach isgenerally applicable to studies of complex RNA-containing molecules.

INTRODUCTION

Elucidation of the mechanism of ribosome-catalyzed proteinsynthesis is a challenging task for contemporary biochemistry.Continuing efforts are focused on determining the structure ofribosomes (1–5) and on identifying key components involvedin substrate recognition and catalysis of the peptidyl trans-ferase reaction (6–14). Advances in cryoelectron microscopyhave resulted in the publication of 15 Å resolution structures of70S ribosomes, and structures at 10 Å resolution now appearfeasible (1,2). Recent X-ray diffraction results have been evenmore dramatic, resulting in structures of the 30S and 50S subunitsat 5.5 and 5.0 Å, respectively (3,4). However, even at theseresolutions, the precise placement of ribosomal componentswithin the ribosome structure can be a daunting task. Chemicalcrosslinking can provide information that is helpful in thisregard.

We have been using radioactive photolabile oligonucleotideprobes (PHONTs) targeted toward functionally importantrRNA sites to identify proteins and individual nucleotides site-specifically photolabeled from these sites as a way of generatingcrosslinking information useful for the development ofdetailed structural models (15–21). The PHONT approach

offers several important advantages for such an exercise. Fit allows the targeting of sequences of particular functionalstructural significance throughout the ribosome structuSecond, it is readily applicable to the examination of the natuof the substantial conformational changes that the ribosoundergoes during the overall cycle of protein biosynthesThird, the crosslinks formed provide defined upper limdistances for the separation of the linked components withthe ribosome, given by the length of the tether arm. As tlength of this arm can be varied, the approach can be useidentify components both immediately neighboring the targsite or somewhat further away. Fourth, sample preparatemploying readily synthesized PHONTs and intact ribosomis straightforward; an important attribute given the largnumber of constraints required for model building. Morgeneral considerations of the PHONT approach, which canapplied to study any complex RNA-containing molecu(e.g. spliceosomes, RNase P and ribozymes), are presenteCoopermanet al. (22,23).

Here we targetEscherichia coli23S rRNA nt 2248–2258.These nucleotides fall within the so-called ‘P-loop’ (nt 2246–225that is a vital part of the peptidyl transferase center. It contathree invariant guanosine residues, G2251–G2253, of which tG2252 and G2253, are protected from kethoxal modificatiby P-site bound tRNA (24). In addition, mutational analyshas shown that G2252 forms a Watson–Crick pair with C74the P-site bound tRNA (25,26); mutations at G2251 aboliboth binding of wild-type tRNA fragments and peptidyl transferase activity (27); and there is evidence for direct or indireinteraction of the 2250-loop with U2585, which falls within thcentral loop of domain V that defines another part of thpeptidyl transferase center (28,29). Specifically, we uPHONT1 (Fig. 1), a 2�-O-methyloligoRNA complementary tont 2248–2258 and containing ap-azidophenacyl group attached ta phosphorothioate bridging nucleotide complementary to2252–2253, to identify ribosomal components within 17–19of these nucleotides.

MATERIALS AND METHODS

Except as specified below, all materials were obtained andmethods were carried out as described previous(15,18,20,22). 50S subunits were activated at 37�C for 20 minin the presence of 10 mM Mg2+ immediately prior to thebinding and photocrosslinking experiments.

*To whom correspondence should be addressed. Tel: +1 215 898 6330; Fax: +1 215 898 2037; Email: cooprman@pobox.upenn.eduPresent address:Yuri Bukhtiyarov, DuPont Pharmaceuticals Co., Experimental Station E400/3410, Wilmington, DE 19880, USA

Nucleic Acids Research, 1999, Vol. 27, No. 224377

fied

ur0S

ate

%

ed

n

lvedol,ht

seive

-

H

as

M%

taryto

ngt/

,

ed

eeryas

nd

,

Materials

All oligonucleotides were synthesized using phosphoramiditechemistry on a Milligen Biosearch Cyclone automated DNAsynthesizer.

Synthesis and purification of PHONT1, 2�-O-methyloligoRNA2258-53/2(SAz)-48 (Fig. 1)2�-O-methyloligoRNA 2258-53/2(S)-48 (0.1 �mol), gaccgC(S)cccag, which contains a phospho-rothioate group bridging the sixth and the seventh nucleotides,was reacted with 2�mol of 4-azidophenacylbromide (Sigma)in 50 �l of 100 mM sodium bicarbonate containing 50%dimethylsulfoxide (DMSO). Reaction was allowed to proceed atroom temperature for 1.5 h, after which the mixture was dilutedwith water to 160�l and extracted twice with 300�l of isobutanol.The product, PHONT1, was purified by reverse phase HPLC on anoctadecylsilica column (SynChropak RP-P, MICRA Scientific,Inc.) using a linear gradient from 5 to 40% acetonitrile in 100 mMtriethylammonium acetate, pH 7.4. Starting material and PHONT1eluted as two pairs of diastereo-mers, with the product pair elutinglater. Both isomers of the product were collected and used in equalamounts for the photoaffinity labeling experiments. The corre-sponding phenacyl derivative, 2�-O-methyloligoRNA 2258-52/3(S�)-48, was prepared in a similar manner by using bromoaceto-phenone (Sigma) as alkylating reagent.

Methods

Non-covalent binding of oligonucleotides to 50S ribosomalsubunit.50S subunits (15 pmol) were combined with varyingamounts of 32P-2�-O-methyloligoRNA in the presence of40 mM Tris–HCl (pH 7.6), 60 mM KCl and 0.45 mM MgCl2 ina total volume of 25�l. The mixture was incubated at 37�C for5 min and on ice for 15 min. The MgCl2 concentration wasthen raised to 10 mM, and the mixture was incubated at 37�Cfor 5 min and on ice for 2 h. The samples were diluted with0.5 ml of buffer and applied onto Millipore 0.45�m nitrocellu-lose membrane filters type HAWP that had been pre-washedwith 5 ml of the binding buffer containing 10 mM MgCl2. Thefilters were washed three times with 1 ml of the buffer, air-driedand counted for radioactivity in EcoLite scintillant. Bindingmeasurements were performed in triplicate.

Photoaffinity labeling of 50S ribosomal subunit and isolationof labeled 23S rRNA.PHONT 1 (300 pmol) was radiolabeled

at the 5�-end with 250�Ci of [�-32P]ATP using T4 polynucleo-tide kinase (Promega) (30). The radiolabeled probe was puriby solid-phase extraction on a Sep-Pak C18 cartridge (30) andrecovered in ~50% yield. The product was divided into foequal amounts, and each was incubated with 150 pmol of 5subunit in 75�l of T50K50M0.3 buffer (50 mM Tris–HCl,pH 7.6, 50 mM KCl, 0.3 mM MgCl2) at 37�C for 5 min and lefton ice for 15 min. The Mg2+ concentration was increased to10 mM and the samples were incubated at 37�C for 5 min andkept on ice for 30 min. The solutions were irradiated in 250�lPyrex glass inserts for autosampler vials (Kimble/Kontes)4�C for 4 min using 3000 Å mercury lamps (Rayonet). Thribonucleoprotein complex was precipitated with 2 vol of 102-mercaptoethanol in ethanol at 4�C, and the precipitate waswashed with cold 70% ethanol. The dried pellet was dissolvin 24�l of T50K50M10 buffer (50 mM Tris–HCl, pH 7.6, 50 mMKCl, 10 mM MgCl2), and 23S rRNA was precipitated by additioof 67�l of 100 mM MgCl2 in 90% acetic acid (0�C, 45 min).The procedure was repeated, and the dried pellet was dissoin 100 mM sodium acetate, pH 5.2, extracted with phenphenol–chloroform and chloroform, and precipitated overnigat –20�C with 2 vol of ethanol. Samples prepared for revertranscriptase assays were photolyzed with non-radioactPHONT1 at higher probe/subunit ratios (10:1 or 20:1).

RNase H cleavage of [32P]PHONT 1-labeled 23S rRNA.Fivepicomoles of the RNA was hybridized with 10 pmol of oligodeoxyribonucleotides or DNA/2�-O-methylRNA hybrid oligo-nucleotides by heating at 55�C for 5 min and incubating at32�C for 15 min in 8�l of T25N125D1.25 buffer (25 mM Tris–HCl,pH 7.6, 125 mM NaCl, 1.25 mM DTT). Mg2+ was adjusted to10 mM and the RNA was digested with 2 U of RNase(Promega) for 30 min at 32�C. To these samples 8�l of 95%formamide containing 50 pmol/�l of 2�-O-methyloligoRNA2258-48, 0.1% of bromophenol blue and xylene cyanol wadded, the samples were heated at 65�C for 10 min and separatedon an acrylamide–7 M–urea 90 mM Tris–borate, pH 8.3, 2 mEDTA gel (the acrylamide varied between 5 and 10depending on the desired size range) preheated to 45�C byapplying a 20–25 mA current. The added excess complemenunlabeled parent oligonucleotide reduced background duenon-covalent PHONT1 in the urea–PAGE analysis.

Sequence scanning. Sequence scanning was carried out usiAMPLIFY (authored by Bill Engels, www.wisc.edu/genestesCATG/amplify ).

Reverse transcriptase primer extension. The procedureemployed followed Muralikrishna and Wickstrom (31). Typically5�-32P-labeled primer oligoDNA (0.5–1 pmol, 1.5� 106 c.p.m.,3 �l), prepared using T4 polynucleotide kinase as describabove, was annealed to rRNA (0.6 pmol, 1�l) by incubation in50 mM Tris–HCl (pH 8.3), 60 mM NaCl and 10 mM DTT for5 min at 70�C. Samples were frozen in isopropanol/dry icbath (1 min) and then kept on ice for at least 30 min, aftwhich time they were microfuged briefly to bring down anliquid from the sides of the tube. To each annealing mix wadded extension mix (6�l), containing 10 U of avian myelo-blastosis virus (AMV)–reverse transcriptase (Promega) aaffording the following final concentrations in 10�l totalvolume: each dNTP (0.32 mM), 50 mM Tris–HCl (pH 8.3)

Figure 1. PHONT 1. The distances between the photogenerated nitrene andthe 2-amino positions of the complemented nt G2252 and G2253 in 23S rRNAare 17–19 Å.

4378 Nucleic Acids Research, 1999, Vol. 27, No. 22

esan

).

ithuserlts

ine,h

thndSforer

ly

sA

1)ongbyphy-48–

gnce

nts.the

is

s

a

Aere53182,

the

01.s’eo-

dion

50 mM KCl, 10 mM MgCl2, 10 mM DTT and 0.5 mM spermidine.For sequencing samples, one ddNTP was added to a finalconcentration of 0.16 mM. Samples were incubated at 47�C for30 min then quenched by addition of 5�l of 100% formamide/0.1%xylene cyanol FF/0.1% bromophenol blue loading buffer. After10 min at 70�C, samples were loaded on an 6% polyacrylamideurea sequencing gel and electrophoresed at 1400–1500 V for1–1.5 h until bromophenol blue reached the bottom of the gel.The temperature of the gel during the run was maintained at45–55�C.

Chemical footprinting. The procedure followed Moazedet al.(32) with slight modifications. Samples (75�l) containing 50Ssubunits (150 pmol) either in the absence or presence (3–10-foldmolar excess) of complementary oligonucleotides were incubatedwith either dimethyl sulfate (DMS) (1�l) or kethoxal (8�l of37 mg/ml kethoxal in 20% ethanol) at 4�C for 2.5–3 h withgentle shaking. Reactions were stopped by addition of either75 �l of 1 M Tris–acetate (pH 7.5), 1 M 2-mercaptoethanol,1.5 M NaAc and 0.1 mM EDTA (DMS reaction) or 20�l of0.5 M potassium borate (pH 7.0) (kethoxal reaction).Following incubation on ice for 10 min, subunits were ethanolprecipitated, washed with 70% ethanol and rRNA wasextracted by phenol–chloroform extraction for analysis byreverse transcriptase primer extension.

RESULTS AND DISCUSSION

Non-covalent binding of 2′′′′-O-methyloligoRNAs to the 50Ssubunit

The 11 residue 2�-O-methyloligoRNA 2258-48 binds non-covalently to 50S subunits in a biphasic manner, with a highaffinity site stoichiometry of ~0.4/50S subunits, and a secondclass of sites of lower affinity (Fig. 2). The shorter 2�-O-methyl-oligoRNA 2255-48 shows a lower high affinity site stoichiometry(~0.1/50S) and no evidence of lower affinity site binding. Thehigher binding stoichiometry of the longer probe led us todesign the photolabile 2�-O-methyloligoRNA 2258-52/3(SAz)-48,

PHONT1, in which the phosphorothioate between nucleotidcomplementary to G2252 and G2253 is derivatized withazidophenacyl group. PHONT1 binds to 50S subunits in abiphasic manner similar to its parent 2�-O-methyloligoRNA,but with a reduced high affinity site stoichiometry (~0.2/50S

Site-specific photoincorporation of PHONT 1

Photolysis of non-covalent complexes of 50S subunits wPHONT 1 were generally performed at excess 50S versPHONT1 (for an exception see below), in order to maximizphotoincorporation from the high affinity site. Evidence fosite-specific crosslinking was sought by comparison of resuobtained with PHONT1 alone, and with PHONT1 in the presenceof either excess competing 2�-O-methyloligoRNA 2258-48,GACCGCCCCAG, which should show a large reductionphotoincorporation, or of the mismatched oligonucleotidGACAC-AACCAG (mismatched bases are underlined), whicshould show little effect on photoincorporation. Although boreverse phase high performance liquid chromatography aSDS–PAGE analysis showed photoincorporation into 50proteins, neither analysis provided unambiguous evidencespecific labeling of any protein (data not shown). On the othhand, 23S rRNA, but not 5S rRNA, was site-specificallabeled, as described below.

Partial localization of regions in 23S rRNA labeled byPHONT 1 bound to its target site via azide-dependentphotoincorporation, using RNase H cleavage

Partial localization of the photo-crosslinking sites waachieved by RNase H cleavage of the labeled 23S rRNhybridized with a series of deoxyoligonucleotides (Tablethat, used together, generated fragments ~150–400 nt lover the whole length of 23S rRNA. Fragments excisedRNase H were analyzed by urea–PAGE and autoradiogra(Fig. 3). The highest labeling of 23S rRNA was found in fragments 2015–2305, which includes the targeted sequence 222258, and 2305–2501, the 3�-neighboring sequence. This labelinwas site-specific, as shown by the results obtained in the preseof complementary or mismatched 2�-O-methyloligoRNA. Noother radiolabeled fragments were observed in these experimeAn example of the characteristic negative results seen withregions of the 23S rRNA other than between 2015 and 2501shown in Figure 3, lanes 1675–1882.

It is important to verify that photoincorporation into region2015–2305 and 2305–2501 arises from PHONT1 bound to itstarget site, rather than from a ‘fortuitous’ site, defined assequence within 23S rRNA having substantial (>7 out of 11 nt)sequence overlap with nt 2248–2258. Accordingly, 23S rRNwas scanned for possible ‘fortuitous’ sequences. Five wfound; two with three mismatches versus 2248–2258 (2521–2and 2636–2646) and three with four mismatches (772–71158–1168 and 1512–1522). 2�-O-methyloligoRNAs comple-mentary to these five sequences were then compared with2258-48 2�-O-methyloligoRNA for their effects on PHONT1photoincorporation into regions 2015–2305 and 2305–25The results (Fig. 4) clearly show that none of the five ‘fortuitouoligonucleotides are as effective as the target site oligonucltide in inhibiting PHONT1 photoincorporation into regions2015–2305 and 2305–2501.

To verify that photo-crosslinking to regions 2015–2305 an2305–2501 is azide specific, we compared photoincorporat

Figure 2. Non-covalent binding of32P-2�-O-methyloligoRNAs to 50S subunits.PHONT 1 (circle); 2�-O-methyloligoRNA 2258-48 (diamond); 2�-O-methyl-oligoRNA 2255-49 (square). The nitrocellulose filter binding assay wasperformed as described in Materials and Methods.

Nucleic Acids Research, 1999, Vol. 27, No. 224379

cehene. Inthee a).

toA

tecteetryre

the

ingheont

te

theof

results obtained with32P-labeled PHONT1 with those obtainedwith 32P-labeled samples of (a) 2�-O-methyloligoRNA 2258-48,(b) 2�-O-methyloligoRNA 2258-53/2(S)-48 and (c) 2�-O-methyl-oligoRNA 2258-52/3(S�)-48. The three standard photoincor-poration experiments shown above [(i) photolysis experiment,(ii) plus added excess complementary 2�-O-methyloligoRNA-2258-48 and (iii) plus added mismatched oligonucleotide; seeFig. 3] were carried out for each of these four oligonucleotides.Experiments were also carried out with samples that had beenpre-photolyzed prior to incubation in the presence of 50Ssubunits to test for possible ‘dark reaction’ occurring betweenthe photo-activated oligonucleotides and the targeted region on23S rRNA. From Figure 5B it is clear that of the four oligo-nucleotides only PHONT1 shows significant photoincorpor-ation into region 2305–2501. The lack of photoincorporationof 2�-O-methyloligoRNA 2258-52/3(S�)-48 provides unam-biguous evidence that photoincorporation within nt 2305–2501is azide-dependent.

The situation is less clear-cut in region 2015–2305 (Fig. 5A).While PHONT1 gives by far the largest extent of site-specificphotoincorporation, all three other oligonucleotides also showapparent site-specific photoincorporation, reaching levels ashigh as 25–30% (as determined by autoradiogram scanning) ofthat obtained for PHONT1. It is probable that some of such‘photoincorporation’ arises from one or more non-azide photo-dependent chemical reactions along the entire length of theoligonucleotide bound to its target site. In addition, some of theobserved incorporation probably reflects incomplete removalof non-covalently bound material, since, as is clear from thelanes labeled (–), pre-photolyzed PHONT1 [and pre-photolyzed2�-O-methyloligoRNA 2258-52/3(S�)-48 as well] shows alower incorporation level than 2�-O-methyloligoRNA 2258-48, inaccord with the relative binding stoichiometries seen in Figure 2.

Identification of the major labeled site within the 2015–2305fragment

Evidence that labeling in the 2015–2305 region takes pladominantly between nt 2225 and 2305 is provided by tgeneration of a32P-labeled fragment of ~85 nt in length whePHONT1-labeled 23S rRNA is digested with RNase H in thpresence of the cDNA probes 2230–2221 and 2310–2301addition, the same digestion carried out in the presence ofcDNA probes 2020–2011 and 2230–2221 fails to generat32P-labeled fragment of ~200 nt in length (data not shownIdentification of the major labeled site within the 2230–22212310–2301 fragment was accomplished using 23S rRNisolated from PHONT1-labeled 50S as a template for AMV–reverse transcriptase. In these experiments, the ability to dea PHONT1-dependent stop or pause (below, for simplicity, wuse stop to indicate stop or pause) depends on the stoichiomof incorporated PHONT, unlike the RNase H experiments wheeven quite low levels of incorporation are detectable due tohigh specific radioactivity of the32P-labeled PHONT. Accordingly,we used a 10-fold molar excess of (non-radioactive) PHONT1over 50S subunits. Only one light- and PHONT1-dependent stopwas seen within the 2210–2352 region, at C2260, correspondto photoincorporation into U2259, immediately adjacent to ttarget site (Fig. 6). This result is an important demonstratithat PHONT1 binds to its target site during photolysis, buprovides no structural information for the 50S subunit.

Identification of the major labeled sites within the 2305–2501fragment

Identification of photoincorporation sites within this fragmenis obviously of greater interest for defining the structure of thpeptidyl transferase center. For this purpose we modifiedRNase H procedure to allow more precise localization

Table 1.Oligonucleotides used for RNase H cleavage of 23S rRNA

a2�-O-methylribonucleotides are underlined. All others are deoxynucleotides.

Oligonucleotidesfor partial localization

Sequencea Oligonucleotidesfor more preciselocalization

Sequencea

222–213 TTCTTTTCCT 2333–2324 UGCCATTGCA

455–447 GTTCACTAT 2355–2346 CGCTCGCAGU

677–669 TTTCGGGTC 2360–2351 CGTCACGCUC

811–803 AACCAGCTA 2374–2365 GCACCUGCTC

1051–1042 CTGTCTGGGC 2379–2370 CUUTCGCACC

1274–1265 TATCGTTACT 2380–2371 GCTTUCGCAC

1460–1451 ACAACCGTCG 2400–2391 CCACCGGAUC

1680–1671 ATTTTGCCTA 2413–2404 CUUCCATUCA

1857–1848 CAATTAACCT 2431–2422 AUCCGTTGAG

2020–2011 TTCAATTTCA 2450–2441 TATCCCCGGA

2230–2221 CACTGTCCGC 2460–2451 ATCAGCCUGU

2310–2301 GTCCGACCAG 2500–2491 AGGTGCCAAA

2505–2497 CATCGAGGU 2505–2497 CATCGAGGU

4380 Nucleic Acids Research, 1999, Vol. 27, No. 22

nds 2an

eenens ine oftoled

d

ling

112

nd

s.

oes

labeled nucleotides, and utilized primer extension by reverse

transcriptase as well.In the experiments presented in Figures 3–5 we used

oligoDNAs to create RNase H sites, with the result that theRNA fragments excised were microheterogeneous, sinceRNase H cuts at several places within each oligoDNA–23SrRNA heteroduplex. Such heterogeneity was not a problem forlocalization of labeling within >100 nt or so, but more preciselocalization requires a more precise cutting tool. This wasprovided by hybrid oligonucleotides containing four con-secutive deoxynucleotides with the remainder of the residuesbeing 2�-O-methyloligo-ribonucleotides. Such molecules,hybridized to 23S rRNA, create unique RNase H cutting sites(33), resulting in excised RNA fragments of precise length.Application of this approach to 50S subunits photolabeled with32P-PHONT1 showed that photoincorporation occurs at fourlocations in this region, all falling between nt 2355 and 2450(Fig. 7). The first, occurring within 2330–2376 (lane 4) can bemore precisely localized to within nt 2355–2360 by the

absence of a labeled fragment in lanes 1 (2330–2355) a6 (2360–2376) and the presence of such a fragment in lane(2330–2360) and 5 (2355–2376). Two other labeled sites cbe localized between 2376 and 2400 (lane 7) and betw2400 and 2428 (lane 10). Finally, there is a fourth site betwe2428 and 2450, as shown by the presence of labeled bandlanes 11 (2428–2450) and 12 (2428–2460), and the absenca labeled band in lane 15 (2460–2505). It is importantemphasize (see below) that all four of these sites are labesite-specifically, since added 2�-O-methyloligoRNA-2258-48

Figure 3. RNase H cleavage of [32P]PHONT1-labeled 23S rRNA using oligoDNAs. Photolabeling experiments were carried out with an ~4-fold excess of50S subunits over [32P]PHONT1 either in the absence of added 2�-O-methyl-oligoRNA [lanes (–) and (+)], or in the presence of a 10-fold excess over 50Ssubunits of 2�-O-methyloligoRNA 2248-58 (GACCGCCCCAG) (lanes CH) orof mismatched 2�-O-methyloligoRNA 2248-58 (GACACAACCAG, mismatchedbases are underlined, lanes MM). Asterisks over the bar at top indicate the majorlabeled fragments: 2015–2305 and 2305–2501. The (–) lanes represent ‘darkreaction’ control experiments in which PHONT1 was pre-irradiated for 4 minat 4�C prior to dark incubation with 50S subunits, carried out as described inMaterials and Methods. Lanes C, no nucleotides were hybridized to 23S rRNAduring RNase H digestion. In other lanes, RNase H digestion followed hybridizationwith the following added cDNAs: 2501, 2505–2497; 2305–2501, 2310–2301and 2505–2497; 2015–2305, 2020–2011 and 2310–2301; 1675–1852, 1680–1671and 1857–1848. Numbers in the left margin indicate sizes (in nt) of DNA markers.

Figure 4. [32P]PHONT 1 photoincorporation into fragments 2015–2305 an2305–2501. Comparative effects of added target site and ‘fortuitous’ site 2�-O-methyloligoRNAs. Procedures were as described for Figure 3. Photolabeexperiments were carried out either in the absence of added 2�-O-methyl-oligoRNA (lanes 1), or in the presence of the indicated concentration of 2�-O-methyloligoRNA 2248-58 (lanes 2–5), or of 2�-O-methyloligoRNAs, added at20 �M, that are complementary to the ‘fortuitous’ sites 2531–252(UUCAGCCCCAG, lanes 6), 2646–2636 (GCCCCCCUCAG, lanes 7), 1522–15(UCACGCCUCAG, lanes 8), 1168–1158 (CGCUGCCGCAG, lanes 9) a782–772 (UUCACCCCCAG, lanes 10). (A) Fragment 2015–2305; (B) fragment2305–2501. Arrows in the right margin indicate migrations of DNA size marker

Figure 5. RNase H cleavage of the 23S rRNA photolyzed in the presenceffour [32P]2�-O-methyloligoRNAs complementary to nt 2258–2248. Procedurand lane definitions are as in Figure 3. Lanes 2258-48, 2�-O-methyloligoRNA2248-58; lanes 2258-(S)-48, 2�-O-methyloligoRNA 2258-53/2(S)-48; middlelanes, DNA and RNA ladders; lanes 2258-(S�)-48, 2�-O-methyloligoRNA2258-53/2(S�)-48; and lanes PHONT1, 2�-O-methyloligoRNA 2258-53/2(SAz)-48. (A) Fragment 2015–2305; (B) fragment 2305–2501. Arrows in theright margin indicate migrations of DNA size markers.

Nucleic Acids Research, 1999, Vol. 27, No. 224381

,w

es

ary-.iso-is

nt

1–hat

ntsas

for

ide–ual

completely suppresses PHONT1 photoincorporation into the2305–2501 fragment (Fig. 3).

Primer extension (Fig. 8, lane 4) of PHONT1-labeled 23SrRNA shows stops at seven positions (1, A2358; 2, C2359; 3,G2405; 4, A2406; 5, A2407; 6, A2412; and 7, U2431) that arenot seen in 23S rRNA isolated from (i) control 50S subunits(lane 1), (ii) 50S subunits incubated with PHONT1 in the absenceof irradiation (lane 2), (iii) 50S subunits irradiated in the absence ofPHONT1 (lane 3) or (iv) 50S subunits irradiated in the presence ofthe non-photolabile 2�-O-methyloligoRNA 2258-2252/3(S)-2248(lane 5). In the data shown (Fig. 8), cDNA complementary tont 2457–2442 was used as primer, but similar results wereobtained using cDNA complementary to nucleotides 2493–2477 and 2443–2427. Stops 1 and 2, 3–6, and 7 are fullyconsistent with the site-specifically labeled sites 2355–2360,2400–2428 and 2428–2450, respectively. No stop is seen withinlabeling site 2376–2400, but PHONT1-dependent stops withinthis site would be difficult to detect because of the large numberof stops seen in control 23S rRNA.

Despite this gratifying consistency between the two sets ofresults, there is also an apparent inconsistency that we do not asyet fully understand. In contrast to the suppression of labelingof the 2305–2501 fragment by added 2�-O-methyloligoRNA2258-48 (Fig. 3), added non-photolabile 2�-O-methyloligoRNA2258-53/2(S)-48 (or added 2�-O-methyloligoRNA 2258-48,data not shown), in ratios of 1:1, 2.5:1 and 5:1 with respect toPHONT 1 (Fig. 8, lanes 6–8), does not markedly reduce theintensities of the stops at positions 1–7.

The most important difference between the two experimentsis in the ratio of PHONT1/50S subunits during photolysis,which was 0.25 and 10 in the experiments using RNase Hdigestion (Fig. 7) and primer extension (Fig. 8), respectively.Considering both this difference and the heterogeneity inbinding noted in Figure 2, a plausible explanation for theapparent inconsistency can be advanced, based on threereasonable assumptions: (i) at a low PHONT1/50S ratio,

PHONT1 binds to its target site only in high-affinity subunitswhereas at a high ratio it binds to this site in both high and loaffinity 50S subunits; (ii) both types of 50S–PHONT1complexes form photocrosslinks to the same nucleotidwithin 23S rRNA but, at a PHONT1/50 S ratio of 10, theoverall photocrosslink yield is higher in the low affinity ribo-somes; and (iii) the concentration of added complementoligonucleotide is sufficient to block labeling in the highaffinity complexes but not in the low-affinity complexesFurther experiments will be required to test the validity of thexplanation. We note that high concentrations of oligonucletides might introduce artifacts into primer extension analys(i.e. stops or pauses not directly arising from PHONT1photoincorporation) that could also contribute to the appareinconsistency.

Summarizing, PHONT1 site-specifically photoincorporatesinto four sites in 23S rRNA, nts 2355–2360, 2376–2400, 2402428 and 2428–2450, and there is evidence, albeit somewambiguous, that such photoincorporation reflects labeling ofG2357, A2358, U2404, G2405, A2406, A2411 and A2430well as an unidentified site (or sites) within nts 2376–2400.

Figure 6. Primer extension analysis of PHONT1-labeled 23S rRNA. The2352–2210 region. All reaction mixtures contained 0.15 nmol of 50S subunit,and the indicated quantity (in nmole) of PHONT1. Photolysis was as indicated.RNA was prepared for primer extension as described in Materials and Methods.The primer used was oligo-DNA 2493–2477. Lanes U, C, G and A are sequencingproducts generated from control 23S rRNA in the presence of ddATP, ddGTP,ddCTP and ddTTP, respectively.

Figure 7.RNase H cleavage within fragment 2305–2501 of [32P]PHONT1-labeled23S rRNA using hybrid oligonucleotides. Procedures were as describedFigure 3. PHONT1-labeled rRNA (5 pmol) was hybridized with the indicatedpairs of DNA/2�-O-methyloligoRNA hybrid oligonucleotides (10 pmol each,see Table 1), digested with 2 U of RNase H and separated on a 10% acrylamurea gel. Note that labeled fragments migrate with apparent sizes (in nt) eqto the sum of excised fragment and PHONT1 (11 nt). Numbers in the left andright margins indicate sizes (in nt) of DNA markers.

4382 Nucleic Acids Research, 1999, Vol. 27, No. 22

is3S

otons12n)ofthes-As

na,

s.23Sly.(+)

nso

ed

at

Does 2258-48 probe binding to 50S subunits induceconformational change?

Since we seek to generate photocrosslinks that will be useful inconstructing and evaluating models of the peptidyl transferasecenter within the 50S subunit, it is important to assess whetherPHONT 1 binding to the 50S subunit causes structuraldistortions outside of the expected local distortion at itsbinding site. For this purpose we examined (Fig. 9) the effectof binding of 2�-O-methyloligoRNA 2258-53/2(S)-48 onchemical footprinting (DMS and kethoxal) patterns in regionsof 23S rRNA [nt 750–1130, 1780–2215 and 2258–2650; thegap is caused by the binding of 2�-O-methyloligoRNA 2258-53/2(S)-48 to nt 2248-58] that either include the crosslinkingsites described above or that have been implicated eitherdirectly or indirectly in peptidyl transferase activity (5–13).Most significantly, added 2�-O-methyloligoRNA 2258-53/2(S)-48 has no effect on the chemical modification of nt 2258–2650, which includes the region from 2350 to 2435 (Fig. 9A)and that contains all of the nucleotides into which PHONT1

photoincorporates site-specifically. Thus, none of thphotoincorporation can be attributed to distortions in the 2rRNA structure.

Similarly, no changes were seen for nt 750–970 (data nshown). However, a series of changes were seen in regi990–1057 (Fig. 9B), 2005–2019 (Fig. 9C) and 2169–22(Fig. 9D), as well as at nt A1927 and A1928 (data not showand A2071, A2088 and A2090 (Fig. 9C). Interestingly, eachthe changes depicted in Figure 9B–D are also induced bybinding of other PHONTs directed toward the peptidyl tranferase center, targeting nt 2448–2458 and 2604–2612.discussed in greater detail elsewhere (S.N.Vladimirov, Z.Druzi

Figure 8. Primer extension analysis of PHONT1-labeled 23S rRNA. The2434–2290 region. All reaction mixtures contained 0.06 nmol of 50S subunit,and the indicated quantity (in nmole) of 2�-O-methyloligoRNA. Photolysiswas as indicated. The primer used was oligo-DNA 2457–2442. Lanes U, G, Cand A are sequencing products generated from control 23S rRNA in the presenceof ddATP, ddCTP, ddGTP and ddTTP, respectively.

Figure 9. Primer extension analysis of the effect of 2�-O-methyloligoRNA2258-53/2(S)-48 on chemical modification of 23S rRNA within 50S subunit(A) Lanes U, G, C and A are sequencing products generated from controlrRNA in the presence of ddATP, ddGTP, ddCTP and ddTTP, respectiveSequencing lanes are omitted from other panels for clarity. Lanes (–) anddenote the absence or presence of 2�-O-methyloligoRNA 2258-53/2(S)-48 duringincubation of 50S subunits; lanes UNMOD, DMS and KETH denote incubatioof 50S subunits in the absence of chemical modification, in the presencefDMS or in the presence of kethoxal, respectively. The oligo-DNA primers usfor extension were as follows. (A) 2457–2442. (B) 1115–1099. (C) 2125–2109.(D) 2250–2234. Arrows indicate major sites of 2�-O-methyloligoRNA 2258-53/2(S)-48-induced change, including the target region for 2�-O-methyl-oligoRNA 2258-53/2(S)-48 in (A). Double and triple arrows indicate stopsconsecutive nucleotides. Arrows are aligned with stops in DMS lanes.

Nucleic Acids Research, 1999, Vol. 27, No. 224383

nginge,le

t

nceo

rk

)

R.Wang and B.S.Cooperman, submitted for publication), thesechanges are likely to arise from some combination of confor-mational change induced by complementary oligonucleotidebinding to the 50S subunit and of physical proximity betweenthe PHONT-targeted sites and nucleotides showing alteredreactivity.

The ‘P-loop’ subdomain and the peptidyl transferase center

The crosslinks we observed (Fig. 10) demonstrate a physicalproximity (17–19 Å) between nt 2252 and 2253 within the P-loopof 23S rRNA (attached to helix 80), a key part of the bindingsite for the 3�-terminus of P-site bound tRNA (see above), andfour other sites within region V of 23S rRNA falling between(i) nt 2355 and 2360 (most likely nt G2357 and A2358) in theloop attached to helix 86, (ii) nt 2376 and 2400, in the regiondefined by helices 87 and 88, (iii) nt 2400 and 2428 (mostlikely U2404-A2406 and A2411) a region including helix 88and its attached loop and (iv) nt 2428 and 2450 (most likelyA2430) falling between helices 74 and 88.

These results show a remarkable overlap with those ofGregory and Dahlberg (34), who have demonstrated that mutationswithin the P-loop (at nt G2251, G2252 or G2253) result inmajor enhancements in the chemical reactivities of severalnucleotides within 70S ribosomes, including G2357, G2361,U2408–U2410, G2428 and U2431, as well as with a directcrosslink between nt 2258 and 2425 previously reported byStiegeet al.(35). They thus strongly buttress the conclusion ofGregory and Dahlberg (34), based as well on phylogenetically-determined tertiary interactions (36), that the P-loop forms partof a structural subdomain extending from G2238 to A2433.

On the other hand, we did not see any crosslinks to either the2530 loop, the A loop (U2552–C2556) or the 2585 region in23S rRNA, although mutations within the P-loop inducealtered chemical reactivities in each of these 23S rRNA sites(27,34). This may indicate that the latter changes arise fromallosteric effects, rather than from the direct physical proximitywe demonstrate between the P-loop and nucleotides fallingbetween 2357 and 2430. Alternatively, it is possible that theorientation of the azido group within the heteroduplex formedbetween the RNA and PHONT1 does not allow crosslinks tothese other sites to be formed, since modeling the heteroduplexstructure makes clear that only a limited region of the 50Ssubunit is accessible to the photogenerated nitrene. Such anexplanation might also account for the lack of labeling of L27,which has been crosslinked to several nucleotides in the P loopsubdomain (37). Here it is worth noting that the distancebetween the phosphate backbone of the photoprobe and thephotogenerated nitrene nitrogen is only 8.9 Å. Further, attach-ment of the phenacyl moiety in the middle of the oligonucle-otide, as opposed to either the 3�- or 5�-termini, limits itsfreedom of motion. It would be interesting to determinewhether other PHONTs targeted to the 2250 loop willcrosslink within the 50S subunit at sites other than thosedescribed above, following the logic of experiments recentlycarried out on the 530 loop of 16S rRNA (21).

There is abundant evidence that the ribosome is a flexiblestructure, undergoing large-scale conformational changes as itproceeds from one functional state to another (38,39). As X-raystructures achieve increasingly fine resolution, we expect thatPHONT and other site-directed photo-crosslinking studies(40,41) will change focus, from generating photocrosslinks to

aid in the construction of a static ribosome structure to usicrosslinks as a way of defining conformational changes occurrin different functional states of the dynamic ribosome. Herthe intrinsic ability of the PHONT approach to sample availabconformations in solution from functionally significantargeted sequences will be of particular utility.

ACKNOWLEDGEMENTS

We acknowledge with thanks the excellent technical assistaof Nora Zuño in several aspects of this work and Hyuk-SoSeo for his help with preparation of the manuscript. This wowas supported by NIH grant GM-53146.

REFERENCES

1. Frank,J. (1998)J. Struct. Biol., 124, 142–150.2. Harms,J., Tocilj,A., Levin,I., Agmon,I., Stark,H., Kölln,I., van Heel,M.,

Cuff,M., Schlünzen,F., Bashan,A., Franceschi,F. and Yonath,A. (1999Structure, 7, 931–941.

3. Clemons,W.M.,Jr, May,J.L., Wimberly,B.T., McCutcheon,J.P.,Capel,M.S. and Ramakrishnan,V. (1999)Nature, 400, 833–840.

4. Ban,N., Nissen,P., Hansen,J., Capel,M., Moore,P.B. and Steitz,T.A.(1999)Nature, 400, 841–847.

5. Mueller,F. and Brimacombe,R. (1997)J. Mol. Biol., 271, 524–544.6. Cooperman,B.S., Weitzmann,C.J. and Fernandez,C.L. (1990) In

Hill,W.E., Dahlberg,A., Garrett,R.A., Moore,P.B., Schlessinger,D. andWarner,J.R. (eds),The Ribosome: Structure, Function and Evolution.American Society for Microbiology, Washington, DC, pp. 491–501.

7. Dontsova,O., Tishkov,V., Dokudovskaya,S., Bogdanov,A., Döring,T.,Rinke-Appel,J., Thamm,S., Greuer,B. and Brimacombe,R. (1994)Proc. Natl Acad. Sci. USA, 91, 4125–4129.

Figure 10. Identification of crosslinks. The portion of Domain V of the 23SrRNA showing major labeling by PHONT1. Arrows point to crosslinks identifiedby primer extension analysis.

4384 Nucleic Acids Research, 1999, Vol. 27, No. 22

8. Cooperman,B.S., Wooten,T., Romero,D.P. and Traut,R.R. (1995)Biochem. Cell Biol., 73, 1087–1094.

9. Brimacombe,R. (1995)Eur. J. Biochem., 230, 365–383.10. Noller,H.F., Green,R., Heilek,G., Hoffarth,V., Huttenhofer,A., Joseph,S.,

Lee,I., Lieberman,K., Mankin,A., Merryman,C., Powers,T., Puglisi,E.V.,Samaha,R.R. and Weiser,B. (1995)Biochem. Cell Biol., 73, 997–1009.

11. Garrett,R.A. and Rodriguez-Fonseca,C. (1996) In Zimmermann,R.A. andDahlberg,A. (eds),Ribosomal RNA: Structure, Evolution, Processing andFunction. CRC Press, Boca Raton, FL, pp. 327–355.

12. Green,R. and Noller,H.F. (1997)Annu. Rev. Biochem., 66, 679–716.13. Baranov,P.V., Kubarenko,A.V., Gurvich,O.L., Shamolina,T.A. and

Brimacombe,R. (1999)Nucleic Acids Res.,27, 184–185.14. Khaitovich,P., Mankin,A.S., Green,R., Lancaster,L. and Noller,H.F.

(1999)Proc. Natl Acad. Sci. USA,96, 85–90.15. Muralikrishna,P. and Cooperman,B.S. (1991)Biochemistry, 30,

5421–5428.16. Muralikrishna,P. and Cooperman,B.S. (1994)Biochemistry, 33,

1392–1398.17. Alexander,R.W., Muralikrishna,P. and Cooperman,B.S. (1994)

Biochemistry, 33, 12109–12118.18. Muralikrishna,P. and Cooperman,B.S. (1995)Biochemistry, 34, 115–121.19. Muralikrishna,P., Alexander,R.W. and Cooperman,B.S. (1997)

Nucleic Acids Res., 25, 4562–4569.20. Alexander,R.W. and Cooperman,B.S. (1998)Biochemistry, 37,

1714–1721.21. Wang,R., Alexander,R.W., van Loock,M., Vladimirov,S.N.,

Bukhtiyarov,Y., Harvey,S.C. and Cooperman,B.S. (1999)J. Mol. Biol.,286, 521–540.

22. Cooperman,B.S., Alexander,R.W., Bukhtiyarov,Y., Vladimirov,S.N.,Druzina,Z., Wang,R. and Zuño,N. (1999)Methods Enzymol., in press.

23. Cooperman,B.S., Bukhtiyarov,Y., Vladimirov,S.N., Druzina,Z., Wang,R.and Seo,H.-S. (1999) In Garrett,R.A., Douthwaite,S.R., Liljas,A.,

Matheson,A.T., Moore,P.B. and Noller,H.F. (eds),The Ribosome.Structure, Function, Antibiotics and Cellular Interactions.American Society for Microbiology, Washington, DC, in press.

24. Moazed,D. and Noller,H.F. (1989)Cell, 57, 585–597.25. Samaha,R.R., Green,R. and Noller,H.F. (1995)Nature, 377, 309–314.26. Samaha,R.R., Green,R. and Noller,H.F. (1995)Nature, 378, 419.27. Green,R., Samaha,R.R. and Noller,H.F. (1997)J. Mol. Biol., 266, 40–50.28. Steiner,G., Kuechler,E. and Barta,A. (1988)EMBO J.,7, 3949–3955.29. Porse,B.T. and Garrett,R.A. (1995)J. Mol. Biol., 249, 1–10.30. Sambrook,J., Fritsch,E.F. and Maniatis,T. (1989)Molecular Cloning

A Laboratory Manual,2nd Edn. Cold Spring Harbor Laboratory Press,Cold Spring Harbor, NY, pp. 11–39.

31. Muralikrishna,P. and Wickstrom,E. (1989)Biochemistry,28, 7505–7510.32. Moazed,D., Stern,S. and Noller,H.F. (1986)J. Mol. Biol.,187, 399–416.33. Inoue,H., Hayase,Y., Iwai,S. and Ohtsuka,E. (1988)Nucleic Acids Symp.

Ser., 19, 135–138.34. Gregory,S.T. and Dahlberg,A.E. (1999)J. Mol. Biol., 285, 1475–1483.35. Stiege,W., Atmadja,J., Zobawa,M. and Brimacombe,R. (1986)

J. Mol. Biol., 191, 135–138.36. Gutell,R.R. (1996) In Zimmermann,R.A. and Dahlberg,A.E. (eds),

Ribosomal RNA: Structure, Evolution, Processing and Function inProtein Biosynthesis. CRC Press, Boca Raton, FL, pp. 111–128.

37. Wower,J., Wower,I.K., Kirillov,S.V., Rosen,K.V., Hixon,S.S. andZimmermann,R.A. (1995)Biochem. Cell Biol., 73, 1041–1047.

38. Wilson,K.S. and Noller,H.F. (1998)Cell, 92, 131–139.39. Agrawal,R.K., Heagle,A.B., Penczek,P., Grassucci,R.A. and Frank,J.

(1999)Nature Struct. Biol., 6, 643–647.40. Mundus,D. and Wollenzien,P. (1998)RNA, 4, 1373–1385.41. Sergiev,P., Dokudovskaya,S., Romanova,E., Topin,A., Bogdanov,A.,

Brimacombe,R. and Dontsova,O. (1998)Nucleic Acids Res., 26,2519–2525.