Eur. J. Biochem. 205, 1115-1121 (1992)

0FEBS 1992

Expression, purification and biochemical characterisation of the human immunodificiency virus 1 nefgene product Vera WOLBER', Hans RENSLAND'. Birgit BRANDMEIER'. Martin SAGEMANN2. Ralf lIOFFMANN3, Hans Robert KALBITZER' and Alfred WITTINGHOFER Max-Planck-Institute for Medical Research, Heidelberg, Federal Republic of Germany European Molecular Biology Laboratories, Heidelberg, Federal Republic of Germany ' University of Saarbriicken, Federal Republic of Germany (Received December 12, 1991/January 29, 1992) - EJB 91 1641

The human immunodeficiency virus 1 (HIV-1) nef gene encoded by the HIV-1 isolate lymphadenopathy-associated virus type 1 was expressed in Escherichia coli under the control of the lac promoter. The protein is found mainly in the soluble part of the bacterial lysate; a simple twocolumn purification scheme has been developed allowing isolation of the recombinant protein without using denaturing agents. Analysis of the circular dichroism spectra reveals that the purified protein is folded and has a helix content of 16% and a P-pleated sheet content of 31%. GTPase activity and binding of guanine nucleotides were measured for Nef and compared with the results obtained under identical experimental conditions for p2lraSCc, which represents a typical, well-characterized guaninenucleotide-binding (GNB) protein. Within the limits of error, native Nef does not show GTPase activity and does not bind guanine nucleotides strongly (association constant, K,,, < 5 x lo3 M - ' 1. An upper limit for the association constant of Nef for ATP was determined by equilibrium dialysis as 5 x lo3 M-I. Nef can be autophosphorylated by ATP; under the experimental conditions used, 1-2% of the protein become phosphorylated. Correspondingly, our Nef preparation shows a low, but significant, ATPase activity. In conclusion, Nef is not a member of the GNB protein family, but a possible role as a protein kinase cannot be excluded.

Human immunodeficiency virus (HIV) shows unusual complexity in its genomic structure and gene regulation relative to other animal retroviruses. Beside the classic structural genes of retroviruses (gug, pol, env), the 9.2-kb HLV genome comprises at least six other genes whose products have been serologically and/or genetically identified: neL tat. rev, vif, vpu and vpr. The nefgene is located near the 3' end of the viral genome, partially overlapping the U3 region of the long terminal repeat sequence. As the gene is conserved in HIV and simian immunodeficiency virus (SIV), it is supposed to have an important function in the life cycle of the virus. However, the function of its gene product, Nef, is still unclear. Deletions or mutations of the nefgene from an HIV-1 infectious provirus were reported to enhance virus replication and cytopathic effects in vitro [l -61. These observations suggest that the nef gene product negatively regulates HIV production. However, in other systems Nef fails to show an effect on HIV replication [7-91. h vivo, Nef is essential for efficient viral replication

and progression of acquired immunodeficiency syndrome (AIDS) in rhesus monkeys infected with SIV [lo]. Amino acid sequence comparison has suggested that Nef may be a guanine-nucleotide-binding(GNB) protein or a protein kinase [ll - 131. No quantitative adenine-nucleotidebinding studies or ATPase activity tests have been reported so far. However, there have been conflicting reports regarding the GNB protein and GTPase activities of HIV-1 Nef [12161. In order to reveal or exclude possible functions of the Nef protein, it appears necessary to isolate native Nef protein and study its nucleotide-binding and NTPase activity under controlled experimental conditions using well-defined biochemical methods. Moreover, the expression and purification of large amounts (that is, several hundred milligrams) of native protein are prerequisites for the determination of the threedimensional structure of the protein by X-ray crystallography or NMR spectroscopy.

Currespondence to H. R. Kalbitzer. Max-Planck-Institut fur medizinische Forschung, Abt. Biophysik, Jahnstr. 29, W-6900 Heidelberg, Federal Rcpublic of Germany Fax. +49 6221 486 351 Abbreviations. GNB, guanine-nucleotide-binding;HIV, human immunodeficiency virus; SIV, simian immunodeficiency virus; AIDS, acquired immuno deficiency syndrome; mant-GDP, 3'-O-(N-methylanthraniloyl) guanosine diphosphate. Enzymes. Alkaline phosphatase (EC 3.1.3.1); polynucleotide kinase (EC 2.2.1.78); restriction endonucleases (EC 3.1.21.4).

MATERIALS AND METHODS [8-3H]GDP (485 GBq/mmol, Amersham Buchler) was diluted to the desired specific radioactivity (200 - 300 cpm/ pmol) with unlabelled GDP (Pharma Waldhof). [ Y - ~ ~ P J G T P and [ Y - ~ ~ P I A(300 T P -400 GBq/mmol; Amersham Buchler) was diluted with unlabelled nucleotide (Pharma Waldhof) to the desired specific radioactivity (300 -400 cpm/pmol). Cibacron blue F3GA (Ciba-Geigy) was coupled to Sepharose CL6B (Pharmacia) as described [17]. Ultrogel AcA54 was

1116

from Pharmacia LKB. Isopropyl-tho-B-D-galactoside was from Serva (Heidelberg). Nitrocellulose filters (type BA85, 0.45 pm) were from Schleicher & Schiill, HPLC-grade acetonitrile was obtained from Merck. All other reagents were of the highest purity available. Plasmids and Escherichia coli strains Plasmid pUCl8nef containing the 1.4-kb Hind111 fragment of pNL4-3 [18] was donated by R. Braun. The E. coli strain W31101acIq ( F , hsdR-, hsdM+, l a d q ) is a luc-repressor-overproducing strain [19], which was obtained from E. Amann. The p21'"'C expression system and the procedure used for isolation of the recombinant protein have been described previously [20, 211. Cloning methods Restriction endonucleases, alkaline phosphatase and polynucleotide kinase were products of Boehringer Mannheim or BRL Gibco and were used following the laboratory manual of Maniatis et al. [22]. The sequence of the final ptacnef construction was verified by dideoxynucleotide sequencing.

E. coli culture and protein purification The cells were grown in an 50-1 fermenter at 30°C. The culture medium was described previously [20]. Induction with 50 mM isopropyl-thio-p-D-galactosidelasted 12 - 15 h and was initiated when the absorbance (660 nm) of the cell culture was approximately 0.7. The soluble extract of E. coli cells was applied to a pseudoafinity Cibacron blue F3GA column (800 rnl), equilibrated with buffer A (64mM Tris/HCl, pH 7.5, 1 mM NaN3, 0.5 mM dithioerythritol, 0.01 mM phenylmethylsulfonyl fluoride, 1 mM EDTA, 0.05 M NaCl). The column was developed with a 4-1 linear gradient of 0.05 - 1 M NaCl in buffer A and the protein was eluted by 0.45 M NaCl. It was concentrated by ammonium sulfate precipitation (40% ammonium sulfate, 45 min). After centrifugation at 40000 x g for 30 min, it was dissolved in 15 ml buffer B (buffer A with 0.1 mM GDP and 200 mM NaCl added), clarified by centrifugation and eluted from an AcA54 gel-filtration column. Filter-binding assay 2.5 pM protein (Nef or p21'"'C) with 10 pM [8-3H]GDP (200-300 cpm/pmol) or 10 pM [ Y - ~ ~ P I G(300-400 TP cpm/ pmol) in buffer C (64mM Tris/HCl pH 7.6, 1 mM dithioerythritol, 10 mM EDTA, 1 rnM NaN3), in a total volume of 25 pl, were incubated for 30 min at 0°C. 2 pl 1 M MgC12 was then added. The mixtures were subsequently filtered through nitrocellulose, washed three times with 2 ml rinse buffer (64 mM Tris/HCl, pH 7.6, 5 mM MgC12, 100 mM KCl, 10 mM NH,Cl, 1 mM 2-mercaptoethanol) and the filterbound radioactivity was determined by liquid scintillation counting. HPLC determinations A C18reversed-phase column (0.4 cm x 25 cm) filled with Shandon Hypersil was obtained from Abimed. The system consisted of a Beckman 280B pump and an ultraviolet 111 detector (254 nm) from Latek. The column was run at ambient temperature with a flow rate of 2.5 ml/min with 50 mM so-

dium phosphate buffer, pH 6.5, containing 0.2 mM t-butylammonium bromide, 8% (by vol.) acetonitrile and 0.2mM NaN,. The column was calibrated with solutions of appropriate nucleotides of known concentration. The amount and identity of protein-bound nucleotide was determined by applying the protein sample directly to this column and eluting with the above buffer. GTPase and ATPase assays GTPase was assayed using the calibrated HPLC system described above. 5 pl of the reaction mixture (SO pM Nef or p21'"'CC,2 mM nucleotide in buffer D (50 mM Hepes, pH 7.5, 2 mM MgC12, 1 mM dithioerythritol) was injected onto the HPLC column after incubation for various times at room temperature, and the increase in GDP and the decrease in GTP was analyzed. ATPase activity was determined similarly, but with ATP replacing GTP. Autophosphorylation assay 100 pM Nef protein was dissolved in buffer D containing 500 pM [y-32P]ATPand 500 pM [ Y - ~ ~ P I G(800 T P cpmlpmol). After incubation for various times, 20 ml of the solution was filtered through nitrocellulose and washed three times with 2 ml rinse buffer containing 50 mM ADP. The filter-bound radioactivity was determined by liquid scintillation counting. Binding studies with fluorescent analogues Fluorescence measurements were performed with an SLM Smart 8000 photon-counting spectrofluorometer (Colora, Lorch) equipped with a PH-PC9635 photomultiplier. Fluorescent adenine and guanine nucleotides, synthesized according to John et al. [23], were donated by R. s. Goody. 5 pM nucleotide-free p21'"'C, prepared as described by John et al. [23], and 5 pM Nef were each allowed to react with an equimolar amount of 3'-O-(N-methylanthraniloyl)-guanosine diphosphate (mant-GDP) in buffer D in a final volume of 1 ml. Analytical gel fdtration Analytical gel filtration was performed with two HPLC columns in series (Bio-Rad TSK-250, both 300 mm x 7.5 mm) using an ISCO 2350 pump and ultraviolet detection (280 nm). The column was eluted with 200 mM sodium phosphate, pH 6.5, 5% (by vol.) acetonitrile, at 1 ml/min, and calibrated using a mixture of markers with known molecular mass (USB marker kit 30180). Protein sequencing The N-terminal sequence of recombinant Nef was determined by Edman degradation using an Applied Biosystems 473A protein sequencer. The phenylthiohydantoin derivatives of amino acids were analysed on-line by a Microbare HPLC and processed on a Macintosh IICX. Equilibrium dialysis 65 pM Nef and p21 laSC were each dissolved in 200 p1 buffer D containing the nucleotide ([8-3H]ATPor [8-3H]GDP) and dialysed against the same volume of nucleotide-containing

1117

ptac nef

nef Ola 5'-~UITTCTAT~TUjC AAGTGGTCCI\I\RRGTAG~TffiTTffiA~GC~-3 . nef Olb 3 . ~ T ~ A C C G ~ A C C A ~ ~ C A ~ C A C ~ C ~ A C ~ G - 5 . ncf 02a

5' - G T M C G G ~ G I \ I \ T G A G A C G ~ A ~ ~ - ~ T ~ ~ T C - 3 .

ncf 02b 3 ' - A C G T C A T T C C C ~ T C A ~ ~ ~ C T C f f i C ~ T C G T C T A C C C ~ C C C T f f i T C G T A G ~ - S

Fig. 1. The nef expression vector. The nef expression vector and oligonucleotides linkers used to connect the EcoRI site before ATG with the XhoII site in the coding region. Amp', ampicillin resistance gene; ori, origin ofreplication; tac, tacprornoter; SD, Shine/Dalgarno sequence; t o , transcription terminator of phage 1 amida.

buffer D for 24 h. The nucleotide concentrations were varied in the range 12.5- 100 pM. Circular dichroism and absorption spectroscopy CD measurements were performed on a Jobin Yvon CD6 spectropolarimeter interfaced to a NEC personal computer for data collection and manipulation. The instrument was calibrated with d-10-camphorsulfonic acid. Depending on the protein concentration, rectangular cuvettes with 0.01 cm and 0.1 cm path lengths were employed. Molar mean residue ellipticity values (0,)were expressed using a mean amino acid residue mass of 133.31 Da. All spectra were recorded in the range 190-260 nm; an averaged of 30-90 scans gave satisfactory signal/noise ratios. The measurements were performed in 20 mM sodium phosphate Suffer, pH 7.2, at room temperature with varying protein concentrations. The protein concentration was determined according to Gill and von Hippel [24]. Optimal experimental results were obtained with a protein concentration of 2.0 mg/ml. The spectra were analysed according to Provencher and Glockner [25].

RESULTS Construction of the expression vector To produce Nef in E . coli, we used the expression vector pKM-tacI, which has been used successfully for producing soluble p21ra'C (261. The Nef-encoding sequence was isolated from pUC18nef [18], which contains the 1.47-kb Hind111 fragment from positions 8175-9646 of pNL4-3 [27], and inserted into the expression vector ptacras [21] in the correct orientation. Part of the nefsequence encoding the N-terminus and containing the 5' untranslated nucleotides were eliminated by EcoRIIXhoII digestion. The DNA sequence, encoding the authentic Nef protein, was regenerated and connected to the Shine/Dalgarno sequence of the vector, by inserting synthetic

Fig.2. Induction and purification of Nef. (A) Nef in induced and uninduccd cells. Lane M, molecular mass markers (kDa); E. coli lysate of uninduced cells (lane 1) and induced cells (lane 2). (B) SDS/PAGE of fractions obtained during the purification of Nef. Lane M, molecular mass markers (kDa); soluble part of the E. coli lysate (lane I), blue Sepharose pool (lane 2) and AcA54 pool (lane 3). Table 1. Purification of recombinant Nef from E. coli. Purification step

Total protein

Nef content

mg 5050 480 76

YO

Crude extract Nef/pool after blue Sepharose Nef/pool after gel filtration

17 43 > 95

oligonucleotide linkers consisting of two overlapping pieces comprising the coding sequence of the 34 N-terminal residues (Fig. 1). The linker contains several changes to the original sequence: in position 8844, A was changed to C to generate an AvaII restriction site; in position 8871, T was changed to A to generate an unique PstI restriction site; in position 8901, A was changed to G to generate an unique NaeI restriction site. All changes are silent and do not modify the derived amino acid sequence of Nef. The nefsequence of the resulting plasmid ptacnef was verified by dideoxynucleotide sequencing. The construct was transformed into W3 110lacIq. In medium containing high concentrations of carbenicillin (100 mglml), the rac promoter is repressed. Expression of Nef could be induced with 50 - 100 pM isopropyl-tho-P-Dgalactoside to produce amounts of protein as large as 16.6% of the soluble E. coli protein (Fig. 2). On SDS/PAGE, Nef has an apparent molecular mass of 26 kDa. This molecular mass is somewhat higher than the 23 367 Da expected from the amino acid composition, which is in agreement with the observations of other authors [14]. However, in HPLC gel-filtration experiments, performed as described in Material and Methods, Nef migrates as a monomer of 23 kDa, which is very close to its expected molecular mass. Purification of Nef When the cells were grown at 37°C most of the Nef protein was insoluble, but by growing the bacterial cells at 30°C most

1118

190

195

200

205

210

215

220

225

230

235 240 X(nm1

Fig. 3. CD spectrum of Nef. CD spectrum of 2.0 mM Nef dissolved in 20 mM phosphate buffer, pH 7.2. The molar mean residue ellipticity OMis depicted as function of the wavelength.

of the Nef protein was found to be soluble. The soluble cell extract was purified in a simple two column procedure using a pseudoaffinity Cibacron blue F3GA column and an AcA54 gel-filtration column (see Materials and Methods). The purity of the Nef protein was estimated to be greater than 95% (Fig. 2, Table 1). The purified protein is recognized in Western blots with polyclonal anti-Nef antibodies from rabbits and 10 out of 12 sera from AIDS patients in different stages of disease (data not shown). By direct amino acid sequencing of the Nterminus of the purified protein, the sequence Gly-Gly-LysTrp was obtained, i.e. the N-terminal methionine is removed during the expression of Nef in E. coli.

nucleotides bound to the native protein would be found in the above experiment, since they would be dissociated during the purification (although we use GDP in the elution buffer during the gel-filtration step). The binding of nucleotides to Nef was therefore investigated by methods usually employed for measuring the interactions between proteins and small molecules; nitrocellulose-filter-binding assay, observation of the interaction with fluorescent analogs and equilibrium dialysis. In the case of Nef, no nucleotide binding could be detected with these methods (Figs 5 and 6 ) . From equilibrium dialysis experiments using 0.1 mM Nef, an upper limit of the binding constant for Mg2+ . GDP can be estimated as 5 x lo3 M-'.

Ultraviolet spectroscopy and circular dichroism

GTPase activity

Due to its high content of aromatic residues (7 Phe, 7 Tyr, 7 Trp. i.e. is more than 10% of all amino acid residues), Nef shows a very strong ultraviolet absorbance; a molar absorption coefficient at 280 nm of 44190 M-' . cm-' can be obtained from our experimental data. Analysis of the CD data gives a helix content of 16% and a fl-sheet content of 31% (Fig. 3). However, it is important to note that the high content of aromatic amino acids makes the analysis of the data for small wavelengths difficult and could have influenced the accuracy of the secondary structure determination.

GNB studies GNB proteins usually have a high affinity for GDP and GTP ranging over 108-10'2 M-' [28]. As a consequence, these proteins, when isolated under non-denaturing conditions, usually contain 1 equivalent bound nucleotide/mol protein. The amount and identity of bound nucleotides can conveniently be monitored by reverse-phase HPLC [29].When 10 p1 0.4 mM Nef is directly applied to a C18 reversed-phase column and eluted with appropriate buffer, no guanine, or any other nucleotide, is eluted (Fig. 4). As a control, p2lraSC was analyzed under identical experimental conditions and showed the binding of 1 equivalent GDP/mol; no bound GTP can be observed under these conditions. because all GTP is hydrolysed during the isolation of the protein. If the binding of nucleotides to the Nef protein was much weaker than the binding of GDP or GTP to p2lraSCc, no

The most crucial reaction of GNB proteins is their ability to hydrolyse GTP, which ensures that they can switch between the active GTP-bound and the inactive GDP-bound forms. Although the reaction is rather slow in the absence of effectors, it can still be conveniently measured. Since Nef apparently does not bind guanine nucleotides with an appreciable binding constant, it might still have a turnover GTPase activity. This activity was measured using the HPLC system described above and compared with the GTPase activity ofp2lraSC.Incubation of GTP with p21ragC results in a slow hydrolysis of the nucleoside triphosphate, but no significant hydrolysis of GTP by Nef could be observed (Fig. 7). Adenine nucleotide binding and ATPase activity After its preparation, native Nef contains no bound adenine nucleotides (Fig. 4), indicating that it does not have a high affinity for these substrates. Weaker binding of nucleotides could possibly be determined by equilibrium dialysis. However, within the limits of error, no ADP binding could be detected with this method, and an upper limit for the association constant was estimated as 5 x 10' M-'. The ATPase activity of Nef was determined using a similar method to that for to its GTPase activity. Our Nef preparation shows a low ATPase activity (Fig. 8), but it is not clear if this is really an intrinsic ATPase activity of Nef or is due to the presence of trace amounts of other proteins with high ATPase activities.

1119 C

A

-I (min)

t (min)

-

-t (min)

Fig.4. HPLC analysis of the nucleotide content of isolated Nef and p21"'C. (A) Nucleotide-standard mixture containing more than 1 mM guanosine (Guo), 4 mM GMP, 39 mM GDP, 24 rnM guanosine 5'-[B,y-imino]triphosphate (GppNHp) and 32 mM GTP; (B) analysis of 10 mM Nef; (C) analysis of 10 mM p21"'C. Elution times (rnin) are indicated. Absorbance was measured at 254 nm; the absorbance profiles of protein solutions are enhanced fourfold relative to those of standards. 16000 1 4000

-2-:

]

12000

-

'5 ._ 10000 I

-

GTP GDP C

0

8 8 C

u

0

5 E

a000 -

U

5

6000

-

4000

-

n

L

2 !._

.-5

P

LL

0

Nef

Lm GDP

Oh Nef

pZlra5c

p21'"C

Fig.5. Binding of GTP and GDP to Nef and p21""C determined by nitroceUubfilter-binding assay. Nucleotide-binding analysis on nitrocellulose filters with 2.5 pM Nef and 2.2 pM p21r"'C, using 10 pM [y-"P]GTP or [8-3H]GDP in the presence of 10 mM EDTA. 350

450

0.2 0.0

1

400

0.4

ii -

5w

550

GTP Oh

GDP 36h

GTP 36h

GDP 74h

GTP 74 h

Fig. 7. GTPase activity of Nef. The GTPase activity was determined by incubating the proteins in the presence of GTP (and GDP in a low initial concentration) for various times at room temperature. The nucleotide content was analysed using a calibrated HPLC system. The samples contained 2 mM nucleotide in buffer D (O), 2 mM nucleotide, 50 pM Nef in buffer D (m) and 2 mM nucleotide. 50 pM p2lraSCin buffer D (N). The mean of three different measurments is depicted.

room temperature are shown in Fig. 9. 1-2% of the protein is phosphorylated after 5 h a n d can be shown to be Nef (or a n o t h e r protein with t h e same mobility as Nef on SDSjPAGE gels) by autoradiography. Incubation for longer times ( > 12h) leads to no significant increase in phosphorylation. Under identical experimental conditions, no significant (< 0.1YO) autophosphorylation with GTP could be observed.

wavelength ( nrn )

Fig. 6. Binding of mant-CDP to Nef. Emission spectra of 5 pM mantGDP with 5 pM p21""C in buffer D, 5 pM mant-GDP in buffer D (2) and 5 pM mant-GDP with 5 pM Nef in buffer D (3). Excitation wavelength. 370 nm; temperature, 25°C.

Autophosphorylation by GTP and ATP Incubating our Nef preparation with ATP results i n phosphorylation of the protein; the kinetics of this reaction at

DISCUSSION I t was reported by Guy et al. [12, 131 that Nef has GNB and GTPase activity and is probably a member of the class o f GNB proteins. The protein used f o r these experiments was expressed in E. coli. Since i n their expression system the protein was mainly f o u n d in the insoluble p a r t o f the cells, it had to be solubilized with t h e potentially denaturing detergent

1120

In

3

m

:

1

.

(min)

-

I

-1

'

(min)

-

ATP + N e f

AXP-standard

(0h)

3

_:1 ;1

-!

-t

ul

(min)

-m

-

c

4

-t

(mln)

ATP t Nef

(39 h)

-

I

-t (rnin)

-

ATPcontrol

(39 h)

Fig. 8. ATPase activity of Nef. The ATPase activity was determined by incubating Nef in the presence of ATP. The nucleotide content was analysed using a calibrated HPLC system. Samples contained 2 mM ATP in buffer D or 2 mM ATP with 50 pM Nef in buffer D. AXP, adenine nucleotide standards. Incubation times are given in parentheses.

25000

r

--

21000

17000

0

.-0

U

,E, 13000 c

3

-f

9ooc

woc

0

80

160

240

t(min)

Fig. 9. Autophosphorylatioo of Nef by ATP. 65 pM Nef protein was dissolved in buffer D containing 500 pM [y-32P]ATP. After incubation. for various times, the solution was filtered through nitrocellulose and washed with rinse buffer containing 50 mM ADP. The filterbound radioactivity was determined by liquid scintillation counting.

SDS. Recently, Nebreda et al. [IS] reported experimental results, obtained with isolated Nef protein, where no GNB or GTPase activity of Nef could be detected. During the isolation of their protein, denaturing conditions were used, and the

missing GNB and GTPase activity could result from denaturation of the protein. We describe the construction of an E. coli expression system which can be used to produce large amounts of native Nef in a soluble form. Using the purification procedure described above, it is possible to isolate the protein without using denaturing agents which may perturb the native protein conformation. The CD spectrum shows that the protein structure has a helix content of approximately 16% and a P-pleated sheet content of 31YO.Using two different methods ([30]; Kabsch and Sander, unpublished results) the predicted protein content is 11-17% j-pleated sheet, 20-35% u helix and 55 - 63% non-regular structure (Sander, C., personal communication). It cannot be expected that these studies (as in general all studies of secondary structure) give accurate results, but the general trend may be correct. It is interesting that the two methods predict a relatively low content of regular secondary structure elements (> 55%) in line with our experimental results. A typical GNB protein such as p2lraSCusually contains 1 equivalent bound GDP/mol protein after purification, but we have shown by HPLC that Nef protein is nucleotide free after purification. Moreover, by equilibrium-dialysis experiments, filter-binding assays and experiments with fluorescent nucleotide analogs, we have shown definitively that Nef has no appreciable affinity for guanine nucleotides. In addition, within the limits of error, Nef has no intrinsic GTPase activity compared to p2lraSC,i. e. a second essential property of typical GNB proteins is missing. Our experimental results demonstrate that Nef (or more precisely Nef from the subclone of pNL4-3) is not a GNB protein. This is in line with investigations by other groups [14- 161. The myristylation of the Nterminus apparently does not affect the GNB activity [16].

1121 Sequence similarities have also been used to support the idea that Nef constitutes a GNB protein [12, 131. The GNB protein superfamily contains three stretches of highly conserved amino acid sequence which are involved in nucleotide binding: (a) Gly-Xaa-Xaa-Xaa-Xaa-Gly-Lys-Ser, (b) Thr(Xaa)20- 25-Asp-Xaa-Xaa-Glyand (c) Asn-Lys-Xaa-Asp. Sequence (a) is absolutely conserved in all proteins of that family and also in a number of adenine-nucleotide-bindingproteins e. g. myosin heavy chain, adenylate kinase and F1 ATPase. In the case of Nef, Lys-Gly-Gly-Leu-Xaa-Gly was proposed as homolog for sequence (a). However, as noted by Nebreda et al. [15],reversing a sequence creates a different three-dimensional structure. The lysine residue in sequence (a) which is here in a different position, is important for catalysis [32]. Sequence (b), whch is important for metal-ion coordination, is not found in Nef. One could speculate that Trp-Arg-Phe-Asp may replace sequence (c), which is involved in the binding of the guanine base. In summary, our experimental results and an analysis of the proposed sequence similarities provide no evidence of Nef being a GNB protein. Since Nef is not a GNB protein, what other function could it have? Sequence similarities suggest a possible role as a protein kinase [ll]. That Nef has a certain affinity for nucleotides is suggested by its binding to blue Sepharose, which is known to bind to the active center of virtually all nucleotide-binding proteins [33]. As mentioned above, no nucleotide is bound to Nef after purification, i. e. Nef has no high-affinity binding site for nucleotides. The binding constant for ATP has been estimated to be lower than 5 x lo3 M-I. However, ATP-utilizing enzymes do not need very h g h affinities for ATP, because the intracellular ATP concentration is rather high (in the millimolar range). Autophosphorylation of Nef by GTP or ATP was reported earlier; however, the phosphorylation yield was very low (