Transcriptional (p40x) and post-transcriptional (p27x-III) regulators are ...

Proc. Natl. Acad. Sci. USA Vol. 84, pp. 3653-3657, June 1987 Biochemistry

Transcriptional (p40x) and post-transcriptional (p27x-III) regulators are required for the expression and replication of human T-cell leukemia virus type I genes (pX gene function/gag expression/spllcing/trans-activation)

JUN-ICHIRO INOUE, MITSUAKI YOSHIDA*,

AND

MOTOHARU SEIKI

Department of Viral Oncology, Cancer Institute, Kami-Ikebukuro, 1-37-1, Toshima-Ku, Tokyo 170, Japan

Communicated by Hidesaburo Hanafusa, February 18, 1987 (received for review December 17, 1986)

The pX sequence of human T-cell leukemia ABSTRACT virus type I codes for three products: p40X, p27x-m, and p21m-M. p40" is a transcriptional trans-activator that activates not only the viral long terminal repeat but also cellular genes for interleukin 2 and its receptor. p27'-m and p2lx-H, are not required for transcriptional activation, and their functions were unknown. Cotransfection experiments with defective human T-cell leukemia virus type I proviruses and various pX expression plasmids revealed that p27"-"', in addition to p40", was required for gag gene expression. Furthermore, it was shown that p27""' induced accumulation of a high level of unspliced viral gag mRNA. These results indicate that p27'-m is a post-transcriptional modulator of viral RNA whose transcription has been fully activated by p4OX.

tion from the LTR of the proviral genome and that p27"-"l is a post-transcriptional modulator that controls the level of gag mRNA.

Human T-cell leukemia virus type I (HTLV-I) (1, 2) is closely associated with a T-cell malignancy, adult T-cell leukemia (2-4), and is thought to be involved in the mechanism of leukemogenesis (5). Our work involves the role of the pX region (6), located between the env gene and 3' long terminal repeat (LTR), in transcription of the proviral genome and certain cellular genes. The pX sequence in HTLV-I contains four overlapping open reading frames (ORFs), I-IV (6). The following three pX proteins have so far been identified: p40x encoded by ORF IV (7-10) and p27x-"', and p21"-"', encoded by ORF III (11). p40x has been shown to be a trans-acting transcriptional activator that promotes transcription from the LTR (12-18) responding to the enhancer (19-21); thus, p40x is indispensable for efficient viral gene expression (22). p40x also activates cellular genes for interleukin 2 and its receptor (23, 24), which may explain the polyclonal growth of infected T cells that are possibly involved in the development of adult T-cell leukemia. p27`-"' and p21-"I are not required for these transcriptional activations (16, 18, 25), and their functions have not been well characterized. However, it has been suggested that these two pX proteins are important in viral gene expression in view of the observation that p27""I is localized in the nucleus (11). The fact that p27`"' and p40x are encoded by a single mRNA species (25, 26) led us to the hypothesis that p27X-""' operates in the same process as p40x or in a closely associated process. We have reported (27) preliminary evidence that both p40x and p27`"' are required for expression of the gag gene in a cell line containing proviruses. However, we could not analyze the mechanism of activation because of complications with the integrated proviruses. In this work, we have analyzed the function of p27x-"II in gag gene expression, which was induced by cotransfection of proviral DNA and pX expression plasmids. The results showed that expression of p40x alone could induce transcrip-

intact only the coding sequences for gag and protease. The protease gene, however, has an amber mutation in its coding sequence and thus cannot code for an active protease (28). A Sma I-Pvu II fragment of the thymidine kinase gene containing the transcriptional terminator (29) was inserted downstream of the 3' LTR. The plasmid pGAGdb contains a provirus with a large deletion between the Bgl II site (bp 2760) and the Sma I site (bp 8306) in the 5' region of the 3' LTR. The expression plasmid pMTCXdb and its mutants were described (16, 18, 24, 25) and are illustrated in Fig. 1B. Briefly, a cDNA sequence of 2.1-kilobase (kb) pX mRNA was truncated at exon 1 and inserted downstream of the metallothionein promoter in the wild-type pMTCXdb (16, 18, 24). pMTCXds-ATG1 has a deletion in the 5' region including the first ATG, pMTCXdb-ATG4 has a base replacement at the fourth ATG codon, and pMTCXds-ATG1&4 (25) has a deletion and a base substitution at the first and fourth ATG codons, respectively. pMTCXsn-IVter has a termination codon induced by a cytidine to thymidine point mutation in ORF IV at the position 73 bases from the initiator ATG codon. Transfection. FL cells (6 x 105 cells), a human amnion cell line, were seeded on 60-mm plates and transfected the following day with 6 ,ug of the defective provirus and with 3 ,ug of pX expression plasmid by the calcium phosphate method as described (13). After 40 hr, cells were harvested. Some cells were examined by indirect immunofluorescence staining, and protein and RNA were extracted from other cells. Analysis of Protein and RNA. Viral antigens in the transfected cells were characterized by the blotting procedure using serum of an adult T-cell leukemia patient or rabbit antiserum against synthetic peptides. Cytoplasmid RNA was extracted with vanadyl complexes according to the method of Berger and Birkenmeier (30). RNA containing poly(A) was

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

MATERIALS AND METHODS Plasmids. The plasmids pGAGdm and pGAGdb, both containing a defective HTLV-I provirus, were derived from the whole provirus clone XATK-1 (6). The plasmid pGAGdm contains a provirus with two deletions: one deletion from the Bgl II [base pairs (bp) 2760 from the 5' end of the provirus genome] to the BamHI site (bp 6098) and the other from the Mlu I (bp 7480) to the Sma I (bp 8360) site as illustrated in Fig. LA. These deletions encompass most of pol and the 5' region of env, as well as most of the pX coding sequence, leaving

Abbreviations: HTLV-I, human T-cell leukemia virus type long terminal repeat; ORF, open reading frame. *To whom reprint requests should be addressed.

I;

LTR,

3654

Biochemistry: Inoue et al. A

Proc. Natl. Acad. Sci. USA 84 (1987)

".-

exon1 _-. O_

exon2 W

cEL= LTR

< °

s

rtc a93

gag

exon3

,, .

I~~~~~~~~~~~~" ra 4< ' E3 v0 r

o E

v

LRgag

pol

pr

pX mRNA

~i~ll-0TK

AATK-1

p

env

pGAGdm

V~~~~~~~~~~~~~~~~

I

pGAGdb exonl m*

gag

Sia

I 2

1

B

Probes

Bg III 3

5

4

aml lexon2i

pX mRNA ii~-. ,&- A aa

tel:

I I I I

p40

8

9 kb

exon 3

II

p27X-111

p20xX

7

6

--O.

I

LLTR _nrl

,v lv x_

9 .9 L\Xr: ... 1. 1.

.............. .......... . . .

pMTCXdb

ACG(III) v

i

.

::.......

pMTCXds-ATG 1 &4

TAA (IV)

v

"Im

U

L----i

-. I * I v

....... ......

......

.....

.......... ..........

..........

.........

pMTCXsn-IV ter

FIG. 1. Construction of defective HTLV-I provirus and expression plasmids containing pX sequence. (A) XATK-1 is the original provirus isolated from a leukemia patient (6). Open bars in the HTLV-I genome represent ORFs or the LTR, and prt is the coding sequence for viral protease (28). Closed arrows with S.D. or S.A., sites for splicing donor or acceptor, respectively. Open triangles under the exons, ATG codons for initiation of each pX protein. TK, transcriptional terminator of the thymidine kinase gene. Closed bars, regions used for hybridization probes. x in prt, amber mutation. (B) Horizontal thick arrows represent pX mRNA and horizontal thin arrows indicate initiation and termination site of each product. Three constructions for pX expression are illustrated. MT, the metallothionein promoter. Open triangles, active initiation codon ATG; closed vertical arrows, modified codon as indicated. pMTCXsn-IVter has a repeat of the 3' half of exon 2 and a termination codon in ORF IV 73 bases from the 5' end of ORF IV.

isolated on oligo(dT)-cellulose and analyzed by the blotting procedure after denaturation with 2.2 M formaldehyde. The RNA on the filter was hybridized with a 32P-labeled DNA fragment containing exon 1 (bp 354-616), the gag sequence (bp 851-1370), or the Sma I-Bgl II fragment (bp 848-2760) of pGAGdb (see Fig. 1A). Hybridization was carried out at 420C in buffer containing 50% (vol/vol) formamide and 4x SSC [lx SSC = 0.15 M NaCl/0.015 M sodium citrate, pH 7.5). The filter was finally washed in 0.1 x SSC for 15-20 min at 650C.

RESULTS Effect of p27x-m on gag Expression. We have suggested (27) that p27x-111, p21X-4II, or both are required for expression of the gag gene in the integrated proviral genome. To study the function of p27x-111, eliminating the possibility of artifacts arising from the provirus integrating in an unex-

pected manner into the cellular genome, we conducted experiments using the transient assay in which pX expression plasmids and proviral DNA were cotransfected. We constructed a defective provirus, pGAGdm, carrying an intact gag gene, a mutant protease gene (see ref. 28), and two deletions encompassingpol, env, and the pX coding sequence (Fig. LA). Thus, pGAGdm can code for the gag precursor, PrS3, but not for active protease, polymerase, envelope, or pX proteins. In addition, this provirus contains the transcriptional termination signal of the thymidine kinase gene to terminate a possible readthrough at the 3' LTR. pGAGdm was cotransfected into the FL cell line with pX expression plasmids (Fig. 1B), and the viral proteins were analyzed 2 days later. pGAGdm alone failed to induce expression of a detectable level of gag precursor PrS3 (Fig. 2AI, lane f), whereas cotransfection with the wild-type plasmid pMTCXdb gave a strong band of 53 kDa (Fig. 2A1, lane a). The molecular size

_ -~p27

Biochemistry: Inoue et al.

A

_

-

_--

A2

-.

a

5

A3

B1

a b c d e f g

Z

-

I

0.

lk

..

-

0

Pr53

x

a b c d e f

-~4;W

B2

p4O

x-II


B3

-.

w..

*_.-

-

Pr53

p40

__III

0_4-4WO

p27

FIG. 2. Protein blot analysis of PrS3 induced by pX expression in transient assay. Defective provirus pGAGdm (A) or pGAGdb (B) was cotransfected with 3 ,ug of pMTCXdb (lane a), of pMTCXdsATG1&4 (lane b), of pMTCXds-ATG1&4 + pMTCXsn-IVter (lane c), of pMTCXsn-IVter (lane d), of pMTCXds-ATG1 (lane e), of Escherichia coli DNA (lane f), or of pMTCXdb-ATG4 (lane g); and proteins (30 ,ug) were subjected to immunoblot analysis. All lanes were on the same autoradiogram. To avoid artifactual contamination into the next lane, the lanes were electrophoresed in various orders and then reassembled in the same order in this figure. Al, A2, Bl, and B2 were probed with an adult T-cell leukemia patient serum as the first antibody; A3 and B3 were probed with rabbit antibodies against C-terminal peptide III of p27x-m (11). The p27x-.II in lane d of A3 and B3 was expressed by a plasmid pSRaCXsn-IVter in which the coding sequence was directed with a strong promoter SRa (simian virus 40 enhancer plus HTLV-I LTR).

of 53 kDa was identical to gag precursor identified (31). The specificity of Pr53 was confirmed by monoclonal antibody to p19 (data not shown). As expected from the presence of an amber mutation in the protease gene, Pr53 was not processed into p24, p19, and p15. The dense bands observed in lower molecular weight regions are those of cellular proteins nonspecifically reacting with this particular patient's serum; the exact same bands were seen in untransfected cells (data not shown). Thus, it is clear that the 53-kDa band is the gag precursor protein, and it can, therefore, be concluded that expression of the gag gene is dependent on pX functions supplied in trans by the pX expression plasmid. To analyze the contribution of each pX protein, mutants of the pX expression plasmid were tested. Cotransfection with the p40" expression plasmid pMTCXds-ATG1&4 failed to induce significant expression of Pr53 (Fig. 2A1, lane b). Expression of p40" in this experiment was similar to the level detected in lane a as ascertained by protein blot (Fig. 2A2, lane b) and also by trans-activation of pLTR-CAT (data not shown), thus it is clear that p40" alone was not sufficient for efficient expression of Pr53. Cotransfection of pMTCXsnIVter, which codes for only p27"-II and p21"-"'1 but not for p40", also failed to induce Pr53 expression (Fig. 2AI, lane d). For further confirmation of this conclusion, a complementation experiment was carried out: Cotransfection of pMTCXds-ATG1&4 and pMTCXsn-IVter with pGAGdm induced PrS3 expression as efficiently as the wild-type plasmid (Fig. 2A1, lane c). The effects of p27"-.11 and p21"-"' were distinguished by the use of pMTCXds-ATG1, which can code for p40" and p21"" but not for p27"-"'. With pMTCXdsATG1, only a low-level expression of PrS3 was observed (Fig. 2A1, lane e). On the other hand, pMTCXdb-ATG4 coding for p40x and p27`"' but not for p21"I induced PrS3 as efficiently as with the wild-type plasmid (Fig. 2A1, lane g). Thus, it was concluded that p40" and p27x-III are primarily required for gag expression. In these experiments, a similar level of p27x-III expression was confirmed by rabbit antibod-

3655

ies against the C-terminal peptide of p27"-11' (Fig. 2A3), except in Fig. 2AI, lane d. To ascertain the effect of p27""', another strong promoter SRa, which was constructed from the simian virus 40 enhancer and HTLV-I LTR (unpublished data), was used to express p27"-", and the particular result was shown in Fig. 2A3, lane d. Using this extract, absence of the Pr53 band was confirmed as in Fig. 2A1, lane d (data not shown). Effect of p27'm on the Viral RNA Synthesis. For determination ofthe mechanism ofgag gene induction by p27"11I, the levels of viral RNA synthesized in transfected cells were analyzed. Cytoplasmic RNA containing poly(A) was isolated 2 days after transfection and analyzed by the blotting procedure with various DNA probes. In cells transfected with pGAGdm and the wild-type pX expression plasmid, an RNA band of 4.6 kb was detected with a probe containing the gag sequence (gag probe) (Fig. 3A, lane a). The 4.6-kb RNA was concluded to be genomic-sized RNA because (i) it was the same size as the provirus construct and (ii) the RNA contained the gag and R sequences (as shown below). In contrast, only a very faint band of4.6-kDa RNA was detected in cells in which either p40" or p27"-'1 was expressed (Fig. 3A, lanes b and d-f). Coexpression of p271-III and p40" by independent plasmids, however, induced the 4.6-kb RNA (Fig. 3A, lane c). Evidently, the 4.6-kb genomic RNA was observed concomitantly with expression of PrS3. Therefore, we concluded that both p27"-'1 and p40" are required for PrS3 expression and that expression is regulated at the mRNA level. B abc de f g

A abcde f g

9

*

4.6-

_

2.4 -

-

1.2

**

D

kb *

* -

_

kb 3.4.-

3.22.4 -

4.6.2.4-

o

0

E abcdef

abcd ef

3.2-

kb

kb

kb 4.6-S

3.4_

C 1 234 5

a

oo

Sa

1.2

.-SOS

F ab cd ef

kb -

3.4_ 3.2-*

FIG. 3. Viral RNA expression in cells cotransfected with defective provirus DNA and pX expression plasmid. FL cells were cotransfected either with defective provirus pGAGdm (A, B, and C) or with pGAGdb (D, E, and F) and with one of the pX plasmids. Combinations of the pX expression plasmids (lanes a-g) are the same as in Fig. 2. In C, cells were transfected with 3 ,ug of pMTCXdsATG1&4 and the following amounts of pMTCXsn-IVter: O ,g (lane 1), 0.3 ug (lane 2), 1 ,ug (lane 3), 3 ,g (lane 4), and 6 ,ug (lane 5). Cytoplasmic RNA was prepared, and an equal amount of RNA was applied for blot analysis. The filter was probed with a 32P-labeled DNA fragment of gag probe (A and D), exon 1 probe (B, C, and E), or Sma I-Bgl II probe (F). The 2.4-kb band is the transcript of the pX expression plasmids.

3656

Biochemistry: Inoue et al.

For detection of all the viral RNA species transcribed from the LTR, the filter was probed with a DNA fragment containing the exon 1 sequence of the pX mRNA, which corresponds to bases + 1 (cap site) to + 132 (the first splicing donor) (see Fig. 1A). Three RNA bands of 4.6, 2.4, and 1.2 kb were detected in cells transfected with pGAGdm and the wild-type pX plasmid (Fig. 3B, lane a). The two RNA bands of 4.6 and 1.2 kb were derived from pGAGdm: the 4.6-kb RNA is a primary transcript that is identical to that detected with the gag probe (Fig. 3A, lane a), and the 1.2-kb RNA corresponds in size exactly to RNA spliced between the first donor site in the LTR and the second acceptor site, which are both retained in the defective provirus (Fig. LA). In addition, the 1.2-kb RNA hybridized with the exon 1 and pX probes, but not with the gag or thymidine kinase probes. These properties indicate that the 1.2-kb RNA is spliced at the expected donor and acceptor sites. The 2.4-kb RNA was derived from the pX expression plasmid, since it was present in cells transfected with the pX plasmid alone. The presence of spliced 1.2-kb RNA was found to parallel p40" expression (Fig. 3B, lanes a-c, e, and g) but was unrelated to the expression of p27`"' and p21"I (Fig. 3B, lanes b, d, and g). These results are consistent with the previous conclusion that p40x alone is sufficient for transcriptional activation. In contrast to the 1.2-kb RNA, the 4.6-kb RNA was detected only in cells in which both p27x-"I' and p40x were expressed by the wild-type or complementary pX plasmids (Fig. 3B, lanes a, c, and g) but not in cells with p27x-"' alone (Fig. 3B, lane d). However, we could not detect a significant level of 2.4-kb RNA derived from pMTCXsn-IVter. To confirm the effect of p27"`T1, we expressed p27"`1' using another strong promoter SRa. As shown in Fig. 2A3, lane d, p27"`1' was detected without p40x in a level similar to that in lane a. However, we could not detect any profound effect on the viral RNA, which confirmed the result in Fig. 3B, lane d. Therefore, it can be concluded that p27x-"' alone is not sufficient and that both p27"-I" and p40x are required for accumulation of the 4.6-kb RNA. Furthermore, when the quantity of the p27x-III expression plasmid was increased relative to that of the p40x expression plasmid, the intensity of the 4.6-kb band increased with a concomitant decrease in that of the 1.2-kb band (Fig. 3C. With an excess of p27`-"', the transcripts were almost all the 4.6-kb RNA. These observations indicate that p27""' modulates the post-transcriptional process that eventually increases the level of unspliced gag mRNA. From these results, the process affected by p27x-"' could be splicing, stabilization, or transport specific to the gag mRNA or any combination of these processes. Requirement of p27xm for Viral RNA Without Splicing Acceptor Signals. The observation that only unspliced gag RNA was dependent on p27""I strongly suggested that the splicing process was a target of p27""'. To test this hypothesis, we constructed another defective provirus pGAGdb, which carries a large deletion between the Bgl II and Sma I site encompassing pol, env, andpX (Fig. 1A). Thus, pGAGdb does not retain a splicing acceptor signal used for viral gene expression. As shown in Fig. 2B, expression of Pr53 was dependent on both p27""' and p40x (Fig. 2B, lanes a and c) as was observed with pGAGdm, which had a splicing acceptor signal (Fig. 2A). Weak expression of Pr53 was also observed with p21"" and p40x as was the case with pGAGdm (Fig. 28, lane e).

The RNA pattern in cells transfected with pGAGdb and pX expression plasmid was similar to that observed with pGAGdm when probed with the gag probe; presence of the gag mRNA (3.2 kb) was dependent on the expression of both p27""' and p40x (Fig. 3 D and E, lanes a and c). One minor difference was that with pGAGdb, the 3.2-kb RNA was always associated with a faint band of 3.4-kb RNA. The


larger band represents RNA that had been read through from the 3' LTR and was terminated at the thymidine kinase signal. In this provirus, no spliced small viral RNA was detected with the exon 1 probe (Fig. 3E). These results may suggest that the splicing process itself is not a direct target of p27""'1. Splicing at a cryptic acceptor signal in the pGAGdb proviral sequences seems to be unlikely because the exon 1 probe, which can detect the 5' and 3' ends ofthe transcripts, did not detect any RNA species other than genomic 3.2- and 2.4-kb RNA. The latter was a transcript of the pX expression plasmid (Fig. 3E, lanes a-e). The possibility that RNA spliced at the cryptic acceptor is coincidentally 2.4 kb and comigrated with the pX transcript was ruled out by the fact that a DNA fragment from Sma I to Bgi II, which covers almost all the sequence of the defective provirus, did not detect any species of spliced RNA (Fig. 3F). However, this evidence cannot rigorously exclude the possibility that splicing occurs at a cryptic site. For example, RNA spliced at a cryptic site might have been rapidly degraded, thus undetectable in the blotting analysis.

DISCUSSION Here we have shown that the second pX protein p27"-I" is required for gag protein expression. Complementation assays of the HTLV-I genes clearly showed that transcription of integrated or unintegrated provirus DNA was fully activated by p40" alone but that gag gene expression was dependent on both p27"-"' and p40x. The appearance of gag mRNA was observed concomitantly with gag protein expression. Therefore, it is evident that p27""' operates at post-transcriptional levels that finally affect the level of gag mRNA but not at the translational level. When the defective proviral construction retained a splicing acceptor signal, the maximum level of spliced viral mRNA was detected with p40" alone. The expression of p27""' induced an increase in the level of gag mRNA that was not spliced and a concomitant decrease of spliced RNA. The simplest explanation for this observation could be that p27"-" suppressed the splicing of the viral transcript. However, another defective construction that had a deletion of the splicing acceptor signal also showed p27""' dependency for gag gene expression, and we detected no evidence for splicing at a cryptic acceptor signal. These observations suggested that p27""' affected the levels of unspliced viral RNA, whether they could eventually be spliced or not. Therefore, the direct target of p27""' may not be the splicing process itself. Although post-transcriptional regulations are not well understood, it seems likely that individual posttranscriptional processes may affect each other in such a way that one alteration modulates more than one different following or preceding processes. If this is the case, acceleration of transport of unspliced viral RNA or its stabilization before splicing or degradation may also explain the observed results that both RNA with or without splicing signals were modulated by p27""'. In fact, a mutation at a splicing signal in the ,l-globin gene of a /8-thalassemia patient reduced both the unspliced mRNA as well as the spliced mRNA levels (32). Thus, at this stage, it seems difficult to specify the direct target of p27""'. The observations on these defective proviruses should not be artificially restricted to these constructions because the constructions for gag expression and the splicing signals were identical to those used in HTLV-I replication. Furthermore, in a preliminary experiment, a whole provirus DNA carrying a mutation in the p27""' coding sequence produced only spliced pX mRNA and did not express any gag protein. Therefore, the requirement of one viral product p27"-"', for expression of the other viral structural proteins leads us to postulate an early and late stage in viral replication as in DNA

Biochemistry: Inoue et al. tumor viruses: Initial transcription of the proviral genome depends on cellular function, and all viral RNA is spliced into

pX mRNA, eventually producing a low level of p40x. This initial quantity of p40x further activates transcription of the proviral genome and produces high levels of pX mRNA

without p27"`"' or with low levels of p27`"', but no unspliced viral RNA yet accumulates. When enough pX mRNA has accumulated to produce a sufficient level of p40x for efficient transcription, p27x-"' has also accumulated, since these two proteins are encoded by a single species of mRNA (25). Thus, accumulated p27`411 induces production of a high level of unspliced viral mRNA for gag and probably also for env proteins. Thus, p27`-"' initiates the late phase in viral replication in which gag, pol, and env proteins are produced efficiently. At the same time, p27X-"'1 reduces the level of spliced pX mRNA resulting in a reduction of the level of p40X. Less p40x leads in turn to further reduction of viral gene transcription. This hypothesis suggests that HTLV-I has a self-regulatory mechanism in its replication and that p27"-"' and p40" are the mediators of this regulation. Therefore, a balanced expression of p27xIII and p40x is vital for efficient replication of HTLV-I. In this respect, the mechanism of expression in which these two pX proteins are encoded independently by a single mRNA is highly significant in self-regulation of replication. For the regulatory gene (ORF III) coding for p27`1', we will adopt the term tel (trigger for expression of late gene). The fact that all viruses in the HTLV family have the second ORF that codes for the second pX protein in addition to transcriptional activator (33, 34) strongly suggests a similar mechanism of regulation by the tel gene throughout the HTLV family. Human immunodeficiency virus, also called human T-cell lymphotropic virus type III/lymphadenopathy-associated virus, was shown to have two post-transcriptional regulators, tat-III (35) and art (36), required for viral gene expression and replication. However, neither of these regulators seems to affect the level of viral mRNA according to Sodroski et al. (36). On the other hand, Feinberg et al. (37) reported that a post-transcriptional regulator trs, which is apparently the same locus as the art locus described by Sodroski et al. (36), regulates splicing of the viral transcript. The transcriptional activator p40x of HTLV-I, also termed the tat-I gene product, is completely different in its mode of function from the tat-III of human immunodeficiency virus despite the similarity in terminology. The tel-i gene is also functionally different from the tat-III but may have some similarities to the trs gene of Feinberg et al. (37). We conclude from this work that HTLV-I replication requires two functions at different levels of transcription and post-transcription. These are mediated by p40x and p27x-III (tel gene product), respectively, and the tel gene exerts a self-regulatory mechanism in its replication. We thank Ms. A. Hikikoshi, M. Higashi, and Y. Hirayama for their technical assistance. This work was supported in part by a Grantin-Aid for Special Project Research on Cancer Bio-Science from the Ministry of Education, Science and Culture of Japan.

Poiesz, B. J., Ruscetti, F. W., Gazdar, A. F., Bunn, P. A., Minna, J. D. & Gallo, R. C. (1980) Proc. Natl. Acad. Sci. USA 77, 7415-7419. 2. Yoshida, M., Miyoshi, I. & Hinuma, Y. (1982) Proc. Nail. Acad. Sci. USA 79, 2031-2035. 3. Hinuma, Y., Nagata, K., Hanaoka, M., Nakai, M., Matsumoto, T., Kinoshita, K., Shirakawa, S. & Miyoshi, I. (1981) Proc. Natl. Acad. Sci. USA 78, 6476-6480. 4. Wong-Staal, F., Hahn, B., Manzari, V., Colonbini, S., Fran1.


5. 6. 7. 8.

9. 10. 11.

3657

chini, G., Gelman, E. P. & Gallo, R. C. (1983) Nature (London) 302, 626-628. Yoshida, M., Seiki, M., Yamaguchi, K. & Takatsuki, K. (1984) Proc. Natl. Acad. Sci. USA 81, 2534-2537. Seiki, M., Hattori, S., Hirayama, Y. & Yoshida, M. (1983) Proc. Nail. Acad. Sci. USA 80, 3618-3622. Kiyokawa, T., Seiki, M., Imagawa, K., Shimizu, F. & Yoshida, M. (1984) Gann 75, 747-751. Lee, T. H., Coligan, J. E., Sodroski, J. G., Haseltine, W. A., Salahuddin, S. Z., Wong-Staal, F., Gallo, R. C. & Essex, M. (1984) Science 226, 57-61. Miwa, M., Shimotohno, K., Hoshino, H., Fujino, M. & Sugimura, T. (1984) Gann 75, 752-755. Slamon, D. J., Shimotohno, K., Cline, M. J., Golde, D. W. & Chen, I. S. Y. (1984) Science 226, 61-65. Kiyokawa, T., Seiki, M., Iwashita, S., Imagawa, K., Shimizu, K. & Yoshida, M. (1986) Proc. Natl. Acad. Sci. USA 82,

8359-8363. 12. Sodroski, J. G., Rosen, C. A. & Haseltine, W. A. (1984) Science 225, 381-385.

13. Fujisawa, J., Seiki, M., Kiyokawa, T. & Yoshida, M. (1985) Proc. Natl. Acad. Sci. USA 82, 2277-2281. 14. Sodroski, J. G., Rosen, C. A., Goh, W. C. & Haseltine, W. A.

(1985) Science 228, 1430-1434.

15. Felber, B. K., Paskalis, H., Kleinman-Ewing, G., Wong-Staal, F. & Pavlakis, G. N. (1985) Science 229, 675-679. 16. Seiki, M., Inoue, J., Takeda, T., Hikikoshi, A., Sato, M. & Yoshida, M. (1985) Jpn. J. Cancer Res. 76, 1127-1131. 17. Cann, A. J., Rosenblatt, J. D., Wachsman, W., Shah, N. P. & Chen, I. S. Y. (1985) Nature (London) 318, 571-574. 18. Seiki, M., Inoue, J., Takeda, T. & Yoshida, M. (1986) EMBO J. 5, 561-565. 19. Fujisawa, J., Seiki, M., Sato, M. & Yoshida, M. (1986) EMBO J. 5, 713-718. 20. Paskalis, H., Felber, B. K. & Pavlakis, G. (1986) Proc. Natl.

Acad. Sci. USA 83, 6558-6562. 21. Shimotohno, K., Takano, M., Teruuchi, T. & Miwa, M. (1986) Proc. Natl. Acad. Sci. USA 83, 8112-8116. 22. Chen, I. S. Y., Slamon, D. J., Rosenblatt, J. D., Shah, N. P., Quan, S. G. & Wachsman, W. (1985) Science 228, 54-58. 23. Greene, W. C., Leonard, W. J., Wano, Y., Svetlik, P. B., Peffer, N. J., Sodroski, J. G., Rosen, C. A., Goh, W. C. & Haseltine, W. A. (1986) Science 232, 877-880. 24. Inoue, J., Seiki, M., Taniguchi, T., Tsuru, S. & Yoshida, M. (1986) EMBO J. 5, 2883-2888. 25. Nagashima, K., Yoshida, M. & Seiki, M. (1986) J. Virol. 60,

394-399.

26. Seiki, M., Hikikoshi, A., Taniguchi, T. & Yoshida, M. (1985)

Science 228, 1531-1534. 27. Inoue, J., Seiki, M. & Yoshida, M. (1986) FEBS Lett. 209, 187-190. 28. Inoue, J., Watanabe, T., Sato, M., Oda, A., Toyoshima, K., Yoshida, M. & Seiki, M. (1986) Virology 150, 187-195. 29. Mcknight, S. L. (1980) Nucleic Acids Res. 8, 5949-5964. 30. Berger, S. L. & Birkenmeier, C. S. (1979) Biochemistry 18, 5143-5149. 31. Hattori, S., Kiyokawa, T., Imagawa, K., Shimizu, F., Hashimura, E., Seiki, M. & Yoshida, M. (1984) Virology 136, 338-347. 32. Fukumaki, Y., Ghosh, P. K., Benz, E. J., Reddy, U. B., Lebowitz, P., Forget, B. G. & Weissman, S. M. (1982) Cell 28, 585-593. 33. Shima, H., Takano, M., Shimotohno, K. & Miwa, M. (1986) FEBS Lett. 209, 289-294. 34. Sagata, N., Yasunaga, T. & Ikawa, Y. (1985) FEBS Lett. 192, 37-42. 35. Rosen, C. A., Sodroski, J. G., Goh, W. C., Dayton, A. I., Lippke, J. & Haseltine, W. A. (1986) Nature (London) 319, 555-559. 36. Sodroski, J., Goh, W. C., Rosen, C., Dayton, A., Terwilliger, E. & Haseltine, W. (1986) Nature (London) 321, 412-417. 37. Feinberg, M. B., Jarrett, R. F., Aldovini, A., Gallo, R. C. & Wong-Staal, F. (1986) Cell 46, 807-817.