The efficiency of RNA 3'-end formation is determinea by the distance between the cap site and the poly(A) site in spleen necrosis virus Kouichi Iwasaki and Howard M. Temin

McArdle Laboratory for Cancer Research, The University of Wisconsin, Madison, Wisconsin 53706 USA

The efficiency of RNA 3'-end formation of spleen necrosis virus (SNV) is determined by the distance between the cap site and the poly(A) site. When the distance between the cap site and the poly(A) site was shorter than 500 bases, only 3-9% of the RNA was polyadenylated at the SNV poly(A) site. However, when the distance between the cap site and the poly(A) site was 1400 bases or more, 70% of the total RNA was polyadenylated at the SNV poly(A] site. In contrast, the poly(A) signal sequences of the thymidine kinase {tk) and SV40 late genes functioned at high efficiency, even with a distance between the cap site and the poly(A) site that was short enough to inactivate the SNV poly(A) signal. Therefore, this distance-dependent inactivation of RNA 3'-end formation is specific for SNV sequences and perhaps for related retroviruses. This finding explains the difference between the 5' and 3' poly(A) sites in many retrovirus RNAs.

{Key Words: Retrovirus; RNA polyadenylation]

Received July 30, 1990; revised version accepted October 12, 1990.

Eukaryotic mRNA undergoes post-transcriptional pro­cessing. One of the post-transcriptional steps is the 3'-end formation of primary transcripts. The 3'-end forma­tion involves two reactions: cleavage of the transcript and addition of the 200- to 300-base polyadenylation chain (for review, see Bimstiel et al. 1985; Humphrey and Proudfoot 1988; Manley 1988). At least tvv o se­quence elements are known to be necessary for 3'-end formation in cells and in cell-free systems: the AAUAAA sequence and the downstream element (for review, see Bimstiel et al. 1985; Humphrey and Proud-foot 1988; Manley 1988). The AAUAAA sequence is lo­cated 10-30 bases upstream from the cleavage/polyad-enylation site [poly(A) site] and is well-conserved among genes. The downstream element is located 10-50 bases downstream from the poly(A) site and consists of a GU-rich sequence. It is less conserved among genes.

In retroviral replication, viral RNA is expressed from proviral DNA with cellular enzymes and undergoes post-transcriptional processing in a manner similar to the mRNA of cellular genes. The 3'-end formation of re­troviral RNA is also carried out by cellular enzymes and requires the AAUAAA and downstream poly(A) signals (Bohnlein et al. 1989). However, the AAUAAA and downstream poly(A) signals appear twice in the viral RNA of some species of retroviruses. In these retro­viruses, it is necessary to transcribe through the first poly(A) site to express full-length viral RNA. This group of retroviruses includes murine leukemia viruses, avian

reticuloendotheliosis virus/spleen necrosis virus (SNV), murine mammary tumor virus, and human immunode­ficiency virus (HIV) (Weiss et al. 1985), as well as retro-transposons, such as Ty elements in yeast (Elder et al. 1983; Boeke et al. 1985; Clare and Farabaugh 1985).

In viruses of this group, the AAUAAA sequence is lo­cated in the R region; the poly(A) site is at the junction between the R and U5 regions; and the downstream ele­ment is located in the U5 region (see Fig. lA) (Weiss et al. 1985; Bohnlein et al. 1989). The U3 region contains enhancer and promoter elements. Therefore, viral RNA is initiated at the 5' end of the R region in the 5' long terminal repeat (LTR) and is polyadenylated at the 3 ' end of the R region transcribed from the 3 ' LTR; conse­quently, a poly(A) site near the 5' end of the RNA re­mains unprocessed. The question we approached was how these retroviruses differentiate between the 5' and 3 ' poly(A) sites.

It was proposed that the U3 region, which is tran­scribed from the 3 ' LTR and is present only at the 3 ' end of the RNA (see Fig. lA), leads to the functional differ­ence between the retroviral 5' and 3 ' poly(A) sites (Benz et al. 1980; Dougherty and Temin 1987). However, we have shown that this hypothesis is unlikely (Iwasaki and Temin 1990). Therefore, to understand why retroviral RNA is polyadenylated at the 3 ' poly(A) site and not at the 5' poly(A) site, we proposed three alternative hy­potheses: (1) The poly(A) signal sequence in retroviral RNA is inefficient and allows readthrough at both

GENES & DEVELOPMENT 4:2299-2307 £ 1990 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/90 $1.00 2299

Cold Spring Harbor Laboratory Press on September 6, 2021 - Published by genesdev.cshlp.orgDownloaded from

Iwasaki and Temin

CS

yMMM/MA m{_

tk poly(A) site

•AAAA

pJD214Hy

pSNV(wt)

pSNV(AU3)

pSNV(Appt/U3)

pSNV(AE)

6 pSNV(AE/U3)

7. pSNV(AE/ppt/U3)

8. pSNV(SV40)

U3 R U5 U3 R U5 fujNengthRNA

read-through RNA

hygro probe E ppt U3

-AAAA

Sizes of Processed RNA(kb) RNA(%)

2.2/--

SV40

2.2/3.4

1.8/3.0

1.7/2.9

1,9/3.1

1.4/2.7

1.3/2.6

2.2/3.4

96

85

83

92

81

82

>97

k b 7 . 5 -

4.4—

3 f^

a u ^ U >

> > z z CO C«

z z

^Nittiiili 1 2 3 4 5 6 7 8

.Okb

-TCTAGAGTCGA B 1 TTGCTCGGCC TCGAGGTCGG GGTCGCCGTC CTACACATTG TTGTTGTGAC GTGCC

60 ^CCCA

poly(A) site 61 [ 12C GATTCGAATC TGTAATAAAA CTTTTTTTTT TCTGAATCCT CAGATTGGCA GTGAGAGGAG

ATTTTGTTCG TGGTGTTGGC TGGCCTACTG GGTGGGCGCA GGGATCttgg tggcgtgaaa

Figure 1. Experiments to measure the efficiency of RNA 3'-end formation at the SNV 3' poly(A) site in dif­ferent contexts. [A] Constructs used in the experiments. pfD214Hy contains cis sequences required for retroviral rephcation and can repUcate efficiently in the presence of SNV proteins (Dougherty and Temin 1986). pSNV(wt) contains pJD214Hy and the herpes simplex virus thymi­dine kinase [tk] gene fragment. The rest of the pSNV series plasmids were constructed based on pSNV(wt) (see Materials and methods). In pSNV(SV40) the R and

U5 regions in the 3' LTR of pJD214Hy were replaced with the SV40 late RNA 3'-end formation sequence (Wickens and Stephenson 1984). (ppt) Polypurine tract; (E) encapsidation sequence; (open regions) LTR sequences; (thin lines) non-LTR regions of the SNV genome; (right-hatched region) hygrcf gene; (stippled region) tk gene; (left-hatched region) SV40 late RNA 3'-end formation sequence; (broken lines) RNAs; (crosshatched box) probe used in Northern blot analyses; (B) BamHl site; (C) Clal site; (S) Sad site; (X) Xhol site; (thick lines) structures of constructs; (vertical broken lines) boundaries of regions. [Right] Numbers of expected sizes of RNAs (kb), sizes of full-length RNA compared to sizes of readthrough RNA, and percentages of processed RNA [i.e., full-length RNA divided by total RNA in D17 cells (the mean values of at least two experiments)]. [B] Sequence of the R and U3 regions in pSNV(wt). The sequence starts 9 bases upstream from the 5' end of the R region in the SNV LTR. An Xhol linker was inserted into the Aval site at the U3/R boundary in the 3' LTR of pSNV(wt) (pJD214Hy does not contain this Xhol site); insertion of this linker had no effect on RNA production (Iwasaki and Temin 1990). The sequences created by this treatment are indicated in italics (bases 9-20). The original 5' end of the R region is indicated by a bold letter at base 22 (Shimotohno et al. 1980). When this LTR sequence was used for initiation of a transcript, initiation occurred at base 13 as shown by a horizontal arrow, determined by an RNA protection assay (data not shown). All of the sequences used in the experiments of Figs. 2 and 5 were inserted into the Xhol site at base 15, and all of these sequences contained the sequences shown above the inverted triangle. pR/U5(90) contained no insertion at this Xhol site (plasmid 9 in Fig. 2A). The AATAAA poly(A) signal sequence is underlined. The poly(A) site is located at base 102, determined by an RNA protection assay (Iwasaki and Temin 1990). The tk gene sequence is shown as lowercase letters after base 167. (C) Northern blot analysis. D17 cells were transfected (Kawai and Nishizuwa 1984) with the indicated constructs, and RNA was isolated (Berger and Birkenmeier 1979) at 48 hr after transfection and subjected to Northem blot analysis (Maniatis et al. 1982). The DNA fragment used as the probe is shown in Fig. lA. Transfection efficiency was normalized to levels of chloramphenicol acetyltransferase (CAT) activity (Gorman et al. 1982), resulting from cotransfection with pCMV-CAT (Stinski and Roehr 1985) (data not shown). Relative expression levels were obtained by microdensitometory scanning of Northem bands. The sizes of marker RNAs are shown at left. The numbers at the bottom of each lane correspond to the numbers of the transfected constructs shown in Fig. lA.

poly(A) sites; (2) viral genomic sequences outside of the U3, R, and U5 regions (designated non-LTR regions) are responsible for the functional difference between the 5' and 3 ' poly(A) sites; and (3) the short distance betv^^een the cap site and poly(A) site prevents RNA 3'-end forma­tion at the 5' poly (A) site.

Here, we report experiments to test each hypothesis. Our experiments support the third hypothesis. The effi­ciency of RNA 3'-end formation of SNV is determined by the distance between the cap site and the poly(A) site. In contrast, the poly(A) signal sequences of the thymi­dine kinase [tk] and SV40 late genes function even with a distance between the cap site and the poly(A) site that

is short enough to inactivate the SNV poly(A) signal. Therefore, this distance-dependent inactivation of RNA 3'-end formation is specific for the SNV poly(A) signal sequences and perhaps for related retroviral poly(A) signal sequences.

Results

The poly(A) site of SNV is used efficiently

To test the first hypothesis, that the poly(A) signal se­quence in retroviral RNA is inefficient and allows readthrough at both poly(A) sites, and also to measure the efficiency of RNA 3'-end formation at the 3 ' poly(A)

2300 GENES & DEVELOPMENT

Cold Spring Harbor Laboratory Press on September 6, 2021 - Published by genesdev.cshlp.orgDownloaded from

Distance-dependent polyadenylation of SNV RNA

site of SNV, we constructed plasmids pSNV(wt) and pSNV(SV40) (plasmids 2 and 8 in Fig. lA). pSNV(wt) contains the 5' LTR, the primer binding site (pbs), the encapsidation sequence (E), the hygromycin phospho­transferase B gene {hygrcf), the polypurine tract (ppt), the 3' LTR, and the herpes simplex virus thymidine kinase gene (tic). The region from the 5' LTR to the 3' LTR is identical to an SNV vector, }D214Hy (plasmid 1 in Fig. lA), which replicates efficiently in the presence of SNV proteins (Dougherty and Temin 1986). For convenience, an Xhol linker was inserted at the U3/R boundary in the 3' LTR of pSNV(wt), which pJD214Hy does not contain; insertion of this linker had no effect on RNA production (Iwasaki and Temin 1990). This sequence is shown in Figure IB. pSNV(SV40) is identical to pSNV(wt), except for replacement of the R and U5 regions in the 3' LTR (15-166 bases in Fig. IB) by the SV40 late poly(A) signal sequence. With these two constructs, RNA transcribed through the 3' poly(A) site of SNV or through the SV40 poly(A) site can be detected. D17 (a dog osteosarcoma cell line) and chicken embryo fibroblast (CEF) cells were transfected with these constructs. Both cells are permis­sive for replication of SNV (Watanabe and Temin 1983). Forty-eight hours after transfection, RNA was isolated, digested with DNase I to remove contaminating plasmid DNA, and subjected to Northem blot hybridization. The percentage of processed RNA at the SNV 3' poly(A) site (full-length RNA) compared with total RNA (full-length plus readthrough RNAs) was 96% in D17 cells and 90% in CEF cells [lane 2, SNV(wt) in Fig. IC; data of CEF not shown]. The percentage for the SV40 poly(A) site was >97% in D17 cells [Fig. IC, lane 8, SNV(SV40)1. [The percentage for the SV40 poly(A) site in SNV(SV40) was not determined in CEF cells.] Thus, in the context of this vector, the poly(A) signal sequence of SNV is effi­cient.

Deletions of the viial genome do not decrease the level of full-length RNA

The second hypothesis, that viral genomic sequences outside the U3, R, and U5 regions (designated non-LTR regions) are responsible for the functional difference be­tween the 5' and 3' poly(A) sites, can be divided into two possibilities: (1) the non-LTR regions inhibit RNA 3'-end formation at the 5' poly(A) site; and (2) the non-LTR regions enhance RNA 3'-end formation at the 3' poly(A) site. To test these h3T)otheses, we deleted various re­gions of the SNV genome from pSNV(wt) (plasmids 3-7 in Fig. lA) and measured the level of RNA transcribed from these vectors by transfection and Northem blot analysis.

If the non-LTR regions inhibit the RNA 3'-end forma­tion of the 5' poly(A) site, these deletions should de­crease the level of total RNA by activating the function of the 5' poly(A) site. However, the relative levels of total RNA (full-length and readthrough RNAs) tran­scribed from these vectors were approximately the same as that of total RNA from SNV(wt) (Fig. IC, lanes 2-7). This result suggested that these deletions in the SNV

genome did not significantly decrease the level of total RNA in D17 and CEF cells (data of CEF not shown). Therefore, it is unlikely that the non-LTR regions inac­tivated the RNA 3'-end formation function at the 5' poly(A) site to allow expression of full-length RNA.

Deletions of the U3 region slightly decreased the per­centage of full-length RNA over total RNA from 96% to 85% in D17 cells (Fig. IC, lane 3). The deletion of the polypurine tract (ppt) and U3 regions decreased the per­centage of full-length RNA only 2% more (Fig. IC, lane 4). The deletion of the encapsidation sequence (E) region alone decreased the percentage of full-length RNA from 96% to 92% (Fig. IC, lane 5). The percentage of full-length RNA containing deletions of the E and U3 re­gions [SNV(AE/U3)], and of the E, ppt, and U3 regions [SNV(AE/ppt/U3)] were 81% and 83%, respectively (Fig. IC, lanes 6 and 7). Therefore, it is possible that the non-LTR and/or U3 regions enhance RNA 3'-end formation at the 3' poly (A) site to some limited extent.

Lengthening the distance between the cap site and the poly(A) site activates 3'-end formation of SNV RNA

We then constructed plasmids to test the third hy­pothesis, that the short distance between the cap site and the poly(A) site prevents RNA 3'-end formation at the 5' poly(A) site. The length of the R region was in­creased by insertion of the neo' gene fragment (Fig. 2A). The neo^ gene fragment was inserted into the Xhol site, as shown in Figure IB. These insertions did not seem to change the levels of transcription initiation, because the levels of total RNA (5' RNA and full-length RNA) tran­scribed from all the constructs used in our experiments were similar to one another (see Figs. 2B and 5B, below).

pR/U5(1423), in which the poly(A) site was located 1423 bases downstream of the cap site, gave two tran­scripts, 5' RNA and full-length RNA, detected with the 3' neo probe (Fig. 2B, lane 14 in the 3' neo probe panel). The percentage of 5' RNA relative to total RNA (5' RNA and full-length RNA) was 71% in this construct. When the distance from the cap site to the poly(A) site was 1042 bases [R/U5(1042)], the ratio of the RNAs was changed slightly (Fig. 2B, 61%, lane 13). In R/U5(471), the percentage of 5' RNA relative to total RNA was 8% (Fig. 2B, lane 10).

To detect full-length RNA for all of the constructs, a TK probe was used (Fig. 2B, TK probe panel). The level of full-length RNA of R/U5(90) was almost the same as that of R/U5(471) (Fig. 2B, lanes 9 and 10).

Similar deletions were made in the neo' gene fragment from the 3' end instead of the 5' end (Fig. 2A, II and 12). Transfection and Northem analyses of these 3'-deletion constructs were performed. When the distance between the cap site and the poly(A) site was 1023 bases, the per­centage of 5' RNA relative to total RNA was 51% (Fig. 2A, plasmid 12; Northem hybridization data not shown). When the distance between the cap site and the poly(A) site was 497 bases, the percentage of 5' RNA rel­ative to total RNA was <3% (Fig. 2A, plasmid 11; Northem hybridization data not shown).

GENES & DEVELOPMENT 2301

Cold Spring Harbor Laboratory Press on September 6, 2021 - Published by genesdev.cshlp.orgDownloaded from

Iwasaki and Temin

U3 neo gene R U5

5'RNA

r^-poly(A) site

full-length RNA

Probes ^ S ^ 5'neo 3'neo TK

9. 10.

*11.

*12.

13.

14.

*15.

pR/U5(90) pRAJ5(471)

pR/U5(497)

pRAJ5(1023)

pRAJ5(1042)

pRA)5(1423)

pR/U5(2586)

• AA.^A

Sizes of RNA(kb)

— 0.1/1.4 — 0.5/1.8 — 0.5/1.8 — i.0/2.4 — 1.0/2.4

— 1.4/2.8

— 2.4/3.8

*, Northern data not shown .Okb

MA (%)

8

<3 51

61

71

66

B •

k b

7 .4 — 4 .4 —

2 .4 —

1.4 —

3' neo

^ *rt •ii

ai

9

w*

«o 3 tf

•

10

probe

»s Tt o V) 3 tf

• • • • •

•

13

1 f*i r) ^ v> 3 «

«»

14

TK probe

• . — ^ 1—1

^ ^ «« »r» ;2 s ^ ^

m^

9 10

1 N f*5 3 !3 1 - ^ 1 - ^

»n in 5 3 tt OS

M

13 14

Figure 2. A short distance between the cap site and the poly(A) site inactivates RNA 3'-end formation function of SNV. [A] Con­structs used in the experiments. The numbers in parentheses in the names of the constructs represent the numbers of bases between the cap site and the SNV poly(A) site. pR/U5(25861 contained both the necf and hygrc/ fragments. (Left-hatched box) nee/ gene fragment; (right-hatched box) hygro' gene; (crosshatched boxes) probes used for Northern blot analyses. {Right] The percentage of processed RNA (i.e., the percentage of 5' RNA divided by total RNA in D17 cells (see Fig. 6j. Other symbols are as described in the legend to Fig. 1(A). {B] Northern blot analyses. Dl7 cells were transfected with the constructs shown [see the legend to Fig. 1(C)].

To eliminate the possibility that this distance-depen­dent RNA 3'-end formation depended on the specific na­ture of the fragment inserted into the R region, we in­serted the hygTcy gene fragment instead of the ne<y gene fragment to extend the length of the R region. The hy-gT(/ gene fragment was deleted from the 5' and 3 ' ends of the hygrcf fragment (the structures of these constructs are not shown). The relative ratios of 5' RNA to total RNA with these constructs were determined by Northern hybridization analysis as described above. The percentages of 5' RNA over total RNA were 58% for 1123 bases from the cap site to the poly(A) site, 54% for 936 bases, 44% for 858 bases, < 3 % for 377 bases, and 9% for 265 bases (20, 19, 18, 17, and 16, respectively, in Fig. 6, below; Northern hybridization data not shown). Therefore, the efficiency of 3'-end formation was deter­mined by the distance between the cap site and the poly(A) site and was independent of the sequence in­serted into the R region.

pR/U5(2586) was constructed to extend the length of the R region to 2586 bases, by insertion of both the hy-gr(f and necf gene fragments (Fig. 2A, plasmid 15). The insertion of the hygicf and neo" gene fragments gave a percentage of 5' RNA over total RNA of 66% (plasmid 15 in Figs. 2A and 6, below; Northern hybridization data not shown). Therefore, extending the length of the R re­gion >1400 bases (Fig. 2A and 6, below, plasmid 14) did not further enhance RNA 3'-end formation at the SNV poly(A) site.

Because the Northern analyses described above did not reveal the \QV&\ of 5' RNA from R/U5(90), which contains an intact LTR (Fig. 2A, plasmid 9), we decided to use the polymerase chain reaction (PCR) to measure the level of 5' RNA of R/U5(90) (Fig. 3A). It has been reported that PCR can be used to measure DNA or RNA levels in a quantitative manner (Delidow et al. 1989; Wang et al. 1989). The same preparation oi RNA used in the experiment of Figure 2B was subjected to reverse

transcription in the presence of a d(T)25_3o primer to copy the RNA into cDNA. Immediately after mixing all components, one-fifth of the reverse transcription mix­ture was removed and combined with 2 pi,Ci of [^^P]dCTP. The incorporation of ^^P was assessed by TCA precipitation to estimate the amount of cDNA synthesized in the reaction. The same amount of cDNA from each sample was then subjected to PCR with each set of primers (Fig. 3A).

To measure the levels of full-length RNA, the TK-2 and d(T) PCR primers and the TK-3 probe were used (Fig. 3A). The sizes of the PCR products were determined by agarose gel analysis. Because of the poly(A) tail of mRNA, the agarose gel analysis showed a smearing pat­tern from 100 to 1000 bases, in which a peak of the signal was —300-500 bases (data not shown). In Figure 3B, panel b (hybridization with the TK-3 probe), the one-fifth and one-twenty-fifth dilutions of R/U5(1423) cDNA prior to PCR indicated that the levels of the full-length PCR products were proportional to the levels of the original cDNA. When the level of the full-length product of R/U5(I423) was taken as 1.00, the levels of the full-length PCR products were 0.72 for R/U5(1042), 2.35 for R/U5(471), and 0.72 for R/U5{90). We found some variation between PCR and Northern analyses. The intensities of the full-length bands in the Northern analysis were 1.0 for R/U5(1423), 1.5 for R/U5(1042), 4.0 for R/U5(471), and 3.0 for R/U5(90) (Fig. 2B, lanes 14, 13, 10, and 9, respectively, in the TK probe panel). In the PCR analysis, the full-length RNAs from all of the con­structs were detected at a range between 0.72 and 2.35.

To measure the levels of 5' RNA, we used the R21-1 and d{T) PCR primers and the R51 probe (Fig. 3A). The sizes of the PCR products were determined by agarose gel analysis. The agarose gel analysis showed a smearing pattern from 100 to 1000 bases, in which a peak of the signal was —300-500 bases (data not shown). In Figure 3B, panel a (hybridization with the R51 probe), the one-

2302 GENES & DEVELOPMENT

Cold Spring Harbor Laboratory Press on September 6, 2021 - Published by genesdev.cshlp.orgDownloaded from

Distance-dependent polyadenylation of SNV RNA

U3 U5 tk polv(A) site

_1 5'RNA

R21-1 (RSI)

AAAA TTTT

full-length RNA

B a mi-l <— >TTTT

R/VS ^ N <^ (1423)

o t * 'f • ' * ^ •H i- l

m m w ft^ „ a ..

TK-2

TK.2<- • ^ T T T T

R/U5 (1423)

(TK-3)

pR/VS _ilf23L

-AAAA * TTTT

pH/P •» , ! - . «

3 S S

Figure 3. PCR to detect the 5' RNA of R/U5(90). |A) Scheme of primers and probes used in the PCR exper­iments. The positions of the primers (R21-1, TTTT, and TK-2) and probes (R51 and TK-3) indicate the ap­proximate positions in the RNAs. The names of the probes are in parentheses to distinguish between the primers and the probes. (Open regions) LTR se­quences; (stippled regions) tk gene; (broken lines) RNAs; (solid bars) primers; (hatched bars) probes; (double-headed arrows) regions amplified by PCR; (zigzag double lines) boundaries of skipped regions. Primers and probes: R21-1 is 21 bases long and is lo­cated from base 30 to base 50 in Fig. IB. R51 is 51 bases long and contains 48 bases complementary to the minus strand of the R region (base 104 to base 57 in Fig. IB). Therefore, R51 does not overlap the R2I-I and d(T) primers. TK-2 is 21 bases long and is located 58 bases upstream from the major poly(A) site of the tk gene. TK-3 is 21 bases and is located 11 bases up­stream from the major poly(A) site of the tk gene. TK-3 is complementary to the plus strand of the tk region and does not overlap the TK-2 and d(T)

primers, (fi) Slot blot analyses of the PCR products. The RNA used in Fig. 2B was subjected to cDNA synthesis (Maniatis et al. 1982) using AMV reverse transcriptase and an oligod(T)25_3o primer. This cDNA was subjected to PCR amplification (Delidow et al. 1989; Wang et al. 1989) and slot blot hybridization (Maniatis et al. 1982). The names of the constructs, the dilution magnitudes, and the primers used for amplification are shown at the top of the columns. The dilution magnitudes indicate the dilution of the cDNA prior to PCR to verify that the levels of PCR products were proportional to the levels of the original cDNA. Oligonucleotide probes are listed at left. pR/U5(1423) DNA and pH/P DNA (the c-rel DNA in pBluescripf) were included as positive and negative controls for the slot blot hybridization, respectively. Two separate experiments were performed. The results of these experiments were consistent with each other. Failure of the TK-3 probe to hybridize to PCR products from R21-1 and d(T) primers (panel a, TK-3 probe) indicated that the PCR product from R21-1 to the tk poly(A) site was not detectable in these experiments (see A and Discussion).

TK-3 Itllll I RSI II

fifth and one-twenty-fifth dilutions of R/U5(1423) cDNA prior to the PCR again verified that the levels of the 5' RNA PCR products were proportional to the levels of original cDNA. When the signal of the 5' RNA of R/U5(1423| was taken as 1.00, the signal of the 5' RNA of R/U5(1042) was 1.00, that of R/U5(471) was 0.10, and that of R/U5(90) was 0.02. In the Northern analysis, the relative levels of the 5' RNA were 1.00 for R/U5(1423), 2.20 for R/U5(1042), and 0.28 for R/U5(471) (Fig. 2B, 5' RNAs in lanes 14, 13, and 10, respectively, in the 3 ' neo probe panel). This PCR experiment showed that the 5' RNA of R/U5(90) was expressed at a level as low as the 5' RNA of R/U5(471).

One possible explanation for the observation that the levels of shorter 5' RNAs were lower than the levels of longer 5' RNAs was that the shorter RNAs were less stable than the longer 5' RNAs. To check the stability of these RNAs, we performed an actinomycin D chase ex­periment (Fig. 4). D17 cells were transfected with pR/U5(1423) and pR/U5(471) (Fig. 2A, plasmids 14 and 10, respectively). Thirty-six hours after transfection, ac­tinomycin D was added to the medium of the cells. After different times of incubation in the presence of actino­mycin D, RNA was isolated and subjected to Northern analysis (Fig. 4). The data from this Northern analysis were plotted, and the decay curves for the RNAs were biphasic (curves not shown). One phase was seen for up to 8 -15 hr and had a steep slope; the other phase was seen after 15 hr and had a gradual slope. The half-lives of the RNAs were estimated from the slopes of the first phase. The half-lives of the full-length RNA and 5' RNA

of R/U5(1423) were 7 and 8 hr, respectively. The half-life of the 5' RNA of R/U5(471) was 10 hr, and the half-life of the full-length RNA of R/U5(471) was 11 hr. Therefore, no significant differences in half-lives were found be­tween the 5' RNA and the full-length RNA of R/U5(1423) and between the 5' RNA and the full-length RNA of R/U5(471). This observation indicated that the low levels of the 5' RNAs from the constructs having short R regions were not the result of instability of these RNAs.

Short distance between the cap site and the poly(A) site in the tk and SV40 poly(A) signals did not inactivate RNA 3'-end formation

We then tested whether the distance-dependent func­tion of RNA 3'-end formation is a general phenomenon or is specific to the SNV poly(A) signal. To test these hypotheses, we made constructs containing the tk (Wagner et al. 1981) and SV40 late poly(A) signal se­quences (Fig. 5A, Wickens and Stephenson 1984). When the distance from the cap site to the poly( A) site of the tk gene was 1421 bases, the percentage of 5' RNA relative to total RNA (5' RNA and full-length RNA) was 85% (lane 28 in the 3' neo probe panel of Fig. 5B). When the distance from the cap site to the poly (A) site was 1031 bases, the percentage of 5' RNA relative to total RNA was 87% (Fig. 5B, lane 27). When the distance from the cap site to the poly(A) site was 460 bases, the percentage of 5' RNA relative to total RNA was still 81% (Fig. 5B, lane 26). When the distance from the cap site to the

GENES & DEVELOPMENT 2303

Cold Spring Harbor Laboratory Press on September 6, 2021 - Published by genesdev.cshlp.orgDownloaded from

Iwasaki and Temin

kb R/U5(1423)R/U5(471)

• 0 8 lS22f'0 8 1522 • h o u r s

nr numbers

Figure 4. Northern blot analysis to measure the half-life of 5' RNA and full-length RNA of pR/U5(1423) and pR/U5(47I). D17 cells were transfected with pR/U5(1423) and pR/U5(471). Thirty-six hours after transfection, actinomycin D was added to the medium of the cells. After various times of incubation in the presence of actinomycin D, RNA was isolated and sub­jected to Northern analysis using the 3' neo probe (see Fig. 2A1. Decay curves were obtained by microdensitometry scanning of Northern bands.

poly(A) site of the tk gene w as only 83 bases, the level of full-length RNA of tk{83] was not significantly increased compared with the three other tk constructs (Fig. 5B, lane 25 in the TK probe panel). Therefore, shortening the distance between the cap site and the poly(A) site had little effect on the RNA 3'-end formation function of the tk sequence (Fig. 5B).

A similar phenomenon was observed with constructs that contained the SV40 late poly(A) signal sequence (Fig. 5A, 21-24; Northern hybridization data of these constructs not shown). The only difference observed be­tween the tk and SV40 poly(A) signal sequences was the percentage of RNA processed at these poly(A) sites. The SV40 poly(A) signal was stronger than the tk signal se­quence. With the SV40 poly(A) signal, the percentage of RNA processed at the SV40 poly(A) site was 92-97% in all of the constructs (Figs. 5A and 6, plasmids 22-24).

Discussion

We tested three hypotheses to explain why retroviral RNA is polyadenylated at the 3 ' poly(A) site and not at the 5' poly(A) site: (1) The poiy(A) signal sequence in retroviral RNA is inefficient and allows readthrough at both poly(A) sites; (2) non-LTR regions in the viral genome are responsible for the functional difference be­tween the 5' and 3 ' poly(A) sites; and (3) the short dis­tance between the cap site and the poly(A) site prevents RNA 3'-end formation at the 5' poly(A) site.

To test the first hypothesis, that the poly(A) signal se­quence in retroviral RNA is inefficient and allows readthrough at both poly(A) sites, (1) the level and half-life of the RNA processed at the 3 ' poly(A) site compared with the full-length RNA was measured by transfection and Northern analysis and by actinomycin D chase ex­periments; and (2) the level of the 5' RNA from an intact LTR construct was measured by using PCR. Our results were (1) the poly(A) signal of SNV was efficient (Fig. IC);

(2) the lower level of the shorter 5' RNA was not the result of RNA instability (Fig. 4); and (3) a PCR experi­ment showed that the 5' RNA of the intact LTR con­struct [pR/U5(90)] was expressed at a low level (Fig. 3B). Therefore, we conclude that this hypothesis is unlikely.

The second hypothesis, that the non-LTR regions are responsible for the functional difference between the 5' and 3 ' poly(A) sites, was tested with constructs con­taining deletions of various sequences of the SNV genome (Fig. lA). The deletion of the SNV genomic se­quences did not decrease the level of total RNA com­pared with the parental vector [SNV{wt)] in D l 7 and CEF cells (Fig. IC; CEF data not shown). This deletion de­creased the levels of RNA processed at the SNV 3 ' poly(A) site slightly (Fig. IC; CEF data not shown). How­ever, if this phenomenon completely explained the func­tional difference between the 5' and 3 ' poly(A) sites in SNV, the relative amounts of RNA that are processed at the 5' poly(A) site and that pass through the 5' poly(A) site would be 80% and 20%, respectively. However, we found the RNA that was processed at the 5' poly{A) site and the RNA that passed through the 5' poly(A) site were <10% and >90%, respectively (Fig. 3). Therefore, even though the non-LTR regions and/or the U3 region change the percentage of RNA processed at the 3'

RL5 ik polylA) site

_J_RNA

full-leneth RNA

Sizes of Processed RNA(kb) RNA (%) 0.1/1.4

92

97

>97

81

87

85

. Northern data not shown

D 3' neo probe TK probe -J r

k b

4 . 4 —

2 . 4 —

1 .4—

0.3-

— ^ Tt

I I I 5 — ~ •<»

£ 1^ r) , ^ vC « • * M^

s — — 5 • 5 •^ •jj ^

l^ 2 5 2 6 2 7 2 8 14 2 5 2 6 2 7 2 8 1 4

Figure 5. A short distance between the cap site and the poly(A) site did not inactivate RNA 3'-end formation in the tk and SV40 sequences. {A] Constructs used in the experiments. The pSV40 series contained the SV40 poly(A) signal sequence (Wickens and Stephenson 1984) instead of the R and U5 se­quence. The ptk series contained the tk poly(A) signal sequence (Wagner et al. 1981) instead of the R and U5 sequence [see the legends to Figs. 1(A) and 2(A)1. [B] Northern blot analysis with constructs containing the tk poly(A) signal sequence. The de­scription of this figure is the same as that of Figs. IC and 2B.

2304 GENES & DEVELOPMENT

Cold Spring Harbor Laboratory Press on September 6, 2021 - Published by genesdev.cshlp.orgDownloaded from

Distance-dependent polyadenylation of SNV RNA

100 -

80 -

60 -

40 -

20 -

0 -

22 K

26

_23___J4 sy4Q

27 • 14

1 3 ^ 1 . ,Q/b20 14

P°12 / • 18tf 12

9 16 1 0 /

•17 " n I I r — T — T — I — I — 1 — r -

- ^ R/U5

1 ' • T 1

1.0 2.0 Distance between cap and

poly(A) sites (kb)

3.0

Figure 6. Summary of Northern analyses. Percentage of pro­cessed RNA was defined as the amount of 5' RNA divided by the amount of total RNA (5' RNA and full-length RNA) in each construct (for designation of RNAs, see Figs. 2A and 5A). The data plotted in this graph were the mean values of at least two experiments. Probes for constructs containing the 5' deletions of the neo^ gene (plasmids 10, 13, 14, and 26-28), the 3' dele­tions of the necf gene (plasmids 11, 12, 14, and 22-24), the 5' deletions of the hygrd' gene (plasmids 15, 17, 19, and 20), and the 3' deletions of the hygr<y gene (plasmids 15, 16, 18, and 20) were the 3' neo, 5' neo, 3' hygro, and 5' hygro probes, respec­tively (see Figs. 2A and 5A). (O, D, and A) Data from experi­ments with D17 cells; (•) data from CEF cells. The number near each point on this graph represents the number of the con­struct in Figs. 2A and 5A. The structures of the pR/U5 series containing the hygrc/ gene instead of the necy gene to extend the length of the R region are not shown (16-20). The results of some of the pR/U5 and ptk series in D17 cells are shown in Figs. 2B and 5B, but those in CEF and the results of the other constructs in Dl 7 cells are not shown. The data from pR/U5(90) (plasmid number 9) were obtained from the PCR experiments (Fig. 3B).

poly(A) site of SNV to some small extent, these regions cannot be responsible for the low expression of the 5' RNA.

The results of Northern hybridization and PCR anal­yses supported the third hypothesis, that the short dis­tance between the cap site and the poly(A) site prevents RNA 3'-end formation at the 5' poly(A) site. When the distance between the cap site and the poly(A) site was shorter than 500 bases, 3 - 9 % of the RNA was processed at the 5' SNV poly(A) site. As the distance became larger, the amount of RNA processed at the 5' SNV poly(A) site increased. At 1423 bases between the cap site and the poly(A) site, the percentage of RNA pro­cessed at the 5' SNV poly(A) site was 70% compared with total RNA. Extending the length of the R region >1423 bases did not further enhance RNA 3'-end forma­tion at the 5' SNV poly(A) site, hi contrast, the poly(A) signals of the tk and SV40 late genes functioned at high efficiency even with a distance between the cap site and the poly(A) site that was short enough to inactivate the SNV poly(A) signal. Therefore, this distance-dependent inactivation of RNA 3'-end formation seemed to be spe­cific for the SNV poly(A) signal sequence.

In the PCR analysis, we expected that a PCR product

extending from the R21-1 primer to the poly(A) site of the tic gene would be generated along with the PCR product between the R21-1 primer and the poly(A) site at the 3 ' end of the R region (Fig. 3A). However, the product from the R21-1 primer to the poly(A) site of the tk gene was not detected (Fig. 3B, slot blot of TK-3 probe in panel a). This observation could be explained by the low elongation efficiency of long sequences by reverse transcription and PCR. It has been reported that the re­verse transcription reaction is inefficient for long mRNA (>500 bases) when total RNA is used (Williams and Mason 1985). To avoid this problem of elongation efficiency, we used another set of primers, TK-2 and cl(T)25_3o, to detect full-length RNA (Fig. 3A).

Recently, Russnak and Ganem (1990) reported that se­quences upstream from the AAUAAA poly(A) signal are responsible for increasing the level of RNA processed at the 3 ' poly(A) site of hepatitis B virus (HBV) and that the upstream sequences can, to some extent, be replaced by the U3 sequences of SNV and HIV. They reported that the U3 sequences of SNV and HIV increased RNA pro­cessed at the HBV poly(A) site from 5% to 40 -50% rela­tive to total RNA (processed RNA and readthrough RNA), although the HBV sequence itself gave >95% of RNA processed at the HBV poly(A) site.

The percentage of RNA processed at the 3 ' poly(A) site of SNV(wt) was 96% in DI7 cells (Fig. IC, lane 2). The deletion of the U3 region decreased the percentage of full-length RNA from 96% to 85% in D17 cells (lane 3 in Fig. 1). The observation by Russnak and Ganem (1990) might reflect the difference of the levels of processed RNA between SNV(wt) and SNV(AU3) in the case of the SNV sequence.

During preparation of this manuscript, Sanfacon and Hohn (1990) published data suggesting that proximity to the promoter inhibits recognition of the cauliflower mo­saic virus poly(A) signal. These results are consistent with ours, indicating that a pararetrovirus and a retro­virus utilize similar mechanisms to avoid RNA 3'-end formation at the 5' poly(A) site. However, our data sug­gested that the poly(A) signals of the tk and SV40 genes behaved differently from the poly(A) signal of SNV.

We propose that the R and U5 regions of SNV RNA contain signals for RNA 3'-end formation and an addi­tional signal for the distance-dependent function. This additional signal may be a binding sequence for a protein that blocks RNA 3'-end formation.

Materials and methods

Nomenclature

Plasmid names are preceded by the letter p [e.g., pR/U5(90)], whereas transcripts and viruses made from these plasmids are not [e.g., R/U5(90)|.

RNAs that are polyadenylated at the first poly(A) site [e.g., polyadenylated at the poly(A) site of the 5' LTRj are designated 5' RNAs. RNAs that are transcribed through the first poly(A) site and polyadenylated at the second poly(A) site are desig­nated full-length RNAs. RNAs that are transcribed through the first and second poly(A) sites and are polyadenylated at the third poly(A) site are designated readthrough RNAs.

GENES & DEVELOPMENT 2305

Cold Spring Harbor Laboratory Press on September 6, 2021 - Published by genesdev.cshlp.orgDownloaded from

Iwasaki and Temin

Cells

D17 cells and CEF cells were grown as described previously (Watanabe and Temin 1983). For tests of RNA stability, cells were treated with 10 ixg/ml of actinomycin D (Sigma) (Temin 1963). Transfection of cells was performed as described (Kawai and Nishizawa 1984).

Plasmids

Recombinant DNA techniques were as described (Maniatis et al. 1982). All constructs in Figure lA were cloned mto the PstI-£coRl sites of pUC19, except pJD214Hy. pSNV(wt) con­tained an Xi^ol linker at the U3/R boundary in the 3 ' LTR (see Fig. IB), which pJD214Hy did not contain; this linker had no effect on RNA production (Iwasaki and Temin 1990). The se­quences used for cloning are defined as follows: (E) Bases 576-915 [BamHl-BamUl fragment); (ppt) bases 2017-2077 {CM-Sacl fragment); (U3 in the 3 ' LTR) bases 2078-2511 [Sacl-Xhol fragment); |R/U5 in the 3 ' LTR) bases 2512-2663; (the hygicf gene) bases 912-2017 {Xbal-Clal fragment) in pJD214Hy (Dougherty and Temin 1986). The tk fragment com­prises bases 1200-3455 in pFG5 (Colbere-Garapin et al. 1979); the SV40 sequence comprises bases - 59 to -I- 79 in pSVL - 171/ 4-79 (Wickens and Stephenson 1984).

pR/U5(90) was a subclone of pSNV{wt), running from 12 bp upstream from the 5' end of the 3 ' LTR through the entire tk fragment. All constructs in Figures 2A, 5A, and 6 were cloned into the Hind i and £coRI sites of pUC19. The 1325-bp frag­ment of the nety gene was defined as 942-2267 bp of pJD214Neo (Dougherty and Temin 1986). All of the necf and hygrc/ fragments of the constructs in the pR/U5, pSV40, and ptk series contained an Xhol site at the 5' end and a Sail site at the 3 ' end, which were created during cloning. (Structures of the pR/U5 series containing the hygrc/ gene fragment are not shown.) These fragments were cloned into the Xhol site of pR/U5(90), which was located 2 bases downstream from the cap site (see Fig. IB).

The SV40 poly(A) signal sequence contained bases - 5 9 to -h79 of pSVL-141/-h79 for the pSV40 series constructs (Wickens and Stephenson 1984). The tk poly(A) signal sequence contained bases - 7 9 to -^57 of pTKl for the ptk series con­structs (Wagner et al. 1981). For the SV40 constructs, the poly(A) site was designated as + 1 . For the tk constructs the major poly(A) site was designated as -f 1, and a minor poly(A) site was located 11 bases downstream from the major poly(A) site (Zhang and Cole 1987).

RNA analyses

Cellular RNA was isolated as described previously (Berger and Birkenmeier 1979) with minor modifications. Cells were scraped, washed with TD buffer [140 mM NaCl, 5 mM KCl, 25 mM Tris-Cl, and 0.7 mM Na2HP04 (pH 7.5)], and incubated at 4°C for 5 min in 0.3% NP-40, 10 mM Tris-Cl (pH 7.5), 10 mM NaCl, 1.5 mM MgClj, 3 mM CaCli, and 10 mM ribonucleoside-vanadyl complex (BRL). After precipitation of nuclei, the super­natant fraction was added to one-fifth volume of 2.5% SDS, 500 mM Tris-Cl (pH 9.0), and 50 mM EDTA, extracted with phenol, precipitated with ethanol, and treated with RQl DNase (Pro-mega) in the presence oi RNasin (Promega) as described by the supplier.

Northern blot analysis was performed with DNase-treated RNA (20 |xg) as described (Maniatis et al. 1982). Labeling of the DNA fragments to make radioactive probes was performed with a random priming procedure (Feinberg and Vogelstein 1984). The 5' and 3 ' neo probes are fragments from the 5' end to

the Eagl site and from the Ncol site to the 3 ' end of the ne(/ gene fragment, respectively. The 5' and 3 ' hygro probes are fragments from the 5' end to the £coRI site and from the Hindi site to the 3 ' end of the hygrcf gene fragment.

PCR

Ten micrograms of total RNA was subjected to the reverse transcriptase reaction, as described (Maniatis et al. 1982). One-fifth of the reverse transcriptase mixture was separated from the rest of the mixture and incubated with P^P]dCTP (2 |xCi) (Delidow et al. 1989). Incorporation of ^^P was monitored by TCA precipitation. Two to 10 ng of cDNA, as calculated from the incorporation of ^^P, was subjected to PCR. For each PCR, the same amount of cDNA from each sample was used. PCR was performed for up to 22 cycles with the TK-2 and d(T) primers or 25 cycles with the R21-1 and d(T) primers. At this number of PCR cycles, the levels of products were proportional to the levels of original cDNA (see Fig. 3B). The concentration of MgClj was 0.5 mM for R21-1 and d(T)25_3o primers, and 1.0 mM for TK-2 and d(T)25_3o primers. The temperatures for dena-turation, annealing, and extension for both sets of primers were 94''C, 45°C, and 72°C, respectively (Ohara et al. 1989). Six mi­croliters of the PCR products was denatured and blotted to ni­trocellulose, as described (Maniatis et al. 1982). The probe oli­gonucleotides were end-labeled by T4 polynucleotide kinase, as described (Maniatis et al. 1982). Hybridization of the nitrocel­lulose with •' P oligonucleotides was performed at 47°C, as de­scribed (Maniaris et al. 1982). pR/U5(1423) DNA was linearized with £coRI, and pH/P DNA was linearized with Hindlll. These plasmids were included in each slot blot hybridization as posi­tive and negative controls, respectively. The intensities of the signals were measured by microdensitometry.

R21-1, CCTACACATTGTTGTTGTGAC, is located 9 bases downstream from the 5' end of the R region. R51 is 51 bases long and contains 48 bases complementary to the minus strand of the R region. R51 does not overlap the R21-1 and d(T) primers. For details about the locations of primers and probes in the PCR analysis, see Figure 3A. TK-2, GAAACACGGAAG-GAGACAATA, is located 58 bases upstream from the major poly(A) site of the tk gene. TK-3 is 21 bases long and is located 11 bases upstream from the major poly(A) site of the tk gene. TK-3 is complementary to the plus strand of the tk region and does not overlap the TK-2 and d(T) primers.

In Figure 3B, panel a, hybridization with the TK-3 probe did not detect any signals. In panel b, hybridization with the R51 probe detected only very weak signals.

Acknowledgments

We thank B. Fritz, K. Krebsbach, J. Rein, and R. Wisniewski for technical assistance; and J. Dahlberg, M. Hannink, W.-S. Hu, V. Pathak, G. Pulsinelli, I. Riegel, J. Ross, S. Yang, and J. Zhang for helpful comments on the manuscript. This work was supported by grants CA-07175 and CA-22443 from the U.S. PubUc Health Service. H.M.T. is an American Cancer Society Research Pro­fessor.

The publication costs of this article were defrayed in part by payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 USC section 1734 solely to indicate this fact.

References

Benz, E.W., R.M. Wygro, B. Nadal-Ginard, and D, Dino. 1980. Moloney murine sarcoma proviral DNA is a transcriptional

2306 GENES & DEVELOPMENT

Cold Spring Harbor Laboratory Press on September 6, 2021 - Published by genesdev.cshlp.orgDownloaded from

Distance-dependent poiyadenylation of SNV RNA

unit. Nature 288: 665-669. Berger, S.L. and C.S. Birkenmeier. 1979. Inhibition of intrac­

table nuclease with ribonucleoside-vanadyl complexes: Iso­lation of messenger ribonucleic acid from resting lympho­cytes. Biochemistry 18: 5 I43-5I49 .

Bimstiel, M.L., M. Busslinger, and K. Strub. 1985. Transcription termination and 3 ' processing: The end is in site! Cell 41: 349-359.

Boeke, J.D., D.J. Garfinkel, C.A. Styles, and G.R. Fink. 1985. Ty elements transpose through an RNA intermediate. Cell 40:491-500.

Bohnlein, S., J. Hauber, and B.R. Cullen. 1989. Identification of a US-specific sequence required for efficient poiyadenyla­tion within the human immunodeficiency virus long ter­minal repeat. /. Virol. 63: 421-424.

Clare, J. and P. Farabaugh. 1985. Nucleotide sequence of a yeast Ty element: Evidence for an unusual mechanism of gene expression. Proc. Natl. Acad. Sci. 82: 2829-2833.

Colbere-Garapin, F., S. Chousterman, F. Horodniceanu, P. Kourilsky, and A.-C. Garapin. 1979. Cloning of the active thymidine kinase gene of herpes simplex virus type 1 in Escherichia coh K-12. Proc. Natl. Acad. Sci. 76: 3755-3759.

Delidow, B., f.J. Peluso, and B.A. White. 1989. Quantitative measurement of mRNAs by polymerase chain reaction. Genet. Anal. Techniques 6: 120-124.

Dougherty, J.P. and H.M. Temin. 1986. Fiigh mutation rate of a spleen necrosis virus-based retrovirus vector. Mol. Cell Biol. 168: 4387-4395.

. 1987. A promoterless retroviral vector indicates that there are sequences in U3 required for 3 ' RNA processing. Proc. Natl. Acad. Sci. 84: 1197-1201.

Elder, R.T., E.Y. Loh, and R.W. Davis. 1983. RNA from the yeast transposable element Tyl has both ends in the direct repeats, a structure similar to retrovirus RNA. Proc. Natl. Acad. Sci. 80: 2432-2436.

Feinberg, A.P. and B. Vogelstein. 1984. A technique for radiola-beling DNA restriction endonuclease fragments to high spe­cific activity. Anal. Biochem. 137: 266-267.

Gorman, CM. , L.F. Moffat, and B.H. Howard. 1982. Recombi­nant genomes which express chloramphenicol acetyltrans-ferase in mammalian cells. Mol. Cell. Biol. 2: 1044-1051,

Flumphrey, T. and N.J. Proudfoot. 1988. A beginning to the bio­chemistry of poiyadenylation. Trends Genet. 4: 243-245.

Iwasaki, K. and H.M. Temin. 1990. The U3 region is not neces­sary for 3 ' end formation of spleen necrosis virus RNA. /. Virol. 64: 6329-6334.

Kawai, S. and M. Nishizawa. 1984. New procedure for DNA transfection with polycation and dimethyl sulfoxide. Mol. Cell. Biol. 4: 1172-1174.

Maniatis, T., E.F. Fritsch, and J. Sambrook. 1982. Molecular cloning: A laboratory manual. Cold Spring Harbor Labora­tory, Cold Spring Harbor, New York.

Manley, J.L. 1988. Poiyadenylation of mRNA precursors. Bio-chim. Biophys. Acta. 950: 1-12.

Ohara, O., R.L. Dorit, and W. Gilbert. 1989. One-sided poly­merase chain reaction: The amplification of cDNA. Proc. Natl. Acad. Sci. 86: 5673-5677.

Russnak, R. and D. Ganem. 1990. Sequences 5' to the poiya­denylation signal mediate differential poly(A) site use in hepatitis B viruses. Genes Dev. 4: 764-776.

Sanfacon, H. and T. Hohn. 1990. Proximity to the promoter in­hibits recognition of cauliflower mosaic virus poiyadenyla­tion signal. Nature 346: 81-84.

Shomotohno, K., S. Mizutani, and H.M. Temin. 1980. Sequence of retrovirus provirus resembles that of bacterial transpos­able elements. Nature 285: 550-554.

Stinski, M.F. and T.J. Roehr. 1985. Activation of the major im­mediate early gene of human cytomegalovirus by cis-acting elements in the promoter-regulatory sequence and by virus-specific trans-acting components. /. Virol. 55: 431-441.

Temin, H.M. 1963. The effects of actinomycin D on growth of Rous sarcoma virus in vitro. Virology 20: 577-582.

Wagner, M.J., J.A. Sharp, and W.C. Summers. 1981. Nucleotide sequence of the thymidine kinase gene of herpes simplex virus type 1. Proc. Natl. Acad. Sci. 78: 1441-1445.

Wang, A.M., M.V. Doyle, and D.F. Mark. 1989. Quantitation of mRNA by the polymerase chain reaction. Proc. Natl. Acad. Sci. 86 :9717-9721.

Watanabe, S. and H.M. Temin. 1983. Construction of a helper cell line for avian reticuloendotheliosis virus cloning vectors. Mol. Cell. Biol. 3: 2241-2249.

Weiss, R., N. Teich, H. Varmus, and J. Coffin. 1985. RNA tumor viruses. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York.

Wickens, M. and P. Stephenson. 1984. Role of the conserved AAUAAA sequence: Four AAUAAA point mutations pre­vent messenger RNA 3 ' end formation. Science 226: 1045-1051,

Williams, J.G. and P.J. Mason. 1985. Primer extension in hy­bridization in the analysis of RNA. In Nucleic acid hybrid­ization: A practical approach (ed. B.D. Hames and S.J. Higgins), pp. 152-160. IRS Press, Oxford.

Zhang, F. and C.N. Cole. 1987. Identification of a complex as­sociated with processing and poiyadenylation in vitro of herpes simplex virus type 1 thymidine kinase precursor RNA. Mol. Cell. Biol. 7: 3277-3286.

GENES & DEVELOPMENT 2307

Cold Spring Harbor Laboratory Press on September 6, 2021 - Published by genesdev.cshlp.orgDownloaded from

10.1101/gad.4.12b.2299Access the most recent version at doi: 4:1990, Genes Dev. 

  K Iwasaki and H M Temin  between the cap site and the poly(A) site in spleen necrosis virus.The efficiency of RNA 3'-end formation is determined by the distance

  References

  http://genesdev.cshlp.org/content/4/12b/2299.full.html#ref-list-1

This article cites 28 articles, 16 of which can be accessed free at:

  License

ServiceEmail Alerting

  click here.right corner of the article or

Receive free email alerts when new articles cite this article - sign up in the box at the top

Copyright © Cold Spring Harbor Laboratory Press

Cold Spring Harbor Laboratory Press on September 6, 2021 - Published by genesdev.cshlp.orgDownloaded from