Splice site skipping in polyomavirus late pre-mRNA processing.

0022-538X/91/126637-08$02.00/0

Copyright © 1991,AmericanSocietyforMicrobiology

Splice Site Skipping in Polyomavirus

Late

Pre-mRNA

Processing

YING LUOAND GORDONG. CARMICHAEL*

Department ofMicrobiology, University of ConnecticutHealth Center,

Farmington,

Connecticut 06030

Received 1 July 1991/Accepted 27 August 1991

Polyomaviruslate nuclearprimary transcripts contain tandemrepeatsof the late strand of the viralgenome,

as a result of inefficient transcription termination and polyadenylation. Pre-mRNA processing involves the

splicing ofshort noncoding late leaderexonstoeach other(removing genome-length introns)andthesplicing of the last leader to acoding bodyexon (such asforthemajor virion structural protein, VP1). Asa result,

cytoplasmicmRNAscontain 1to 12 tandem leaderexonsattheir5' ends thatarefollowed byasingle coding

exon. To understandmoreabout how polyomavirusexonsare spliced together, we studiedadouble-genome construct consisting oftwotandem but nonidentical polyomavirus late transcription units. The alternating leaderexonsaredistinguishable fromoneanother but retainidenticalflanking RNA-processing signals,asfor

thealternatingVP1exons.Wetransfected thisconstructandderivatives of it intomousecells anddetermined which leaderexons aresplicedtowhich others and whichVP1exons areutilized. Results showed that leader

exons arealmostneverskippedduring splicing andarespliced sequentiallytooneanother. On the other hand, VP1exons were often skipped, with the VP1 exonclosest to the polyadenylation site splicing tothe nearest upstream leaderexon. Splice site replacementexperiments showed that VP1 exon skipping isnot dueto a relativeweaknessofits3'splicesiteortoanysequenceupstreamof theVP13'splice site. Exon skipping is also not the result ofsequences within the VP1 exon. Rather, VP1 3' splice site skippingcan be eliminated by

replacing theinefficientlate polyadenylation signal withanefficientone, orbyinsertinga5'splicesite between the VPI3' splicesiteand thelatepolyadenylation site. Thus,sequencesthatcomposethedistal border of the VP1 exon caninfluence usageof theupstream3' splice site.

Polyomavirus late pre-mRNA molecules contain optional

splice sites and polyadenylation sites. Polyomavirus late

transcription termination andpolyadenylation are both

inef-ficient, leadingto theproduction ofaheterogeneous

collec-tionof nuclearprimarytranscripts, some of which are many

times the size ofthe viral genome (1-3, 37, 38). Figure 1 illustrates the current model for how one such giant pre-mRNA molecule is processed into a mature cytoplasmic messageforthemajor viral structural protein, VP1. At the 5' ends of lateprimary transcriptsisanoncoding exon, the late leader. Ingiant transcripts,this exon appearsmultiple times.

DuringRNA processing, leaders are spliced to each other,

skipping internal VP1 exons and removing genome-length

introns.Inaddition, for mVP1 molecules a body splice site is chosen to attach the final late leader exon to a terminal

coding exon. Thus, maturelate cytoplasmic messages

con-tain,attheir 5' ends, 1 to 12tandemcopies of the late leader

exon that arefollowed by a single coding exon (20, 21, 25,

36).The averagelate message contains threetofour tandem late leader units (36). Aninteresting featureof this system is that thesplice sites andpolyadenylation sites thatarechosen

are identicalto those thatareskipped.

Wehave beeninterested in understanding how exons are

spliced to each otherduring polyomavirus late pre-mRNA

processing. Tobeginto address thisissue, we have made a

series ofconstructs containing a duplicated viral genome, but with distinct late leader and VP1 exons. We have

focusedontheexpressionof the message for VP1, themajor capsidprotein,becauseitaccountsfor 80to85%of late viral message (21, 35). We show here that,following transfection ofour constructs into mouse cells, late leader exons were spliced sequentiallytoeach other inprimary transcriptsand

* Correspondingauthor.

were rarely skipped. VP1 exons, however, were often

skipped, with the 3'-terminal VP1 exon splicing to the nearest upstream leader. Splice site and polyadenylation signalreplacement experiments showedthat VP1 exon skip-ping isnotdue toarelatively weak 3' splice site that isoften

ignored. Rather,skippingof the VP13'splicesite isstrongly influencedbythe strength of the nearest downstream poly-adenylation site and by the presence of a downstream 5'

splice site. These results suggest that exon skipping in

polyomavirus isnormallyduetoinefficientrecognition ofthe

latepolyadenylation site andprovideinvivo supportfor the exondefinition modelrecently proposed byRobberson et al.

(32).

MATERIALSANDMETHODS

Materials. Restriction enzymes, T4 DNA ligase, DNA

polymerase I large fragment (Klenow enzyme), T4 DNA polymerase, and T4 polynucleotide kinase were from New EnglandBiolabs andwereusedassuggestedbythesupplier.

RNase T2 was from Bethesda Research Laboratories. [a-32P]dATP, [t-32P]UTP,[y-32P]ATP, and avian myeloblas-tosis virus reverse transcriptase were from Dupont, NEN

Research Products. Polyomavirus strain 59RA has been

describedpreviously (17, 18).This strain is very similarin its

nucleotide sequence to that published for strain A3 (14). Plasmidp43.25.67 iswild-type polyomavirus strain A2 with an XhoI linker at the PvuII site at nucleotide (nt) 5128, inserted at the BamHI site into pAT153 (39), and was generouslyprovided byA.Cowie andR. Kamen. All recom-binant plasmids used here were propagated in Escherichia

coliJM83.

Cell culture andtransfections. Mouse NIH 3T3 cells were obtained from the American Type Culture Collection. The techniquesfor their

propagation

and transfection havebeen 6637

on November 10, 2019 by guest

http://jvi.asm.org/

leader-leader splice leader-body splice

L <V L VP1

5' VXselAAzoeZZA_

LL VPI

_ AAAZZn2M-AAAAAAAA

FIG. 1. Model for the processing of polyomavirus late giant primary transcripts and the production of multiple, tandem, non-translated late leader unitsonmaturemRNAmolecules. Because of inefficienttermination by RNA polymeraseII,late viral transcripts are heterogeneous in length, withmany representing multiple cir-cuits of the viral genome. The example shown is for atranscript created bytwo passes through the viral late region. During pre-mRNAprocessing,aleader-body splice joins the 3'-most noncoding late leader exon (L) to a message-coding body (here mVP1). In addition, leader-leader splicing removes a genome-length intron, yieldingafinal mVP1messagewithtwotandem leader unitsatits 5' end.

3'ss 5'ss

I L

SLM

1-13

1-13 5'ss

3'ss

AP1

described elsewhere (10). Briefly, all transfections were

performed byamodification of the calcium phosphate

trans-fection procedure of Chen and Okayama (11). Cells were

seeded ata concentration of 106 cells per 150-mm plate in

Dulbecco's modified Eagle medium supplemented with 10% bovine calfserumplusL-glutamine and penicillin and strep-tomycin at 37°C in 5% CO2 approximately 48 h prior to

transfection. Sixteen hours prior to transfection, the cells

wereserumstimulatedby the addition of fresh medium. The

total amount of DNA per transfection was 40 p.g. Before transfection,recombinantplasmidswerecutwithEcoRIand diluteligated with T4 DNA ligase as described previously (6).

Harvesting of cytoplasmic RNA. Cellswereharvested44to 48 h after initiation of transfection. Each plate was rinsed

with lx phosphate-buffered saline-minus (PBS-) (without

Mg2+ and Ca2+), and RNA was harvested as described elsewhere (10). Briefly, cells were scraped into PBS- and

pelleted by centrifugationat1,000 x gfor3 min. Thepellet wasresuspendedinPBS- andcentrifuged againat1,000 xg

for 3min.The cellpelletwasresuspendedin 3 ml ofNonidet P-40 lysis buffer and left on ice for 10 min. Nuclei were

removed by centrifugation at 1,000 x g for 3 min. To the supernatant(the cytoplasmicRNAfraction)wasadded 1.42

g of guanidine isothiocyanate powder. RNA was separated from DNA by centrifugation through a5.7 M cesium chlo-ridecushion for 20hat110,000x g(12).ThepelletedRNAs were drained well and resuspended in 300

RI

of 10 mM

Tris-Cl (pH7.0)-i mMEDTA(pH 7.0)-0.2%sodium

dode-cyl sulfate and ethanol precipitated. RNAs were

resus-pended in100 Il of distilled H20.

Construction of plasmids. All cloning was done by

stan-dardprocedures.Thestructuresof themajor plasmidsused inthis study areshown schematically in Fig. 2.

(i) Double-genome constructs. The construction of plas-mids pl-13 and pl-16isdescribed in detail elsewhere (27a).

Constructs 1-13 and 1-16 containduplicated viral genomes,

but withdistinguishablelate leader and VP1exons. Figure2 illustrates schematicallythe intron-exon structuresof these constructs. Plasmidpl-135'ss is thesameaspl-13,excepta

late leader 5'splicesitewasinserteddownstream of the VP1

3'ss 5'ss

WT

3'ss

I vp1' _M

man _ v zzzgzj</ "AA I^ |AAsmmn s

5'ss

son _ r //

//////AA,,,,,o....

1-16 son m Ui;///z7AA~§AA.

L-LVP1 ..

FIG. 2. Schematic diagram of the major constructs used in this study. Constructs were made as described in Materials and Methods. Before transfection, each was digested with EcoRI, which liberated the polyomavirus sequences from the plasmid backbone, and recircularized with T4 DNA ligase. 1-13 contains two copies of the viral late region, but the first copy has a substituted late leader exon sequence(SLM[6]), while the second has a truncated VP1-coding exon (VP1'). 1-135'ss is thesame as1-13,exceptthe late leader 5' splice site has beeninsertedinto the coding region of VP1, 80 nt downstream of the VP1 3' splice site.1-16is the same as1-13,except fordeletion ofsequencesincludingoneintact early coding region, replication origin, late promoter, and late leader exon. L-LVP1 (8) is a single-leader mVP1cDNAwhichalsocontains an upstream single-leader exon (flanked by its normal splice sites) and an intron of 165 nt. The important splice sitesin theseconstructs are indicated. There is a minor 3' splice site (not shown) between the late leader and the VP1 3' splice site for theproductionofmessages encoding VP3. AA, polyadenylation signal AAUAAA.

on November 10, 2019 by guest

http://jvi.asm.org/

[image:2.612.56.283.75.353.2] [image:2.612.142.459.504.645.2]

3' splice site. To do this, we inserted an 87-bp fragment into the EcoRV site (14 bp downstream of the VP1 3' splice site). This fragment contained 44 bp upstream of the mutant SLMY (6) 5' splice site and 15 bp downstream of the leader 5' splice site. A 14-bp linker was added at both ends, giving anXhoIsite upstream and a ClaI site downstream of the new 5' splice site. In this 81-bp fragment, only 21 bp derive from the wild-type polyomavirus genome, 6 bp upstream and 15 bp downstream of the leader 5' splice site.

(ii) Construct L-LVP1. Plasmid L-LVP1 was made by inserting a 165-bp fragment containing the late leader exon and flanking sequences (including the late promoter just upstream of the late leader 3' splice site) into a genome in which the VP1 intron had been precisely removed by oligo-nucleotide-directed mutagenesis. This construct was orig-inally made for other experiments in the laboratory and has been described in detail elsewhere (8).

PCR assays. Oligonucleotide primers were made by using a Milligen/Biosearch Cyclone DNA Synthesizer and the phosphoramidite chemistry.Oligonucleotide 275 (5'-TATCA CCGTACAGCCTTG) anneals across the VP1-late leader splice junction, with the terminal 3 bases annealing to the late leader and the rest annealing to the VP1-coding exon. Oligonucleotide 278 (5'-CCTGACATTTTCTATTTTAAG)

anneals to late cDNA molecules immediately upstream of the late leader exon. Oligonucleotide PE (5'-CTATGACT GTTGCCC) anneals to late mRNA just inside the VP1 exon. Oligonucleotide 279 (5'-TCCTAGATGAAAATGGAG) an-neals to the truncated VP1' exon but not to the wild-type VP1 exon.

The polymerase chain reaction (PCR) (34) was performed following reverse transcription of the RNA.Oligonucleotide 275, 279, or PE was added to the RNA in a buffer recom-mended by Perkin-Elmer Cetus. AMV reversetranscriptase (22 U) was added and allowed to incubate for 1 h at 45°C. TaqI DNA polymerase and oligonucleotide 278 were then added, and PCR continued for 30 cycles. Each cycle had three sections. The first was a 2-min incubation at 92°C. After a 30-s ramp time, the incubation was continued for 1

min at 45°C. Finally, after a 0.5-min ramp time to 72°C, incubation wascontinued for 1.5min. A 10-min incubation at 72°C followed the final cycle. Reaction products were re-solved on 6% polyacrylamide-urea gels, and bands were detected byautoradiography. It is important to note thatthe PCRamplification of RNA performed in ouranalyses accu-rately and reliably reflected the RNA amounts in our sam-ples. Suchverification has previously been presentedby our laboratory (20).

RESULTS

Constructs and experimental design. We constructed a

series of plasmids that contain duplicatedviral genomes, but withdistinguishable late leader and VP1 exons. Each of the constructs used contains at least one intact early coding

region and DNA replication origin, so each produces early

viral proteins and replicates in transfected mouse cells. Figure 2 illustrates schematically the intron-exon structures of the constructs used in the work described here. The primary construct is 1-13: it has two distinguishable leader exons and two distinguishable VP1 exons. The wild-type

(WT) and substituted (SLM) late leaders are of different sequence and differ by 6 nucleotides in length. We have shownpreviously that a virus with the SLM leader used here grows like wild type and is notdefectivein latepre-mRNA

processing (4, 6). TheVP1' exon contains a deletion within

the

wild-type

VP1 exon and is 81 nucleotides

long.

The

splice

sites

bordering

the

wild-type

and SLM leader exons

are

identical,

as arethe sequences

flanking

theVP1andVP1' exons.Thisconstruct was

designed

such thatafter transfec-tion into mouse

cells,

primary

transcripts

with

alternating

wild-type and SLM leader exons and

alternating

VP1 and

VP1' exons are

produced.

Since late promoter

activity

requires

only

a very small

region

just

upstream ofthe late leaderexon

(10,

33a),

late

transcription

from1-13caninitiate upstream of either the SLM or

wild-type

leader exon.

Construct I-135'ssis thesame as I-13 exceptthata5'

splice

site

(derived

fromthe late

leader)

has been inserted into the VP1 exon, 80 nucleotides downstream ofthe VP1 3'

splice

site. Construct1-16is relatedto1-13 butcarriesadeletion of

the second

(wild-type)

late leader

region.

L-LVP1 is a

genome

containing

alateleaderunitupstream ofa

prespliced

leader-VP1 cDNA.This constructmimicswild

type

butwith the VP1 3'

splice

site

replaced

witha leader3'

splice

site.

For all

experiments,

polyomavirus

single-

or

double-ge-nome inserts from these constructs were excised from the parent vector backbones and recircularized

by

dilute

liga-tion. Transfectionofthese recombinant viral genomes into

permissive

mouseNIH 3T3 cells mimics the infection proc-ess: late viral gene

expression

proceeds

in a manner that

produces

giant

primary

transcripts.

Leaderexons are

spliced

sequentially

tooneanother in

giant

primary

transcripts.

Because mature mVP1 molecules can containupto12 tandemleader unitsattheir5' ends

(20),

we

asked how leader exons are chosen in the late

pre-mRNA

processing

pathway.

Can leader exons be

skipped during

leader-to-leader

splicing

in the same way that VP1 exons appear to be

skipped,

or are

multiple

leaders in

giant

transcripts

spliced

sequentially

to one another? This was examined

by

a PCR assay which has been shown

by

usto

accurately

and

reliably

reflect the initialamount ofRNA in our

samples

(34).

Briefly,

an

oligonucleotide

complementary

to a

leader-to-body

splice

junction (275)

was annealed to

cytoplasmic

RNA isolated from transfected cells and

ex-tended

by

reverse

transcription.

Thena

32P-labeled

oligonu-cleotide

(278)

which binds to the 5' end of the reverse

transcript

was

added,

and the PCRwasused to

amplify

the

products.

The labeled

oligonucleotide

anneals

immediately

upstream of the late leader exon. The sizes of the bands

observed after

gel

electrophoresis

reveal the distribution of

leaders at the 5' ends ofthe messages

(20).

Figure

3 shows theresultsofsuchaPCR

experiment

using

oligonucleotide

275,

which binds to the

junction

of the leader-to-VP1

splice

as wellas tothe

junction

of the leader-to-VP1'

splice.

Inthis

experiment,

messages

containing

one

andtwotandem leader unitswereobserved for wildtypeand constructsI-13and1-16.

Wild-type

viral RNAwasused here

to

provide

size markers forone and two tandem

wild-type

leader units

(each

unit is 57

nt).

1-16

only

encodes SLM

leaders,

which are 6 nucleotides shorter than

wild-type

leaders and which can be used as additional size markers.

Thisaccountedfor thealteredleader ladderobserved forthis construct and allowed its use for size markers. Because

therearetwolatepromotersin

1-13,

we

expected

toseeboth

wild-type

and SLM leader

signals

attheone-leader level for thisconstruct. This was observed. At the two-leader

level,

however,

the

major

band seen was intermediate in size

compared

with two

wild-type

leaders or two SLM

leaders,

indicating

that most messages contained one

wild-type

leader exon

spliced

to one SLM leader exon. If

wild-type

leaderswere

spliced

together

(skipping

SLM

leaders)

in1-13 RNA

processing,

or if SLM leaders were

spliced

together

on November 10, 2019 by guest

http://jvi.asm.org/

275

VP1orVP1'

278

0

Pt

A 2 __.",--

1-..--leader structures

1 zi.

SLM

B

1-13 K .AAA A

1-16 A.

FIG. 3. PCR assay to determine whether leader exons splice

sequentiallytoeach other. Theoligonucleotides used for PCR (275 and 278) are shown at the top. The assay was performed as described in thetext, usingcytoplasmic RNA isolated from mock-transfected cells and cells mock-transfected with religated DNAs froma

wild-type viralgenome(WT),1-13,or1-16.Reaction productswere resolved on a 6% polyacrylamide-7 M urea gel and detected by autoradiography.Quantitationwasperformed withaBetagen

Beta-scope 603 Blot Analyzer. The expected positions of leader

struc-tures are shown on the left. Black boxes represent SLM leader units, which are 6 nt shorter than wild-type (WT) leader exons, shown as white boxes. At the bottom is an illustration ofagiant primary transcript, with sequential leader splicing indicated.

(skipping wild-type leaders),then bands wouldappearatthe

positions of molecules with two wild-type or two SLM

leaders. Such material accounted for less than 2% of the signal as determined with a Betagen Betascope Blot

Ana-lyzer, leadingustoconclude that the 1-13 leaderexonsmust

besplicing inasequentialfashion. Results atthethree-and

four-leader positions (data not shown) are also consistent with sequential leader splicing, but those results are more complex, asa result of the multiplepromoters that exist in

our constructs. Thus, although VP1 exons are frequently

skipped during leader-to-leader splicing, late leader exons are spliced sequentially to one another and are rarely skipped.

VP1exons arespliced tothenearest upstreamleaderexon.

Although leader-to-leader splicing does not involve leader exon skipping, the possibility remained that leader exons could be skippedduring leader-to-bodyexon splicing. Since we know the SLM leader isupstreamof the VP1 exon and

[image:4.612.321.531.73.363.2]

the wild-type leader precedes the VP1' exon in primary transcripts from construct 1-13, then by measuring which leader is splicedto which exon, we can deduce whether a

FIG. 4. (A) PCR analysis of VP1 and VP1' exon splicing in constructs1-13and 1-16. Theanalysiswasasdescribed in thelegend

toFig. 3, exceptoligonucleotides 278 and279were used for VP1' spliced products, and oligonucleotides278 and PE were used for VP1 splicedproducts. Oligonucleotide 279only binds tothe VP1'

exon,allowingustoanalyzethesplicingof thisexonindependently

ofthesplicingofthewild-type (WT)VP1exon,whilePEbindsonly

totheVP1exon.Theinterpretationof leaderstructuresisshownto

the right. (B) Schematic representation of some of the spliced products inferredfrom the datashown in panelA. Construct 1-13 containstwolatepromoters, eachimmediatelyupstream ofaleader

exon. Only products that derive from the upstreampromoter are

shown.Since the 1-13genomeiscircularly permuted,the constructs derivingfrom the second promoterareanalogous.

VP1 or a VP1' exon is only spliced to the leader exon

immediately upstream of it. The PCR technique allowed such an analysis andwas done forboth the VP1 and VP1' exons. Atypicalresult is shown inFig. 4A. Wild type(WT)

and1-16wereused for sizemarkers,aseachcontainsonlya

single type ofleader exon (SLM). At the one-leaderlevel, almostall thesignalobserved for1-13wasattheSLM leader

position when using a VP1-specific probe and at the wild-type position when using a VP1'-specific probe. This indi-catesthat in thisconstruct, most VP1 orVP1' splicingisto the immediately upstream leader exon. Again, the leaders werenotskipped during pre-mRNA splicing, while the VP1 and VP1' exons were both skipped. Since the VP1' exon retainsonly81 ntderivingfrom thewild-typeVP1 exon [14

ntadjacenttothe VP1 3'splice siteand 67ntjustupstream of the poly(A) site], exon skipping is not likely to be the result ofsequences within the late viral coding region that inhibit theuseof theVP1 3'splice site.Wecannotexclude,

however, that thesesequencesmightsomehow influence the useof the VP1 3'splicesite.Figure4Billustratessomeof the

splicing eventsdemonstratedby thesedata.

leader

structures

E==

-=

2 78

C') co

0 v

--

-WT =

SLM _

= WI

_ AIM

gp

-x

L) Cl) co

0 v-

v-.7 -1 -1

--II.-

-

I--I

SL WT .. SLM

-7 Em"M AA

on November 10, 2019 by guest

http://jvi.asm.org/

[image:4.612.72.269.74.372.2]

Exons are not skipped because the VP1 3' splice site is weaker than the late leader 3' splice site. Figure 4A also shows that leader splicing in1-16can occur toeither the VP1 or the VP1' exon, indicating that exon skipping does not have to involve splicing only toa downstream leader exon. This construct contains only an SLM leader, which is upstream of a VP1 exon. The distal VP1' exon has no leader immediately upstream of it. As seen in Fig. 4A,however, the VP1' exon can still be spliced efficiently to the SLM leader exon, following skipping of the upstreamVP1 exon, indicat-ing that VP1 exon skippindicat-ing does not result from the presence of a downstream leader exon and is therefore not likely to be the result of especially strong leader splice sites that out-compete the VP1 3' splice site.

A possible explanation for VP1 exon skipping in polyoma-virus would be that the VP1 exon has an inherently ineffi-cient 3' splice site which is poorlyrecognized by the splicing machinery of the cell. To determine whether this model is correct, we analyzed construct L-LVP1. In L-LVP1, the VP1 3' splice site has been replaced with the late leader 3' splice site. Thus, a 3' splice site that is frequently skipped has been replaced with one that is rarely skipped. In addi-tion, this construct lacks almost all viral sequences that normally exist upstream of the VP1 exon. Cytoplasmic RNAs from L-LVP1 and wild-type polyomavirus were sub-jected to PCR analysis, and theresults are shown in Fig. 5. Bands at the one-, two-, and three-leader positions are shown. Clearly, L-LVP1 generated late messages with mul-tiple leader structures, so the modified VP1 exon of this construct must be skipped. Theimportantresult here is that the ratio of mRNAs with two leaders to those with three leaders for L-LVP1 is the same as the ratio ofmRNAs with one leader to those with two leaders for wild type. Quanti-tation of these results with a Betagen Betascope Blot Ana-lyzer revealed that in this experiment the frequency of exon skipping for wild type was about 40%, while the frequency for L-LVP1 was about 25%. In other experiments, the frequency of splice site skipping in wild type and L-LVP1 has been virtually identical. Weconclude from these exper-iments thatVP1 skipping does not occur because the VP1 3' splice site is weaker than the late leader 3' splice site, or because sequencesupstream of the VP1 3' splice site inhibit its use.

Late polyadenylation signal strength affects exon skipping.

Since VP1 exon skipping appears not to result from

rela-tively strong or weak splice sites or from intron or exon

sequences flanking the VP1 3' splice site, we asked whether

skipping might be influenced by the late polyadenylation

signal.We thereforeanalyzed a viral variant, ins5, provided

by N. Acheson. This mutantcontains two tandem polyade-nylation sites in the late transcription unit: one is the wild-type late polyadenylation site, and the other, just up-stream, is adifferent, moreefficient rabbit 3-globin polyade-nylation site (23). Wild-type or mutant viral genomes were

transfected into mouse cells as in earlier experiments, and the resultant cytoplasmic RNA was subjected to a leader

PCR assay. The results are shown in Fig. 6. As expected,

wildtypeproduced alargeproportion of VP1 messages with

multiple leaders. In contrast, however, mutant ins5 pro-ducedonly afew latemVP1 molecules which had more than

a singleleader, demonstrating that an efficient

polyadenyla-tion signal can suppress exon skipping. Quantitation with a Betagen Betascope Blot Analyzer revealed that ins5

pro-ducesabout95% single-leadermolecules in this assay. These results for ins5 are in good agreement with those reported

0 0. ..4

:^ AhI2~AA

.6

M 4

.

4b

L-LVP1

JflE2AA

AA

WT WT VP1

m I A

FIG. 5. PCRassay to determinewhether VP1 exon skipping is dueto arelativelyweak VP1 3' splice site or to a relatively strong leader 3' splice site. Cells were transfected with religated DNAs fromwild-typeconstruct pPyBS (Py)orplasmidL-LVP1.

Cytoplas-mic RNA was isolated44 hlater and subjected toPCRanalysis as

describedin thelegend toFig. 3.The positions expected formVP1 molecules with one, two, or three tandem leader units areindicated, and splicing events that generate the structures observed for L-LVP1 are shown to the right of the bands. Band intensities were determined withaBetagen Betascope BlotAnalyzer. The mocklane here shows an artifactual bandjust below the position of

single-leader mVP1 molecules. WT, wild type.

earlierby Lanoix and Acheson (23) forthe polyadenylation efficiency of this mutant.

A 5' splice site within the VP1 exon can prevent exon

skipping. The suppression of exon skipping by an efficient polyadenylation signalisdifficult tointerpret. Efficient

poly-adenylation is usually associated with efficient transcription

termination, as has been reported even for mutant ins5(23). Thus, the results shown in Fig. 6 might be the trivialresult of limited splicing choices in short transcripts. However, VP1 exon skipping could also be suppressed by another type of mutation. Construct 1-13 5'ss is identical to 1-13, except a

small DNAfragment comprising the late leader 5' splice site has been inserted downstream of the VP1 3' splice site. PCR analysis witholigonucleotide279 (specificforthe VP1'exon)

was used to examine exon skippingfrequency in this mutant, and theresults are shown inFig. 7. Strikingly, the VP1exon

in this construct is now almostnever skipped, asevidenced by the virtual absence of a band representing two tandem leaders spliced to the VP1' exon. Rather, the SLM leader is spliced efficiently to the VP1 3' splice site, and the inserted 5' splice site is used to splice to the downstream wild-type leader exon (top band in the I-13 5'ss lane). As would be expected, the VP1' exon is still efficiently skipped (data not

shown).

on November 10, 2019 by guest

http://jvi.asm.org/

[image:5.612.347.540.77.358.2]

3 leaders

2leaders

1 leader

FIG. 6. Effect of poly(A) site strengthon VP1 exon skipping. Cells were mock transfected or transfected with religated DNAs from pPyBS (wildtype)orins5. CytoplasmicRNAwasisolated 44

h later and subjectedtoPCRanalysisasdescribed in the legendto

Fig. 3.The positionsexpected for mVP1 molecules withone,two,

orthreetandem leaderunitsareindicated.

Takentogether withtheabove data,weconclude thatVP1

exon skipping isalmostcertainlynotduetoaweak3' splice site orto inhibitory flanking sequences. Rather, use of the

VP1 3'splice siteappearstobe directlycoupledtotheuseof adownstreamRNAprocessing signal,eithera polyadenyla-tion signal or a 5' splice site.

DISCUSSION

Polyomavirus late pre-mRNAprocessing is unusual. Late viral nuclear primary transcripts containtandem repeats of

(0

LI)

C _ 1~ am

1-13 - - AAA m___AA

.--SL M WT VP<.

1-135'ss . AA rl ~AA--A

FIG. 7. Effect of a downstream 5' splice site on VP1 exon

skipping.Cells were mocktransfected or transfected withreligated

DNAs from I-13or I-135'ss. Cytoplasmic RNA wasisolated 44 h

laterandsubjectedto PCRanalysisasdescribed in thelegendtoFig.

3,using oligonucleotides278 and279,the latter of which is specific

for the VP1'exon.Interpretationsof theproductsare shownbeside

thelanes. WT,wildtype.

the late strandof the viral genome.

During

processing,

short

noncoding 5' late leader exons are

spliced

to each

other,

removinggenome-length introns which containunused

poly-adenylationsitesand 3'

splice

sites. We have been interested in understanding how this

processing

is

accomplished

and how splice sites and

polyadenylation

sites are

skipped.

In

thisstudy,weshowed thatleaderexons are

spliced

sequen-tially to one another in

giant

primary

transcripts and,

like most internal exons in known pre-mRNA

molecules,

are

rarely skipped during splicing (Fig. 3 and 4). On the other

hand,

VP1-coding

exons areoften

skipped.

Hereweshowed that whenaVP1 3'

splice

site isselectedforuse,it is

spliced

to thenearest upstreamlate leader unit

(Fig. 4).

HowareVP13'splicesites

skipped?

Aslate leaderexons appear to be

spliced

sequentially

to one another in

giant

primarytranscripts, it is reasonableto assumethat the 5' and 3' splice sites

flanking

the leader are

efficiently

selected in vivo. This ledus toconsider that the VP1 3'

splice

site

might

be recognized less

efficiently

(compared with the leader 3'

splicesite), leadingtofrequentexon

skipping.

This

possibil-ity was ruled out,

however,

when we could

effectively

replacethewild-typeVP1 3'

splice

site withthe late leader 3'

splice siteand still observe VP1 exon

skipping

atthe same

frequency as for wild type

(Fig.

5).

Similarly,

when the

downstream leader exon was

deleted,

as inconstruct

1-16,

VP1exons were still

skipped, by

leader

splicing

to adistant downstream VP1 3'

splice

site

(Fig. 4).

These results indi-cated that intron sequences upstream of the VP1 3'

splice

site and VP1exonsequences

likely play

no

important

role in theobservedexon

skipping.

Since we were unable to affect VP1 exon

skipping by

exchanging splice sites,

we

compared

skipping

in several constructsthatdifferedintheirlate

polyadenylation

signals.

Mutant ins5 was constructed and

analyzed

previously

by

Lanoix and Acheson(23) andcontainstwo

polyadenylation

signals

in the late

region,

one

being

the

wild-type

polyoma-virus signal and the other

being

from the rabbit

P-globin

gene.In agreementwith the results

reported

by

Lanoixand Acheson (23), wefound that mostlate messages from ins5 have asinglelate leaderunit at their 5' ends

(Fig. 6). Thus,

VP1 exons are

rarely

skipped

inthismutant.

Normally,

polyomavirus

late

transcription

termination is inefficient (38). It is

possible

that the

fragments

inserted to

provide

additional

polyadenylation signals

in mutant ins5 contain

transcription

termination

signals

aswell.Whilemore

experiments

areneededto

rigorously

rule outthis

possibil-ity, the sequences added in this construct are

short,

andan

efficient

polyadenylation

signal

isnotsufficient for

transcrip-tion termination.

Transcription

termination

requires

both

polyadenylation

and

other,

downstream sequences

(13, 23,

27, 40). Levittetal.(26) demonstrated

that,

whenit is

placed

withinan

intron,

anefficient

synthetic

polyadenylation

signal

closelyrelated totherabbit

,3-globin

polyadenylation signal

used

in

ins5 still does not prevent

transcription reading

through until the end of the intron several hundred bases

away.

The suppression ofexon

skipping

seen withmutant ins5

couldbe mimickedby

placing

a5'

splice

site downstream of

the VP13' splice site andbefore the inefficient late

polyade-nylation site (Fig. 7). This

again

demonstrated

clearly

that

exon

skipping

could beaffected

by

changing

the

signals

that compose thedistalborderofthisexon. Inother worknowin progress, we areplacinga5'

splice

site atvarious

positions

within the VP1 exon to examine the effect ofexon size on

splice site skippingin this system.

Howcould

polyomavirus

lateexon

skipping

resultfroman

on November 10, 2019 by guest

http://jvi.asm.org/

[image:6.612.134.222.78.243.2] [image:6.612.97.263.457.660.2]

inefficient late polyadenylation site? Our working model, consistent with the results presented here, is that there must be coordinate recognition and use of the 3' splice site and polyadenylation signal that demarcate the borders of the terminal coding exon of pre-mVP1 molecules. In this model, recognition of a polyadenylation site for use is normally associated with the use of the upstream VP1 3' splice site-neither site is used unless both are used. In giant polyomavirus late primary transcripts, the polyadenylation site is inefficiently recognized; when this polyadenylation signal is not used, the upstream VP1 3' splice site is also

ignored. This model is also consistent with work reported elsewhere (27a), in which we examined 3'-end formation in a number of polyomavirus constructs including the double-genome constructs used in the study described here. In that work we showed that, when the wild-type late polyadenyla-tion signal was duplicated, the signal closest to the VP1 3' splice site was normally chosen, even though transcription proceeded past the downstream site, allowing the efficient accumulation of messages with multiple leaders. However, insertion of a 3' splice site just upstream of the second polyadenylation site allowed the use of both polyadenylation signals. Also, deletion of a late polyadenylation signal in a double-genome construct did not interfere with use of the upstream VP1 3' splice site, showing that VP1 3' splice site choice is independent of polyadenylation site choice. In other words, VP1 3' splice site usage, while normally cou-pled to polyadenylation site use, can still occur if the VP1 exon is placed in a different context.

Our findings lend in vivo support for the exon definition model of splice site selection proposed recently by Robber-son et al. (32). In that model, both boundaries of an exon must be recognized, ordefined, before RNAprocessing can occur at either one. Thus, forinternal exons, a 3' splice site and a downstream 5' splice site must both berecognized; for terminal exons, apolyadenylation signal could replace a 5'

splice site. For a terminal exon, if thepolyadenylation signal was not recognized, then the terminal exon would not be

properly defined, and the VP1 3' splice site could not be

used. Although one might envision regulation of VP1 exon skipping as occurring at either end of thisterminal exon, i.e., through an inefficient VP1 3' splice site or through an inefficient late polyadenylation site, the data presented here strongly suggest that the exon skipping observed in polyo-mavirus is due to the inefficiency of the latepolyadenylation

site.

The modelpresented for the coordinate choice of a splice site and apolyadenylation site isconsistent with a number of published observations on alternative splice site and poly-adenylation site choice in naturally occurring complex tran-scription units. In such systems, there isgrowing evidence that polyadenylation and splicing are not independent pro-cesses, i.e., regulation at either end of a terminal exon can affect exon skipping or polyadenylation site choice. In one example of cell-specific regulation, studies on the switch between secreted and membrane forms of immunoglobulin

M ,u heavy chain have suggested that this switch isregulated at the level ofpolyadenylation site choice (15, 19, 22, 28, 33). In the immunoglobulin M system, choice of alternative polyadenylation signals leads to altered splicing pathways, and there may be competition between splicing and poly-adenylation (30, 31). Incontrast, the expression ofcalcitonin and the calcitonin gene-related peptide, although controlled in acell-specific manner from a complex transcription unit containing two polyadenylation sites, appears to be regu-lated at the step of splice siterecognition, which in turn leads

totheuseofthenearestdownstreampolyadenylationsignal (7, 16, 24).

Other results froma number ofrecent studies have been

interpreted as demonstrating that splicing dominates over polyadenylation. Ifanefficientpolyadenylationsite is

placed

within an intron, then itcan be skipped; if the same siteis

placed within an exon, or if surrounding splice sites are

mutated, then it is used (5, 9, 26). Niwa et al. (29) have

suggestedthatplacementofapolyadenylation site withinan

introninactivates polyadenylation notbecause the polyade-nylation machinery cannot compete with the splicing

ma-chinery, but because it cannotinteract with it, reflecting an

underlying polarity of pre-mRNA processing signals. Our

results are consistent with this view. ACKNOWLEDGMENTS

We thank N. Achesonforproviding mutant ins5 and N. Barrett, E. Carmichael, R. Hyde-DeRuyscher, and A. Roome for helpful

comments onthismanuscript.

This work was supported by grantCA 45382 from the National CancerInstitute.

REFERENCES

1. Acheson, N. 1976. Transcription during productive infection with polyomavirusandSV40. Cell 8:1-12.

2. Acheson, N. 1984. Kinetics andefficiencyofpolyadenylation of late polyomavirus nuclear RNA: generation of oligomeric poly-adenylated RNAs and theirprocessing intomRNA. Mol. Cell. Biol. 4:722-729.

3. Acheson, N. H. 1978. Polyoma giant RNAs contain tandem repeats ofthe nucleotide sequence of the entire viral genome. Proc. Natl. Acad. Sci. USA 75:4754-4758.

4. Adami, G., and G. G. Carmichael. 1987. Thelength butnotthe sequenceof thepolyoma virus late leaderexonisimportant for both late RNA splicing and stability. Nucleic Acids Res. 15: 2593-2610.

5. Adami, G., and J. R. Nevins. 1988. Splice site selection domi-nates over poly(A)site choice in RNA production fromcomplex adenovirus transcription units. EMBO J. 7:2107-2116. 6. Adami, G. R., and G. G. Carmichael. 1986. Polyomavirus late

leader region serves an essential spacerfunction necessary for viability and late gene expression. J. Virol. 58:417-425. 7. Amara, S. G., R. M. Evans, and M. G. Rosenfeld. 1984.

Calcitonin/calcitonin gene-related peptide transcription unit: tissue-specific expression involves selective use ofalternative polyadenylation sites. Mol. Cell. Biol. 4:2151-2160.

8. Barrett, N. L., G. G. Carmichael, andY. Luo. 1991. Splice site requirement for theefficientaccumulationofpolyoma virus late mRNAs. Nucleic Acids Res. 19:3011-3017.

9. Brady, H. A., and W. S. M. Wold. 1988. Competition between splicing and polyadenylation reactions determines which ade-novirus region E3 mRNAs are synthesized. Mol. Cell. Biol. 8:3291-3297.

10. Cahill, K. B., andG. G. Carmichael. 1989. Deletion analysis of the polyomaviruslate promoter: evidenceforbothpositive and negative elements in the absence of early proteins. J. Virol. 63:3634-3642.

11. Chen, C., and H. Okayama. 1987. High efficiency transforma-tion of mammalian cells by plasmid DNA. Mol. Cell. Biol. 7:2745-2752.

12. Chirgwin, J. M., A. E. Przybyla, R. J. MacDonald, and W.J. Rutter. 1979. Isolation of biologically active ribonucleic acid from sources enriched in ribonuclease. Biochemistry 18:5294-5299.

13. Connelly, S., and J. L. Manley. 1988. A functional mRNA polyadenylation signal is requiredfortranscription termination by RNA polymeraseII. Genes Dev. 2:440-452.

14. Deininger, P. L., A. Esty, P. LaPorte, H. Hsu, and T. Fried-mann. 1980. The nucleotide sequence and restriction enzyme sites of the polyoma genome. Nucleic Acids Res. 8:855-860. 15. Early, P., J. Rogers, M. Davis, K. Calame, M. Bond, R. Wall,

on November 10, 2019 by guest

http://jvi.asm.org/

and L. Hood. 1980.Two mRNAs can beproduced from asingle immunoglobulin gene by alternative RNA processing pathways. Cell20:313-319.

16. Emeson, R. B., F. Hedjran, J. M. Yeakley, J. W. Guise, and M.G.Rosenfeld. 1989. Alternativeproduction of calcitonin and CGRP mRNA is regulated at the calcitonin-specific splice acceptor. Nature(London)341:76-80.

17. Feunteun, J., L. Sompayrac, M. Fluck, and T. L. Benjamin. 1976. Localization of gene functions in polyoma virus DNA. Proc. Natl. Acad. Sci. USA68:283-288.

18. Freund, R., G. Mandel, G. G. Carmichael, J. P. Barncastle, C. J. Dawe, and T. L. Benjamin. 1987. Polyomavirus tumor induction in mice: influences of viral coding and noncoding sequences on tumorprofiles.J. Virol.61:2232-2239.

19. Galli,G., J. W. Guise, M. A.McDevitt, P. W. Tucker, and J. R. Nevins.1987.Relative positionand strengthsofpoly(A)sitesas well as transcription termination are critical to membrane ver-sus secreted >.-chain expression during B-cell development. GenesDev. 1:471-481.

20. Hyde-DeRuyscher, R. P., and G. G. Carmichael. 1990. Poly-omavirus late pre-mRNA processing: DNA-replication-associ-ated changes in leader exon multiplicity suggest a role for leader-to-leader splicing in the early-late switch. J. Virol. 64: 5823-5832.

21. Kamen, R., J. Favaloro, and J.Parker. 1980. Topography of the three late mRNAs of polyoma virus which encode the virion proteins. J. Virol. 33:637-651.

22. Kemp, D. J., G. Morahan, A. F. Cowan, and A. W. Harris. 1983. Production of RNA for secreted immunoglobulin chains does notrequire transcriptiontermination 5' to the M exons. Nature (London)301:84-86.

23. Lanoix, J., and N. Acheson. 1988. Arabbit ,-globin polyadenyl-ation signal directs efficient termination of transcription of polyomavirusDNA. EMBO J. 7:2515-2522.

24. Leff, S. E., R. M. Evans, and M. G. Rosenfeld. 1987. Splice commitment dictatesneuron-specific alternativeRNA process-ingincalcitonin/CGRPgeneexpression. Cell 48:517-524. 25. Legon, S., A. J. Flavell, A. Cowie, and R. Kamen. 1979.

Amplification in the leader sequence of late polyoma virus mRNAs. Cell 16:373-388.

26. Levitt, N., D. Briggs, A. Gil, and N. J. Proudfoot. 1989. Definition ofan efficient synthetic poly(A) site. Genes Dev. 3:1019-1025.

27. Logan, J., E.Falck-Pedersen,J. E. Darnell, and T. Shenk. 1987. Apoly(A) addition siteand a downstream termination region are required for efficient cessation oftranscription by RNA poly-merase II inthe mouse Pmaj-globin gene. Proc. Natl. Acad. Sci. USA84:8306-8310.

27a.Luo, Y., and G. Carmichael. 1991. Splice site choice in a complextranscription unitcontaining multipleinefficient

poly-adenylation signals. Mol. Cell. Biol. 11:5291-5300.

28. Mather, E. L., K. J. Nelson, J. Haimovich, and R. P. Perry. 1984.Mode ofregulationofimmunoglobulinmuand delta chain expressionvariesduring B-lymphocytematuration. Cell 36:329-338.

29. Niwa, M., S. D. Rose, and S.M.Berget. 1990. In vitro polyade-nylation is stimulated by the presence ofanupstream intron. GenesDev.4:1552-1559.

30. Peterson, M. L., E. R. Gimmi, and R. P. Perry. 1991. The developmentallyregulated shift frommembranetosecreted ,um RNA production is accompanied by an increase in cleavage-polyadenylation efficiencybutnomeasurablechangeinsplicing efficiency. Mol. Cell.Biol. 11:2324-2327.

31. Peterson, M. L., and R. P. Perry. 1986.Regulation ofproduction ofALm and >L mRNArequires linkage ofthe poly(A) addition sites and isdependentonthelength ofthe

.Us.Um

intron. Proc. Nati. Acad. Sci. USA83:8883-8887.

32. Robberson, B. L., G. J. Cote, and S. M. Berget. 1990. Exon definition may facilitate splice site selection in RNAs with multiple exons. Mol. Cell. Biol. 10:84-94.

33. Rogers, J., N. Fasel,and R. Wall.1986. A novel RNA in which the 5' end is generated by cleavage at the poly(A) site of immunoglobulin heavy-chain secreted mRNA.Mol. Cell. Biol. 6:4749-4752.

33a.Roome, A., and G. Carmichael.Unpublisheddata.

34. Saiki,R. K., S. Scharf, F. Faloona, K. B. Mullis, G. T.Horn, H. A. Erlich, and N. Arnheim.1985.Enzymaticamplificationof P-globin genomic sequences and restriction site analysis for diagnosisof sicklecellanemia. Science 230:1350-1354. 35. Tooze, J.,ed.1980. Molecularbiologyoftumorviruses, 2nd ed.

DNA tumor viruses. Cold Spring Harbor Laboratory, Cold SpringHarbor, N.Y.

36. Treisman, R. 1980. Characterization ofpolyoma late mRNA leader sequences by molecular cloning and DNA sequence analysis. NucleicAcids Res. 8:4867-4888.

37. Treisman, R., and R. Kamen.1981. Structure ofpolyomavirus late nuclear RNA. J. Mol. Biol. 148:273-301.

38. Tseng, R. W., and N. H. Acheson. 1986. Use ofa novel S1 nuclease RNA-mapping technique to measure efficiency of transcription termination on polyomavirus DNA. Mol. Cell. Biol. 6:1624-1632.

39. Tyndall, C., G. LaMantia, C. Thacker, and R. Kamen. 1981. A region ofthe polyoma virus genome between the replication origin andlateprotein codingsequences isrequired in cis for bothearlygeneexpression and viral DNAreplication. Nucleic AcidsRes. 9:6231-6250.

40. Whitelaw, E., and N. J. Proudfoot. 1986.a-Thalassaemia caused byapoly(A) site mutation revealsthattranscriptional termina-tion is linked to 3'endprocessing inthe human 4globingene. EMBO J. 5:2915-2922.

on November 10, 2019 by guest

http://jvi.asm.org/