Tobacco Documents Online

MOLECULAR AND CELLULAR BIOLOGY .4 publication o~ tl~ Amerfcan Society lot MicrobfoMg'y Manuscript Number: MCBI445.92 Title: "A unique ribonucleoprotein complex..." 21 December 1992 Am~ma ~ty Ior Micrcblol~y 132~ Maasaghu.~t~ Av~ntte, N.W. W~f~tm, IX~ ~00~¢171 PI~: (202) 7~7.~600 Dr. Sally A. Amero Dept. of Molecular and Cellular Biochemistry Strich School of Medicine Loyola University Medical Center 2150 S. First Avenue Maywood, IL 50153 FAX (708) 215-8523 Dear Dr. Amero, All three referees saw much that was good in this manuscript, but felt that the data for HRB/PEP interactions should be clearer, and the basis for differential sedimentation of hnRNP and PEP- containing complexes should be further investigated. Under the circumstances I would like to strongly encourage you to resubmit an expanded manuscript addressing the major comments of all three referees. Three complete copies of the revised manuscript would have to be sent directly to the ASM, together with a detailed covering letter stating that this is a resubmission of MCB 1445/92 and indicating your response to each comment of each referee. The ASM would then automatically forward the resubmitted manuscript and cover letter to me, and I would forward the new manuscript to the same referees for scrutiny. Sincerely, Alan M. Weiner Editor in Chief cc: ASM Publications 40000027

A Unique Ribonucleoprotein Complex Assemble on Ecdysone-Responsive Sites in Dros Running Title: The PEP Ribonucleoprotein Complex in Drosophila Sally A. Amero1°, Joel W. Hockensmith2, Gopa Raychaudhuri3"4, and Ann L. Beyer4 "Department of Molecular and Cellular Biochemistry, Loyola University Chicago, Stritch School of Medicine, Maywood, IL 60153. Corresponding author. 2Department of Biochemistry, University of Virginia School of Medicine, Charlottesville, VA 22908. 3Present address: Hepatitis Section, Laboratory of Infectious I~iseases, NIAID, National Institutes of Health, Bethesda, MD 20892. 4Department of Microbiology, University of Virginia School of Medicine, Charlottesville, VA 22908. *TEL: 708-216-3365 FAX: 708-216-8523 1 40000028

ABSTRACT The Protein on Ecdysone Puffs (PEP) is associated preferentially with active ecdysone-inducible puffs on Drosophila polytene chromosomes and contains motifs characteristic of transcription factors and RNA-binding proteins (Amero, S.A., Elgin, S.C.R., and Beyer, A.L. Genes Dev. 5:188-200, 1991). Using cytological and biochemical approaches, PEP was found to be integral to a chromosomal ribonucleopr'otein (RNP) complex. Its distribution on polytene chromosomes was similar to that of the HRB proteins - which are basic Drosophila hnRNP proteins (Raychaudhuri, G., Haynes, S.R., and Beyer, A. L. Mol. Cell. Biol. 12:847-855, 1992) - with PEP sites comprising a large subset of HRB protein sites. In sucrose density gradients the PEP RNP complex is large, reasonably abundant, and nondiscrete; it sediments differently than the RNP complex containing the HRB proteins, suggesting that the PEP and HRB RNP complexes can exist independently. Possible associations between the complexes were revealed by the highly-specific retention of portions of PEP and of all the major HRB isoforms on an anti-PEP antibody column; RNAse digestion released a specific subset of HRB polypeptides. These observations lead us to suggest that a PEP/RNA complex assembles preferentially on ecdysone-regulated genes in Drosophila, presumably to expedite the transcription and/or processing of these transcripts. Once assembled, PEP and HRB proteins may interact via both protein-protein and protein-RNA interactions. 40000029

5 Frr~. 2.'~L Accd. 3eL U3A Vcl. $9, loP. S-P3-3-,~413. $Cl~tCmbcr 1~92 Independent deposition of heterogeneous nuclear ribonucleoproteins and small nuclear ribonucleoprotein particles at sites of transcription (ml~lA spllcing/RNA processing)" SALLY A. AMERO*t, GOPA. RAYCHAUDHURI*:[', CYNTHIA L. CASS*, ~VALTHEK J. VAN VENROOLI§, WINAND J. HABETS§¶, ADRIAN R. KRAINERt[, AND ANN L. BEYER='** "De~rlmen~ of Microblology, U~ivcn;ty of Yh'~n~. School of Medicine, Charlottcsvil]co VA 22908: |Dcpaxlment of B~och~m|stry, University of N~jmegcn. Nijr~e.~en. The Netherlands: and ~Cold Spring H~r ~¢o~,'Co~ Spring ~r, ~ 11724 Communicated by Oscar L. Miller, Jr.. June 15. I~2 (recMved for review Ap~! 16, l~2) ABSTRACT The major nuclear ribonucleoproteins (RNPs) involved In pre.mRNA processing are classified in broad terms either as small nuclear RNPs (snRNPs), which are major participants in the splicing reaction, or heterogeneous nuclear RNPs (hnRNPs), which traditionally have been thought to function in general pre-mRNA packaging. We obtained antibodies that recognize these two classes of ~ in Drosophila melanogaster. Using a sequential immunestaining technique to compare directly the distribution of these RHPs on Droxophila polytene chromosomes, we found that the two patterns were very similar qualitatively but not quantitatively, arguing for the independent deposition of the two RNP types and supporting a role for hnRNP proteins, but not snRNPs, in g~neral tran~ript packaging. Both heterogeneous nuclear ribonucleoproteins (hnRNPs; reviewed in refs. 1 and 2) and small nuclear ribonucleopro- loins (snRNPs; reviewed in ref. 3) are deposited cotranscrip- tionally on eukaryotic RNA polymerase II transcripts (4-8). Whereas the major basic hnRNP proteins have been consid- ered traditionally to function in general pre-mR.NA packaging (2, 9), they have been proposed recently to be specific splicing cofactors or to be preferentially associated with splice junction sequences (10-15). snRNPs are major partic- ipants in the splicing reaction (3) but have been implicated recently in general packaging as part of a previously assem- bled unitary processing complex also containing hnRNPs (5, 6). The various proposals predict different amounts and ratios of the two protein types on nuclear pre-mRNA molecules at chromosomal sites of transcription, which is the issue we have addressed by sequential immunostaining. The core hnRNP proteins (A, B, and C proteins of 32-45 kDa) were originally identified as the major proteins that are associated with newly synthesized pre-mRNA (in the form of 30-50S RNP particles) when it is extracted from nuclei (reviewed in refs. I and 2). This observation, together with their nuclear abundance, their ability to bind single-stranded nucleic acids regardless of sequence, and their helix- dcs~bilizing properties, led to the notion that these core hnRNP proteins are involved in general pre-mRNA packag- ing. much as the histones are involved in the general pack- aging of DNA (1, 2). However, more recent investigations of hnRNP proteins, using in vitro splicing or in vitro RNA binding studies, have suggested that these proteins play a role in the splicing reaction (10-12), that they bind with high al~nity to sequences at 3' splice sites (13,14). and that they ~re deFendent on snRNFs for acquisition of a crosslinkable association with RNA (13). These in v[tro studies have led to a reappraisal of the independent structural role of hnRNP proteins in pre-mRNA packaging towards a view that they are a few of the many required cofactors for splicing. The simplest version of this view would predict a constant stoi- chiometry of snRNPs and the core hnRNP proteins on pre-mRNA, in amounts that correlate with the number of splicing signals. Another recently proposed model would also predict a constant stoichiometry of snRNPs and hnRNP proteins on pre-mRNA, but in amounts that correlate with RNA length rather than with splicing signals (5, 6). The unitary processing complex proposal (5, 6) predicts codeposition and constant stoichiometry of hnRNP proteins and snRNPs on transcripts and is based on cytological observations of oocyte contents of the newt Notophthalmus vlrldescens. First, hnRNP pro- teins and snRNPs (plus other splicing factors) occur in the same nuclear extrachromosomal complexes: the B snurpo- seines, and second, these same components occur on almost all lampbrush chromosome loops in amounts that correlate with RNA mass distribution on that loop (5), Although reports of snRNPs at loci thought not to have introns [Chi. ron.omus polyten~chromosome Balbiani rings (4) and newt histone gene-contalning lampbrush loops (5)] appear to sup- port this model, it is now known that the Balbiani ring genes contain introns (16, 17), and splicing signals may occur on extremely long (hundreds of kilobases) readthrough tran- scription units on lampbrush chromosomes (18). The original views that the major hnRNP proteins associate promiscuously with pre-mRNA, while snRNPs are deposited specifically at splice sites, are supported by many in vitro RNA binding studies (e.g., refs. 8 and 19; reviewed in refs. 1-3), by analysis of RNA sequences associated with tracted hnRNP complexes (reviewed in ref. 1), and by electron microscopic visualization of active genes (20, 21). Thus the abundance of snRNPs at a given transcriptionally active site would reflect the number of introns and the strength of their splicing signals, whereas the abundance of hnRNP proteins would be a function of RNA length, leading to site-specific ratios of hnRNP proteins to snRNPs. Our observations of such site-specific ratios and of intense hnRNP staining at the highly transcribed puff sites support these original predictions. Abbreviations: hnRNP, heterogeneous nuclear ribonucleoprotein; snRNP, small nuclear fibonucleoprotein. tPresent address: Department of Molecular and Cellular Biochem- istry, Loyola University Medical Center, Maywo~d, 1L 60153. $Present address:. Laborat~r] of Mo!ecular Genetics. NationoJ In- stitute of Child Health and Human Development, National Insti- tutes of Health. Bethesda, MD 2(]$92. ~Preseat address: Organon Telmika B.V., Bosei~d 15, PO Box 84, 52~9 AB Boxtel, The Netherlands. **To whom re~rint req-es~s sho~Id b-. a~Idressed. 40000030

8-~10 Bicr.h~-m~stry: ~ero e,r ~/. MATEP, J.4J.S AND METHODS Antibodi~s. Four antibodies were u~ed, all of which have been described pre~,~iously. The aati-Drosap~'la hnRNP an- tibody was a rabbit antiserum s~dfic for ~he major basic A/B-ty[:-: hp..P, NP proteins (22). The o~h.~r three were mouse monoclanal antil~odies: 4G3, Sl:~Cific for the mammalian U2 sn.RNP B" protein of 28 kDa (23); anti-m3G cap anti~o:ly, which recognLzes the snRNA-spec~ic cap (24); and Y1D2. specific for th= Drosophila nuclear protein PEP, which is found primarily on ¢cdysan¢-regulated puffs (25). [mmunablatting. Nuclear extract was prepared as de- scribed (26) from logarithmic-phase Drosopldla Schneider 2 cells. The nuclear extract was fractionated in SDS]12% pulyac~'!amide gels, and the separated proteins were either s:ai~:¢c ',vi~h Coomassie blue or electroblotted onto n|troccl- lu/ose and probed with undiluted culture superuatant from h.vbddoma cells secreting antibody 4G3. The primary anti- body was detected by using biotinylated horse anti-mouse IgG and the VectaStaln ABC immunoperoxidase detection system (Vector Laboratories), according to the manufactur- er's instructions. Immunopre©lpil~flon. Immunopre.clpitation experknents utilized protein A-agarose beads (BRL) incubated with rabbit anti-mouse IgG (Zymed) and undiluted hybridoma cell pcrnatants. The loaded beads were incubated in Drosophila Kc cell nuclear extract (26) at 4°C overnight, and washed with 10 mM Tfis, pH 7.5/504) mM NaCl/0.05% Nonidct P-40 (4G3 antibody) or 20 mM Hopes, pH 8.0/150 mM NaCI/0.05% Triton X-100 (anti-m~G antibody) prior to. phenol extraction. Total nuclear RNA samples were prepared by phenol extrac- tion of the Kc cell nuclear extract. The RNA samples (either total or immunoprecipitated) were fmctionatad in 10% poly- acrylamldc denaturing gels and visual~.ed by silver staining. Immunofluoreseence on Polytene Chrom(~somes. Immuno- fluorescence assays on polyten¢ chromosomes from third- insmr D. melaaogaster larvae were performed as described (25, 27), using a 45% acetic acid/3.7% formaldehyde solution to fix the chromosomes. All antibodies were incubated for 2 hr at room temperature on the slides containing the squashed polytenc chromosomes. The mouse monocloaal antibodies were detected with a fluoresceinated goat anti-mouse anti- body (ICN; 1:1000 dilution), and the rabbit anti-hnRNP antibodies were detected with a goat anti-rabbit antibody conjugated to rhodamine (ICN; 1:$00 dilution). Thus, when the same chromosome set was stained with two different antibodies (one raised in mice and one raised in rabbits), the secondary detection systems were noncrossreactive and nonoverlapping. The localization of the t'wst antibody on the chromosomes was recorded by photography using UV illu- mination appropriate for either rhodarnine or fluorescein. The coverslip then was removed from the slide containing the chromosome squash and the slide was washed three times for $ rain, with agitation, in 5~ mM Tris, pH 7.6/725 mM NaCI at room temperature. The antibody staining procedure was repeated, as was photography with different UV illumination. For each experiment, the secondary antibodies were shown m produce background levc|s of fluorescenc~ in the absence of the primary antibody. RESULTS As a probe for the deposition of hnRNP proteins, we used an antiserum raised against the major basic A/B-type hnRNP "'HRB'" proteins from Drosophila (22, 28, 2~). These 3~- to 41-kDa b~sic proteinsare coisolated with Drosophila nuclear poly(A)+ RNA and share all properties tested, including sequence sknilarity and s~ngle-sh-m~ded nucleic acid-binding properties, ~ith mamma~Jan A/B-type hnR~P pro:eLu~ (22). Prcc. Natl. Acad. Sci. USA ~) (1~92) This antikody also re,:oga~ze-~ HeLa c¢~ A/B-~P ~ on ~uaoblots (22). For a s~ protc~ probe, w~ ~d B" protein of 28 ~a on ~unobla~, ~fl ~unop~cigi- rotes o~y U2 s~P (~). Howewr, t~s ~y n~ed two Drosophila nuclc~ pro~ins of 35 ~munoblo[s ~ig. In, l~c 2) ~d precipimled ~o philo ~A s~cics (Fig. Ib, Imc 3) ~t wore ~so p~cipi- rated by ~ ~ti~y s~c~c [o th~ sn~A-s~c~c t~ethylgu~osinc (m~G) ~p (Fig. lb, lane 2). ~ evidence • ~ ~esc two m~G~ap~d ~A s~cies represent philo UI ~d U2 s~NAs ~ ~e follow,g: cx~tcd (30), approx~tely 1~ md 165 aucleofides] on e[cc~ophoretic million relativ~ [o kno~ Hc~ ccH s~As (dam no~ shown), thek ~sitian of to endogcnoas 5S ~A ~d ~A (Fig. lb) ~d m~ cell s~As (see ref. 31). ~d ~e~ ch~ctefistic apace ~ the two l~gcst abun~at ~d s~P ~fib~y-precipi~l~ nuc[c~ ~A species in this re, on of 10~ ~ly~w~de den~ufing g~Is (Fig. lb; e.g., comp~c fig~v I of rff. ~I). Thus, ~s antibody, which mcog~zes o~y protein in m~m~i~ cells, app~eatly ~so reco~izes a U1 sn~P protein in Droxophila. Wc propose ~[ t~s second protein is the fly U1 sn~P A protein, b~ed on sequence identity (32) plus aatigcnic relatedness b~twcen ~c hum~ U1 sn~P A protcin human U2 sn~P B~ protein (28 ~@. reactive Drosophila protein species on the immunoblot (Fig. la, lane 2) of 28 ~a would represent ~c fly protein, whereas the 3G~a species would represent a b 1 2 1 2 3 F=~. 1. Recognition of Drosophila proteins and s~s by the anti-B" human snRNP protein antibody. (a) Schneider 2 cell nuclear extract was fractionated in SDS/12% polyacrylamide gels, which were either stained with Coomassic blue (lane l) or electroblottcd onto nitrocellulose and probed with undiluted cell supematant from hybridoma cells secrering antibody 4G3 (lane 2). Two bands of 36 and 28 kDa, tentatively idenfi/icd asDrosophila A and B" snlT~P pro~dns (see text), arc recognized. (b) RNA samples were prepared from Kc cell nuclear extract either by phenol extraction (L~ne 1), by noprecipitation with anti-m~G cap antibody (lane 2), or by immuno- precipitation with anti-human B" protein antibody 4433 (lane 3). The samples were frartiormted in 10~ polyacryla~de denaturing gels and visualized by sliver st~ng. The top two arrowheads to the oflane 1 poin~ m R.HA species identified as U2 and U1 snP.NA by comparison with migration of mamm~an snRNAs (not sho~)and with previously reported dectrophoretic n,Jgra~oas cf snP, NAs (e.g., 30, 31). The U2 and U1 ass|~v.ments are firm bk~! their sLze, ab,_,n~.~.nce, and reproducible relative migration. renm~ning arrowheads ~entLfy tentatively ~e small RNA spcci~ U4, 5S, US, and U6 (in descending order); these latter ~=nts are n~t directly relevant to the resuIts reperted here. Two abundant ceI~ular RNA species, 5S RNA (labeled as i:~::~ed ,~.ufl th= tRNA po9nt~t£on (12rge szr.~ ~ L~= b~t~cm cf t~ £el.), a.~ 40000031

L :h :d in td ~s a* ~s 'c) -- Biochemistwl: Amero et ,:L U1 snRNP A protein. This antibody thus is weft suited as a vrobe for sn~P deposition, since U1 and U2 sn~Ps are ~o~ in p~icipate in the e~ly ~ges of spli;e si~ r~c~- ninon (3), We dc~s~d a sequential indirec~ immunofluorescent ~tain- ing p~edu~ wffh nonoveflaFpmg detection systems to ~mp~ directly the distnbutionq o~ the snRNP protein~ ~[g. ~) and ~e hnRNP proteins fFig. 2M on the sam~ set of ~l~en~ ch~mosomes. Both patterns ~ere similar to the t~i~ ~1yme~se II transcription pattern ¢e.g.. tel 25k wi~h s~ning of ~sibl~ puffs and interband regions but no staining 0f ~xtmchromosom~ fE) or nucleol~ (Nu) deb~s fs~e Fig. ~). StOning ~th an anti-m~G cap monoclonal antibody (24} Pr,,c. Natl. Acad. ScL USA ¢59 flff~2) 8411 , Fig. 2 fand.e ~ pr~ ~lu=ed a similar" p~ffand [nterband sta[nL~g, at expected if the snRNP protein pattern indeed ~p~sen~ the d~stdb~t~n of intact sn~P ~:le~. ~e ~c~ of the immunoflu~rescence assay ~s sho~ by the limited dietdbution pattern of anzther nuclear protein, PEP, ~ htch was found primarily on the active, ¢cdysone-~g~ated ~uflk { rcf. 25: Fig. 2 d and e). Vie note also that sn~Ps ~d hnRNPs were distributed diffe~ntly ~ an inta~ pol~ene nucleus IN in Fig. 2 a~L Although ~th ~s ~cuned non-nucleolar regions of the in~ct nucleus, which is la~ely occOpied by polytene chromosom~s,.~ patterns we~ dif- ferent in specifics, such as the fine s~c~ing of the hn~Ps {Fig. 2b). It is not clear whether the patterns noted here Fro. 2. Sequential staining of polytene chromosomes with snRNP- and hnRNP-speeific antibodies. The chroinosome set in a-c was first stained with the anti-snRNP 4G3 antibody, which was detected with a fluoresceinated goat anti-mouse anfik, ody arid then photographed onder t~ase-contra~t (c) and UV illuinination (eL The staining procedure was repeated with affinity-purified anti-hnRNP antiserum, which v,-as det,'cted v,4t~ a goat anti-rabbit antibody conjugated to rhodamine Ihl. Photography was repeated with different UV illumination. P'Am of arrows hdieate representative nearby chrnmo~omal sites that display ve~" different staining ratios with hnRNP and snRNP antibodl~s. Groups ofthree e~fe~r arrow~ i~dicate representative specific l~ci {also identified in Fig. 3L vt hich exhibit the same nntibody.specific staining levels regardless d'th.-order ofantibedy stz~ning trcf. Ibis figure with Fig. 3). Proceeding from the ~elomere to/he chr~ina~center on th,' left arm of the ~! ~mmosecne OL). these loci fall within the standard I:olytene ehromo~omc mup po~ition~ 61A-filF (l~beled 61). 62B-62F (labeled 62L and 7~A-79D (labeled 79). The bands at 63A a~d T/E are shown a_s landmark~. E. extrachromo~omal debris: H, historic locus at 39DE; bl, nucleus; N~, nucI~Ias; P, represen~ative puffs. In b. small re~ns of sever:d chr,~mo~ome~ were lifted off the slide when the cover s]~p was removed l~d~r to the second staining (unlabeled arrowhead~). Also ~h~,.vn: UV i]himina~on td) or Fha~e-contrr~t (e) m".er~,raph efchmm~omes stained *,':Lh m~r.oe!a~.al ~_nu~ody YID2. x~-*cifie for PEP f251; UV illuminafir~n t O or pha,~e-contra~t (g) micro~:h cf chmmosemes stunned ~th w.~cc..'=r~l zntikody ~st tL- m,G-~p s.re."ific to snRNPs 124). .q~.-,-?s .-'v. x;. - 40000032

8412 B~achemistry: Amero et aL unique to the po]ytene nucleus or are a result of fixatior, conclir~ons for chromosome squast,.~g (see refs. 34 and 353. In comparing the snRNP and hnRNP patterns on FoIytene chromosomes (Fig. 2 a and b), which essentially were idea- tical qualitatively, numerous quantitative differences could he seen. These were particularly noticeable as differences in snR-NP/hnRNP signal ratios at neighboring sites (paired arrows, Fig. 2 a-c) and were reproducible on other chromo- some sets from the same larva (data not shown). Similar quantitative d~ferences were seen when the order of anti- body staining was reversed (Fig. 3). The reproducibility of the staining patterns with the two antibodies was shown by mapping specitic representative loci (labeled 61, 62, and 79 in Figs. 2 and 3; see Fig. 2 legend) on the left arm of the third chromosome. They were found to have the same relative staining levels with the two antibodies regardless of the order ofantibody staining. With both staining regimens, the hnRNP proteins were abundant in puffs (P in Figs. 2b and 3a), whereas the snRNP proteins sometimes were abundant in puffs but frequently were not (Figs. 2a and DISCUSSION In this study, we tested several proposed models for the involvetaent of snRNPs and hnRNP proteins in the packaging and processing of pre-mRNAs. Our data indicate that snRNPs and major hnRNP proteins are deposited indepen- dendy on nascent RNAs. The data do not supp(~rt models in which snRNPs and hnRNP proteins are codeposited as part of a unitary processing/packaging complex (5.6). The sug- gestion from in vitro studies (10-15) that hnRNP proteins may be localized preferentially at splice sites for a specific func- tion in the splicing reaction is not supported by "our obser- vation that hnRNP protein levels do not correlate with snRNP levels. Finally, the abundance of hnRNP proteins at puffed sites is consistent with a general packaging function for these proteins. Prac. NatL Acad. ScL USA 89 (1;q32) The preferenti~l amlflificar2oa of r.h~ hnRNP si~ h co~elat~ wi~ ~A ~s, s~ce p~ s~ is ~owa ~:r~ ~ a ~ne~vn cf~ ~sgfipt l=n~ ~d s=gngh (3~. Because ~s ~6-~P ~ti~y reco~ a f~y ef ap~ro~mately nine s~ b~ig proteus which ~¢ eneofled by ~o ~erent gen~ (28, ~), we not d~tingu~h whet~er in~vid~ A]B-ty~ ~ have sequence preferences in vivo ~ ~ey do in vitro (14, but we could conclude ~t ~¢ to~ ~o~t of~=seprot=~ at a given site coffelates roughly ~th ~scfipfion~ Funhe~ore, in ano~er recent study, when mon~lo~ ~tib~ies to ~ee ~erent Drosophila ~B-ty~ ~P proteins were lo¢~zed on ~l~en¢ chromosomes, ~ e~ibited intense stai~ng of p~ed sites, ~ no obvious protein-specific loe~i~tion The sn~P protein (Fig. 2a) ~d s~A (Fig. 2~ si~s we~ not ampl~ed in ~ny puffs but ~¢ not neces~y expected to be amp~ed to ~e levels obse~ed for proteins if snRNPs ~e binding o~y to splic~g si~s. The bulk of the ~A mass in !o~ t~sc~p~ (which wo~d expected in most p~s) is ¢ont~buted by long in~ons; number of snRNP bin~ng sites (splice junctions) world incr¢~¢ less dr~atie~ly th~ ~e number of~P prot¢~ binding siteS, assuming that h~P proteins b~d ~A ~ a nonspec~e and stoiehiome~e f~on (1, 2, 9, 38), Addb tioa~ly, sn~P bin~ng e~ciency on nascent transcripts may v~y with the streng~ of individu~ splicing si~. Exa~nation of the histone (intronless) gene locus (H, Fig. 2 a~) w~ not info~ative in assessing whether the s~P sign~ represented splice-site hieing, since it was not s~ed above background levels by either ~tibody, presumably due to the cessation of DNA synthesis, ~d ~us historic ~A synthesis, in these late-stage polyten¢ c~omosomes. Our results indicate that hn~P proteins and s~Ps not present in stoichiome~¢ amounts at ~ sites of~scfi~ tion on Drosophga polyten¢ chromosomes, consls~nt with their independent deposition. ~ ~temative expiration is that the two RNP types ~e code~sited but ~e ~en Fro. 3. Sequential staining of polyrmr,- chromosomes with hnRNP- an~l snRNP-sT-.ceIHe andhodles. Chromosome staining was as except that the order of antibody staining was rcver,.;ed: ~,~nd-k.nRNP was used first (a), followed by and-snRNP (4G3) (b). Phase-eot~trast v~ew of same c~omosome set is shown in c. Pairs of arrows indic:zte representative tz.~hy chromosomal sites tlmt display yew dLff,'rent stai.~ reties with rE= two an~ib~ies. P, representative puffs. Specillc representative loci h~ve been rtmpped on the lift arm ofth,- third (se~ Fig. 2 l~geed}, ~wing the regrc4uci~ility of ste.ln~ng ~ith tlz.e two aat~cd2es, re'g~dless el'the oral= of smirdag. Note that the shown here ar.d in Fig. 2 ~e from som-.whzt di~'erent larval developmental stages, as i~digated by tl-.e two large puffs (P) at 74E to 75B ca~L that are ~tive oa ~ ~et shown in this figure, Izut r.ot ca the set skown in Fig. 2. Tltem are t~eraus o~=r differences in the status of specie sites (as i~'.di~ted by st~2ni=g ~ith ~e hnRNP ~L~u~<xly, wl'2eh Ls ess~t~)- H=nfi:~l to l:<:,lymer~se II ~.n/a"c-~dy S.A.A., t~',F.a~s~ed ~ork}, ~cla~g ~ extra she in th-. 79 cluster in this fig~,rc as cem~e~ ~th the s~me region ~ Fig. 2. 40000033 simp mien 21). 1 and~ Dros atth retie, retie, ofh reins Thes then affin tion. whic the k thest th¢ < may that imp¢ nucb COnS ing t nucl corn facil in vi pack ing i obse met~ envl splic func the proF sele~ "Fr (A.L Ame (S.A Natk

"B~ch=mislry: Amero et ,~i " " pendently susoept~]c to dissc~ation. We favor th~ former, simp:ar explanation her ~au~e it is consistent with electron mi~oscopic observatioas in which the tarli-_st RNA pack- aging steps were visualized on nascent RNA and cviden~ for. specific panicle deposition was s~sn only at splice sites (20, 21}. If snRNPs are indeed codeFosited with hnKNP proteins on amphibian oocytc lampbrash chromosom= loops (refs. 5 and 6; sec Introduction), this phenomenon does not extend to Drosophila. As shown here at the gent FOl~ulation level and as supported previously by observations of individual genes at the electron microscopic level (20, 21), RNP abundance reflects RNA mass whermas snRNP abundance probably reflects early splicing activity at an active locus, The amo.unt of hnRNP protein seen is not that predicted by rel~orts suggesting the "'specific'" association of major hnRNP pro- • tcins with 3' splice sites (14, 15) or the "more stable'" association of those hnRNPs in the vicinity of snRNPs'(13). These in vitro studies predict a positive correlation I~:tween the abundance of stable hnRNPs and snRNPs if these affinity" binding sites arc the only targets for their deposi- tion. However, the nuclear abundance of hnRNP proteins, which ensures their excess over these high-affinity sites, and the known RNA-binding and helix-destabilizing properties of these proteins (1, 2, 9, 38) argue for binding ofhnRNPs along the entire transcript length in rive. Although their binding may be nonspecific in terms of sequence recognition, the role that they play in managing long transcripts is probably very important. If left naked, RNA will not only be accessible to nuclease attack but will also adopt complex higher-order structures that have a high probability of masking the short consensus sequences recognized by specific RNA-procesSo ing factors. By binding to the nascent transcript within 100. nucleotides or less of its emergence from the.polymerase complex {39), these helix-destabilizing proteins presumably facilitate the very rapid, cotranscriptional splicing observed in Wvo (e.g., refs. 21 and 40). Moreover, a role in general packaging of pr~-mRNAs does not preclude a specific splic- ing function for core hnRNP proteins, as suggested by a recent in vitro'splicing study in which specific effects were obsetwcd when hnRNP A1 protein was present in stoichio- metric excess over the pre-mRNA substrate (12). One can envision, in fact, that specific effects ofhnRNP proteins on splicing may be a direct result of their general packaging funetionl for example, effects of hnRNP protein binding on the kinetics of splice site recognition or on the flexibility properties of nascent transcripts might influence spllcc-sitc selection. This work was sul~ported by the Nationat Institutes of Health {A.L.B. and S.A.A.), the National Science Foundation (A.L.B.}, the American Cancer Society ($.A.A.L and the March of Dimes {S.A.A.) and in part by a Cancer Center Support Grant from .National Cancer Institute. A.L.B. is the rocil~ient of an American Cancer Society Faculty Research Award. 1. Chung, S. Y. & WooteF. L {1986} Proteins I, 195-210, 2. LcStourgeon. W.M., Barnett, S.F. & Northingtono $. figs0} in ~'he £uka~.otic Nucleus° eds. Stm-ss, P. & Wilson, (Telford° Caldwdl. NJ}. Vol. 2. I~l~. 477--.~02. 3. Steitz. I. A.. Black. D. L., C,~rke, V.. Parker. K. A,, Krnmer, A.. Frendeway. D. & Keller, W. {1988} in Structure and Proc. NatL Acad. ScL USA 89 (1~92) 8413 P~Hcles. ~. BimsE=l. M. L. (Sg~ger. BerI~n}, ~p. I~LN. B~oL 1~, ~5~3. 6. Gall. J. G. (1~1) S6~nc~ ~ 14~-1~. 7. Fake, S., Lesen G. & M~n, T. ~. (~8~ ~. Cell B~oh 1153-~. & M~in, T. E. (1~) C~omoaama ~ 9. Conway. G., W~ley. J.. B~. T. & ~Stou~eon, W. M. (19~8) Mol. CdL Biol. ~ I0. Choi, Y. D., G~ws~, P. J., Sh~, ~. A. & ~uss, G. (19~) Science ~1, 153~1S9. 11,. Sie~kowska. H., Szer, W., Fu~. P. J. & Kole, R. Nucleic Acids Res. 14, 5241-5~4. 12. Mayeda, A. & K~ner, A. R. (1~} Cell ~, 365-375. May~nd. S. H. & Pede~on, T. (I~) Nucleic Acids Rea. 18, 3307-3318. 14. Swanmn. M. S. & D~ss, G. (1988) EMBOd. 7, 3519-3529. Buvoli. M., Cobianehi, F,, Biamonfi, G. & ~va, S. Nucleic Acids Res. 18, 6595~. 16. H~g, C., ~eholt, B. & Wiesl~der, L. (1988) d. Mol. Biol. 2~, 655-~. 17. Paulsson. G.. Len~hl. U., GMll. L. Ericsson, C. & Wieslander, L. {1~) I. MoL Biol. 211, 331-~9. 18. D~z, M. O.. Ba~acchi-~lone, ~., Mahon, K. A. & Gall, A G. (1981) Cell ~, ~9-659. 19. Retd, R. (I~) Prec. Natl. Aead. Sol. USA ~, 20. Osheim, Y. O., Miller, O. L., Jr. & Beyer, A. L. (198~ 143-151. 21. Beyer, A. L. & Osheim, Y. O. (1988) Genes Dev. 2, 75~765. Raychaudhufi, G., Haynes, 8. R. & Btyer, A. L. (1~2) Cell. Biol. 12, ~7~55. Habets.W.J.,H~t,M. H.,~ong, B. A. W.,vanderKemp, A. & van Ven~ij, W. J. (1989) d. Immunol. 143, 25~. 24. K~iner, A. R. (1988) Nucleic Acids Res. 16~ 9415-~29. Ame~, S. A., El#n, S. ~. R. & Beyer, A. L. (1~1) Genes Dev. $~ 188-2~. 26. Di~am, J. D., Lebovi~, R. M. & R~der, R. G. (1983) cleie Acids Res. ll~ 1475-1~9. 27. James, T. C. & El#n, S. C. R. (1980 Mol. Cell. B~ol. 6~ 3862-38~. Haynes, S. R., Ray~haudhufi, G. & Beyer, A. L. (1~) Mol. Cell. Biol. 10~ 316-3~. 29. Haynes, S. R.,Johnson, D., Rayehandhu~, G. & Btyer, A. L. fl~l) Nucleic Acids Res. 19, 30. Reddy, R. {1988) Nucleic Acids Res. 1~, Suppl. RTI-R85. Mount, 8. M. & Steitz, J. A. (198D Nucleic Acids Res. 9~ 6351~3~. 32. Sillekens, P. T. G., Ha~, W. L, ~ijer, R. P. & v~ r~ij, W. J. {1~ EMBO d. 6, ~3. Habtm, W., H~t, M., B~n~n, P., Luh~n, R. & v~ V~n~ij, W. 2. (1985) ~MBO d. 4, ~45-~50. M. 8~e~r, D. L. fl~) Prec. Natl. Aead. $eL USA ~, 147-~L 35. ~o-Fons~, M., TolM~, D., Pe~rkok, R., Bambino, g. M. L., Mo~es. A., Brenner, C.. ~o~. P. D., M. R.. Hue. E. & ~mond. A. I. (I~D ~MBOL 10,195-2~. 36. Simon, J. A., Sutt~. C. A., Lo~ll. R. B., Closer, R. L. Us. J. T. (1985) Celt 41,805-817. Matunis. M. J.. Matunls. E. L. & ~uss, G. (1~2)J. Cell Biol. 116, 245~6. Cobianehi. F., Karl, R. L., Willlam~. K. R., No~, V. 39. E~mn, L., G~ham, !., ~fi~ths, A. & ~n, !. (I~) ~ 393~1. ~. LeMaire, M. F. & ~nmmtl, C. 8. (I~} MoL Cell. Biol. 40000034

The Origin of Nuclear Receptor Proteins: A Single Precursor Distinct from Other Transcription Factors Sally A. Amero*, Robert H. Kretsinger, Nancy D. Moncdef, Keith R. Yamamoto, and William R. Pearson Department of Microbiology (S.A.A.) University of Virginia School of Medicine Charlottesville, Virginia 22908 Department of Biology (R.H.K.) University of Virginia Charlottesville, Virginia 22901 Virginia Museum of Natural History (N.D.M.) Martinsville, Virginia 24112 Department of Biochemistry and Biophysics (K.R.Y.) University of California at San Francisco San Francisco, California 94143-0448 Department of Biochemistry (W.R;P.) University of Virginia School of Medicine Charlottesville, Vlrginia 22908 Nuclear receptor proteins regulate transcription under the influence of hormones or other small ligands, These proteins bind to specifio DNA sequences, termed hor- mone response elements (HREs), that reside close to hormone-responsive genes (reviewed in Ref. 1). This signal transduction process involves three dlscemible domains in the nuclear receptor proteins (reviewed in Refs. 2, 3)--the N-terminus, the central DNAobinding domain comprised of two zinc finger motifs, and the C- terminal IIgand-bindlng domain; only the N-termlnus is not well conserved. In general, virtually nothing is known about the origin of transcriptional regulatory factors. With respect to the nuclear receptor proteins in particular, two evolutionary histories have been proposed. The first assumes inde- pendent origins for the different domains. By this view, ligand-binding segments with functions in bioenergetJcs and intermediary metabolism ~came fused to a DNA- binding motif to produce transcription factors, and the ceil acquired the= abiZity to respond at the transcriptional level to fluctuations in its physiological state (1). The second mo~el [mp~cates a single, muRi-domaln precur- sor that initially meal!areal a s!mpte s.~gnal transduct~cn mechanism (perhaps s~m~ar to that em;~Ioyed by the modem thyroid receptors) ~d subsequent;y acquired increasingly complex functions (4, 5). Based on protein sequence comparison and evolutionary analysis, we believe the second model is correct. Our results sug- gest that all known nuclear receptors diverged from a single common ancestor. While this ancestor may have been formed by a domain-joining event, we find no evidence for subsequent exon-shuffling or for homology between the nuclear receptors and any other transcrip- tion factors. We have investigated the evolution of the nuclear receptor gene family using established computer algo- rithms to detect sequence similarity indicative of ho- mology (common evolutionary ancestry). These studies differ from searches for analogous sequences (6), which may share a common property (e.g. periodic cysteines and histidines) without sharing common ancestry. We used the amino acid sequences of the DNA-bindIng domains (77 amino acid residues) from several nuclear receptors to search the PIR prote~n sequence databeseI by Smith and Waterman (7) and FASTA analyses (8)z. in every case, the highest similarity scores belonged to the nuctear receptor sequences in the database. For exampIe, the DNA-bi~ding dom~n of the 40000035

Vc46 r~-o. 1 rat glucocorticoid receptor (#A27284)3 produced scores of 434 against itself, 300-400 with most of the other steroid and g;ucccorticoid receptors, and 209 with the Drosophila egon protein (#S06010), the lowest scoring receptor. As a score of 125 v~ou~d be convincing evidence of homology for a sequence of this length, this analysis indicates homology for all the DNA-blnding domains in the nuclear receptor family. Proteins other than nuclear receptors produced scores lower than 125; in particular, no other classes of zinc finger pro- teins produced high scores. Additional searches of the PIR library with the zinc-coordinating ONA-binding do- mains in the transcription factors Spl (C.,H~ zinc fingers; #A29635) and GAL4 (Cs zinc fingers; #A05022) iden- tified no nuclear receptors in the top 200 sequences. Thus, we conclude that the zinc finger sequences in nuclear receptors all arose from a common precursor, but that they are unrelated to those in any other tran- scription factor. The DNA-binding domains in nuclear receptors are comprised of two functionally distinct zinc finger motifs (10-17). Comparisons of the two zinc fingers that com- prise the nuclear receptor DNA-binding region suggest that these motifs are themselves not homologs. Simi- larity scores for all of the first zinc fingers ranged from 223-91; scores for the second fingers ranged from 244-84. In contrast, scores for the first finger/second finger pairs ranged from 46-24, Thus, the two fingers in the DNA-binding domain of the nuclear receptors appear to have evolved as a single, discrete structural unit rather than by the duplication of an ancestral finger sequence. This conclusion Is consistent with recent crystallographic data (18), An evolutionary tree for the nuclear receptors (Fig. 1) was derived from the mutational distances between pairs of aligned DNA-blndlng domains (19, 20). Despite the fact that information about ligands or ligand-binding sequences was not included in this analysis, at least 4 of the 10 subfamilies ~dentified reflect common ligand- binding specifictties: the glucocorticoid-like (G), estro- gen-like (E), retlnoi¢ acid (R), and thyroid receptors (T), The ligand-binding specificlties of receptors in the other subfamilies are not known. The same subfamilies were identified In trees produced by parsimony analyses (20). Mammalian sequences appear in 9 subfam~ies, white Drosophila sequences appear in 3. This broad disper- sion suggests that a receptor-like protein preexisted in the organism that was the ance~or of arthropc~s and vertebrates. The conclusion that the receptor proteins sh~re a single precursor is supported by similar observations of the l!g~d-binding domains. In addition to I!ga~d binding, this C-term".nai segment conta.~ns a prote~n Inactivation function (21) and binding sites for the heat shock protein h~pe0 (22, 23), thought to be important for signal transduct~on (24). Searches of the PiR database using the sequence of the ligand-b~nd~ng dcm~.;n of the rat a $, e.g. $H~J.~JAD.=I, gIuccco~coid r~ptor or vitamin D receptor (VDR) (#A31761) yielded high s[mi!arity scores among atmost all of the nuclear receptor sequences. For examp:e, with the rVDR ligand-b~nding domain (330 amino acid residues), sequences in subfamily D scored 1400- 1577; subfamily R, 310; subfamily T, 200-280; subfam- ily E, 127-260; and subfamily G, 98-155. Only the Drosophila receptors in subfamily K were not among the top 200 sequences (scores from 41-53). While the low similarity scores for the ligand-binding domains in the K subfamily might suggest the acquisition of a novel ligand-binding domain in thL~ subfamily, the high simi- larity scores for the ligand-binding domains in the other nine subfamilies and the observation that the K-family DNA-binding domains are also among the most diver- gent suggest a more parsimonious explanation: the K- family nuclear receptors are simply distantly related, as the length of the K-subfamily branch in Fig. 1 indicates~. Many protein families display a similar range of quence diversity (25). The subfamilies found in the DNA-binding domains are evident also in the ligand-binding domains (Fig. 2). We compared the similarity relationships among the DNA-binding domains to those in the ligand-binding domains, using separate measures of similarity to ac- commodate the difference in conservation and length. The value plotted for the DNA-bind[ng domains is the percent of the best similarity score possible for any pair of sequences; this closely approximates a percent iden- tity values. The value plotted for the ligend-blndlng domain is the raw similarity score. With only a few exceptions, the densities of highest similarity indices are mirror images in the two domains~strong evidence for concerted evolution of these two domains without discernible shuffling. We conclude that the nuclear receptors diverged from a common ancestor that contained both a DNA- binding domain with two nonhomologous zinc fingers and a ligand-binding domain. The odgins of these do- mains themselves remain open questions, as no ho- mologous, nonreceptor sequences were found for either type of sequence, and neither domain was found in the absence of the other. These observations sug- gest that the nuclear receptors do not share a common ancestor with other tranecdption factors, zinc finger proteins, or ligand-blnding proteins, and that the zinc ion has been adopted at several times during the course of evolutionary history to stabilize protein structural motifs. We speculate that in the time since the appear- ~ Other l~gand-bind;ng domains obtain low sIm~=dty scores wh.=n compared w;th some receptors. For examp;e, c~mp~- son of the l!g~nc~-b;nd:ng demons from hVDR End hTR25 yIetds a low sc~re (51), ~.S WOLItd be exl:~ctEd from the p3th th3t co~ec~ these two receptors In F~g. 1. However. the hTR25 [;gEnd-b:n~ng dorn~n obta;ns h~h s:m? ~.~J~ty scores (136--180) wh+n compazed w;th mernt:ers of subfam~y Y. s F~" ex~m~!~, the rVDR DNA-~ng don'~n h3s a of 441 v;hen comp~ed ~th [tee:f, r~t e~A.3 t~as a score 4E0 ~g~nst [t~Ef. The two sequ.=n=es I-,.~ve a ps;r/,",se ,~f 270. Thus rVDR &-.=l r-e,"~B st,--~re 270Itr;.r4441,4E0) 61% 40000036