HIV-1 Tat Raises an Adjuvant-free Humoral Immune Response Controlled by Its Core Region and Its Ability to Form Cysteine-mediated Oligomers*

Proteins are poor immunogens that require an adjuvant to raise an immune response. Here we show that the human immunodeficiency virus, type 1 Tat protein possesses an autoadjuvant property, and we have identified the determinants and the molecular events that are associated with this unusual property. Using a series of chemically synthesized Tat101 derivatives, we show that the core region controls the autoadjuvant phenomenon independently of the B-cell recognition and T-cell stimulation that are associated with epitopes respectively located on the N-terminal region and the cysteine-rich region. We also show that cysteine-mediated oligomerization is a key molecular event of the adjuvant-free antibody response. In particular, a Tat dimer formed by the oxidation of two cysteine residues, at position 34 only, raises an adjuvant-free antibody response that is comparable with that observed with the wild-type protein. Unlike the parent protein, the Tat dimer has no transactivating activity and remains homogeneous for several weeks in solution. This construct might be of value for the design of an adjuvant-free Tat-based vaccine. Furthermore, we suggest that the specific autoadjuvanticity determinant of Tat could be used to provide other proteins with adjuvant-free immunogenicity.

Most free proteins injected in a soluble form in humans or animals are poor immunogens and can induce B-or T-cell tolerance or rapid proliferation followed by rapid death of Ag-specific T-cells. They usually become immunogenic when they are mixed with an adjuvant. It is not completely understood how adjuvants are able to convert a tolerogenic stimulus into an immunogenic one. The process seems to be related to an increase in immunogen half-life through the so-called "depot effect" (1) and to a series of immunological events that include induction of inflammation and of inflammatory cytokines (2,3), improvement of Ag delivery to Ag-presenting cells (4), increase in Ag processing and presentation through the induction of major histocompatibility complex and/or costimulatory molecules (5), and induction of the production of immunomodulatory cytokines (6). Because most proteins cannot trigger such complex events, it is no surprise that, alone, they do not induce an immune response.
The Tat (transcriptional transactivator) protein of HIV-1 2 is a regu-latory protein that is produced early after infection and that is essential for viral replication (7,8). This molecule is released in the extracellular milieu (9,10), where it exhibits numerous biological activities. In particular, Tat was suggested to induce angiogenesis (11)(12)(13), chemotaxis of monocytes (14), and secretion of proinflammatory cytokines such as IL-1, IL-6, and tumor necrosis factor ␣ (15). Furthermore, Tat was proposed to target monocyte-derived dendritic cells and enhance their maturation, function, and Ag-specific T-cell responses (16). Also, a recent report showed that Tat can reprogram immature dendritic cells to express chemoattractants for activated T-cells and macrophages, but in contrast to the previous paper this work showed that Tat cannot induce their maturation and induction of proinflammatory cytokines (17). Although somewhat contradictory, these various observations suggest that Tat expresses activities that are reminiscent of those triggered by adjuvants, and we may wonder whether they could be associated with the Tat-specific immune response that is observed in HIV-1infected humans (18 -21). Therefore, we investigated whether Tat can induce an immune response in the absence of adjuvant. This hypothesis was supported by a report showing that the repeated intradermal immunization of one monkey with low doses of a biologically active 86-aa-long Tat protein, in the absence of adjuvant, induced a specific Th1 response and cytotoxic T-cells without Ab production (22). To document the possibility that Tat may possess immunoadjuvant properties, we have analyzed the humoral immune response raised against Tat, in mice. In most studies, truncated forms of Tat are used that come from laboratory isolates and encode sequences of 86 aa. However, primary isolates encode Tat sequences that are 15 aa longer. Because this additional C-terminal region could contain immunogenic sites that might influence the Tat immune response, we selected a Tat of 101 aa and synthesized it by chemical means. We show that a synthetic 101-aa Tat molecule is indeed able to induce an autoadjuvant immune response. Then we searched for the determinant that is associated with this uncommon property. We show that introduction of substitutions in the N-terminal part, the core region (aa 39 -48), or the cysteine-rich region (aa [22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37] affect the Ab response. However, only substitutions introduced in the core region abrogated the Ab response without affecting the major B-cell epitope region and/or the T-cell stimulating capacity of Tat. Furthermore, we observed that cysteine-mediated oligomerization represents a molecular event that is required in the process of the adjuvant-free Ab response. In particular, we observed that a dimer formed by oxidation of two cysteine residues at position 34 only raises an autoadjuvant Ab response that is comparable with that of the wildtype protein. Because this derivative has no transactivating activity and remains homogeneous for several weeks in a PBS buffer, we suggest that it could be considered for the design of an adjuvant-free Tat-based vaccine. Furthermore, the specific autoadjuvanticity determinant of Tat identified in this work may represent a novel component to provide other proteins with adjuvant-free immunogenicity.

Synthesis of Tat Proteins and Peptides
Synthesis of Tat and of its derivatives was performed using the Fmoc/ tert-butyl strategy on an Applied Biosystems 433A synthesizer. The chemical procedure used 0.1 mmol of Fmoc-Asp(OtBu)-PAL-PEG-PS resin, a 10-fold excess of each aa, dicyclohexylcarbodiimide/1-hydroxy-7-azabenzotriazole, and diisopropylethylamine/N-methyl pyrrolidone. Incorporation of Gln 54 was performed twice manually. Cleavage and deprotection were achieved using a mixture of trifluoroacetic acid/triisopropylsilane/water (9.5/0.25/0.25, v/v/v). The crude material was precipitated twice with cold tert-butyl methyl ether and subsequently dissolved in 15% aqueous acid. The crude protein was then purified by HPLC on a Jupiter C4 column. The S(tBu) moieties were then removed from the cysteines using degassed 0.1 M phosphate buffer, pH 8.5, containing 6 M urea and dithiothreitol (50 eq/Cys). After completion the mixture was acidified to pH 2.2 and purified by HPLC on a C4 column, and the fully reduced Tat proteins were kept freeze-dried at Ϫ20°C.
Eighteen 15-mer peptides representing the sequence of Tat101 with an overlap of 5 aa residues were synthesized on a 357 Advanced ChemTec multiple peptide synthesizer using the Fmoc strategy. The peptides were synthesized using a rink amide resin and were N-terminally acetylated using acetic anhydride. The peptides were cleaved and deprotected as described above and then purified on a Vydac C18 reverse phase column (Hesperia, CA). The synthesized peptides and proteins were characterized by mass spectrometry and aa analysis. The mass found for each synthesized Tat protein is shown in the supplemental data section.

Transactivation Assay
The HIV-1 LTR sequence chosen for this study comes from the HIV-1 AVR-2 isolate (GenBank TM accession number K02007). The sequence covering positions Ϫ137 to ϩ58 relating to the RNA start site was synthesized by the method described by Stemmer et al. (23). The reporter plasmid pLTR-G was constructed by inserting the synthetic HIV-1 LTR upstream of the promoterless enhanced green fluorescent protein (EGFP) gene within the multiple cloning site of pEGFP-1 (Clontech). HeLa cells were transfected with pLTR-G and subsequently cloned by limiting dilution. The cell line overexpressing EGFP with the highest efficiency in the presence of Tat was then selected for the transactivation assay. In the transactivation assay, the transfected cell line was incubated with either Tat101 or Tat101C(22,37)S or Tat101R(52,53)Q in the presence of 100 M chloroquine. After 3 h at 37°C, 10% fetal bovine serum containing Dulbecco's modified Eagle's medium was added. 45 h later, the cells were harvested, and fluorescence intensity was determined by fluorescence-activated cell sorter analysis.

Immunization of Mice
Groups of four BALB/c mice (IFFA CREDO, France) were injected twice subcutaneously at the tail base with 100 l of a PBS solution containing 5 g of Tat101 or Tat101 derivatives. Blood samples were collected 14 and 28 days after the second injection. Groups of four SWISS mice were injected with 5 g or 50 g of Tat101 according to the same protocol. To assess the T-cell response, BALB/c mice were injected twice with the different Ags following the procedure described above.

T-cell Stimulating Assay
Ten days after immunization of mice, spleens were harvested and suspended in a proliferation medium containing 1% fetal bovine serum. The cells (5 ϫ 10 5 cells/well) were cultured at 37°C with serial dilutions of the different Ags. The presence of IL-2 in culture supernatants was determined after a 24-h period by measuring the proliferation of an IL-2-dependent cytotoxic T-cell line, using methyl-[ 3 H]thymidine ([ 3 H]TdR; 5 Ci/mmol; Amersham Biosciences). Proliferation of the cells was assessed after 3 days and an 18-hour pulse with methyl-[ 3 H]thymidine. The data are expressed in cpm.

Ab Titration by Enzyme Immunoassay
Ab Titer-Enzyme-linked immunosorbent assay plates were coated overnight with either Tat101 or Tat101 derivatives (0.1 g/well) in 0.05 M phosphate buffer, pH 7.4, at 4°C. The plates were then saturated with 0.1 M phosphate buffer, pH 7.4, containing 0.3% bovine serum albumin. Individual antisera were serially diluted in the same buffer containing 0.1% bovine serum albumin and incubated in the wells overnight at 4°C. Binding of Abs was assessed using a goat anti-mouse IgG peroxidase conjugate and 2,2Ј-azinobis(3-ethylbenz-thiazoline-6-sulfonic acid). The titers were defined as the highest serum dilution giving an absorbance value of 0.6 above the negative control. For this control we used pooled sera collected before immunization of mice.
Mapping of the B-cell Epitopes-Enzyme-linked immunosorbent assay plates were coated overnight with the 18 overlapping peptides (1 g/well) as described above. Antisera were serially diluted in 0.1 M phosphate buffer, pH 7.4, containing 0.1% bovine serum albumin and incubated in the wells overnight at 4°C. Binding of Abs was assessed using a goat anti-mouse IgG peroxidase conjugate and 2,2Ј-azinobis(3-ethylbenz-thiazoline-6-sulfonic acid).

Protein Sizing Analysis Using Lab-on-a-Chip Technology
The chip-based separations were performed on an Agilent 2100 bioanalyzer using the Protein 50 Plus LabChip kit (Agilent Technologies, Waldbronn, Germany). All of the chips were prepared according to the protocol provided with the kit. Briefly, the channels of the chip were filled with a mixture of a sieving matrix and a fluorescent dye for detection. Tat101 and its derivatives were incubated for different times in the absence or presence of an excess of reducing agent, cysteine oxidation was stopped by the addition of an excess of maleimide, and the mixtures were subsequently boiled in the presence of SDS. The samples were then heat-denatured at 95-100°C for 7 min before loading them onto the chip. The chip was then placed in the bioanalyzer, and each sample was sequentially separated and detected by laser-induced fluorescence detection (670 -700 nm) within 45 s. After completion of the chip run, the data were displayed as an electropherogram and as a gel-like image.

Statistical Analysis
Student's t tests were performed using Lumière software.

RESULTS
Selection of the Tat Sequence-We determined a theoretical consensus sequence containing the amino acids most represented in 66 sequences of Tat variants found in different HIV-1 isolates and reported in the Swiss Protein Database and the TrEMBL Database between 1999 and 2000 (24). The sequence that was closest to the theoretical consensus sequence was found in the infectious clone 320.2A 2.1 (25). We synthesized it chemically and replaced its N-terminal methionine by a nonoxidizable isosteric norleucine (see Fig. 4), because it is known that spontaneous oxidation of methionine can occur during purification.
Intrinsic Immunogenicity of Tat101-We investigated whether Tat and other unrelated protein Ags, such as hen egg lysozyme (HEL), ovalbumin, glutathione S-transferase, and viral protein R from HIV-1, could trigger an adjuvant-free immune response in BALB/c mice. We injected subcutaneously 5 g of each Ag in PBS, and 14 or 28 days after the second immunization we observed that Tat101 had raised a high Ab response, whereas the four other Ags triggered no response (Fig. 1A). Therefore, the other proteins do not possess adjuvant-free immunogenic properties. Then we examined the anti-Tat Ab response in outbred mice, by injecting four SWISS mice with either 5 or 50 g of Tat101 in PBS. A clear humoral immune response was elicited upon administration of the highest dose of Tat protein (Fig. 1B), indicating that the ability of Tat to raise a humoral immune response in the absence of adjuvant is not related to the genetic background of the mice and is clearly dose-dependent.
Because Tat86 was previously described as being poorly immunogenic in the presence of adjuvant (26), we compared the immunogenicity of Tat101 in the presence or absence of Alum. In these experiments, a single intraperitoneal injection of Tat101 previously mixed with Alum raised an anti-Tat antibody response (8302 Ϯ 4212) close to that with two adjuvant-free injections of Tat101 (4895 Ϯ 2733), indicating that Tat101 is substantially immunogenic in the presence of adjuvant. The reasons for the differences between our observations and those of previous reports are not clear but might be related to the differences between the Tat molecules, both in terms of lengths and amino acid sequences.
To examine whether Tat also elicits a T-cell response in mice in the absence of adjuvant, we immunized BALB/c mice with two subcutaneous injections of Tat101 solubilized in PBS, collected spleens 10 days after the second injection, and measured the ability of splenocytes to secrete IL-2 in vitro in the presence of Tat101. As shown in Fig. 2A, cells from Tat-immunized mice produced IL-2 when incubated in the presence of Tat101 but not in the presence of an irrelevant Ag (HEL). In contrast, no IL-2 secretion was observed with spleen cells from naive mice (not shown). Surprisingly, however, although capable of secreting IL-2, the splenocytes from Tat-immunized mice did not proliferate in the presence of Tat (Fig. 2B). The reasons for the discrepancy between IL-2 production and proliferation remain unclear but could be related to Tat toxicity and/or to its ability to induce apoptosis in cells (27). Nevertheless, the results indicated that, in the absence of adjuvant, Tat can raise specific T-cells that are functional for IL-2 secretion.
Mapping the Tat B-cell Epitopes in the Absence of Adjuvant-To characterize the B-cell epitope(s) of Tat101, we synthesized eighteen 15-mer overlapping peptides encompassing the Tat polypeptide chain. The peptides were subsequently coated on enzyme-linked immunosorbent assay plates, and we assessed their binding to a nonimmune serum and to a pooled serum from BALB/c mice immunized with Tat101 in PBS. As shown in Fig. 2C, the nonimmune serum revealed no Ab binding to the coated plates. In contrast, Abs present in sera from Tat101immunized mice bound mainly peptides 1 and 2 and to a lesser extent peptide 18 (Fig. 2D), suggesting that region 1-20 of Tat101 is the B-cell immunodominant region in BALB/c mice when Tat is injected in the absence of adjuvant.
Identification of the Molecular Regions That Control Tat Autoadjuvant Property-Because we had no obvious starting guidelines to identify the molecular determinants involved in the autoadjuvant property of Tat101, we modified the Tat regions that govern its biological properties, and we examined the effect of these modifications on the capacity of Tat101 to raise Ab in the absence of adjuvant.
Because the main activity of intracellular Tat is to transactivate the viral genome, we first investigated whether or not the determinant associated with transactivating activity overlaps the determinant that is responsible for the autoadjuvant effect. Because positions 22, 37, 52, and 53 were previously described to be important for the transactivating activity of Tat86 (28,29), we synthesized two Tat101 derivatives in which cysteines 22 and 37 were replaced by serine residues (Tat101C(22,37)S) in the first derivative and arginine residues 52 and 53 were substituted by glutamine residues (Tat101R(52,53)Q) in the other. The transactivating activity of these derivatives was examined using an assay where HeLa cells were first transfected with a plasmid coding for the LTR sequence of HIV-1 and for the sequence of EGFP. Transfected cells were subsequently cloned by limiting dilution and transiently transfected with a plasmid coding for wild-type Tat101. The HeLa cell line that overexpressed EGFP with the highest efficiency was selected for the assessment of the transactivating activity in the presence of 100 M chloroquine. As shown in Fig. 3A, wild-type Tat101 transactivates efficiently as it increased EGFP expression by the HeLa cell line. In contrast, neither Tat101C(22,37)S nor Tat101R(52,53)Q caused such an increase in EGFP expression, indicating that these two derivatives are devoid of transactivating property, in agreement with previous reports (28,29).
Then we immunized groups of four BALB/c mice, in the absence of adjuvant, with either 5 g of Tat101 or its derivatives in PBS and measured the anti-Tat Ab response 14 days after the second injection. As shown in Fig. 3B, Tat101C(22,37)S and Tat101R(52,53)Q were as immunogenic as wild-type Tat101, suggesting that the determinant associated with the autoadjuvant property is different from the one that is involved in transactivating activity.
Tat has multiple extracellular activities, and at least five determinants have been associated with these properties. To understand whether or not one of these regions controls the autoadjuvanticity of Tat, we prepared a series of five Tat derivatives by introducing a series of substitutions in these five regions (Fig. 4A). Because Asp 5 and Pro 6 contribute to the mechanism of immunosuppression of CD26-dependent T-cell growth (30,31), we respectively replaced Asp 5 and Pro 6 with Ile and Leu in the first Tat derivative (Tat101D5I,P6L). Then, because the cysteinerich region (aa 22-37) is known to contribute to numerous activities, including inhibition of lymphocyte proliferation (32), inhibition of apoptosis (33), monocyte chemotaxis (14), and angiogenesis (34), we FIGURE 2. T-cell response of BALB/c mice immunized with Tat101 and identification of its B-cell epitopes. A, the animals were injected twice at 14-day intervals. 10 days after the second injection spleens were collected, and splenocytes were challenged with serial dilutions of Tat or HEL. After a culture period of 24 h, each supernatant was assayed for its capacity to stimulate incorporation of [ 3 H]thymidine in an IL2-dependent cytotoxic T-cell line. B, proliferation of the cells was assessed after 3 days and an 18-h pulse in the presence of [ 3 H]thymidine. C and D, the B-cell epitopes of Tat101 were mapped using either a control serum (C) or pooled sera raised against Tat101 in PBS (D). Using an enzyme immunoassay, the sera were tested for their ability to bind overlapping 15-mer peptides representing the sequence of Tat101.
replaced all the cysteines, highly conserved in HIV-1 isolates, with serine residues in the second derivative (Tat101C(22-37)S). The core region (aa 38 -48) contributes to monocyte chemotaxis (14) and angiogenesis (34), but the residues involved in these activities are not yet identified. Therefore, we synthesized a Tat derivative (Tat101core) in which the hydrophobic and charged residues of the core were substituted by the polar uncharged residue, glutamine, whereas the polar serines and threonine were replaced by alanine, as shown in Fig. 4A. The basic region is also involved in numerous activities, including the inflammatory response (35), neurotoxic activity (36), translocation (37), and cellular internalization through heparan sulfate proteoglycan binding (34). Therefore, we prepared a fourth derivative (Tat101polyQ), in which the five lysines and the three arginines of the basic region (aa 49 -57) were replaced by 8 glutamines. Finally, because the RGD motif contributes to Tat binding to the integrins ␣5␤1 and ␣v␤3 (38), it was replaced by a KGE nonbinding sequence (39) in the derivative (Tat101RGD,KGE).
We immunized groups of four BALB/c mice, in the absence of adjuvant, with 5 g of either Tat101 or its derivatives and monitored the anti-Tat Ab response after the second injection. As shown in Fig. 4B, Tat101polyQ and Tat101RGD,KGE were as immunogenic as the wildtype Tat101, indicating that the basic region and the integrin-binding sequence are unlikely to be involved in the autoadjuvant property. A significant decrease in the Ab response was observed with Tat101D5I,P6L 14 days after the second immunization (p Ͻ 0.05) but not at 28 days. Even more pronounced was the lack of response observed with Tat101core and Tat101C(22-37)S (Fig. 4, B and C). Altogether, these data suggest that the autoadjuvant phenomenon could be controlled by the N-terminal region, the core region, and/or the cysteine-rich region of the molecule.

Do the B-cell Epitopes of Tat and/or Its T-cell Stimulating Capacity Play a Role in the Autoadjuvant Effect
?-To raise a T-cell-dependent humoral immune response, a protein must both be capable of stimulating T-helper cells and possess one or more B-cell epitopes (40). There-fore, we investigated whether the diminution of Ab response observed 14 days after immunization for Tat101D5I,P6L and the lack of immunogenicity found for Tat101C (22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)S and Tat101core could be related to an alteration of one or both of these two required properties.
We first examined whether or not the B-cell antigenic sites were preserved in these derivatives. Using plates coated to similar levels with each of these three derivatives or with wild-type Tat101 or Tat101(polyQ) used as controls (not shown), we examined the binding of each Ag to polyclonal Abs present in an antiserum raised against wild-type Tat101 in mice. As shown in Fig. 5A, the binding capacity of Tat101D5I,P6L was much lower than that of the others and in particular of the control molecules Tat101 and Tat101(polyQ), which had comparable autoadjuvant properties. This observation indicates that one or more of the antigenic binding sites of the Tat101D5I,P6L have been altered by the D5I and P6L substitutions. This was not surprising because we have seen above that the N-terminal region of Tat constitutes its B-cell immunodominant region. Therefore, the low Ab response observed with this derivative may result from an alteration of its B-cell epitopes. In contrast, the Ab binding capacity of the two derivatives Tat101C(22-37)S and Tat101core was virtually identical to that of Tat101 and Tat101(polyQ). Therefore, the antigenic binding sites of these two nonautoadjuvant derivatives are similar to those of the wildtype Tat molecule, indicating that their lack of autoadjuvant property is not due to an alteration of their B-cell epitopes.
We then investigated whether or not the two nonautoadjuvant derivatives Tat101core and Tat101C(22-37)S were still capable of inducing a T-cell response in mice. We injected BALB/c mice with either Tat101core or Tat101C(22-37)S in the absence of adjuvant, and 10 days after the second injection we collected spleens and investigated the ability of splenocytes to secrete IL-2. As shown in Fig. 5B, no IL-2 secretion was observed with splenocytes from mice immunized with Tat101C(22-37)S, indicating that the T-cell stimulating capacity is lost when the 7 cysteines of Tat are replaced by 7 serines. Therefore, the lack of Ab response observed with Tat101C(22-37)S may result from its inability to stimulate T-cells.
More striking was the result obtained with the Tatcore derivative. Thus, splenocytes from mice immunized with Tat101core comparatively produced IL-2 when they were incubated in the presence of Tat101core or Tat101 (Fig. 5C). This result demonstrates that the residues that have been substituted in the core region are not implicated in the T-cell stimulating capacity. Therefore, altogether, our data show that only substitutions of residues from the core region (i) do not modify the T-cell stimulating capacity, (ii) do not affect the B-cell epitopes, and (iii) alter the autoadjuvant response. We therefore conclude that one or more residues of the core region control autoadjuvanticity of Tat in a manner that is dissociated from its otherwise necessary B-cell recognition capacity and T-cell stimulating capacity. This conclusion does not rule out the possibility that the determinant associated with the autoadjuvant property could overlap with some B epitopes and T-cell-stimulating determinants.
Is the Autoadjuvant Property of Tat101 Associated with Its Capacity to Oligomerize?-It has been reported that several stable oligomerization forms of Tat86 occur spontaneously upon dissolution in an aqueous buffer (41). Therefore, we wondered whether this natural trend of Tat101 to oligomerize influences its autoadjuvant property.
First, we examined whether Tat101 has a propensity to form oligomers, like Tat86. Mass spectrometry analyses showed that after chemical synthesis and purification using organic solvents, Tat101 and its derivatives adopt a fully reduced monomeric form (not shown). To investigate its form in aqueous solution, we incubated the fully reduced monomeric form of Tat101 in the PBS buffer used for the immunization experiments, treated samples taken at different times with N-ethylmaleimide to block the remaining free cysteines, and analyzed each treated sample using a chip-based protein electrophoretic assay. This assay proved to be comparable in sensitivity, sizing accuracy, and reproducibility to SDS-PAGE plus standard Coomassie staining and superior in resolution and absolute quantitation accuracy (42). As shown in Fig. 6, when it was treated after a few seconds of incubation Tat101 predominantly migrated (lane 2) with an apparent molecular mass of 20.6 kDa, a value that substantially differs from the 11507 Da calculated and experimentally observed by mass spectrometry for wild-type Tat101. A similar discrepancy was previously described for Tat86 (43), but no clear explanation for this phenomenon has been proposed as yet. After 1 h of incubation (lane 3), Tat101 revealed a relatively similar homogeneous migration profile, whereas mass spectrometry analyses showed the presence of three components with closely related molecular masses of 11507, 11505, and 11503 Da. They correspond to the masses of mono- mers having respectively zero, one, or two intramolecular disulfide bridges. After 24 h of incubation, Tat101 migrated with a more complex profile composed of three major bands (lane 4), with apparent molecular masses of 20.1, 31.5, and 46.3 kDa. Unfortunately, the mass analyses did not provide us with clear data concerning the masses of the components present in the solution, so that it was not possible to conclude that the three forms correspond mostly to monomers, dimers, or trimers, whose calculated molecular masses were 11507, 23012, and 34517 Da. However, lane 6 of Fig. 6 shows that the two bands with higher molecular masses vanish when an excess of reducing agent is added to the sample, supporting the view that these two bands correspond to disulfide-mediated Tat oligomers. In further confirmation that oligomerization of Tat101 occurs through formation of disulfide bonds, we found that the mutant in which all seven cysteines were replaced by serine  FEBRUARY 10, 2006 • VOLUME 281 • NUMBER 6 remained strictly in the monomeric form even after 6 days of incubation (Fig. 7A, lane 6). Finally, after 5 days of incubation, mainly the two bands of higher molecular mass persisted, especially the one with an apparent mass of 31.5 kDa (lane 5). Therefore, after dissolution in a PBS buffer, Tat101 has a propensity to make intramolecular disulfide bridges and to evolve toward disulfide-mediated oligomers. Such a propensity might reflect a physiological property of Tat because it was previously shown to multimerize in cells (44,45).

Autoadjuvant Property of Tat
Then we wondered whether all seven cysteines played equivalent roles in the capacity of Tat101 to form oligomers. To approach this question we synthesized four derivatives, Tat101C(25,27)S, Tat101C(30,31)S, Tat101C(22,37)S, and Tat101C (22,34,37)S, in which we replaced two or three cysteines by serine residues, and we examined their oligomeric trends after 6 days of incubation. As shown in Fig. 7A, two major bands, characterized by apparent molecular masses of 20.7 and 31.5 kDa were associated with the migration of the first three derivatives (lanes [3][4][5]. Most likely, the band with the lower molecular mass corresponds to a monomeric form, because it is also observed with Tat101C(22-37)S, in which all cysteines are replaced by serine residues (lane 6). The 31.5-kDa band is dominant, with an intensity that is respectively 2.8-, 3.5-, and 1.98-fold greater than that of the monomeric forms. In solution, the three samples were characterized by a major mass signal that indicated the presence of dominating components of molecular masses of ϳ23 kDa. Therefore, we suspected that the dominant band observed in lanes 3-5 corresponds to dimeric forms of Tat101. In contrast, the fourth derivative Tat101C (22,34,37)S showed a lower propensity to form dimers because the corresponding band was 0.94-fold less intense than the band corresponding to the monomeric one. Because Tat101C (22,34,37)S is the only one of the four derivatives to have Cys 34 replaced by a serine, we conclude that its lower ability to form a dimer is due to this particular substitution, suggesting that Cys 34 plays a critical role in the dimerization of Tat101.
To examine whether or not the capacity of the protein to form a dimer could affect the immunogenicity of Tat101, we investigated the level of Ab response raised against the four derivatives. We found (Fig. 7B) that the Ab responses raised against Tat101C (22,37)S, Tat101C (25,27)S, and Tat101C(30,31)S were not significantly different from that elicited by the wild-type protein, whereas it was at least 5-fold lower (p Ͻ 0.05) with Tat101C (22,34,37)S. Therefore, we conclude that the autoadjuvant property of Tat101 is connected to its capacity to dimerize, which might be controlled by cysteine 34.
A Stable Single Disulfide-mediated Tat Dimer Has Full Autoadjuvanticity-In view of the above observations, we investigated the Ab response of a homogeneous Tat101 dimer in which the two monomers are linked by a disulfide bond formed between two cysteines at position 34. For this purpose, we first synthesized a derivative of Tat101, called Tat101C (22-31;37)S,C34, in which cysteine 34 was maintained while cysteines 22, 25, 27, 30, 31, and 37 were replaced by serine residues. Then we prepared a homogeneous dimer by incubating an aliquot of Tat101C

DISCUSSION
In contrast to previous studies reporting that Tat is an immunosuppressive protein and a poor immunogen (26,32,46,47), we show that Tat101 can elicit a specific Ab response in BALB/c mice, in the absence of adjuvant. This autoadjuvant property is not shared by other proteins such as viral protein R, HEL, ovalbumin, or glutathione S-transferase and is not related to contaminating endotoxins because no such toxin was detected in the preparations of synthetic proteins using the Lymulus amebocyte assay (not shown). Furthermore, an anti-Tat Ab response could also be raised in SWISS outbred mice, indicating that the autoadjuvant property is not restricted to a particular genetic background. Moreover, the adjuvant-free injection of Tat101 elicits specific T-helper cells that are functional for IL-2 secretion in vitro. Altogether these observations indicate that Tat101 has the uncommon property of eliciting a humoral response in mice in the absence of adjuvant.
We have identified the molecular determinants that are associated with the autoadjuvant property of Tat101 using a number of Tat101 derivatives prepared by peptide synthesis using Fmoc methodology. We substituted one at a time each of the five regions that have been previously defined as being functionally important in the molecule. For two of these regions, the basic region and the C-terminal region (which includes the RGD motif), the introduced substitutions caused no effects, suggesting that they are unlikely to be involved in the determinants that are responsible for Tat autoadjuvant property. In contrast, replacements introduced in the three other regions, i.e. the N-terminal region, the cysteine region, and the core region, suppressed the capacity of Tat101 to stimulate an Ab response in the absence of adjuvant. However, the substitutions made in the first two regions altered respectively the major B-cell epitope region and the T-cell stimulating capacity of Tat101, which suffices to account for the inability of the derivatives to trigger an Ab response. However, we cannot rule out that one or both of these regions contribute in addition to the autoadjuvant effect of Tat. More striking was the finding that substitutions introduced in the core region suppressed the Ab response without affecting the major B-cell epitope region and/or the T-cell stimulating capacity of Tat. Therefore, the core region includes elements that control autoadjuvanticity independently of B-cell recognition and T-cell stimulating capacities of Tat101.
A variety of evidence suggests that the autoadjuvant phenomenon may also depend upon a tendency of Tat101 to oligomerize, a propensity that was previously suggested to be physiologically relevant in cells (44,45). The first evidence supporting this hypothesis came from experiments in which we replaced the cysteine residues of Tat101 two by two or three by three. Thus, we observed a clear correlation between their capacity to form dimers and to express autoadjuvanticity. In these deriv-atives, serine replacement of Cys 22 and Cys 25 , Cys 30 and Cys 31 , and Cys 22 and Cys 37 preserved the capacity of Tat101 to predominantly dimerize and to be autoadjuvant, whereas replacement of Cys 34 in addition to Cys 22 and Cys 37 caused a clear decrease in the tendency to form dimers, leading to more monomeric forms together with a decrease in autoadjuvanticity. This observation suggested that Cys 34 may play a key role in dimer formation and hence in the capacity of Tat101 to raise an adjuvant-free humoral immune response. The second piece of evidence supporting the role of oligomerization in autoadjuvanticity was provided by the adjuvant-free immunogenic properties of a homogeneous Tat dimer. This construct, prepared through a disulfide bond between the Cys 34 of two monomers in which all the other cysteines are replaced by serine, was fully able to trigger an Ab response. Therefore, oligomer-  Groups of BALB/c mice were injected twice at 14-day intervals. Blood was collected on days 14 and 28 after the second injection, and the presence of anti-Tat Abs was assessed by an enzyme immunoassay.
ization of Tat101, and in particular dimerization, represents a major molecular event in the process leading to the Ab response in the absence of adjuvant.
From the above observations, we conclude that the determinant by which the Tat protein exerts its autoadjuvant property includes the core region and the capacity of the protein to form a dimer through the 34 -34 cystine. It remains to investigate whether this determinant can be grafted onto other regular proteins to provide them with adjuvantfree immunogenicity. If this is possible, we anticipate that the determinant may open new perspectives to challenge the use of conventional adjuvants.
Oligomerization is known to be an essential parameter for in vivo activity of growth factors and chemokines (48 -51). Although Tat also possesses extracellular chemokine-like (52)(53)(54) and/or growth factor (55) functions, we do not believe that the autoadjuvant effect is related to such activities. Thus, we observed that Tat101 remains able to trigger an Ab response when its CXC and CC chemokine-like motifs (52)(53)(54) are modified. Also, we found that the autoadjuvant effect is preserved when the RGD motif of Tat, involved in activation of the angiogenic program (56), is replaced by a KGE motif. As a consequence, the other activities associated with RGD, including adhesion to various cells (38,39) and inhibition of the engulfment of apoptotic bodies by dendritic cells (57), are also unlikely to be directly involved in the autoadjuvant effect. Also, the multiple basic region-related extracellular activities (12,29,35,36,58) are unlikely to be involved in the phenomenon because suppression of all the positive charges did not affect the autoadjuvant property. Finally, it is worth mentioning that the transactivating activity is not involved in the effect because Tat101polyQ (not shown), Tat101C(22,37)S, and Tat101R(52,53)Q have lost this activity but remain able to trigger an Ab response. It remains to be seen whether one of the other as yet unexplored numerous activities of Tat101 is associated with its autoadjuvant property.
Tat is considered to be an attractive candidate vaccine in the field of HIV infection. In this context, the finding of the humoral autoadjuvant property of Tat could be of interest. However, the wild-type protein has a number of drawbacks. Thus, we and others (41) have shown that it is unstable and forms a heterogeneous mixture after dissolution in buffer. Furthermore, it possesses numerous activities that could trigger adverse side effects after injection in humans. Tat101C (22-31;37)S,C34 dimer has the advantages of (i) raising an adjuvant-free Ab response similar to that provided by the wild-type molecule, (ii) being devoid of any transactivating activity, as with other vaccine candidates (59,60), and (iii) remaining homogeneous for several weeks in PBS buffer. We propose to explore the efficacy of this newly characterized disulfide-mediated dimer of Tat101 as a vaccine.