The chemical synthesis of Tat Eli was performed in a single run using Fast Fmoc chemistry. The synthesized protein had 99 residues and a molecular mass of 11081 (data not shown). Amino acid analysis revealed an amino acid content compatible with Tat Eli, and sequencing of the first five residues from the N-terminus gave a sequence identical to Tat Eli (data not shown). A trans-activation assay was performed and showed that the synthetic protein had trans-activational activity (Figure 1A). This assay closely resembles the natural conditions for extracellular Tat as the synthetic protein was added to the culture, and had to cross the cell membranes before binding to the nucleotide target TAR, triggering trans-activation. We compared the trans-activational activity of this synthetic Tat Eli with both a synthetic subtype B Tat (HXB2(86)) and with a recombinant subtype B Tat. We show that our synthetic B Tat had the same trans-activational activity as the recombinant subtype B Tat, but that Tat Eli had 4.5 fold more activity at the same concentration tested (Figure 1B).

Tat Eli gives a CD spectrum with a main negative band close to 200 nm (Figure 2A). This is similar to the CD spectrum of a random coil peptide model 18. However, a random coil-like CD spectrum is also observed in proteins such as protamine that have a stable structure and only β-turns as secondary structures 19. Furthermore, as one is unable to differentiate between static and dynamic structures using CD, one cannot associate a CD band to random coils 18. Therefore, to evaluate if Tat Eli could be a random coil we measured CD spectra over a range of temperatures under denaturing conditions. Under these conditions a random coil protein should display similar CD spectra at all temperatures tested. We also compared the CD spectra of Tat Eli at different temperatures with those of two other proteins: bovine serum albumin (BSA) and protamine (Figure 2). We observed a decrease in the CD signal intensity for all three proteins when the temperature was raised (Figure 2). According to CD theory, secondary structures have a CD specific signal due to a resonance phenomenon resulting from repetitive and similar Phi and Psi dihedral angles 18. The melting of secondary structures should induce a decrease of CD signal as is illustrated in our experiments with the collapse of the α-helix signal of BSA (Figure 2C). The decrease in CD signal is not due to a reduction in solubility or aggregation, as the absorption spectra were similar at each temperature tested (data not shown). It is interesting to note that the CD signal of Tat Eli and protamine are almost similar at 70°C revealing that these two proteins have probably become random coils. The fact that Tat Eli has CD spectra markedly different at lower temperature indicates that Tat Eli is not a random coil at least at 20°C and CD data analysis (data not shown) reveals the presence of secondary structures such as extended structures (22%), β-turns (31%) and almost no α-helix (5%). Other structures with no repetitive dihedral angles represent 42% of the residues. We then tested the effect of Zn²⁺ on Tat structure as previous reports stated that Tat binds Zn²⁺ through its cysteine-rich region and that binding of Zn²⁺ affects Tat CD spectrum and structure 202122. We tested different molar ratios of Tat Eli to Zn²⁺ from 1:1 through 1:16 and the only effect observed was precipitation of Tat at 1:16 at pH 4.5 and 1:8 at pH 7 (Figure 3). When there is no precipitation, the CD spectra remain similar whatever the Tat:Zn²⁺ ratio. Therefore, the binding of Zn²⁺ does not modify the structure of Tat Eli.

The following spin systems were identified from the TOCSY spectrum: 28/28 Asp, Cys, His, Phe, Ser, Tyr and Trp; 14/14 Gln, Glu and Met; 9/9 Gly, 2/2 Ala, 16/16 Pro, 16/18 Arg and Lys, 2/2 Ile, 7/7 Val and Leu, and 3/3 Thr. The homonuclear ¹H NMR spectra of Tat Eli allowed sequential assignment of all 99 spin systems by exploiting chemical shift similarity to previous 2D NMR assignments of the two short active Tat variants Tat Bru and Tat Mal 1415. Interestingly, the chemical shifts were similar to Tat Bru and Tat Mal 1415. No NOE-back calculation procedure was necessary to assign Tat Eli spin systems as was the case for Tat Bru 14. The sequential assignment for these spin systems was obtained using the space connectivities H^α(i)-H^N(i+1), side chain H^N(i+1) as H^βi)-H^N(i+1) and side chain H^α(i). The unique spin systems corresponding to Trp 11, Phe 38, Ile 45, and Ile 69 were used as starting points and allowed the complete sequential assignment. The aromatic spin systems were identified from the ¹H NOESY spectra using connectivities observed between the aromatic and the β and/or α protons. The ¹H chemical shifts of Tat Eli are listed in Additional file 1. It was not possible to identify the H^N of Gln 72, Lys 89, and Lys 90 (Additional file 1). Although the H^α of the prolines have a low dispersion in the NOESY spectra, we were able to identify all of them using sequential H^δ(i)-H^N(i-1) and H^α(i-1) connectivities. From the proton assignment, we identified 1639 NMR distance constraints out of which 264 were interresidual, 179 were sequential (i, i+1), 34 medium [(i-j) < 5], and 51 long range [(i-j) ≥ 5]. More than 15% of the long-range constraints involved the second exon of Tat, and half of those were in relation to the cysteine-rich region showing that the second exon is essential in Tat Eli folding (Figure 4).

The sequence of Tat Eli is very similar to Tat Mal 15. Therefore, we chose to compare NMR constraints of Tat Eli with Tat Bru that has 25% sequence variation with Tat Eli (Figure 5A) 14. The two proteins have a similar folding despite the fact that there are less NMR constraints for Tat Eli due to its high flexibility. Figure 5B shows the contacts between the different regions of Tat. We can deduce that region III is more stable because it is interacting with regions I, II, IV and V. Interestingly, even if the beginning of region III is highly variable among Tat variants, the sequence between residues 42 and 51 is the most conserved. Moreover, NMR data allow us to confirm the presence of two β-turns (⁹EPWN and ⁴⁵YSIG) although CD experiments indicate that Tat Eli has around 30% of β-turn secondary structures. This could be due to the superposition of spin systems on Tat Eli's spectra. Long distance NMR constraints were identified between the extra residues of the C-terminus and residues of regions II, III and V (Figure 5C).

Model structures of Tat Eli were determined with NMR constraints using a simulated annealing protocol 23. Superimposition of conformers with the lowest van der Waals energies, Coulombic energies, and respect of NMR constraints shows a similar folding (Figure 4). The mean structure was then refined by energy minimization without NMR constraints but with a freeze backbone (Additional file 2). Although we found specific NMR constraints for only two β-turns, model structures reveal that Tat Eli has eight β-turns in agreement with CD data. Figure 6C shows the structure of Tat Eli compared to Tat Bru and Mal (Figures 6A and 6B). In region I, we found two β-turns involving residues ⁹EPWN and ¹⁷QPRT as for Tat Mal 15. The cysteine-rich region (region II) is constituted of two loops which are well exposed to solvent. Region III begins with a loop followed by a β-turn starting from Ile 45. This turn was also found in Tat Bru and Tat Mal, corresponding to a well-conserved sequence among Tat variants 1415. The basic region (region IV) adopts an extended structure similar to Tat Bru and Tat Mal. The glutamine-rich region (region V) is composed of two β-turns involving residue ⁶³QAHQ and ⁷⁰PKQP. The C-terminal region of Tat Eli (region VI) is composed of three β-turns involving residues ⁷⁶QPRG as for Tat Mal, ⁸³GPKE as for Tat Bru and ⁹¹VESE, not present in the shorter Tat variants. Furthermore, the core of Tat Eli is mainly composed of region I, with Trp 11 at a central position that is part of a hydrophobic cluster containing Phe 38 and Tyr 47. This is the same core as in both Tat Bru and Tat Mal 1415.

The primary sequence of Tat Eli is MDPVDPNLEPWNHPGSQPRTPCNKCHCKKCCYHCPVCFLNKGLGISYGRKKRRQRRGPPQGGQAHQVPIPKQPSSQPRGDPTGPKEQKKKVESEAETDP. Tat Eli was synthesized in solid phase using Fast Fmoc (9-fluoenylmethoxy carbonyl) chemistry by the method of Barany and Merrifield 29 using 4-hydroxymethyl-phenoxymethyl-copolystyrene-1% divinylbenzene preloaded resin (0.65 mmol) (Perkin Elmer) on an automated synthesizer (ABI 433A, Perkin Elmer) as previously described 12. Purification was carried out using a Beckman high-pressure liquid chromatography (HPLC) apparatus with a Beckman C8 reverse phase column (10 × 150 mm). Buffer A was water supplemented with 0.1% (v/v) trifluoroacetic acid (Sigma) and buffer B was acetonitrile (Merck) supplemented with 0.1% (v/v) trifluoroacetic acid. Gradient was buffer B from 15–35% in 40 minutes with a 2 ml/min flow rate. HPLC analysis was carried out using a Merck Chromolith™ Performance RP-8e (4.6 × 100 mm) with similar buffers but using a gradient from 10–50% in 15 minutes with a 1.8 ml/min flow rate. Purity of the protein was up to 95%. Amino-acid analyses were performed on a model 6300 Beckman analyzer and mass spectrometry was carried out using an Ettan matrix-assisted laser desorption ionization time-of-flight apparatus (Amersham Biosciences). The synthetic Tat HXB2(86) was previously described 6. Recombinant Tat was obtained through the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH from Dr. John Brady and DAIDS, NIAID 30.

The trans-activation activities of the synthetic Tat proteins were analyzed by monitoring the production of β-galactosidase after activation of lacZ expression in HeLa-P4 cells 31 using a previously described protocol 610. Briefly, 2 × 10⁵ cells per well were incubated in 24-well flat-bottomed plates (Falcon) at 37°C, 5% CO₂, in Dulbecco's Modified Eagles Medium (DMEM) supplemented with 10% (v/v) heat-inactivated fetal bovine serum and 100 μg/ml neomycin (all Invitrogen) After 24 h, cells were washed with phosphate-buffered saline. Tat protein was directly mixed with DMEM supplemented with 0.01% (w/v) protamine (Sigma) and 0.1% (w/v) bovine serum albumin (BSA; Sigma) and added to the cells. After 16 hours at 37°C, 5% CO₂, cells were washed with phosphate-buffered saline, lysed and the β-galactosidase content was measured with a commercially available antigen capture enzyme-linked immunosorbent assay (β-galactosidase ELISA, Roche Diagnostics). Results were normalized using the Bradford reagent (Sigma). Control corresponds to the background β-galactosidase expressed by HeLa-P4 cells in DMEM supplemented with 0.01% (w/v) protamine and 0.1% (w/v) BSA with vehicle and without Tat. Concentrations of Tat used are noted in the figure legend.

CD spectra were measured with a 100 μm path length from 260–178 nm at 10–70°C on a JASCO J-810 spectropolarimeter. Data were collected at 0.5 nm intervals using a step auto response procedure (JASCO). CD spectra are presented as Δε per amide. Protein concentration was 1 mg/ml in 20 mM pH 4.5 phosphate buffer for the three proteins: BSA, protamine, and Tat Eli and in 20 mM pH 7 phosphate buffer for Tat Eli with 0 to 16 molar equivalents of ZnCl₂. The CD data were analyzed with VARSELEC to determine the secondary structure content according to the method of Manavalan and Johnson 32 using a set of 32 reference proteins and an average of 4960 calculations.

Tat samples for NMR (1 mM) were prepared in H₂O/D₂O [9:1] 100 mM phosphate buffer at pH 4.5. The homonuclear ¹H NMR spectra were recorded on a Varian Inova 800 MHz NMR spectrometer operating at 799.753 MHz. ¹H TOCSY spectra 3334 with 80 ms mixing, and NOESY spectra 35 with 50, 100, 150, and 200 ms mixing, were recorded at 20°C with a spectral width of 10999.588 Hz. The water signal was suppressed using weak presaturation (2 s). Data were processed with the Felix 2002 from Accelrys (San Diego, CA).

Molecular modeling was performed using the Insight II 2002 package including Biopolymer, Discover, Homology and NMR-refine software (Accelrys, San Diego, CA). High temperature simulated annealing was carried out according to Nilges et al. 23.