Two Regions Downstream of AATAAA in the Human Antithrombin ...

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1987 by The American Society of Biological Chemists, Inc.

Vol. 262, No. 19, Issue of July 5, pp. 9004-9010,1987 Printed in U.S.A.

Two Regions Downstream of AATAAA in the Human Antithrombin I11 Gene Are Important for Cleavage-Polyadenylation*

(Received for publication, October 24,1986)

Edward V. Prochownik$j, Michael J. Smith$, and Alexander Markhamn From the $Section of Hematology/Oncology, the Department of Pediatrics, and The Committee on Cellular and Molecular Biology, The University of Michigan School of Medicine, Ann Arbor, Michigan 48109 and the 7lZCZ Pharmaceuticals Divisions, Mereside, Alderley Park, Macclesfield, Chesire 5KlO4TG, United Kingdom

We have investigated the sequence requirements for the cleavage-polyadenylation reaction in the human antithrombin 111 (ATIII) gene. A series of 5’-3‘ and 3‘-5’ deletions were produced around the AATAAA site using Ba131 nuclease. Ligation of appropriate pairs of such mutations resulted in the generation of varying sized deletions or duplications of sequences either upstream of, downstream of, or within the re- gion encompassing the poly(A) site. Whereas a large deletion 3‘ to the AATAAA signal abolished cleavage and polyadenylation of ATIII transcripts, smaller dele- tions, all of which were subsets of the large one, did not. This indicated that the ATIII gene contains at least two independently acting poly(A)-cleavage signals 3’ of AATAAA. When one of these signals was eliminated and the other was partially deleted at its 3‘-end, we were able to disrupt the normal spacing between AA- TAAA and the cleavage site without substantially af- fecting the efficiency of the cleavage reaction. This suggested that the distance between AATAAA and the cleavage site is determined by the same sequence which, along with AATAAA, specifies cleavage and polyadenylation. The duplication of regions either up- stream or downstream of AATAAA affected neither the efficiency nor the site of cleavage of the ATIII transcript. When a duplication included a large region containing AATAAA as well as downstream sequences, both sites were chosen for cleavage reactions. With a more delimited duplication, which included AATAAA but not the downstream cleavage signals, both sites were again used. However, two new cleavage sites were now detected. These results suggested that the distance between AATAAA sites and critical down- stream cleavage-spacing sequences may also be impor- tant in determining the site of cleavage. This may in part stem from spatial constraints imposed by RNA- protein complexes which have been postulated to be critical in catalyzing the cleavage-polyadenylation re- action.

The formation of mature eukaryotic mRNAs involves sev- eral modifications of the primary transcript (1). Capping and the excision of introns occur at the 5’-end and within the body, respectively, of newly synthesized molecules. At the 3‘-

* This work was supported by Grant R01-HL33471 and by grants from the Children5 Leukemia Foundation of Michigan and the Amer- ican Heart Association (Michigan Chapter). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked ‘‘aduertkement’’ in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

3 Established Investigator of the American Heart Association. To whom reprint requests should be addressed.

end, processing appears to be somewhat more complex in that most precursor mRNAs are synthesized with long 3’-exten- sions (2-4). An endonucleolytic scission step then cleaves the precursor at a point that will subsequently define the 3’ terminus of the mature molecule. Following polyadenylation, the fully modified transcript undergoes transport to the cy- toplasm.

Recently, the features of eukaryotic mRNAs which specify cleavage and polyadenylation steps have begun to be defined. The canonical sequence AATAAA is found in the 3”untrans- lated region of most mRNAs, 10-30 nucleotides 5’ to the site of poly(A) addition (5, 6). Mutations within this sequence lead to drastic reductions in mRNA cleavages, although those transcripts which are cleaved are still polyadenylated (7-9). That not all AATAAAs are recognized as cleavage-poly(A) signals (10-12) has led to the notion that additional sequences must be involved in this process. Currently, most data are compatible with a model in which AATAAA plus other nearby sequences are recognized by an enzymatic complex that cleaves and polyadenylates the primary transcript in a coupled reaction (13-17).

In this report, we have utilized a region from the 3’-end of the human antithrombin I11 (ATIII)’ gene (18) to study the sequence requirements of the cleavage and polyadenylation reactions in a eukaryotic cellular mRNA. Our approach has involved the deliberate deletion or duplication of regions around the AATAAA and cleavage sites. Our results demon- strate that there exist at least two sites downstream of AA- TAAA in the ATIII gene which can direct the cleavage- polyadenylation process. These regions also appear to be involved in determining the distance of the cleavage site from AATAAA. The deletion of either of these regions affects neither the efficiency nor specificity of transcript cleavage thus indicating that the two regions are functionally identical. Duplication of certain regions around AATAAA unmasks cryptic cleavage-polyadenylation sites which are used as fre- quently as the naturally occurring ones. These findings begin to explain how the spacing between AATAAA and the cleav- age site is determined and also how a single transcript may generate mRNAs with multiple 3’-ends (19, 20).

MATERIALS AND METHODS

Cell Culture and DNA Transfection-COS-1 cells (21) were grown in a 5% CO, atmosphere in Dulbecco’s modified minimal essential medium supplemented with 10% fetal calf serum, 100 units/ml peni- cillin, 100 pg/ml streptomycin, and 2 mM glutamine (GIBCO). One day prior to DNA transfections, lo6 cells were seeded into 100-mm tissue culture dishes. Each plate received 20 pg of the respective pSV2- neo-AT111 plasmid plus 20 pg of a normal human @-globin gene

The abbreviations used are: ATIII, human antithrombin 111; PIPES, 1,4-piperazinediethanesulfonic acid bp, base pair.

9004

Cleavage-Polyadenylation of Human Antithrombin III Transcripts 9005

(2.2%) to serve as an internal control for efficiency of DNA uptake. Calcium-phosphate-DNA transfections were performed as previously described (22) with precipitates being allowed to remain in contact with the cell monolayer for 16 h. Following a 25% glycerokminimal essential medium shock, cultures were maintained for an additional 48 h at which time total cellular RNAs were prepared by the guanidine hydrochloride method (22). In some cases, poly(A)+ mRNAs were purified using Hybond membranes (Amersham Corp.) according to the directions of the manufacturer.

Enzymes, Reagents, and Plasmids-Restriction enzymes, BglII linkers, and nuclease Ba131 were purchased from New England Bio- labs (Beverly, M A ) and used according to the supplier's directions. Bacterial alkaline phosphatase, T, polynucleotide kinase, and Klenow DNA polymerase were purchased from Boehringer Mannheim. T4 DNA ligase was from New England Nuclear. SP6 RNA polymerase was from Promega Biotech (Madison, WI). S1 nuclease was from Sigma. [-p32P]ATP (5000 Ci/mmol), [w3'P]GTP (400 Ci/mmol), and [a-32P]dATP (400 Ci/mmol) were from Amersham Corp.

The plasmid pAT2.5 has been described (18). pSV2-neo (23) was a gift from Dr. M. Imperiale (University of Michigan). All plasmids for DNA transfections were prepared as previously described (24) and purified by CsC1-ethidium bromide equilibrium centrifugation.

Following the construction of pAT2.5-derived poly(A) region du- plications or deletions, the StuI-EcoRI fragments were purified in 1.8% agarose gels and ligated into either HpaI + EcoRI-digested pSV2-neo or into Him11 + EcoRI-digested ml3-derived sequencing vectors. The correctness of each mutant was confirmed by dideoxy sequencing (25) through the duplicated or deleted region using either the m13 universal primer or synthetic oligodeoxynucleotide primers complementary to various regions of the wild-type StuI-EcoRI frag- ments. The same StuI-EcoRI ATIII fragments were also cloned into HincII + EcoRI-digested SP65 (26). These vectors were purified, digested with HindIII, and utilized for the run-off synthesis of uni-

formly labeled high specific activity complementary RNA probes for hybridizations with COS-1 RNAs.

Nucleic Acid Hybridizations-3r2P-Labeled SP65 run-off tran- scripts of each duplication or deletion mutant were synthesized as previously described (26, 27). lo5 dpm of each transcript was hybrid- ized for 16 h with 5 pg of total COS-1 RNA or with 50-100 ng of poly(A)+ mRNA in 80% formamide, 0.4 M NaCl, 40 mM PIPES, pH 6.7, and 1 mM EDTA at 45 "C. In some cases, neo-specific transcripts were measured using a 790-nucleotide-long NdeI-BglIl strand-sepa- rated S1 probe which had been end-labeled with 32P at the BglII site. Neo-transcripts protected an approximately 200-nucleotide-long frag- ment. (3-Globin transcripts were detected using a similarly labeled 1.94-kilobase B a n probe. &Globin gene transcripts protected a 200- nucleotide-long second exon fragment. The reaction conditions were the same as for RNARNA hybridizations except that the hybridiza- tion temperature was 51 "C. All hybridization reactions were digested by the addition of 300 units of S1 nuclease as previously described (28). After incubating 37 "C for 60 min, S1 reactions were terminated, precipitated with an equal volume of isopropyl alcohol, lyophilized, and redissolved in 10 p l of sequencing buffer (29). One-half of each reaction was loaded onto a 6% polyacrylamide-8 M urea sequencing gel and electrophoresed at 2000 V for approximately 90 min. For molecular weight markers we used 32P end-labeled MspI-digested pBR322 DNA. Gels were then transferred to a piece of used x-ray film, wrapped in Saran Wrap, and exposed with an intensifying screen to a piece of Kodak X-AR film overnight.

Construction of Deletions and Duplications around the ATIII Poly(A) Site-To study the influence of sequences around the ATIII AATAAA site on mRNA cleavage, we constructed a series of Ba131- generated mutations which contain varying sized duplications or deletions of the poly(A) site and its surrounding sequences (Fig. 1). A 2.5-kilobase PstI fragment containing the 3'-end of the human ATIII gene (18) was cloned in pBR322 in which the EcoRI site had

P S E K P

" I & L I "

Stu I dlgesl A Eco RI digest

Bal 31 dlgest: Klenow: Bgl II Ilnkers: Bgl II digest. sell-hgallon: translorm MC1061 cells.

FIG. 1. Strategy for the construc- tion of AT111 poly(A) site duplica- tions and deletions. pAT2.5 contains ATIII codons 375-432 (HI, 87 nucleo- tides of 3"untranslated region including the poly(A) site, and approximately 2.2 kilobases of 3'-flanking region (18). 3~131-digested DNAs were ligated to produce duplications or deletions (0) around the poly(A) site. These fragments were then used to replace the SV40 early poly(A) site in pSV2-neo (23). The same StuI-EcoRI fragments were simultane- ously cloned into the HincII + EcoRI- digested vector SP65. These vectors were used in the generation of uniformly la- beled RNA transcripts for S1-type anal- ysis of transfected neo-ATIII RNAs. P, PstI; S , StuI; E, EcoRI; K , KpnI; Bg, 3gLII; H, HindIII; Hp, HpaI. In the bot- tom portion of the figure, the SV40 early promotor and ATIII poly(A) site are in- dicated by hatched and open boxes, re- spectively. Dashed lines represent pBR322 sequences in the original pAT2.5 vector.

c) c-c Bal 31 I Bgl I1 + Kpn digest .cL

Bat 31 Bgl11 + Kpn digest

I Into Hpa I-Eco RI dlgested pSV2-neo Isolate Slul-Eco RI fragment. Llgate

pSV2-ne0 AT 111

9006 Cleavage-Polyadenylation of Human Antithrombin 111 Transcripts - 60

vaL431 LP4JZ -40 - 20 0 20 .m AAG T A A A A T - T T A ~ - ( A C C ~ - T A - ~ A C A C A A D T A A M A T A A A T A C A A A C T A C ~ ~ A ~ ~ A C A ~ T A A A - A ~ ~ - ~ = 90

4-48 -40) -25) 1-13 414 4 2 6 431

' t 60 80 IO0 I20

T C A A A T ~ A C A C A A G G A i r A O O G C A A C A r ~ ~ A ~ - C A C A - A ~ ~ ~ ~ T ~ C ~ M ~ ~ - A C C C ~ D T ~ A C C A C A ~ r ~ ~ ~ A A C A A A A r G : 10

443 73) 483 129)

FIG. 2. DNA sequence around the ATIII poly(A) site. The poly(A) site is underlined, and the normal site of cleavage-polyadenylation at position 24 is denoted by the asterisk. The extent of various Bal31 deletions is indicated by the arrows. Each deletion is designated by the position in the sequence at which Bal31 digestion was terminated. Arrows pointing to the right indicate deletions from the StuI site, whereas those pointing to the left indicate deletions from the EcoRI site. The numberine system is arbitrarily defined with position +1 being the f i r s t nucleotide after AATAAA.

I

been destroyed. This clone (designated PAT 2.5) was digested with either StuI or EcoRI, which cleave upstream and downstream, re- spectively, of the ATIII poly(A) signal. The DNAs were digested with nuclease Bal31 so as to generate a series of overlapping deletions. After the addition of BglII linkers and religation, DNAs were intro- duced into recipient Escherichia coli MC1061 cells, individual clones were selected, and plasmid DNAs were purified and sized in 1.8% agarose gels. The precise extent of Bal31 digestion in selected clones was ascertained by dideoxy sequencing of BglII-EcoRI or BglII-StuI fragments after cloning in the appropriate m13 vectors (Fig. 2). To generate duplications or deletions around the ATIII poly(A) site, the BglII-KpnI fragments from clones which had been subjected to Ba131 digestions from EcoRI site were replaced with the BglII-KpnI frag- ments from clones which had been Bal31-digested from the StuI site. This resulted in the generation of a new set of clones containing deletions or duplications of the poly(A) site and/or its surrounding regions (Fig. 3). The StuI-EcoRI fragments from these clones were then isolated from 1.8% agarose gels and inserted into the vector pSV2-neo (23) from which the SV40 early region poly(A) site had been excised with HpaI and EcoRI. These vectors, containing ATIII poly(A) site duplications or deletions, were then introduced into COS- 1 cells by standard calcium phosphate-mediated transfection. Total cellular RNAs were purified after 48-60 h and used in S1 type analyses (28) to assess the sites of neo-AT111 mRNA cleavage.

RESULTS

Deletions and Duplications Do Not Affect ATIII-neo Tran- script Stability-To determine whether the structural changes which we had introduced into the ATIII 3'-ends quantita- tively affected neo-ATIII hybrid transcripts, we used a single end-labeled neo-specific S1 probe to measure their levels in transfected COS-1 cells. These levels were compared to those originating from a cotransfected normal human @-globin gene which served as a control for transfection efficiency. The levels of neo-ATIII hybrid transcripts were roughly equivalent when compared to @-globin transcripts (not shown). We con- clude that the engineered changes in the 3'-ends of the neo- ATIII plasmids did not affect transcript levels to any sub- stantial degree.

AATAAA and Downstream Sequences Are Required for Efficient Cleauage-We next asked whether elimination of either the AATAAA motif or sequences distal to i t could prevent cleavage of neo-AT111 hybrid RNAs. We therefore tested the deletion mutants -13/2 and 14/129 along with the wild-type StuI-EcoRI fragment depicted in Fig. 3A. In the first instance, 14 bp of DNA containing the AATAAA recog- nition signal have been removed and replaced with an 8-bp BglII linker. In the second case, a 114-bp deletion has been introduced, beginning 14 nucleotides 3' of the AATAAA se- quence. This deletion removes not only the site of mRNA cleavage and polyadenylation but presumably essential down- stream sequences which participate in this process. Each plasmid was transfected into COS-1 cells, and the resultant neo-AT111 RNAs were assayed in an S1 type analysis using a uniformly labeled antisense RNA probe specific for each input plasmid (Fig. 4). In the case of the wild-type plasmid, we observed a 164-nucleotide protected RNA band, indicating

"- "-

"- I "-

FIG. 3. Poly(A) site deletions and duplications used for transfection studies. A, map of the 200 nucleotides surrounding the poly(A) site. The numbered positions correspond to those depicted in Fig. 2 as do the AATAAA site and the cleavage/poly(A) addition site (*). The StuI site is at position -139, and the EcoRI site is at position 225. B, deletions and C, duplications generated by combining the Bal31 deletions shown in Fig. 2. The length of each StuI-EcoRI segment is actually 8 nucleotides longer than depicted here due to the presence of the BglII linker. In C, the solid bores connote the direct repeats of sequences generated by ligating the indicated over- lapping Bal31 deletions.

efficient cleavage at the expected site (position 24 in Fig. 2). With both deletion plasmids, no protected bands were ob- served other than those of the input SP6 probes, indicating a failure of transcribed RNAs to be efficiently cleaved. These results confirmed that AATAAA is an integral component of the cleavage-polyadenylation process and that sequences downstream of it also play an important role (5-9, 13-17).

Localization of Downstream Sequences-To localize regions 3' to AATAAA that are responsible for directing cleavage, we

Cleavage-Polyadenylation of Human Ant i thrombin 111 Transcripts

A B C M - 4 6 2 2 - -527

v 4 4 0 3

C -309

4 2 4 2 / 2 3 8

0 4 2 1 7

e 4 2 0 1 4 1 9 0

d 4 1 8 0

0 -160

0 -147

1 6 4 e

r. 4 1 2 2

-110

U SV40 - AT 111

I64 - FIG. 4. Deletion of AATAAA or downstream regions inhib-

its mRNA cleavage. RNAs from transfected COS-1 cells were hybridized with their respective SP6-generated 32P-labeled probes, treated with S1 nuclease, and analyzed by polyacrylamide-urea gel electrophoresis. End-labeled MspI-digested pBR322 DNA was in- cluded in adjacent lanes as marker ( M ) . correctly cleaved RNA should protect a 164-nucleotide-long labeled probe fragment (large arrow). A, Wild-type ATIII StuI-EcoRI poly(A) fragment. B, -13/2. C, 14/129. D, the 3'-end of the pSV2-neo ATIII wild type plasmid. The thick lines depict the 364-nucleotide-long uniformly labeled ATIII StuI-EcoRI probe and the expected 164-nucleotide-long pro- tected S1 fragment. * indicates the site of endonucleolytic cleavage and polyadenylation. T indicates the SV40 early splice site.

constructed a series of smaller deletions representing subsets of deletion mutant 141129. In the first case, we asked whether elimination of the natural poly(A) cleavage site alone could influence cleavage. Deletion mutant 14/31 removes 16 bp of sequence surrounding the poly(A) addition site at position 24. RNA produced by this mutant nevertheless protected a 160- 162-bp species of labeled SP6 transcript that was 2-4 nucleo- tides shorter than expected (Fig. 5, lane A ) . This indicated that elimination of the natural cleavage site did not prevent efficient cleavage but did alter the accuracy of this process slightly (see "Discussion"). To localize the downstream se- quences involved in cleavage more exactly, we examined three additional deletion mutants, 26173,831129, and 431129. In all three instances, correctly cleaved transcripts were readily detected. This suggested that there must be at least two downstream regions involved in cleavage and that these two domains are essentially identical in both the efficiency of the cleavage reaction they direct and in determining the distance of the cleavage site from AATAAA. Additional information concerning the role of these sequences in the selection of the cleavage site was obtained from the use of mutant 261129 in which the deletion is only 12 nucleotides shorter than the nonfunction 141129 mutant. Following transfection, we de- tected efficient cleavage of 261129 RNAs (Fig. 5, lane E ) . However, the cleavage site was now shifted downstream by

A B

FIG. 5. Cleavage of neo-ATIII deletion transcripts. COS-1 RNAs were hybridized with run-off transcripts derived from their respective SP6 vectors. A, 14/31; B, 26/73; C, 83/129; D, 43/129; E, 26/129. The expected 164-nucleotide protected fragments are denoted by solid arrows. In A and E the open arrows denotes aberrant cleavage products of 160-162 and 179 nucleotides, respec- tively.

15 nucleotides to approximately position 133 (Fig. 2). This experiment indicated that the restoration of ATIII sequences between positions 14 and 26 was sufficient to restore efficient cleavage although insufficient to restore it at its proper site. This required the presence of at least an additional 17 bp (mutant 431129, Fig. 5, lane D ) and suggested that sequences located between positions 26 and 43 were involved in deter- mining the spacing of the cleavage site from the AATAAA signal. This was consistent with the finding with mutant 141 31 which showed nearly normal cleavage (Fig. 5, lane A ) presumably due to the presence of a nearly intact sequence between positions 26 and 43. From these results, we conclude that the two regions determining cleavage spacing of ATIII transcripts are located between positions 14 and 43 and po- sitions 73 and 129 and that sequences between positions 26 and 43 are critical in determining the position at which cleavage will occur.

Duplications around the Poly(A) Site-We next asked whether duplications of sequences around the AATAAA re- gion could influence the choice or the efficiency of utilization of cleavage sites. Mutant 1/83 introduces an 82-bp duplication which includes the true cleavage site a t position 24 as well as the downstream sequences encoding cleavage signals. Tran- scripts of this mutant were efficiently cleaved to produce the expected 164-nucleotide-long S1 product (Fig. 6, lane A ) . Mutant -401-13 includes a 28-bp duplication of sequence just upstream of AATAAA. Transcripts of this mutant were also efficiently cleaved at the natural site. As expected, the cleavage product contained an additional 36 nucleotides of sequence (28-nucleotide-long duplication + 8-nucleotide-long BglII linker) thus giving rise to a 200-nucleotide-long cleavage product (Fig. 6, lane B) . Next, we introduced a 124-bp dupli- cation (construct -40183) that includes the AATAAA site and the natural cleavage site as well as the downstream sequences between positions 14 and 43 required for accurate spacing and cleavage. The results (Fig. 6, lane C) showed that both the upstream and downstream poly(A) sites were utilized in cleavage reactions. Transcripts utilizing the upstream cleavage site were detected about five times as frequently, suggesting either that this site was chosen more frequently or that transcripts cleaved at the downstream site were more rapidly degraded.

Cleavage-Polyadenylation of Human Antithrombin 111 Transcripts

C D

I I I

L

L a c “

FIG. 6. Cleavage of neo-AT111 duplication transcripts. A, 11 83; B, -401-13; C , -40183; D, -25114. I , a total of 5 pg of COS-1 RNA was used for hybridization. I I , approximately 50 ng of poly(A)+ mRNA was used for hybridization.

TABLE I Efficiencies of neo-ATIII transcript cleavage

Mutant Cleaved product”

% Wild type 76 -1312 del <2 141129 del 5 14/31 del 63 26/73 del 52 831129 del 80 431129 del 61 261129 del 48 1/83 dup 70 -40113 dup 69 -40183 dup 83 -25114 dup 85

Cleavage efficiencies were determined by deasitometrically scan- ning autoradiograms to obtain relative values for cleaved product(s) and uncleaved precursor. The efficiency is calculated as the percent- age of total SI-protected transcript that was cleaved. Values represent the average of two to three independent experiments.

Unexpected results were obtained when the duplicated re- gion was more delineated. Mutant -25/14 contains a 40-bp duplication of AATAAA and surrounding sequences. This results in a construct in which only the 3‘ AATAAA is associated with the spacing-cleavage sequences. Both AA- TAAAs, however, served as efficient poly(A) sites. 164- and 215-nucleotide-long S1 products were the result of mRNA cleavages a t sites 21-24 nucleotides downstream of the two AATAAA signals (Fig. 6, lane D). These observations con- firmed our earlier findings that the actual distance of AA- TAAA from its associated spacing-cleavage signals may vary without affecting the cleavage-poly(A) reaction. Furthermore, these results showed that a new AATAAA sequence will be used with high efficiency if it is inserted near the spacing- cleavage signals.

In addition to the two cleavages occurring at the expected locations, two new cleavages were detected. The first occurred at position 6, a point 17 bp upstream of the true cleavage site (147-nucleotide-long S1 product). The second site occurred a t position 58, 34 nucleotides downstream from the predicted site (249-nucleotide-long S1 product). All four cleavage prod-

ucts were polyadenylated (Fig. 6, lane D ) , indicating that the transcripts being detected were the result of the activation of cryptic cleavage sites rather than aberrant mRNA splicing.

Relative Efficiencies of the Cleavage Reactions-Because the uniformly labeled 3”probes were of different specific activi- ties, we could not directly compare the intensities of S1- protected bands to arrive at an estimate of cleavage efficien- cies. However, since the bands corresponding to full-length probe represent precursor neo-AT111 RNA species: it was possible to calculate cleavage efficiency for any given sample by determining the fraction of total transcript represented by the cleaved species (Table I). The results indicated that cleavage efficiency was substantially affected only when both downstream elements (deletion 14/129) or the AATAAA site (deletion -13/2) were eliminated.

DISCUSSION

Poly(A) Deletion Mutants-The cleavage and polyadenyla- tion of primary mRNA transcripts have been previously shown to require both the canonical AATAAA sequence as well as less conserved ones, generally located downstream from the actual point of cleavage (13-17,30). Endonucleolytic cleavage and poly(A) addition appear to be coupled reactions in which both sequences must be recognized by an enzymatic complex in order for efficient processing to occur. In the experiments reported here, we have confirmed earlier results by demonstrating that deletion of the AATAAA recognition site (mutant -13/2) abolishes cleavage of the primary neo- ATIII hybrid transcript (7, 8, 13). We also observed elimina- tion of cleavage by deleting a large region of downstream sequence as well (mutant 14/129). Our results are compatible with the notion that the downstream sequence comprises two noncontiguous but functionally identical elements, either of which can direct the cleavage of the primary transcript as well as direct the distance of the cleavage site from AATAAA. This is based on our studies with mutants 26/173, 83/129, and 43/129 which together eliminate most of the region that is included in the nonfunctional mutant 14/129, yet which all direct the production of appropriately cleaved transcripts. Evidence for the critical sequence of one of these sites is provided by comparing the nonfunctional mutant 14/129 with 26/129 and 43/129. In the last case, appropriately cleaved mRNA is seen. However, as the deletion extends in a more 5’ direction (26/129), the distance of the cleavage site from AATAAA is disrupted even though cleavage still occurs effi- ciently. Eventually, even the ability to cleave is lost (14/129). Since these three mutants all contain identical 3‘ boundaries, these results can only be explained by the elimination of a critical cleavage-spacing sequence located between positions 14 and 43. As deletions proceed in a 3‘-5‘ direction from position 43, the ability to correctly space the cleavage site relative to AATAAA is eliminated and is eventually followed by loss of the cleavage reaction in its entirety. This suggests that the sequence between positions 14 and 26 specifies the actual cleavage process and that sequences between positions 26 and 43 are involved in determining the distance between AATAAA and the cleavage site. In fact, this may provide an explanation for the efficient but slightly inaccurate cleavage seen with mutant 14/31. In this case only 5 bp of “spacing” sequence between positions 26 and 31 have been eliminated. This could provide an adequate although incorrect signal for spacing such that the cleavage occurs 3-4 nucleotides further upstream.

Deletion mut,ants 26/73 and 14/31 both eliminate the crit-

* E. V. Prochownik, unpublished.

Cleavage-Polyadenylation of Human Antithrombin 111 Transcripts 9009

ical cleavage-spacing signals between positions 14 and 43 while continuing to direct the production of cleaved tran- scripts. This indicated that additional sequences between positions 73 and 129 must be involved in the process. Indeed, within this 56-nucleotide-long region we have noted three short regions of homology to the upstream cleavage-spacing sequences between positions 14 and 43 (Fig. 7). I t seems likely that at least one of these short regions is a critical, if not the sole, component of the downstream spacing-cleavage signal. Despite the lack of any extensive homologies to the 14-43 region, it is nonetheless apparent that the 73/129 region is functionally identical.

Poly(A) Duplication Mutants-Studies with duplications around the poly(A) site revealed some unexpected aspects of the cleavage-polyadenylation process. The reiteration of re- gions either immediately upstream or downstream of AA- TAAA (-40/-13 or 1/83) affected neither the site nor appar- ent efficiency of the cleavage process even when the duplica- tion included downstream cleavage sequences. Thus, the du- plication of essential downstream cleavage signals does not affect the site or efficiency of cleavage when only a single AATAAA is present. A much larger duplication (-40/83), which included AATAAA, the natural cleavage site at position 24, and both sets of downstream sequences, gave rise to two transcripts resulting from appropriately spaced scissions at each of the duplicated cleavage sites. The longer transcript might have arisen as a result of readthrough from the first cleavage site. Alternatively, cleavage occurring initially at the downstream site might have inhibited further endonucleolytic scission at the upstream site due to a change from mRNA secondary structure or its immediate transport to the cyto- plasm. Choices among alternate poly(A) sites have been pre- viously described for several genes (10, 19, 20).

Mutant -25/14 contains a much smaller duplication than -40/83, resulting in a closer spacing of the two AATAAA sites in association with only one set of downstream cleavage sequences and the natural cleavage site. Both AATAAA sites are utilized, however, and appropriately cleaved transcripts produced. Additionally, two cryptic cleavage sites are acti- vated and used with high efficiency. We believe that the appearance of these new sites may be related to the spacing of the two downstream elements from the AATAAA se- quences. If this spacing is within some optimal range, as with duplication mutant -40/83 or deletion mutant 26/73, cleavage is precise and occurs at only the natural sites. However, cleavage may become imprecise or promiscuous when this optimal distance is exceeded as in the case of mutant -25/14. Some support for this model has been provided with an additional duplication mutant -41/26. This mutation is sim- ilar to -25/14 except that the distance between the upstream AATAAA and the downstream cleavage elements has been increased by an additional 27 nucleotides. The cryptic cleav-

Position (Firwre - 2) - Sequcncc

18-24 PyCACATPy

125-131 PyCACATPy

37-42 TGC-ATT

71-77 TGCTATT

97-102 TGC-CTT FIG. 7. Homologies among three regions of the ATIII gene

3' to the poly(A) site. Note that deletion mutant 26/73 removes the threonine and glycine residues from positions 71 and 72, respec- tively (Fig. 2). However, these bases are fortuitously resorted by the addition of the BglII linker sequence CAGATCTG.

age sites are still utilized but at a greatly reduced frequency (not shown). These findings provide a potentially useful model for natural systems in which 3'-end heterogeneity can result either from the presence of multiple AAUAAA sites or from a single such site with multiple downstream cleavage points (3, 10, 19, 30, 31).

Requirements for Cleavage-Polyadenylatwn-The experi- ments presented here indicate that the region 3' to AATAAA in the human ATIII gene contains at least two noncontiguous sequences encoding very similar or identical functions in regard to mRNA cleavage. Deletion mutants which eliminate one or the other of these sequences still maintain the ability to cleave primary transcripts at an appropriate distance from the AATAAA. Only when the sequence of both elements is disrupted (mutant 26/129) or when their distance from AA- TAAA exceeds a certain limit (mutant -25/14) do we observe changes in the spacing. Elimination of both these sequences (mutant 14/129) results in a primary transcript incapable of being cleaved. In the case of mutant 14/31, a 5-bp deletion of the spacing sequences between positions 26 and 43 partially alters the spacing accuracy. The closeness of this signal to the AATAAA site may allow it to be dominant over the position 73-129 spacing sequences.

At least two consensus sequences have been previously implicated as being an essential part of the cleavage-polyad- enylation process. Berget (32) has reported that the sequence CAYUG occurs at high frequency at a position immediately 5' or 3' to the cleavage site. An additional region having the consensus sequences YGTGTTYY has been reported in two- thirds of mammalian mRNA 3' termini (16), although the significance of this has been disputed (17, 30). Weak homol- ogies to Berget's consensus occur in both ATIII downstream sequences (positions 39-44, CATTTG, and 84-87, CATG). These regions, as well as the sequence described by Berget, are complementary to areas within the small nuclear RNA U4. Perhaps changes in U4-ATIII mRNA secondary structure are responsible for the activation of the cryptic cleavage sites seen with duplication mutant -25/14. It is interesting that U4 RNAs contain two regions complementary to the sequence CAYUG. This suggests that the duplicated downstream ele- ments in the ATIII transcript might have the potential for simultaneously forming hybrids with both U4 complementary sequences.

Two additional properties of the postulated ATIII upstream spacing sequence deserve mention. First, the region between positions 37 and 47 bears striking homology to an area in the 3"untranslated regions of both human and murine IgM heavy chain mRNAs (30). These regions are capable of forming intramolecular stem-loop structures and are involved in the formation of the secreted form of p chain mRNA. Second, the ATIII sequence between positions 39 and 49 forms a perfect 11-bp palindrome which is in turn contained within a nearly perfect 19-bp palindrome (positions 35-53). The significance of these features is not known; clearly, a substantial amount of palindromic sequence can be eliminated without signifi- cantly affecting mRNA processing (mutant 43/129).

The factors which determine the spacing between the AA- TAAA sequence and the cleavage site are imprecisely under- stood. All the nucleotides may serve as efficient substrates for the polyadenylation reaction (5). Furthermore, in vitro syn- thesized run-off transcripts are efficiently polyadenylated even when the 3'-end is a considerable dist,ance from the naturally occurring cleavage site (33). Thus, the stringent spacing constraints do not appear to result from an absolute requirement of the polyadenylation enzyme complex for a particular 3'-OH terminus, a particular sequence at the im- mediate cleavage site or for a particular distance between

9010 Cleavage-Polyadenylation of Human Anti thrombin 111 Transcripts

AATAAA and the cleavage site. Perhaps spacing is a factor in determining mRNA stability or expression. Nevertheless, our results suggest that the critical determinants of ATIII mRNA cleavage are the presence of two small downstream elements and their distance from AATAAA.

Acknowledgments-We are grateful to Dr. M. Imperiale for his valuable comments and suggestions on the manuscript and to Terry Oliver for secretarial assistance.

REFERENCES 1. Nevins, J. R. (1983) Annu. Rev. Biochm. 5 2 , 441-466 2. Nevins, J. R., and Darnell, J. E. (1978) J. Virol. 25,811-823 3. Nevins, J. M., Blanchard, J. M., and Darnell, J. E. (1980) J. Mol.

4. Salditt-Georgieff, M., and Darnell, J. E. (1983) Proc. Natl. Acad.

5. Fitzgerald, M., and Shenk, T. (1981) Cell 24 , 251-260 6. Proudfoot, N. J., and Brownlee, G. G. (1976) Nature 263, 211-

7. Higgs, D. R., Goodbourn, S. E. Y., Lamb, J., Clegg, J. B., Weath-

8. Montell, C., Fisher, E. F., Caruthers, M. H., and Berk, A. J.

9. Wickens, M., and Stephenson, P. (1984) Science 226 , 1045-1051 10. Aho, S., Tate, V., and Boedtker, H. (1983) Nucleic Acids Res. 11 ,

11. Perricaudet, M., Akusjiirvi, G., Virtanen, A,, and Pettersson, U.

12. Reddy, V. B., Ghosh, P. K., Lebowitz, P., Piatek, M., and Weiss-

13. Conway, L., and Wickens, M. (1985) Proc. Natl. Acad. Sci.

14. Gil, A., and Proudfoot, N. J. (1984) Nature 312 , 473-474

BWl. 144,377-386

Sci. U. S. A. 80,4694-4698

214

erall, D. J., and Proudfoot, N. J. (1983) Nature 3 0 6 , 398-400

(1983) Nature 305,600-605

5443-5450

(1980) Nature 28 1, 694-696

man, S. (1979) J. Virol. 30, 279-296

U. S. A. 82,2949-2953

15.

16.

17.

18.

19.

20.

21. 22.

23.

24.

25.

26.

27. 28.

29.

30.

31.

32. 33.

McDevitt, M. A., Imperiale, M. J., Ali, H., and Nevins, J. R.

McLauchlan, J., Gaffney, D., Whitton, J. L., and Clements, J. B.

Sandofsky, M., Connelly, S., Manley, J. L., and Alwine, J. C.

Prochownik, E. V., Bock, S. C., and Orkin, S. H. (1985) J. Biol.

Setzer, D. R., McGrogan, M., Nunberg, J. H., and Schimke, R.

Zehner, Z. E., and Paterson, B. M. (1983) Proc. Natl. Acud. Sci.

Gluzman, Y. (1981) Cell 23, 175-182 Treisman, R., Orkin, S. H., and Maniatis, T. (1983) Nature 302,

591-596 Southern, P. J., and Berg, P. (1982) J. Molec. Appl. Gen. 1, 327-

340 Maniatis, T., Fritsch, E. F., and Sambrook, J. (1982) in Molecular

Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY

Sanger, F., Nicklen, S., and Coulson, A. R. (1977) Proc. Natl. Acad. Sci. U. S. A. 74, 5463-5467

Melton, D. A,, Krieg, P. A,, Rebagliati, M. R., Maniatis, T., Zinn, K., and Green, M. R. (1984) Nucleic Acids Res. 12 , 7035-7056

Zinn, K., DiMaio, D., and Maniatis, T. (1983) Cell 34,865-879 Favaloro, J., Treisman, R., and Kamen, R. (1980) Methods En-

Maxam, A. M., and Gilbert, W. (1980) Methods Enzymol. 6 5 ,

Danner, D., and Leder, P. (1985) Proc. Natl. Acad. Sci. U. S. A.

Leytus, S. P., Chung, D. W., Kisiel, W., Kurachi, K., and Davie,

Berget, S. M. (1984) Nature 309 , 179-182 Manley, J. L. (1983) Cell 33 , 595-605

(1984) Cell 37,993-999

(1985) Nucleic Acids Res. 13,1347-1368

(1985) Mol. Cell. Biol. 5, 2713-2719

C h m . 260,9608-9612

T. (1980) Cell 22 , 361-370

U. S. A. 80, 911-915

zymol. 65, 718-745

499-550

82,8658-8662

E. W. (1984) Proc. Natl. Acad. Sci. U. S. A. 8 1 , 3699-3702