site-directed mutagenesis of hiv- 1 integrase demonstrates

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1993 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 268, No. 3, Issue of January 25, pp. 2113-2119,1933 Printed in U.S.A.

Site-directed Mutagenesis of HIV- 1 Integrase Demonstrates Differential Effects on Integrase Functions in Vitro*

(Received for publication, July 20, 1992)

Andrew D. LeavittSllI, Lily ShiueS, and Harold E. VarmusS(I** From the $Department of Immunology and Microbiology, §Department of Laboratory Medicine, and [)Department of Biochemistry and Biophysics, The University of California, San Francisco, California 94143-0502

The retroviral integrase (IN) protein is essential for integration of retroviral DNA into the host cell genome. To identify functional domains within the protein and to assess the importance of conserved residues, we performed site-directed mutagenesis of HIV-1 IN and analyzed the mutants in vitro for IN-mediated activities: 3’ processing (at t site-specific nuclease activity), strand transfer (the joining of att site oligonucleotides to target DNA), disintegration (the reverse of strand transfer), and integration site selection. Chang- ing the conserved residue His-16 either to Cys or to Val in a proposed zinc-finger region had minimal effect on IN activities. Alteration of two highly conserved amino acid residues, Asp-1 16 -* Ile and Glu-152 - Gly, each resulted in complete or nearly complete loss of 3’ processing, strand transfer, and disintegration, whereas alteration of another conserved residue, Trp- 235 Glu, had no demonstrable effect on any of the activities in vitro. Two mutants, Asp-64 + Val and Arg-199 -P CysA, each demonstrated differential effects on IN activities. Asp-64 -* Val has no demonstrable strand transfer or disintegration activity yet maintains 3’ processing activity at a diminished level. Arg- 199 + CysA, which lacks part of the carboxyl terminus of IN, has impaired strand transfer activity without loss of disintegration activity. Use of a target site selection assay showed that all of our mutants with strand transfer activity maintain the same integration pattern as wild type IN. We conclude that not all highly conserved IN residues are essential for IN activities in vitro, zinc coordination by the proposed zinc-finger domain may not be required for the activities assayed, alteration of single residues can yield differential effects on IN activities, and target site selection into naked DNA is not necessarily altered by changes in strand transfer activity.

Retroviral integration is one example of how mobile DNA elements are inserted into host cell DNA. Two viral compo- nents are required for integration: (i) attachment (a t t ) sites, short sequences at the ends of viral DNA (1-5), and (ii)

*This work was supported in part by grants from the National Institutes of Health. 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.

li Supported by a Burroughs Wellcome Research Foundation award. To whom correspondence should he addressed Dept. of Im- munology and Microbiology, University of California, San Francisco, HSE 401, San Francisco, CA 94143-0502. Fax: 415-476-0939.

** American Cancer Society Research Professor.

integrase (IN),’ the 32-45-kDa protein encoded by the 3‘ end of retroviral pol genes (6-9). Retroviral IN proteins recognize their cognate att sites, remove (usually) two nucleotides from the 3‘ end of each strand of viral DNA (3’ processing) (10- 13), make a staggered cut in target DNA, join the 3“processed cognate att sites to the newly created 5’ ends of target DNA (strand transfer) ( l l ) , and select integration sites in a non- random manner (14-16). IN can also use the gapped intermediate product of in vitro strand transfer as a substrate to reverse the strand transfer reaction, an activity termed disintegration (17).

Amino acid sequence comparisons of retroviral IN proteins reveal several highly conserved residues and motifs (Fig. 1; Refs. 18 and 19). Related proteins from other mobile genetic elements such as mouse intracisternal A particles, retrotran- sposable elements of yeast and flies, and tobacco plant retro- transposons also share some of these conserved residues (19, 20), suggesting a possible similarity of mechanism among these various mobile genetic elements. Near the amino terminus of all IN proteins, a specific arrangement of histidine and cysteine residues (H-X3-H-X2&-X2-C) resembles zinc- finger domains of some DNA-binding proteins (18-21) and has been referred to as the HHCC region (19). A region located near the middle of IN proteins is characterized by absolutely conserved D and E residues spaced exactly 35 amino acids apart (positions 116 and 152 in HIV-l), a motif that is highly conserved among not only retrotransposases but also the ORF-B proteins of the IS3 family of prokaryotic transposable elements (20). Computer-based structural predictions include a coiled-coil motif that has been postulated to be important for IN oligomerization (21). In addition, a number of other amino acid residues are absolutely conserved among retroviral integrases (18, 21).

In vitro assays for 3’ processing, strand transfer, disintegration, and target site selection provide an opportunity to use mutant IN proteins to assign functional properties to the conserved regions of IN (Fig. 2). Such assays have previously been used with wild type IN to determine nucleotide sequence requirements for att sites in 3‘ processing and strand transfer reactions by cognate IN proteins (14, 22-24). In addition, the assays have been used to demonstrate that: (i) 3‘ processing occurs via IN-mediated nucleophilic attack by water on the phosphodiester bond 3’ of the conserved CA dinucleotide, (ii) specific alcohol compounds (glycerol, 1,2-ethanediol, and 1,2- propanediol) can substitute for water as the attacking nucleophile (22, 25), and (iii) the released dinucleotide has a 5’ phosphate, while the processed viral end has a terminal hy-

The abbreviations used are: IN, integrase; CHAPS, 3-[(3-cholam- idopropyl)dimethylammonio]-1-propanesulfonic acid PCR, polym- erase chain reaction; MLV, murine leukemia virus; PAGE, polyacrylamide gel electrophoresis; RSV, Rous sarcoma virus.

2113

2114 In Vi t ro Analys i s of HIV-1 In t egrase Mutan t s

droxyl group which may serve as the attacking nucleophile in strand transfer (22, 25).

To determine the importance of conserved residues of retroviral IN proteins, we performed site-directed mutagenesis of selected amino acids in HIV-1 IN (Fig. 1). The mutants were analyzed i n uitro for 3’ processing, strand transfer, disintegration, and target site selection. We show that not all highly conserved residues are essential for IN activities i n uitro, zinc coordination by the amino-terminal HHCC region may not be required for i n uitro activity, some mutations preferentially affect a subset of IN-mediated functions, and changes that affect strand transfer activity do not necessarily change selection of target sites in naked DNA.

MATERIALS AND METHODS

Oligonucleotides-Oligonucleotides representing the terminal sequences of HIV-1 DNA were obtained from the University of Cali- fornia, San Francisco, Biomolecular Resource Center and annealed to form the following duplexes used in this report.

H I V - 1 U 5 29/29 5”TTTAGTCAGTGTGGAAAATCTCTAGgGT 3”AAATCAGTCACACCTTTTAGAGATCGTCA

H I V - 1 U 5 27/29: 5”TTTAGTCAGTGTGGAAAATCTCTAGG 3”AAATCAGTCACACCTTTTAGAGATCGTCA

H I V - 1 U 3 27 129 5“TTCTTTGGGACCAAATTAGCCCTTCg 3”AAGAAACCCTGGTTTAATCGGGAAGGTCA

SEQUENCES 1-3

U5 29/29 represents the blunt, unprocessed end of the U5 att site of HIV-1 and U5 27/29 and U3 27/29 represent preprocessed ends of the U5 and U3 HIV-1 att sites, respectively. The 5’ ends of oligonucleotides were labeled with [-y-3’P] ATP (3,000 Ci/mmol, Amersham Corp.) by incubation with T4 polynucleotide kinase (Pharmacia LKB Biotechnology Inc.). Sequenase version 2.0 (United States Biochem- ical Corp.) was used to fill in the 3’ end of the upper strand of HIV- 1 U5 27/29 with [w3*P]TTP (3,000 Ci/mmol, Amersham) placing a radiolabeled phosphate between the two nucleotides (GpT) removed during. 3’ processing. The disintegration substrate was made using the oligonucleotides and annealing conditions of Chow et al. (17).

Mutagenesis-Site-specific mutagenesis of the HIV-1 IN gene was performed as previously described (26). The HIV-1 IN gene from our wild type IN expression vector pBS24UBIN (27) was subcloned into M13 mp19. Synthetic oligonucleotides were designed, which allowed for the desired single amino acid changes and, if possible, a restriction site for screening purposes. Mutants were verified by DNA sequencing, and then cloned back into pBS24UBIN for expression of the mutant HIV-1 IN protein in Saccharomyces cereuisiae.

Protein Expression and Purification-All IN proteins were expressed in S. cereuisiae as an in-frame fusion peptide with the 76- amino acid ubiquitin protein attached to the amino terminus of IN. A naturally occurring hydrolase in yeast cleaves the fusion peptide in uiuo to yield unfused IN with an authentic amino-terminal sequence. The expression system, its application to HIV-1 IN, and purification of HIV-1 IN have been described (14, 27). In brief, the yeast cells were grown a t 30 “C in enriched media supplemented with glucose. The glucose was depleted within 24 h of incubation, thereby inducing expression of the fusion gene from a glucose-regulatable promotor. The cells were incubated for an additional 24 h, pelleted, lysed with glass beads in the absence of salt or detergent, followed by centrifu- gation a t 10,000 X g for 20 min. The pellet was then solubilized in lysis buffer plus 10 mM CHAPS and 1 M NaCl followed by centrifu- gation at 100,000 X g for 1 h, and the supernatant was sequentially chromatographed over a heparin affinity column and a phenyl-Seph- arose hydrophobic interaction column (14). Minor modifications of elution from the columns were applied to various mutants to improve yield and purity. This procedure currently yields 20-25 mg of wild type IN/liter of fermented cells at greater than 95% purity.

Assays for IN Actiuities-The oligonucleotide assays used for 3’ processing and strand transfer (Fig. 2, A and B ) were performed using conditions previously described except that glycerol was used only where indicated, and incubation was a t 30 “C for 45 min (14). The 3’ processing activity was also assayed using att site oligonucleotides radiolabeled between the terminal two nucleotides (G3*pT) removed

during 3’ processing as previously described (22, 25). Target site selection into circular plasmid DNA was performed according to the method of Pryciak and Varmus (15). In brief, pre-processed att site oligonucleotides (U5-27/29 or U3-27/29) served as the donor DNA, pBluescript KS- (Stratagene) served as the circular target DNA, and purified IN provided the integration activity. Sites of integration into pBluescript KS- were determined using PCR amplification followed by electrophoresis of the PCR products on 6% denaturing polyacrylamide gels and visualized by autoradiography using Hyper-film (Amersham) (15). This assay allows for analysis of multiple inde- pendent integration events with single base pair specificity. Amplifi- cation primers included one from the HIV-1 U5 or U3 att site (TTTAGTCAGTGTGGAAAATCTCTAGCAorTTCTTTGGGAC- CAAATTAGCCCTTCCA) and one from pBluescript SK- (CGC- CAGGGTTTTCCCAGTCACGAC). The disintegration reactions (Fig. PC) were performed using the branched oligonucleotide substrate described by Chow et al. (17), but reactions were limited to 20 min a t 30 “C to prevent product formation from reaching a plateau. The products were analyzed by electrophoresis on 15% polyacrylamide gels and visualized by autoradiography using Hyper-film (Amersham).

RESULTS

Generation and Purification of HIV-1 IN Proteins-We generated 11 site-directed mutations in HIV-1 IN to assess the role of amino acids conserved among retroviral IN proteins (Fig. 1). Mutants H12C, H12C/H16C, H16C, and H16V were chosen to evaluate the functional significance of the proposed zinc-finger region (18). Amino acid positions D64, Sal, D116, E152, and W235 were each altered because they are wholly conserved among retroviral INS and may therefore play a pivotal role in specific IN activities. Position 53 of HIV-1 IN corresponds to position 69 of MLV IN, and mutant R69C of MLV (SF1) shows a roughly 100-fold decrease in integration activity in uivo (6). We generated mutant Q53C to compare its phenotype i n uiuo with that of SF1. Arginine at position 199 is a non-conserved amino acid; mutant R199C was generated to compare the effect of a random change of a non-conserved amino acid with that of mutants made by changing conserved amino acids. Clearly, mutants generated by altering other non-conserved residues will not necessarily all have similar phenotypes.

The mutant alleles of HIV-1 IN were expressed in S. cereuisiae. Mutant IN protein production varied within 1-5- fold of wild type IN, except for mutants H12C and H12C/ H16C, which had markedly reduced protein production. In contrast, mutant H16C was produced in amounts similar to wild type IN. We cannot explain the deleterious effect of H12C on protein production. Assays (illustrated in Fig. 2) were performed using protein eluted from the phenyl-Sepha- rose column step in the purification scheme described under “Materials and Methods”.. The purity and gel mobility for each IN protein at that stage of purification are demonstrated in Fig. 3. The mutant IN proteins were assayed in vitro for their ability to mediate 3’ processing, strand transfer, disintegration, and target site selection, using established assays described under “Materials and Methods.” The activities of the IN mutants relative to wild type IN are summarized in Fig. 1. For example, Q53C, the HIV-1 IN mutant made for i n uiuo comparisons with the MLV IN mutant SF1, and not discussed further below, demonstrated wild type levels of 3’ processing and strand transfer activity (Fig. 1). It also had a wild type pattern of integration site selection in naked plasmid DNA (Fig. 9), and nearly wild type disintegration activity (Fig. 5).

Amino Acid Alterations in the HHCC Region Do Not Nec- essarily Impair I N Activity in Vitro-The amino terminus of IN contains a putative zinc-coordinating domain that has been referred to as the HHCC region (18, 19). We made two changes to the second of the two histidine residues, amino

In Vitro Analysis of HIV-1 Integrase Mutants 2115 HHCC ~~

f .

- f . 1 . . .. . I . . 1 * .f f .

H12C H12C H16C H16V 053C D64V S81R Dl161 E152G R199C W235E R199CI nloc

3’PROCESSlNG a a 4‘ 4+ 4+ 2+ a 0 4 4+ 4* 4+

STRAND TRANSFER a a 4+ 4+ 4* 0 a 0 +I- 4+ 4+ 1’

DISINTEGRATION 4+ 4+ 4+ 4+ 3‘ o z+ o 4- 4+ 4+ 4+

FIG. 1. Summary of in vitro activities of HIV-1 IN mutants. HIV-1 IN is depicted as a horizontal bur with the amino terminus at the left and the carboxyl terminus at the right. The location of each mutant is indicated by a vertical line. Rl99CA has the indicated amino acid alteration, but a frameshift mutation produces a truncated protein with 247 amino acids from IN plus a 19-amino acid addition (CRGMHSRHKSSAKKKSKNH) for a total length of 266 amino acids. Asterisks indicate the locations of conserved amino acids among IN proteins (18). Activities are indicated relative to wild type IN under the reaction conditions described under “Materials and Methods”: 4+ = >75% wild type activity; 3+ = 50-75% wild type activity; 2+ = 10-49% wild type activity; I+= 1-9% wild type activity; +/- = trace, could be detected; 0 = undetectable; a = active, but relative activity could not be determined due to poor protein production or contaminating nucleases.

acid position 16. One change was to cysteine (H16C), which may preserve zinc coordination; the other change was to valine (H16V), which should disrupt coordinated zinc binding. Using 5”radiolabeled att site oligonucleotides to assess 3’ processing and strand transfer, both H16C and H16V function similarly to wild type IN (Fig. 4, A and B ) . The disintegration activity of H16C and H16V also mirrors that of wild type IN (Fig. 5).

Others have found that glycerol and 1,2-ethanediol can substitute for water as the attacking nucleophile during 3‘ processing, generating a uniquely migrating adduct composed of the nucleophile complexed with the released dinucleotide (22, 25). Incubation of wild type HIV-1 IN or mutant H16C with HIV-1 U5 att site oligonucleotides demonstrates the expected release of the 3”terminal dinucleotide, GpT (Fig. 4C). In the presence of glycerol or 1,2-ethanediol, wild type IN and H16C can each generate adducts between the released GpT dinucleotide and added glycerol or 1,2 ethanediol (Fig. 4C). In contrast, H16V mediates the release of the GpT dinucleotide but does not allow for the use of glycerol or 1,2- ethanediol as a nucleophile in the 3’ processing reaction, as demonstrated by the lack of adduct formation (Fig. 4C).

We also attempted to analyze two other mutants with alterations in the HHCC region, H12C and H12C/H16C. While these mutants retain some 3‘ processing and strand transfer activity, poor protein production and a co-purifying nuclease activity prevented useful comparison of these activities to those of wild type IN (Fig. 1). The disintegration activity of H12C and H12C/H16C appears similar to that of wild type HIV-1 IN (Fig. 5 ) .

Alteration of Highly Conserved Amino Acids Does Not Al- ways Impair IN Actiuity-Near the center of IN proteins is a highly conserved region characterized by the invariant acidic residues D and E (positions 116 and 152 in HIV-1 IN) sepa- rated by exactly 35 amino acids. Single amino acid substitu- tions, Dl161 and E152G, were made at these residues. Incu- bating these mutant proteins with the appropriate radiolabeled DNA substrate showed no 3‘ processing, strand transfer, or disintegration activity for D1161, whereas trace amounts of product were detected in all three assays of E152G (Figs. 5 and 6). In contrast, changing the highly conserved tryptophan a t position 235 to glutamic acid (W235E) resulted in no demonstrable impairment of 3’ processing, strand transfer, or disintegration (Figs. 5 and 6). Clearly, not all highly conserved amino acids of IN are essential for these in vitro activities.

Another change in a highly conserved amino acid, S81 to R, generated a mutant protein that was highly insoluble and co-purified with a non-specific nuclease activity, preventing definition of the 3‘ processing and strand transfer activities of S81R relative to wild type IN. We were able, however, to determine that S81R maintains some degree of 3’ processing and strand transfer activity, and that disintegration is somewhat impaired relative to wild type IN (Figs. 1 and 5) .

HIV-I IN Mutant D64V Retains 3’ Processing but Not Strand Transfer or Disintegration Actiuity-In addition to altering the conserved residues at D116, E152, W235, and s81, we changed the highly conserved aspartic acid residue at position 64 to valine, creating mutant D64V. Assays with D64V demonstrated somewhat reduced 3‘ processing activity (Fig. 7A ), but no strand transfer activity in either the standard assay (Fig. 7 B ) or the target site selection assay (Fig. 9). Disintegration activity, like strand transfer activity, was essentially undetectable (Fig. 5). The differential effect on IN activities seen with mutant D64V is unique among our mutants with a single amino acid alteration.

A Carboxyl-terminal Truncation Affects Strand Transfer More than 3’ Processing or Disintegration-In addition to altering conserved amino acids, we changed the unconserved arginine at position 199 to cysteine, generating mutant R199C. This mutant demonstrated 3’ processing, disintegration, and strand transfer activities indistinguishable from wild type IN (Figs. 5 and 8). During the generation of R199C, an unexpected mutant (R199CA) was identified due to its greater than expected mobility on SDS-PAGE (Fig. 3). Nucleotide sequencing revealed a frameshift mutation in codon 248, in addition to the desired change in codon 199. The frameshift resulted in 19 missense codons beyond position 247 before reaching a stop codon. R199CA therefore lacks the carboxyl- terminal 42 amino acids of wild type HIV-1 IN, and has acquired 19 unrelated amino acids, for a total amino acid length of 266 (wild type IN = 289 amino acids). With R199CA, 3’ processing is at most mildly impaired relative to wild type IN (Fig. 8A), but strand transfer activity is markedly reduced (Fig. 8B). Disintegration activity is the same as wild type IN (Fig. 5). Thus, strand transfer and disintegration, two IN activities which appear to be simple reverse reactions of one another, are differentially affected in R199CA.

Integration Site Selection Is Not Altered in Any of the Mutants Analyzed-The sites of IN-mediated viral DNA integration into target DNA are influenced by the target DNA

In Vitro Analysis of HIV-1 Integrase Mutants 2116

A @

*- CADT +

+ *-CAB,

c @

*

*

*

OR * [” -b ”

30 -I - I 4. .

30 * ” I FIG. 2. Schematic representat ion of the 3’ processing,

s t rand t ransfer , and dis integrat ion assays. Reaction conditions for each assay are given under “Materials and Methods.” In each reaction depicted above, starting materials are indicated on the left and reaction products are indicated in the middle. IN, integrase protein; linr drawings, annealed oligonucleotide substrates and products. The hold line in each oligonucleotide complex indicates the radiolabeled strand that can be followed by autoradiography, and the asterisk indicates the location of the radiolabel. To the right is a schematic of the reaction products after electrophoresis on a dena- t.uring polyacrylamide gel, followed by autoradiography. The gel shows the radiolabeled products from reactions incubated either without (-) or with (+) IN. During 3’ processing ( A ) , IN removes the two nucleotides 3‘ of the conserved CA thereby shortening the starting radiolabeled oligonucleotide strand (hold arrow) by two nucleotides (stippled arrow). During strand transfer (19). IN mediates the coordinated ligation of the 3”CA of one duplex oligonucleotide to the 5’ end of a new cut in a second duplex oligonucleotide yielding radiolabeled products that are longer (indicated by the bracket) than the starting material (hold arrow). The product of strand transfer, the gapped intermediate, can be fashioned by annealing appropriate oligonucleotides as shown in C. This can be used as the substrate for the disintegration reaction, during which the cut target strand is re- ligated and the incoming viral att site oligonucleotide is released. As presented, this yields a radiolabeled 30-nucleotide strand when begin- ning with a radiolabeled 16-nucleotide strand.

sequence regardless of whether purified IN or nucleoprotein complexes provide the in oitro integration activity (14-16). Integration site selection within a given target sequence, however, is IN protein-specific (15). To investigate if any of our mutants affected integration site selection, we incubated the pre-processed U5 HIV-1 att site oligonucleotide and bacterial plasmid DNA with each mutant IN protein. The products of integration were amplified by PCR using one primer from the incoming att site sequence and another from the target plasmid. Wild type IN and mutant IN proteins that retained strand transfer capacity showed the same preferences for integration sites in an analysis that examined over 300 potential sites in the target plasmid (Fig. 9).

DISCUSSION

To better understand the mechanism of IN activity, we evaluated 12 site-directed mutants of HIV-1 IN with assays for 3’ processing, strand transfer, disintegration, and target

OT-

W -

43-

-29-

18 - I4 -

1 2 3

FIG. 3. SDS-PAGE of 4 5 6 7 0 9 10 11 1 2 1 3

ourif ied wild tYDe a n d m u t a n t IN proteins. The indicated HiV-1 I N proteins-were prepared as described under “Materials and Methods.” Samples were subjected to electrophoresis in 12.5% SDS-PAGE after phenyl-Sepharose column chromatography and were stained with Coomassie Hrilliant Hlue. Each sample represents roughly 15-fold the amount used in 3’ processing and strand transfer reactions and 30-fold the amount used in disintegration reactions. The positions of molecular weight markers are indicated on the left. The arrou, indicates the IN hand in each lane except for R199CA (lane f I), which moves slightly faster than the other I N proteins due to i h shorter length (266 ucrsu.q 289 amino acids).

- B

1 2 3 4 1 2 3 4 1 2 3 4 5 6 7 0 9 1 0

FIG. 4. The amino-terminal HHCC region can be al tered without s ignif icant impairment of in vitro integrase act ivi- ties. A, 3’ processing reactions were performed using oligonucleotide U5-29/29 with the upper strand radiolabeled a t the 5’ end (see “Materials and Methods”). The hold arrow indicates the position of the starting oligonucleotide substrate; the st ippkd arrow indicates the 27-nucleotide product. R, the strand transfer reactions were performed using the pre-processed att site oligonucleotide U5-27/29. with the processed (upper) strand radiolaheled at the 5’ end. The bold arrow indicates the starting oligonucleotide substrate, and the strand transfer products are indicated. C. 3’ processing reactions were performed using U5-27/29 filled in with dGTP and [n - :”P1TP to allow for monitoring of the cleaved terminal dinucleotide. The bold arrow indicates the starting oligonucleotide substrate. and the hold arrowhead indicates the dinucleotide product of correct 3’ processing. The stippled arrow indicates the adduct formed between glycerol and the cleaved dinucleotide; the atipplrd arrouhead indicates the adduct formed between 1,2-ethanediol and the cleaved dinucleotide. fmnr I in A, R, and C represents the substrate incubated in the absence of IN. 1.2-Ethanediol and glycerol were each used at 20%.

site selection. We have found that not all highly conserved amino acids are essential for IN activities in vitro, the proposed zinc-coordinating domain can tolerate certain changes without significant loss of in vitro IN activities, and strand


1 2 3 4 5 6 7 8 9101112131415

FIG. 5. Disintegration activity of wild type and mutant HIV-1 IN proteins. Disintegration reactions were performed according to the method of Chow et al. (17) but incubated for 20 min at 30 “C. Equal amounts of each IN protein, indicated above each lane, were used for each reaction. The radiolabeled 16-nucleotide piece of the disintegration substrate is indicated by the stippled arrow, and the 30-nucleotide product of the disintegration reaction is indicated by the black arrow. The substrate was incubated with either no I N (lane I), wild type IN (lane 2 ) , indicated mutant INS (lanes 3- 1 4 ) , or yeast extract from a mock IN purification (lane 1.5).

A W n

B

I-

W E

1 2 3 4 5 1 2 3 4 5 FIG. 6. Not all highly conserved amino acids are necessary

for in vitro integration activities. Three mutant proteins, each with alterations a t highly conserved amino acids, were assayed for their IN-related activities. 3’ processing ( A ) and strand transfer ( E ) reactions were performed as described in the legend to Fig. 4 ( A and H). Lane 1 in A and H represents the substrate incubated in the absence of IN.

transfer and disintegration can be differentially affected in a given mutant. In addition, all of our mutants with detectable strand transfer activity select target sites in a pattern identical to that of wild type IN.

Altering Conserved Amino Acids Does Not Always Result in Significant Loss of IN Activities-Nine of our 12 HIV-1 IN mutations alter highly conserved amino acids. Three of these amino acids, D64, D116, and E152, are conserved not only among retroviral IN proteins and IN-like proteins of other retroelements, but also in the IS3 family of bacterial transposable elements (19, 20). Such broad amino acid conserva-

1 2 3 1 2 3 FIG. 7. Changing Asp-64 to Val impairs strand transfer

more than 3’ processing. The mutant protein with an alteration at the highly conserved position 64, D64V, was assayed for its IN- related activities. 3’ processing ( A ) and strand transfer ( R ) reactions were performed as described in the legend to Fig. 4 ( A and B ) . Lane 1 in A and R represents the substrate incubated in the absence of IN.

A

*

1 2 3 4

a

f a c 0

4

z 4

1 2 3 4 FIG. 8. A carboxyl-terminal truncation preferentially im-

pairs strand transfer. Two mutant proteins, one with an alteration a t a non-conserved position (H199C) and one with the H199C alteration plus a carboxyl-terminal truncation (R199CS, were ansaved for their IN-related activities. 3’ processing ( A ) and strnntl transfer ( R ) reactions were performed as described in the legend to Fig. 4 ( A and R). Lane I in A and H represents the substrate incuhated in the absence of IN.

tion suggests that these three sites are important for catalytic functions common to these transposases. Mutant Dl161 has no detectable in vitro activity in assays for 3’ processing, strand transfer, and disintegration, while E152G has trace activity for all three functions. For each mutant, all IN- mediated activities are affected similarly. On the other hand, D64V lacks strand transfer and disintegration activity, yet retains 3‘ processing activity at a diminished level. The differential effect on IN functions exhibited by D64V raises the

2118

350 + 300 +

250 .)

200 .)

I50 +

1 0 0 .)

WILD TYPE

In Vitro Analysis of HIV-1 Integrase Mutants

1 2 3 4 5 6 7 8 9 10 11 12 13 14 1516

FIG. 9. Analysis of target site selection by IN mutants. As- says for target site selection were performed as described under “Materials and Methods” using U5-27/29 ott site oligonucleotides as donor DNA, Bluescript plasmid (Stratagene) as target DNA, and the indicated HIV-1 IN protein. Control integration reactions were performed usingwild type IN either without target DNA (lane 1 ), without att site oligonucleotides (lane 2 ) , without IN protein (lane 3), or as a complete integration reaction (lane 4) . The numbers to the left indicate the nucleotide lengths of the PCR amplification products.

intriguing possibility that there is either more than one active site in the IN protein or a single active site with functionally discrete residues.

Our mutants with alterations at amino acids Dl16 and E152 of HIV-1 IN behaved similarly to recently published RSV and HIV-1 IN mutants (28). The authors also changed the RSV IN amino acid that corresponds to HIV-1 IN D64, and reported the mutant (D64E) to be equally deficient in 3’ processing and strand transfer. They did not make the cor- responding mutation a t D64 in HIV-1 IN. I t is not clear if the phenotypic difference between these D64 mutants reflects an unappreciated difference in the assay conditions, a difference due to the V or E substitution a t amino acid position 64, or a real difference in the role of the aspartic acid residue a t position 64 in the two IN proteins.

W235 is also a highly conserved amino acid of IN proteins, but mutant W235E demonstrated wild type activity for all IN functions assayed. Given that most IN mutants analyzed in vivo have an integration defective phenotype (29-31), we were somewhat surprised to find that alteration of a highly con-

served amino acid produced a mutant with unaltered in vitro IN activities. It is possible that the coordinated strand transfer of two viral ends occurring at a single target site in vivo requires IN-IN interactions not required in the in vitro strand transfer assay. Alternatively, some of the conserved amino acids may be responsible for positioning IN within the nucleoprotein complex; alteration of such amino acids may not be detrimental in the in vitro assays. Consequently, we are reconstructing W235E and our other mutants into viruses to assess their phenotypes in vivo.

The HHCC Region of HIV-1 IN Can Be Altered without Loss of in Vitro Activities-The constellation of histidine and cysteine residues conserved in IN proteins has been postulated to be a zinc-finger domain important in DNA binding by IN (18). Direct spectroscopic evidence has recently demonstrated that the HHCC region of HIV-1 IN coordinates zinc, whereas three mutants, H12Q, C40S, and C43S, all fail to do so (32). The functional significance of this region remains unclear (19, 33), and we made four mutants to address this question. Two of the mutants, H12C and H12C/H16C, retained some 3’ processing, strand transfer, and disintegration activities, but detailed analysis was limited by low production and co- purifying nuclease activity. Two other mutations within the HHCC region altered the second histidine in what we pre- dicted might be a conservative (H16C) and a non-conservative (H16V) manner. However, both mutants demonstrated nearly wild type activities in vitro. The putative zinc-coordinating residues of the HHCC region are therefore more tolerant of amino acid changes than would be expected of zinc-coordinating residues in known DNA-binding proteins. Taken to- gether, it seems that the HHCC region coordinates zinc, but zinc binding may not be essential for the IN activities assayed in’oitro. Of interest, one infectious clone of HIV-2 (NIHZ) has the amino acids HHCV instead of the HHCC sequence found in the zinc-coordinating region of all other HIV-1 and HIV-2 isolates (34, 35). Our finding that the HHCC region can be altered without significant loss of IN-related activities may depend on the amino acid chosen for replacement.

Alteration of H16 to V impairs the ability of glycerol and 18-ethanediol to serve as nucleophiles in 3’ processing. In contrast, both compounds can serve as nucleophiles with wild type IN or mutant H16C. The significance of this finding remains unclear, but a conformational change in the protein might prevent larger nucleophiles like glyerol and 1.2-ethanediol from entering the active site.

The Carboxyl-terminal 42 Amino Acid9 Are Necessary for Full in Vitro IN Activity-R199CA, a carboxyl truncation mutant, was inadvertently created while making R199C. R199CA retains full disintegration activity, but strand transfer activity is severely impaired. The differential effects on strand transfer and disintegration are curious, since the two reactions appear to be simple reversals of each other. How- ever, the DNA sequence and structural requirements needed for disintegration are less stringent than those needed for strand transfer.* Furthermore, strand transfer is a tri-molecular reaction in vitro (requiring a single att site oligonucleotide, the target DNA, and the IN protein) and a quatro- molecular reaction in vivo (requiring one att site from each end of the viral genome, target DNA, and IN protein). In contrast, disintegration is a bi-molecular reaction requiring IN to interact with only a single DNA species, the gapped intermediate. Thus, certain changes in IN might have more impact on strand transfer than on disintegration, even if the two reactions are essentially the reverse of each other. Such changes may reside in the DNA binding or recognition site(s)

* S. Chow and P. Brown, personal communication.


of the protein, or perhaps in regions required for IN-IN interactions that are required for strand transfer but not disintegration.

Target Site Selection I s Not Altered i n A n y of the Mutants That Retained Strand Transfer Actiuity-HIV-1 IN and MLV IN show distinct patterns of integration site selection into a common, heterologous plasmid in vitro (15). Analysis of target site selection by the IN mutants presented in this paper showed no alteration from that seen with wild type IN over a roughly 300-base pair region of naked, plasmid DNA. Target site selection is therefore not necessarily influenced by the level of strand transfer activity, and we have yet to identify the residues in HIV-1 IN responsible for its pattern of integration site selection.

Acknowledgments-We thank Pat Brown, Hans-Peter Muller, Pe- ter M. Pryciak, and Karen Vincent for helpful discussions during the course of this work, Peter M. Pryciak for help with the PCR assays, Sam Chow for help with the disintegration assays, and Hans-Peter Muller for critical reading of the manuscript.

Note Added in Proof-Subsequent to the submission of this manuscript, other investigators have published data on the activity of HJV-1 IN mutants that are similar to our findings. These include: van Gent, D. C., Antoinette, A. M., Groeneger, O., and Plasterk, R. H. A. (1992) Proc. Natl. Acad. Sci. U. S. A. 89,9598-9602; Engleman, A., and Craigie, R. (1992) J. Virol. 66, 6361-6369; and Vincent, K. A,, Ellison, V., Chow, S. A., and Brown, P. 0. (1993) J. Virol., in press.

REFERENCES

1. Colicelli, J., and Goff, S. P. (1985) Cell 4 2 , 573-580 2. Colicelli, J., and Goff, S. P. (1988) J. Mol. Biol. 199 , 47-59 3. Fujiwara, T., and Craigie., R. (1989) Proc. Natl. Acad. Sci U. S. A. 8 6 ,

4. Panganiban, A. T., and Temin, H. M. (1983) Nature 306 , 155-160 5. Roth, M. J., Schwartzberg, P. L., and Goff, S. P. (1989) Cell 5 8 , 4 7 3 4 6. Donehower, L. A,, and Varmus, H. E. (1984) Proc. Natl. Acad. Sci. U. S. A.

3065-3069

81,6461-6465

7.

8.

10. 9.

11. 12.

13. 14.

15. 16.

17.

18.

19.

20.

21.

22.

23.

24.

25. 26.

27.

28.

29. 30.

31. 32.

33.

34.

35.

Panganiban, A. T., and Temin, H. M. (1984) Proc. Natl. Acad. Sci. U S. A. Q1 7!2!26"7QQCl

Quinn, T. P., and Grand enett D. P. (1988) J. Virol. 6 2 , 2307-2312 Schwartzberg, P., Colicel%, J., i n d Goff, S. P. (1984) Cell 3 7 , 1043-1052 Brown, P. O., Bowerman B., Varmus, H. E., and Bishop, J. M. (1989) Proc.

Craigie, R., Fu'iwara, T., and Bushman, F. (1990) Cell 6 2 , 829-837 Katzman, M., ka t z , R. A., Skalka, A. M., and Leis, J. (1989) J. Vtrol. 63,

", ," ."""

Natl. Acad. Sci U. S. A'. 86,2526-2529

5.11.9-5.127 Whitcomb, J. M., and Hughes, S. H. (1991) J. Virol. 65 , 3906-3910 Leavitt, A. D., Rose, R. B., and Varmus, H. E. (1992) J. Virol, 6 6 , 2359-

""

Pryciak, P. M., and Varmus, H. E. (1992) Cell 69 , 769-780 Kitamura, Y., HaLee, Y. M., and Coffin, J. M. (1992) Proc. Natl. Acad. Sci.

Chow, S., Vincent, K. A,, Ellison, V., and Brown, P. 0. (1992) Science 2 6 5 ,

Johnson, M. S., McClure, M. A,, Feng, D. F., Gray, J., and Doolittle, R. F.

Kahn, E., Mack, J. ,P. G., Katz, R. A., Kulkosky, J., and Skalka, A. M.

Fayet, O., Ramond, P., Polard, P., Prere, M. F., and Chandler, M. (1990)

Lin. T.-H.. Quinn. T. P.. Grandwnett. D. P.. and Walsh. M. T. (1989)

2368

U. S. A. 89,5532-5536

723-726

(1986) Proc. Natl. Acad. Sci. U. S. A. 8 3 , 7648-7652

(1991) Nuclerc Accds Res. 19 , 851-860

Mol. Microbiol. 4, 1771-1777

ProteiniSiruct. Funct. Genet. g, 156-165

Plasterk, R. H. A. (1992) Nucleic Acids Res. 19,6691-6698

65.5fi24-5fi30

Vink, C., Yeheskiely, E., van der Marel, G . A,, van Boom, J. H., and

LaFemina, R. L., Callahan, P. L., and Cordingley, M. G . (1991) J. Virol.

Bushman, F. D., and Craigie, R. (1991) Proc. Natl. Acad. Sci U. S. A. 8 8 ,

Engelman, A., Mizuuchi, K., and Craigie, R. (1991) Cell 6 7 , 1211-1221 Kunkel, T. A., Roberts, J. D., and Zakour, R. A. (1987) Methods Enzymnl.

Barr, P. J., Sabin, E. A., Ng, C. T. L., Landsber K E Steimer, K. S., Barhurst, I. C., and Shuster, J. R. (1991) in %z&es& Systems and Processes for Recombinant DNA Products (Hatch, R. T., Moreira, C. G.

ton, D. C. T., and Alroy, Y., eds) pp. 51-64, American Chemical Society, Washing

Kulkosky, J., Jones, K. S., Katz, R. A., Mack, J. P. G., and Skalka, A. M.

Roth, M. J. (1991) J. Vwol. 6 5 , 2141-2145 (1992) Mol. Cell. Biol., 12 , 2331-2338

Roth, M. J., Schwartzberg, P., Tanese, N., and Goff, S. P. (1990) J. Virol.

Donehower, L. A. (1988) J. Virnl. 6 2 , 3958-3964 Burke, C. J., Sanyal, G., Bruner,,M. W., Ryan, J. A,, LaFemina, R. L.,

Robbins, H. L., Zeft, A. S., Mlddaugh, C. R., and Cordmgley, M. G. (1992) J. Biol. Chern. 267,9639-9644

Woerner, A. M., M. Klutch, M., Levin, J. G., and Marcus-Sekura, C. J. (1992) AIDS Res. Hum. Retroviruses 8, 297-302

Franchm, G., Fargnoh K. A,, Giombini, F., J a odzinski, L., De Rossi, A. Bosch, M., Biberfeld, G., Fenyo, E. M., Altert, J., Gallo, R. C., and

Myers G., S. F. Josephs and J. A. Berzofsky (1991) Human Retroviruses Wong-Staal, F. (1989) Proc. Natl. Acad. SCL. U. S. A. 8 6 , 2433-2437

Acid Sequences, Los Alamos Natxonal Lagoratory, Los Alamos, NM andkIDS 1991: A Co&ilation and Anal sis of Nucleic Acid and Amino

"."" ""

1339-1343

154,367-382

64,4709-4717

site-directed mutagenesis of hiv- 1 integrase demonstrates

Documents