Communication Vol. 264, No. 24 Issue of August 25, pp. 13975-13978,1989 0 1989 by The American Society fo; Biochemistry and Molecular Biology, Inc.

Printed in U. S. A .

THE JOURNAL OF BIOLOGICAL CHEMISTRY

Co-expression of the Subunits of the Heterodimer of HIV-1 Reverse Transcriptase in Escherichia coli”

(Received for publication, May 4, 1989) Barbara Muller , Tobias Restle, Stefan Weiss, Mathias Gautel, Georg SczakielS, and Roger S. Goody

From the Abteilung Biophysik, Man-Planck Institut fur Medizinische Forschung, Jahnstrasse 29 and the SDeutsches Krebsforschungszentrum, Im Neuenkimer Feld 280, 6900 Heidelberg, Federal Republic of Germany

Expression of the 66-kDa form of human immuno- deficiency virus, type 1 reverse transcriptase in Esch- erichia coli leads to isolation of small amounts of a 2 x 66-kDa homodimer and larger amounts of a heterodi- mer form of the enzyme in which the 66-kDa protein is complexed with its carboxyl-terminally truncated 51-kDa form. The latter arises via proteolysis by con- taminating proteases. The heterodimer, which was characterized by gel filtration (apparent native molec- ular mass of 120-130 kDa), was the most active form of the enzyme (specific activity, 5000 units/mg, cf. e2000 for the 66-kDa fragment). The 66-kDa frag- ment alone was shown to be only partially dimerized, with the activity residing mainly in the dimer fraction. Proteolysis of the 66-kDa form resulting partially in the 51-kDa form led to an increase in reverse tran- scriptase activity. Expression of a truncated version of the protein containing the first 428 amino acids of the reverse transcriptase coding region led to a protein which had low but measurable reverse transcriptase activity (400-500 units/mg). Co-expression of the two proteins on a single plasmid led to expression in a 1:l ratio. The 1:l mixture behaved as a heterodimer, as shown by its chromatographic properties. It is likely that the mechanism for the production of heterodimers in vivo involves cleavage of 66-kDa monomers fol- lowed by rapid dimerization of the 51- and 66-kDa forms to give the heterodimeric form, which is stable toward further proteolysis.

Reverse transcriptase from HIV-1’ is known to be present in virus particles in two forms with molecular masses of about 66 and 51 kDa (di Marzo Veronese et al., 1986; Farmerie et al., 1987). These two species have a common amino terminus (Lightfoote et al., 1986), and the shorter one is thought to arise by proteolytic cleavage of the 66-kDa protein which in turn is initially produced by proteolysis by the viral protease of the gag-pol precursor. The cleavage point for production of the 51-kDa fragment is not known at present, and there is no

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

’ The abbreviations used are: HIV-1, human immunodeficiency virus, type 1; HPLC, high performance liquid chromatography; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; FPLC, fast protein liquid chromatography.

obvious recognition sequence for the viral protease (typically an aromatic residue followed by a proline, although quite different sequences are also cleaved (Darke et aL., 1988)).

The work described in this contribution shows that proc- essing of the 66-kDa fragment to an equimolar mixture of 66- and 51-kDa fragments is catalyzed by traces of bacterial proteases present during isolation of the enzyme and that this behaves as a heterodimer with an apparent molecular mass of between 120 and 130 kDa. In contrast, dimer formation by the 66-kDa form is incomplete. We show that the heterodimer can be produced directly in high yield by co-expression of the subunits in Escherichia coli.

MATERIALS AND METHODS

Cloning of the RT Coding Sequence into the Prokaryotic Expression VectorpKK-233-2: Construction of Pk~midpRT~~-Star t ing from the cDNA clone BHlO (Hahn et al., 1985), the BglII and EcoRI sites at positions 1641 and 4228, respectively (Ratner et al., 1985), were used to clone the BglII/EcoRI cassette into the pEMBL8 (+) vector (Dente et al., 1983) to construct pRT1. To do this the sticky EcoRI and PstI ends were converted to blunt ends using mung bean nuclease. Second, to create an EcoRI restriction site at the 5’ end of the RT coding sequence (double mismatch necessary) and to delete the endonuclease sequence downstream of the RT coding region and introducing a TAA-stop codon simultaneously the site-directed mutagenesis method described by Taylor et al. (1985) was used. In this manner (after converting pRTl by an F1 helper phage lysate into the single- stranded form); pRT2 and -3 were constructed. Next, to create a construct with the recombinant EcoRI site and the correct 3’ end of the RT coding region, a unique EcoRV site (position 2559 in clone BH10) within the RT coding sequence and the HindIII site down- stream of the RT coding region were used, which resulted in the recombinant clone pRT4. Finally, pRT4 was restricted with EcoRI, and the sticky ends were converted to blunt ones using a Klenow fragment. After a HindIII digestion, the RT cassette was cloned into the prokaryotic expression vektor pKK-233-2 (Amman and Brosius, 1985), which was first digested by NcoI followed by a Klenow- mediated blunt end reaction and a HindIII digestion. The result was the plasmid pRT@ which includes a Shine-Delgarno sequence, an exposed ATG start codon, and the RT coding sequence in the correct reading frame. The recombinant reverse transcriptase should consist of the amino acid sequences of the native enzyme except for three more amino acids at the amino terminus (Met, Asn, Ser). DNA sequencing according to the dideoxynucleotide method (Sanger et al., 1977) showed that the construction was correct.

Construction of Plasmid pRT5,-Plasmid pRT% was digested with KpnI and HindIII. The 5.9-kilobase fragment was isolated and ligated with a synthetic oligonucleotide linker, which provided the codon for the carboxyl-terminal amino acid (Gln-428) and a TAA stop codon. Correct insertion of the oligonucleotide was proved by DNA sequenc- ing.

Construction of Plasmid pRT66151-Plasmid pRT6s was linearized with AuaI, and the ends were blunted using Klenow fragment. The 2.5-kilobase BamHI/PuuI fragment from pRTsl was isolated and blunted using T4 DNA polymerase. This fragment contained the promoter and Shine-Dalgarno sequences, the coding region, and the translational terminator and thus represents an independent and complete “expression cassette.” A blunt end ligation with the linear- ized and blunted pRT66 led to pRT66/51, which contained the two expression cassettes in opposite orientations, as shown by restriction analysis. Plasmids containing the two cassettes in the same orienta- tion did not lead to satisfactory co-expression of the two protein fragments.

Isolation of Reverse Transcriptase-Reverse transcriptase was ex- pressed either in E. coli TG1 or 6222 cells containing the plasmids pRT6, pRT,,, or pRT66/b1 and for 6222 cells in addition the lacIq- bearing plasmid pDMI,1 (Certa et al., 1985). Cells were grown to an OD,,, of about 0.5 followed by induction with 70 mg/liter isopropyl-

13975

13976 HIV-1 Reverse Transcriptase 1-thio-fl-D-galactopyranoside for 6 h. After harvesting the cells by centrifugation, the pellet was suspended in 25 mM Tris.HC1 (pH 8.0), 5 mM dithiothreitol, 1 mM EDTA, 0.4 mM phenylmethylsulfonyl fluoride, 1 mM aminobenzoic acid ethyl ester, and 0.1 mM benzami- dine (2.6 ml/g cells). Lysozyme (3.3 mg/ml) and 0.2% Nonidet P-40 (Serva) were added. After incubation for 20 min at 0 "C, the mixture was sonicated and centrifuged for 20 min a t 4 "C (Sorvall SS34 rotor, 20,000 rpm). The supernatant was decanted and stored, and the pellet was extracted for 10 min with 10 ml of 1 M NaCl a t 0 "C with occasional stirring. After centrifugation, the supernatant was com- bined with the first supernatant. Protein was precipitated by the addition of solid ammonium sulfate to 50% saturation. After collect- ing by centrifugation, the pellet was resuspended in about 3 ml/g cells of buffer A (50 mM Tris.HC1, pH 8.2 a t 4 "C, 1 mM dithiothreitol, 1 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, 25 mM NaCI, 6% glycerol). Nucleic acids were precipitated by the addition of strepto- mycin sulfate (40 mg/ml) and removed by centrifugation a t 4 "C (15 min a t 20,000 rpm, Sorvall SS34 rotor). The supernatant was dialyzed overnight a t 4 "C against buffer A containing additionally 1 mM aminobenzoic acid ethyl ester and 0.1 mM benzamidine. Any addi- tional precipitate was removed by centrifugation.

The supernatant was applied to a column of DEAE-Sephacel packed in buffer A, followed by elution with the same buffer a t 4 "C. Reverse transcriptase was eluted at this ionic strength, whereas most of the bacterial proteins remained bound to the column. The reverse transcriptase-containing fraction was then applied to a column of single-stranded DNA-cellulose (Sigma) in buffer A a t 4 "C. Buffer A containing 100 mM and then the same buffer containing 500 mM NaCl were used to develop the column. The 100 and 500 mM fractions were concentrated separately (Ultrafilter, Millipore). These were further purified by gel permeation chromatography on a column of Sephadex G-50 in buffer B (50 mM diethanolamine, pH 8.8 at 25 "C; column run a t 4 T ) . Protein eluting in the exclusion volume was concentrated to a few ml (Ultrafilters) and could be stored frozen at this stage.

Several different columns were used for final purification of the heterodimeric form of reverse transcriptase. For small quantities (about 1 mg), FPLC on an analytical Mono-Q (Pharmacia LKB Biotechnology Inc.) column was the method normally used. This could be scaled up using a column of Fractogel EMD TMAE-650(S) (15 X 1 cm; Merck), which allowed 10-mg scale preparations, or using Q-Sepharose (Pharmacia) a t low pressures, allowing unlimited scale- up. In all cases, the buffer system used was buffer B with a gradient of NaCI. Cation exchange chromatography on Fractogel EMD SO; -6506) was also found to be useful, particularly for reverse transcrip- tase from the plasmid pRTssml.

Analytical Gel Filtration-This was performed using two HPLC columns in series (Bio-Rad TSK-125 followed by a Bio-Rad TSK- 250; both 300 X 7.5 mm) using an ISCO 2350 pump and UV detection (280 nm). The column was eluted with sodium phosphate, 200 mM (pH 7.0) a t 1 ml/min, and calibrated using a mixture of proteins of known molecular weight.

RESULTS AND DISCUSSION

A plasmid construction for the expression of the region coding for the 66-kDa form of HIV-1 reverse transcriptase was prepared which was essentially identical with that re- ported by Larder et al. (1987). As already found by these authors, the main form of reverse transcriptase expressed in E. coli using this system is the 66-kDa form. However, we noticed that the 51-kDa form was also present and that during the course of protein purification the band corresponding to this fragment on PAGE increased in intensity. This is sum- marized in Fig. L4, which shows the relative proportions of the two fragments a t various stages of the purification. In lane 1, the raw extract after cell lysis is shown, and it is clear that the 66-kDa form predominates. However, after the first step in the purification procedure ( i e . removal of most other proteins by passage over DEAE-Sephacel a t relatively low ionic strength), it can be seen that appreciable amounts of the 51-kDa form are present. The next step in the isolation procedure is chromatography on a single-stranded DNA-cel- lulose column, and elution with 100 mM NaCl gives predom- inantly the 66-kDa form, whereas elution with 500 mM salt

94 - 67-

43-

30-

20-

14-

94 - 67 -

43 -

30 -

20 -

14-

n = --

A

B FIG. 1. Purification of reverse transcriptase from E. coli-

carrying plasmids pRTss ( A ) or pRT66,nl ( B ) . Lane 1, total soluble protein; lanes 2-5, reverse transcriptase containing fractions from successive chromatography columns, i.e. DEAE-Sephacel (lane 2), single-stranded DNA cellulose (lane 3 ) , Sephadex G-50 (lane 4 ) , and TMAE-Fractogel (lane 5). Aliquots containing approximately equal amounts of reverse transcriptase were separated on 15% acryl- amide gels in the presence of 0.1% SDS (Laemmli, 1970). Proteins were detected by silver stain (Wray et al., 1981).

gives a mixture of the two forms. The specific activity of the high salt fraction is considerably higher (about 3000-5000 units/mg) than that of the low salt fraction, which shows a large variation from preparation to preparation from several hundred to about 2000 units/mg. FPLC on strongly basic ion exchangers leads to the preparation shown in lane 5 of Fig. lA which elutes as a discrete peak a t about 70 mM salt concentration (from trimethylamino ethyl-Fractogel). This peak has the highest specific activity (about 5000 units/mg). Many other fractions are also eluted from this column, most of which also show reverse transcriptase activity and which consist of 66-kDa protein together with varying amounts of variable length truncated protein. The necessary removal of these fractions leads to a large loss of yield at this final stage in the isolation. The ratio of intensities of the two bands (stained with Coomassie Blue) in the final preparation is 1.27:l (66 kDa:51 kDa), which corresponds to a 1:l molar

HIV-1 Reverse Transcriptase 13977

ratio assuming that both polypeptides stain with equal inten- sity.

These results, together with those from earlier work on native reverse transcriptase from HIV-1, are strongly sugges- tive of a heterodimer structure for the enzyme. It is difficult to imagine any other reason for the production of a 1:1 mixture of the full length protein and its proteolytic cleavage product. According to this interpretation, starting from the 66-kDa protein, proteolysis by as yet unidentified proteases produces a 51-kDa fragment which interacts more strongly, or in a different manner, with the remaining 66-kDa form than is the case for interaction between individual 66-kDa molecules. In this stable heterodimer, the 66-kDa chain is protected from further degradation.

Conclusive evidence for the formation of heterodimers comes from experiments such as those shown in Fig. 2 A , when interpreted together with results reported below on the lack of dimer formation by the 51-kDa protein alone. HPLC gel filtration of the preparation used for lane 5 of Fig. 1A leads to an apparent molecular mass estimate of between 120 and 130 kDa, with no trace of either monomeric form present. The specific activity is unchanged after gel filtration, and it is clear from SDS-PAGE after gel filtration that both the 66- and 51-kDa bands are present in this fraction.

The situation is quite different for the 66-kDa fraction obtained a t low salt from the DNA-cellulose column. Fig. 2B shows that this is only partially dimerized. The activity is mainly in the dimer fraction. On storage at -20 "C, the relative amount of dimer increased over a period of 3 weeks from about 15% to about 35% (the sample was thawed and refrozen several times during this period). Thus, the 66-kDa protein appears to be capable of forming dimers but only very slowly. This is confirmed by the finding that rechromatogra- phy of the monomer and dimer fractions from the HPLC gel filtration column immediately after the first separation showed only minor amounts of cross-contamination. Thus, we conclude that the homodimer is in very slow equilibrium with its monomers.

On standing at room temperature, the 66-kDa protein is slowly partially degraded to the 51-kDa form, with a concur- rent increase in reverse transcriptase activity (data not shown). Preparations of the 66-kDa fragment which have been purified in addition by FPLC do not show this sponta- neous degradation, which supports the idea that traces of proteases, presumably of bacterial origin, are responsible for this processing during purification of the enzyme and that these are removed by FPLC over strongly basic ion exchan- gers. The rate of conversion to the heterodimer can be in- creased by the addition of trypsin to the 66-kDa fragment, but this is not of use for preparative purposes since it is difficult to control the extent of proteolysis.

In order to solve the problem of uncertainty concerning the position of proteolytic cleavage of the 66-kDa form of reverse transcriptase by unidentified proteases, we decided to attempt the co-expression of both subunits of the heterodimer in E. coli. As a first step, a fragment corresponding to the first 428 residues of the region coding for the enzyme was expressed (calculated molecular mass, 50,391 Da). We chose this length partly for technical reasons and partly because it seemed preferable to delete as much of the potentially protease- susceptible region as possible in order to prevent further proteolysis. Large amounts of this protein were obtained on expression in E. coli. It was found to behave anomalously on gel filtration (apparent molecular mass <30 kDa) although its molecular weight under denaturing conditions on gel electro- phoresis was in the expected range. The reason for this

C

0" 0

r

0 5 10

d E 0

I I I I I 15 20

time (min)

A

0 5 10 15 20 25 time (min)

B FIG. 2. HPLC gel permeation chromatography of HIV- 1 re-

verse transcriptase heterodimer ( A ) or 66-kDa protein ( B ) . SDS-PAGE of the separated peaks in B showed them to be indistin- guishable under denaturing conditions, both containing only the 66- kDa band. Reverse transcriptase activity was mainly in the dimer peak (shorter retention time). Conditions: see "Materials and Meth- ods."

behavior is unclear, but it is apparent that the truncated protein does not behave as a dimer. The lack of dimer for- mation is confirmed by the observation that the 51-kDa protein cannot be cross-linked by treatment with dimethyl suberimidate, whereas this occurs readily with the heterodi- mer (data not shown). The 51-kDa protein does, however, have residual reverse transcriptase activity (400-500 units/ mg, cf. about 5,000 units/mg for the heterodimer). This is in qualitative agreement with the results of Tisdale et al. (1988) using a 430-amino acid reverse transcriptase mutant but in contrast to those of Hansen et al. (1988), who saw no activity in the 51-kDa band using enzyme activity gel analysis.

When cell cultures containing induced cells with both pRTs6 and pRTsl were mixed in approximately equal amounts

_ i .

13978 HIV-1 Reverse Transcriptase

before lysis, good yields of heterodimer were obtained, sug- gesting that 66- and 51-kDa monomers can dimerize and that the mechanism of heterodimer formation is not, or at least not only, via homodimers. I t also shows that 66- and 51-kDa monomers, which are presumably in their folded form, are able to interact with each other. Mixing the isolated 51-kDa fragment with a 66-kDa preparation which was largely mon- omeric also led to formation of heterodimers under conditions in which the 66-kDa protein alone showed no tendency to dimerize.

Inclusion of the sequence coding for the 51-kDa fragment (with its own promoter, Shine-Dalgarno sequence, and tran- scriptional and translational terminators) into the same plas- mid as that for the 66-kDa protein resulted in expression of approximately equal amounts of the two proteins (Fig. 1B). It is of int.erest to note that only constructs containing the two expression cassettes in opposite orientations (Fig. 3) gave good expression of both fragments.

In contrast to the situation on expressing the 66-kDa pro- tein alone, the specific activity after the first column (DEAE- Sephacel) was already very high (about 4000 units/mg). The DNA-cellulose and gel filtration columns used routinely for the purification of reverse transcriptase from plasmid pRTS6 did not lead to dramatic increases in specific activity nor did the final anion exchange FPLC column, although there was a definite improvement in terms of removal of impurities (Fig. 1B). HPLC gel filtration showed that the reverse transcrip- tase obtained by coexpression behaved exclusively as a dimer even before the final ion exchange column, in contrast to the situation with reverse transcriptase from plasmid pRTs6. I t is of interest to note that very little material was eluted from the DNA-cellulose column at low ionic strength, in contrast to the situation with reverse transcriptase from pRT66, where a relatively pure 66-kDa fragment was eluted under these conditions. The results indicate that heterodimer formation occurs early in the preparation, probably already in the cells.

The behavior of heterodimer arising from pRT66/51 on anion exchange chromatography is of interest with respect to the length of the shorter fragment in comparison with that which arises by degradation of the 66-kDa protein. The co-expressed heterodimer elutes at considerably lower ionic strength from the column, suggesting a lower net negative charge on the

pRT66/51 8.6 kb

on

FIG. 3. Construct for co-expression of the 51- and 66-kDa subunits of HIV- 1 reverse transcriptase. An expression cassette (dashed line), consisting of the Ptrc promoter (Amman and Brosius, 1985), the coding sequence for the 51-kDa subunit, and the transcrip- tional terminators from the rrnB operon (rrnB TIT*), was cloned into plasmid pRTs6 by blunt end ligation.

molecule. Examination of the sequence in the region of the carboxyl terminus of the shorter subunit indicates that in the next 10 residues (downstream) in the sequence of the full length protein there would be 3 negatively charged residues. Since the reverse transcriptase is only relatively weakly bound by the column, it is likely that this difference in charge explains the difference in chromatographic properties. Thus, the results are in agreement with the idea that the 428-residue protein is shorter than that which is obtained in the main heterodimer fraction starting from plasmid pRTss. I t can be seen from the gels in Fig. 1 that the synthetic 51-kDa fragment is indeed shorter than that arising from degradation of the 66-kDa protein, since the former migrates faster than the latter. This difference in mobilities is helpful in interpreting the experiment described above in which cultures of bacteria containing, separately, the two plasmids pRTs6 and pRTss were mixed before cell lysis. The two 51-kDa fragments (i.e. that arising from direct expression and that arising from degradation of the 66-kDa protein) are easily separated on SDS-PAGE, and it can be seen that only traces of the heavier form are present after the first three steps in the purification, indicating that the heterodimer arises almost exclusively from direct interaction of the initially expressed 66- and 51-kDa proteins.

The only consistent interpretation of the results presented here is that the most active and most stable form of HIV-1 reverse transcriptase is the heterodimer, and from earlier work this appears to be the form present in virus particles. Our results suggest that the mechanism of formation of the het- erodimeric form is via proteolysis of 66-kDa monomers, pre- sumably in vivo by the HIV-1 protease, followed by rapid formation of heterodimers, which are then stable toward further degradation. Our present work is directed toward verifying and extending this mechanism and toward structural characterization of the co-expressed heterodimers, which can be obtained in highly homogeneous form and in high yield from pRT66/51.

Acknowledgments-We thank Ursula Ruhl and Sabine Zimmer- mann for excellent technical assistance, Peter Lang for oligonucleo- tide synthesis, and Natalie Didat (EMBL, Heidelberg) for fermenter runs.

REFERENCES Amman, E., and Brosius, J. (1985) Gene (Amst.) 40, 183-190 Certa, U., Bannwarth, W., Stuber, D., Gentz, R., Lanzer, M., Le Grice, S. F. J.,

Guillot, F., Wendler, J., Hunsmann, G., Bujard, H., and Mous, J. (1985) EMBO J. 5 , 3051-3056

Darke, P. L., Nutt, R., Brady, S. F., Garsky, V. M., Cicarrone, T. M., Leu, C: T., Lumma, P. K., Freidinger, R. M., Veber, D. F., and Sigal, I. s. (1988)

Dente, L., Cesarini, G., and Cortese, R. (1983) Nucleic Acids Res. 11, 1645- Biochem. Biophys. Res. Commun. 156, 297-304

di Marzo Veronese, F., Copeland, T. D., De Vico, A. L., Rahman, R., Oroszlan, 1655

Farmerie, W. G., Loeb, D. D., Casavant, N. C., Hutchinson, C. A,, 111, Edgell, S., Gallo, R. C., and Sarngadharan, M. G. (1986) Science 231, 1289-1291

Hahn, B. H., Shaw, G. M., Arya, S. K., Popovic, M., Hoxie, J. A,, Gallo, R. C., M. H., and Swanstrom, R. (1987) Science 236,305-308

Hansen, J., Schulze, T., Mellert, W., and Moellmg, K. (1988) EMBO J. 7,239- and Wong-Staal, F. (1985) Proc. Natl. Acad. Sci. U. S. A. 82, 4813-4817

942 Laimmli, U. K. (1970) Nature 227,680-685 Larder, B., Purifoy, D., Powell, K., and Darby, G. (1987) EMBO J. 6, 3133-

Lightfoote, M. M., Coligan, J. E., Folks, T. F., Fauci, A. S., Martin, M. A., and 3137

Ratner, L., Haseltine, W., Patarca, R., Livak, K. J., Starcich, B., Josephs, S., Venkatesan, S. (1986) J . Virol. 60, 771-775

Doran, E. R., Rafalski, J. A,, Whitehorn, E. A,, Baumeister, K., Ivanoff, L., Petteway, S. R., Jr., Pearson, M. L., Lautenberger, J. A,, Papas, T. S., Ghrayeb, J., Chanp, N. T., Gallo, R. C., and Wong-Staal, F. (1985) Nature

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

Taylor, J. W., Ott, J., and Eckstein, F. (1985) Nucleic Acids Res. 13,8765-8785 Tisdale, M., Ertl, P., Larder, B. A,, Purifoy, D. J. M., Darby, G. and Powell, K.

Wray, W., Boulikas, T., Wray, V. P., and Hancock, R. (1981) Anal. Biochem.

313;277-284

A . 74,5463-5467

L. (1988) 62,3662-3667

118,197-203