Structural insights into the mechanism of HIV-1 Tat secretion from the plasma membrane

Abstract

Human immunodeficiency virus type 1 (HIV-1) trans-activator of transcription (Tat) is a small, intrinsically disordered basic protein that plays diverse roles in the HIV-1 replication cycle, including promotion of efficient viral RNA transcription. Tat is released by infected cells and subsequently absorbed by healthy cells, thereby contributing to HIV-1 pathogenesis including HIV-associated neurocognitive disorder. It has been shown that, in HIV-1-infected primary CD4 T-cells, Tat accumulates at the plasma membrane (PM) for secretion, a mechanism mediated by phosphatidylinositol 4,5-bisphosphate (PI(4,5)P₂). However, the structural basis for Tat interaction with the PM and thereby secretion is lacking. Herein, we employed NMR and biophysical methods to characterize Tat₈₆ (86 amino acids) interactions with PI(4,5)P₂ and lipid nanodiscs (NDs). Our data revealed that Arg49, Lys50 and Lys51 (RKK motif) constitute the PI(4,5)P₂ binding site, that Tat₈₆ interaction with lipid NDs is dependent on PI(4,5)P₂ and phosphatidylserine (PS), and that the arginine-rich motif (RRQRRR) preferentially interacts with PS. Furthermore, we show that Trp11, previously implicated in Tat secretion, penetrates deeply in the membrane; substitution of Trp11 severely reduced Tat₈₆ interaction with membranes. Deletion of the entire highly basic region and Trp11 completely abolished Tat₈₆ binding to membrane. Our data support a mechanism by which HIV-1 Tat secretion from the PM is mediated by a tripartite signal consisting of binding of the RKK motif to PI(4,5)P₂, arginine-rich motif to PS, and penetration of Trp11 in the membrane. Altogether, these findings provide new insights into the molecular requirements for Tat binding to membranes during secretion.

Graphical Abstract

Introduction

Successful gene transcription and production of the human immunodeficiency virus type 1 (HIV-1) RNA transcripts are critically dependent on the trans-activator of transcription (Tat) protein [1–3]. Tat interacts with the trans-activating response (TAR) element on nascent RNA and recruits cyclin T1, a component of the transcription elongation factor P-TEFb [2–4]. Tat is also implicated in interactions with numerous cellular factors during virus replication [2, 5–7]. Additionally, it is established that the Tat protein plays an important role in the development and progression of HIV-1-associated neurocognitive disorder (HAND) [8–14]. The presence of HIV-1 reservoir in the brain is considered a contributor to an inflammatory milieu and to the neuronal injury in infected individuals [15]. HAND encompasses a group of neuropathological conditions that likely result from the combination of the continued exposure of the central nervous system (CNS) tissue to HIV-1, chronic inflammation, and side-effects of antiretroviral therapy (ART) [11, 16]. In the context of HAND, even if ART were to completely suppress HIV-1 replication, persistently infected microglia/macrophages can actively secrete Tat, which is known to be particularly toxic for neurons [9, 17]. Tat secretion, therefore, represents an attractive target to attenuate HAND. Following secretion, Tat can be absorbed by healthy cells and act as a toxin [18–22]. Central to the deleterious effects of Tat in the CNS is its ability to cross membranes during secretion and uptake, mechanisms that remain to be elucidated.

The length of HIV-1 Tat varies from one subtype to another [23]. Three forms (Tat₇₂, Tat₈₆, and Tat₁₀₁) with similar transcription activity have been identified and widely used in previous studies [23–26]. All three forms contain an acidic and proline-rich N-terminus, a cysteine-rich region, a core, a highly basic region (HBR), and a glutamine-rich segment (Fig. 1). The region spanning amino acids 31–61 has been linked to toxicity-associated functions [3, 23, 27]. However, only Tat₈₆ and Tat₁₀₁ contain the RGD motif (Fig. 1), which interacts with vascular endothelial growth factor and integrin receptors [28]. The RGD motif may help with binding integrin receptors on immune cells and improve Tat transmigration into the CNS. Therefore, Tat₈₆ and Tat₁₀₁ can be concentrated in the CNS, resulting in toxicity of cells and increased likelihood of the development of neurocognitive impairment.

Figure 1. HIV-1 Tat₈₆ sequence and mutants.

(A) Tat₈₆ protein sequence. (B) Mutant constructs used in this study.

Early studies suggested that Tat secretion proceeds via an unconventional pathway, meaning it does not traffic through the endoplasmic reticulum or Golgi apparatus [29, 30]. More recent studies suggested that Tat accumulates at the plasma membrane (PM) of CD4+ T-cells prior to secretion, and that this process is dependent on Tat interaction with phosphatidylinositiol 4,5-bisphosphate (PI(4,5)P₂) [29, 31–35], a signaling lipid on the inner leaflet of the PM that fulfills many cellular functions by acting as a substrate for numerous proteins [36, 37]. It was shown that Tat interaction with PI(4,5)P₂–enriched membranes is mediated by residues in the HBR and Trp11, and that Tat–PI(4,5)P₂ interaction is strictly required for Tat secretion [33]. When present in the extracellular environment, Tat is capable of entering uninfected cells and translocating to the nucleus [21, 38, 39]. During cellular uptake, it was suggested that Tat is endocytosed with the low endosomal pH, triggering a conformational change that facilitates translocation to the cytosol [21, 38].

The precise molecular determinants and the mechanism by which Tat interacts with PI(4,5)P₂ and membranes are not well understood. Previous structural studies of Tat indicated that the protein is disordered in the free state and largely maintains the disordered conformation in complex with partners (e.g., TAR) [4, 40–43]. The intrinsic disordered property and structural plasticity of Tat appear to be critical for its biological function during the virus replication cycle. Elucidating the mechanism of Tat–PI(4,5)P₂ and Tat–membrane interactions at the structural level will lead to a deeper understanding of the mechanisms of Tat secretion, which will better define the functional role of Tat in HAND.

Herein, we employed NMR and biophysical methods to characterize Tat₈₆ interactions with PI(4,5)P₂ and lipid nanodiscs (NDs) as membrane mimetics. We show that the Arg49, Lys50 and Lys51 (RKK motif) constitutes the PI(4,5)P₂ binding site, that Tat₈₆ interaction with lipid NDs is dependent on PI(4,5)P₂ and phosphatidylserine (PS), and that the arginine-rich motif (RRQRRR) preferentially interacts with PS. Furthermore, we show that Trp11, previously implicated in Tat secretion, penetrates deeply in the membrane; substitution of Trp11 severely reduced Tat₈₆ interaction with membranes. Tat–membrane binding was abolished only upon deletion of the entire HBR and Trp11. Our data support a mechanism by which Tat secretion from the PM is mediated by a tripartite signal consisting of binding of the RKK motif to PI(4,5)P₂, Arg-rich motif to PS, and penetration of Trp11 in the membrane bilayer. Altogether, these findings provide novel insights into the molecular requirements for Tat binding to membranes during secretion.

Results

Structural features of the Tat₈₆ protein

Limited structural or biophysical studies have been conducted with the Tat₈₆ protein [43]. Structural studies of Tat and its complexes with HIV-1 TAR or other partners have often utilized either the Tat₇₂ construct or short Tat-derived peptide fragments [4, 40, 41, 44, 45]. Herein, we devised an approach that enabled production of highly pure, homogeneous, and monomeric Tat₈₆ protein as verified by mass spectrometry and size exclusion chromatography coupled with multi-angle light scattering (SEC-MALS) (Fig. S1). The 2D ¹H-¹⁵N HSQC spectrum of Tat₈₆ shows one set of signals corresponding to the number of residues in the protein, indicating that the protein is present in one homogeneous state (Fig. 2). Next, we generated NMR signal assignments for backbone resonances using a standard set of triple-resonance, and ¹⁵N-edited TOCSY and NOESY experiments. The chemical shift assignments for Tat₈₆ were used to predict its order parameters and secondary structure content in TALOS+ (Fig. S2). Consistent with previous studies [43], NMR data analysis indicated that the Tat₈₆ protein is in disordered conformation. Of note, the amide ¹H chemical shifts for Trp11 and Lys12 are atypical compared to those calculated from the Biological Magnetic Resonance Bank (BMRB) database, suggesting a unique conformation for these residues. To further confirm the unstructured nature of the protein, we obtained a far-UV circular dichroism (CD) spectrum for Tat₈₆ (Fig. S3). The CD spectrum displays a negative band at ~200 nm distinctive of a random coil. In summary, our data show that the Tat₈₆ protein is monomeric, homogeneous, and is in a disordered conformation.

Figure 2. NMR spectrum of Tat₈₆.

2D ¹H-¹⁵N HSQC NMR spectrum of the Tat₈₆ protein at 35 °C in 50 mM sodium acetate-d₃ (pH 4.2) and 10 mM TCEP. Peak labels indicate residue numbers. Inset, signal corresponding to the indole NH group of Trp11.

PI(4,5)P₂ binding to Tat₈₆

Native PI(4,5)P₂ has a low critical micelle concentration threshold (~30 μM) in aqueous solution [46, 47], which precludes its use in NMR studies as it leads to severe signal broadening in the NMR spectra. Therefore, inositol 1,4,5-trisphosphate (IP₃), the polar head of PI(4,5)P₂, has often been used in structural and biophysical studies of PI(4,5)P₂–interacting proteins including retroviral matrix (MA) proteins [48–52]. To assess if IP₃ interacts directly with Tat₈₆ in the absence of membrane and/or other lipids, we obtained 2D ¹H–¹⁵N HSQC NMR data on the protein as titrated with IP₃. As shown in Figure 3, a subset of signals corresponding to residues in the HBR exhibited significant chemical shift perturbations (CSPs) upon titration with IP₃, consistent with the role of the HBR in PI(4,5)P₂-mediated Tat secretion [33]. Interestingly, a second group of signals corresponding to residues in the N-terminal (group 2) also exhibited IP₃–dependent CSPs (Fig. 3A–B). To identify the IP₃ binding site, we mapped the CSPs on a low-energy model of Tat₈₆ (PDB ID: 1TIV) determined previously by NMR methods and molecular dynamics (MD) calculations (Fig. 3C) [43]. In this structure, the HBR motif is placed in juxtaposition to the N-terminal domain (residues in group 2). Our NMR data indicate that the CSPs observed for signals belonging to group 2 are likely attributed to the proximal HBR motif. These results suggest that although the Tat protein lacks well-defined secondary structure motifs, it may adopt a favorable fold in which the N- and C-termini are placed in proximity. Altogether, our data revealed that Tat₈₆ binding to PI(4,5)P₂ is mediated by the lipid polar head and the HBR motif, suggesting the electrostatic nature of the interaction.

Figure 3. NMR titration of IP₃ into Tat₈₆.

(A) Overlay of 2D ¹H-¹⁵N HSQC spectra upon titration of IP₃ into 80 μM Tat₈₆ at 35 °C [IP₃:MA = 0:1 (black), 2:1 (red), 4:1 (green), and 8:1 (magenta)] in 50 mM sodium acetate-d₃ (pH 4.2) and 10 mM TCEP. Inset, signal corresponding to the indole NH group of Trp11. Peak labels are colored according to the residue color in the structure shown in C. (B) A histogram of normalized ¹H-¹⁵N chemical shift changes vs. residue number calculated from the HSQC spectra for Tat₈₆ upon titration with IP₃. (C) Cartoon representation of the Tat₈₆ structure (PDB ID: 1TIV) highlighting basic residues (group 1; blue) that exhibited substantial CSPs upon binding of IP₃. Signals of residues highlighted in orange (group 2) are likely perturbed due to their proximity to the IP₃ binding site. Trp11 is shown in magenta sticks.

Next, we further sought to investigate the nature of the Tat₈₆–IP₃ interaction and the contribution of enthalpic and entropic factors. Because the HBR consists of eight basic residues, we assessed whether more than one IP₃ molecule is capable of binding to Tat₈₆. We conducted isothermal titration calorimetry (ITC) experiments which provides information on the dissociation constant (K_d), stoichiometry (n), enthalpy change (ΔH°) and entropic term (TΔS°). As shown in Figure 4, titration of IP₃ into Tat₈₆ yielded an exothermic thermogram. Applying a single set of identical sites model to fit the data yielded the following parameters: K_d = 25.9 ± 5 μM, n = 1.05 ± 0.15, ΔH° = −4.6 ± 1 kcal/mol, and −TΔS° = −1.6 ± 0.6 kcal/mol. The ITC data indicated that Tat₈₆ harbors a single IP₃ binding site, and that the exothermic reaction is indicative of the electrostatic nature of the interaction.

Figure 4. ITC data for binding of IP₃ to Tat₈₆.

ITC data obtained for titration of IP₃ (500 μM) into WT and mutant Tat₈₆ (25 μM) in 50 mM MES (pH 6.5) and 1 mM TCEP. Complete thermodynamic parameters are shown in Table 1. Binding was severely reduced for Tat₈₆ (49–51)A and Δ(49–57) mutants, indicating that the RKK motif is critical for IP₃ binding.

To determine the specificity of phosphoinositide binding to Tat₈₆, and to assess the influence of the number and position of the phosphate groups on the inositol ring, ITC titration experiments were conducted with inositol phosphates containing phosphate groups at biologically relevant positions [37]. Titration of inositol 1,3,4-trisphosphate [I(1,3,4)P₃] or inositol 1,3,5-trisphosphate [I(1,3,5)P₃] into Tat₈₆ yielded exothermic thermograms with minimal heat of binding and poor fitting models, indicating reduced affinity (Fig. S4). On the other hand, titration of I(1,3,4,5)P₄ (an analog of PI(3,4,5)P₃ that is also localized on the PM [37]) into Tat₈₆ yielded a thermogram and thermodynamic parameters (K_d = 30.3 ± 2 μM, n = 1.07 ± 0.07, ΔH° = −8.3 ± 0.6 kcal/mol, and −TΔS° = 2.1 ± 0.4 kcal/mol) that are similar to those observed for IP₃, indicating that phosphates at positions 4 and 5 are important for binding to Tat₈₆.

Identification of Tat₈₆ residues critical for IP₃ binding

To identify the exact residues constituting the PI(4,5)P₂ binding site in Tat₈₆, we constructed a panel of mutants with multiple alanine substitutions in the HBR and Trp11 (Fig. 1). All proteins were expressed and purified as described for the wild type (WT) Tat₈₆ protein. We assessed binding of IP₃ to Tat₈₆ mutants using ITC methods as described above for the WT Tat₈₆ protein. First, we assessed IP₃ binding to a Tat₈₆ mutant in which the RKK motif was substituted with alanine residues [(49–51)A]. The ITC data did not fit any of the binding models as binding of IP₃ to the RKK mutant yielded minimal heat of binding, indicating greatly reduced affinity (Fig. 4 and Table 1). Likewise, binding of IP₃ to (49–53)A and Δ(49–57) mutants, which both lack the RKK motif, was undetectable by ITC (Fig. 4 and Table 1). These results indicated that the RKK motif is critical for binding of IP₃.

Table 1.

Thermodynamic parameters for IP₃ binding to Tat₈₆ protein constructs obtained from ITC data

protein	Kd (μM)	ΔH (kcal/mol)	−TΔS (kcal/mol)	n
WT	25.9 ± 5.0	−4.6 ± 1.0	−1.6 ± 0.6	1.05 ± 0.15
(49–51)A	nd	nd	nd	nd
(49–53)A	nd	nd	nd	nd
(52–57)A	11.0 ± 2.6	−2.1 ± 0.1	−4.7 ± 0.5	0.83 ± 0.10
(55–57)A	15.0 ± 3.2	−3.2 ± 0.1	−3.4 ± 0.3	0.90 ± 0.10
Δ(49–57)	nd	nd	nd	nd
W11A	15.7 ± 4.0	−2.6 ± 0.5	−3.9 ± 0.4	0.93 ± 0.15
W11A/Δ(49–57)	nd	nd	nd	nd

^nd not detectable or data could not be fitted due to very weak binding

Titrations were conducted at 25 °C in a buffer containing 50 mM MES (pH 6.5) and 1 mM TCEP.

Values are average of 2–3 replicates.

Next, we examined whether substitutions in the Arg-rich motif have any effect on IP₃ binding to Tat₈₆. To do so, we assessed binding of IP₃ to Tat₈₆ (52–57)A and (55–57)A mutants, in which the Arg-rich motif is partially or fully substituted with alanines while preserving the RKK motif. Interestingly, fitting the ITC data for both mutants using a single set of identical binding sites model yielded thermodynamic parameters that are similar to those obtained for the WT Tat₈₆ protein, indicating that Arg-rich motif does not play a major role in PI(4,5)P₂ binding (Table 1). Next, we tested whether Trp11 plays any role in IP₃ binding since Trp11 was implicated in Tat secretion [29, 33], and because our NMR data showed that the amide signal corresponding to Trp11 exhibited a significant CSP (Fig. 3). ITC data obtained for IP₃ binding to Tat₈₆ W11A mutant yielded thermodynamic parameters that are virtually identical to those of the WT protein (Table 1), indicating that Trp11 plays essentially no role in IP₃ binding. In summary, these findings indicate that the RKK motif is the main docking site for PI(4,5)P₂ and that the Arg-rich motif and Trp11 play essentially no role in Tat₈₆ binding to PI(4,5)P₂.

Characterization of Tat₈₆ binding to lipid nanodiscs (NDs)

Lipid NDs are increasingly used as membrane mimetics to determine structures of membrane proteins and to characterize protein–membrane interactions [53–58]. One advantage for using lipid NDs is that they can be modified in size and lipid composition. Another advantage is the ability to obtain quantitative measurements of binding to proteins by calculating the ND concentration based on the absorbance of the MSP protein at 280 nm. We have previously utilized lipid NDs as a membrane mimetic to study binding of various proteins [59, 60]. Here, lipid NDs with various lipid compositions of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (PC), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoserine (PS), brain lipid extracts, and PI(4,5)P₂ were prepared. ITC was then employed to obtain the thermodynamic parameters for WT and mutant Tat₈₆ binding to lipid NDs. Titration of WT Tat₈₆ into NDs made of 100% PC produced no detectable heat of binding (data not shown) indicating that neutral lipids alone are not sufficient to support detectable Tat₈₆–membrane binding. To test if inclusion of PI(4,5)P₂ is sufficient for Tat₈₆ binding to NDs, we conducted ITC titration with lipid NDs made of PC:PI(4,5)P₂ (95:5)(Fig. S5). No detectable binding was observed, indicating that at this stoichiometry PI(4,5)P₂ is not sufficient for Tat₈₆ binding.

Next, we examined the effect of PS on Tat₈₆ binding by conducting ITC titrations with PC:PS:PI(4,5)P₂ (75:20:5) NDs. As shown in Figure 5, the resulting thermogram is characteristic of an exothermic binding process. Data were best fit with a model for a single set of identical binding sites and yielded the following parameters: K_d = 2.3 ± 0.4 μM, n = 0.91 ± 0.09, ΔH° = −22.5 ± 1.8 kcal/mol, and −TΔS° = 14.6 ± 2.1 kcal/mol. (Fig. 5 and Table 2). Of note, binding of Tat₈₆ to lipid NDs containing native PI(4,5)P₂ and PS is 10-fold tighter than that measured for IP₃ alone (Fig. 4). To further assess how Tat₈₆ binds to NDs containing native-like membrane composition, we conducted ITC titration with NDs prepared with brain polar lipid extract (Porcine) containing 12.6% PC, 18.5% PS, 33.1% phosphatidylethanolamine (PE), 4.1% phosphatidylinositol (PI), 0.8% phosphatidic acid (PA), and 30.9% unknown lipids. Lipid extract was supplemented with 5% PI(4,5)P₂. The resulting exothermic thermogram fitted with a model for a single set of identical binding sites yielded the following parameters: K_d = 1.3 ± 0.2 μM, n = 2.1 ± 0.1, ΔH° = −9.7 ± 0.4 kcal/mol, and −TΔS° = 1.7 ± 0.4 kcal/mol (Fig. S6). Interestingly, whereas the affinity of binding is only 2-fold stronger than that obtained for PC:PS:PI(4,5)P₂ (75:20:5), the stoichiometry of binding (n = 2) suggests that lipid extracts supplemented with PI(4,5)P₂ are capable of attracting two molecules of Tat₈₆ compared to PC:PS:PI(4,5)P₂ (75:20:5) with n = 1.

Figure 5. ITC data for binding of Tat₈₆ proteins to PC:PS:PI(4,5)P₂ NDs.

Representative ITC data obtained for titration of WT and mutant Tat₈₆ (160–180 μM) into lipid NDs (8 μM) containing PC:PS:P(4,5)P₂ (75:20:5) in a buffer containing 50 mM MES (pH 6.5) and 1 mM TCEP. Data were fitted by applying a single set of identical sites model. Complete thermodynamic parameters are shown in Table 2.

Table 2.

Thermodynamic parameters for Tat₈₆ proteins binding to lipid NDs obtained from ITC data

protein	Kd (μM)	ΔH (kcal/mol)	−TΔS (kcal/mol)	n
PC:PS:PI(4,5)P2 (75:20:5)
WT	2.3 ± 0.4	−22.5 ± 1.8	14.6 ± 2.1	0.91 ± 0.09
(49–51)A	9.2 ± 1.0	−33.1 ± 2.7	26.3 ± 3.8	0.98 ± 0.03
(49–53)A	8.4 ± 0.3	−33.6 ± 2.6	26.7 ± 2.6	0.98 ± 0.05
(52–57)A	22.4 ± 1.9	−41.6 ± 5.7	35.4 ± 3.2	1.20 ± 0.14
Δ(49–57)	22.6 ± 2.3	−39.4 ± 1.0	33.0 ± 0.9	1.09 ± 0.21
W11A	nd	nd	nd	nd
W11A/Δ(49–57)	nd	nd	nd	nd
PC:PS (50:50)
WT	6.0 ± 0.6	−28.9 ± 1.0	21.7 ± 1.0	0.99 ± 0.03
(49–51)A	nd	nd	nd	nd
(52–57)A	nd	nd	nd	nd
brain lipid extract + 5% PI(4,5)P 2
WT	1.3 ± 0.2	−9.7 ± 0.4	1.7 ± 0.4	2.1 ± 0.1
W11A/Δ(49–57)	nd	nd	nd	nd

^nd not detectable or data could not be fitted due to very weak binding

Titrations were conducted at 25 °C in a buffer containing 50 mM MES (pH 6.5) and 1 mM TCEP.

Values are average of 2–3 replicates.

Identification of Tat residues critical for membrane binding

To identify residues critical for overall membrane binding, we assessed binding of a panel of Tat₈₆ mutants to PC:PS:PI(4,5)P₂ (75:20:5) NDs by ITC. First, we tested binding of Tat₈₆ lacking the RKK motif [(49–51)A]. Data show that the affinity of binding is ~4-fold weaker than WT Tat₈₆ (Fig. 5 and Table 2), suggesting that other residues of Tat₈₆ play additional roles in membrane binding. Similar thermodynamic parameters were obtained for a Tat₈₆ mutant lacking the RKK motif and two subsequent arginine residues [(49–53)A] (Table 2). Deletion of the Arg-rich region [(52–57)A] or the entire HBR [Δ(49–57)] reduced the biding affinity by 10-fold (K_d = 22.4 and 22.6 μM, respectively), indicating that both the RKK and Arg-rich motifs are important for binding to PC:PS:PI(4,5)P₂ NDs (Table 2). Next, we assessed the role of Trp11 by conducting binding studies with Tat₈₆ W11A to PC:PS:PI(4,5)P₂ NDs. Intriguingly, binding was severely reduced as indicated by the minimal heat of binding (Fig. 5). Lastly, we performed similar titrations on NDs with PC:PS:PI(4,5)P₂ (75:20:5) or brain lipid extracts supplemented with 5% PI(4,5)P₂ and a Tat₈₆ construct lacking the entire HBR and Trp11 [W11A/Δ(49–57)]. As shown in Figures 5 and S6, binding was completely abolished, indicating that the HBR and Trp11 are both required for Tat₈₆ binding to membranes.

Role of PS in Tat₈₆–membrane binding

We have shown that the RKK motif constitutes the PI(4,5)P₂ binding site and that Tat₈₆–membrane interaction is dependent on the Arg-rich motif and Trp11. Here, we sought to dissect the precise role of PS and Arg-rich motif in membrane binding. First, we assessed whether Tat₈₆ is capable of binding to lipid NDs containing only PC and PS by conducting ITC titrations with PC:PS (50:50) NDs with the total negative charge equivalent to that of PC:PS:PI(4,5)P₂ (75:20:5) NDs (Fig. 6 and Table 2). As shown, the ITC thermogram and thermodynamic parameters for Tat₈₆ titrations into PC:PS (50:50) are similar to those for PC:PS:PI(4,5)P₂ (75:20:5) NDs, suggesting that Tat₈₆ binding to membrane is mediated by charge–charge interactions and that PI(4,5)P₂ binding to Tat₈₆ can be substituted by PS if the overall charge is maintained. To further test if maintaining the charge is sufficient for efficient Tat₈₆ binding to NDs, we conducted ITC titration with lipid NDs made of 90:10 PC:PI(4,5)P₂ (Fig. S5), which has a total negative charge equivalent to PC:PS:PI(4,5)P₂ (75:20:5) or PC:PS (50:50). Interestingly, applying a single set of identical sites model to fit the data yielded the following parameters: K_d = 5.8 ± 0.9 μM, n = 2.05 ± 0.07, ΔH° = −18.2 ± 1.1 kcal/mol, and −TΔS° = 11.0 ± 0.1 kcal/mol. The ITC data revealed that, whereas the apparent binding affinity is ~2-fold weaker than those observed for Tat₈₆ binding to PC:PS:PI(4,5)P₂ (75:20:5) or PC:PS (50:50) (Table 2), the n value indicates that two Tat₈₆ molecules are capable of binding to 90:10 PC:PI(4,5)P₂ NDs. These results suggest that in the absence of PS, PI(4,5)P₂ can bind to the Arg-rich motif.

Figure 6. ITC data for binding of Tat₈₆ proteins to PC:PS NDs.

Representative ITC data obtained for titration of WT and mutant Tat₈₆ (170 μM) into lipid NDs (8 μM) containing PC:PS (50:50) in a buffer containing 50 mM MES (pH 6.5) and 1 mM TCEP. Data were fitted by applying a single set of identical sites model. Complete thermodynamic parameters are shown in Table 2.

Because data above indicated that the RKK motif constitutes the PI(4,5)P₂ binding site, we hypothesized that the Arg-rich motif is the PS-binding motif. To test this hypothesis, we assessed binding of a Tat₈₆ construct containing the RKK motif but lacking the Arg-motif [(52–57)A] to PC:PS (50:50) NDs. ITC data show very minimal heat of binding (Fig. 6); fitting of the titration data did not yield meaningful thermodynamic parameters, indicating that binding was greatly reduced. To assess whether the Arg-rich alone is sufficient for binding to PC:PS NDs, we conducted ITC titrations on Tat₈₆ (49–51)A with PC:PS (50:50) NDs (Table 2). Interestingly, binding was also greatly reduced indicating that the Arg-rich motif alone is not capable of supporting efficient membrane binding. Altogether, our data show that the RKK motif constitutes the PI(4,5)P₂–binding site, that the arginine-rich motif preferentially interacts with PS, and that the RKK motif plays a role in Tat₈₆ binding to PS-enriched membrane even in the absence of PI(4,5)P₂.

Conformational changes of Tat₈₆ and Trp11 penetration into membranes

A variety of membrane mimetics including detergent-based micelles and disk-like bicelles are often used in NMR studies of membrane proteins and membrane-interacting proteins [61–64]. To assess whether the Tat₈₆ protein undergoes conformational changes upon association to membranes, we obtained 2D ¹H-¹⁵N HSQC data of Tat₈₆ in dodecylphosphocholine (DPC) micelles. As shown in Figure 7A, substantial CSPs are observed in the spectrum suggesting that the Tat₈₆ protein undergoes conformational changes upon association to micelles. Interestingly, the signal corresponding to the indole ring of Trp11 as well as Leu8, Glu9, and Lys12 exhibited pronounced CSPs upon Tat₈₆ binding to micelles (Fig. 7A), suggesting that the N-terminus of Tat₈₆ undergoes conformational changes that may allow Trp11 to penetrate the micelle. To assess the type of structural changes in Tat₈₆ upon binding to DPC micelles, we collected a CD spectrum of Tat₈₆ in DPC micelles. Interestingly, CD data show two minima at 208 and 222 nm suggestive of a gain of α-helical features (Fig. S3).

Figure 7. NMR data of Tat₈₆ in micellar solution.

(A) 2D ¹H-¹⁵N HSQC spectra of Tat₈₆ in the free state (black) and in 25 mM DPC (red). (B) 2D ¹H-¹H NOESY spectrum obtained for Tat₈₆ in a buffer containing 50 mM sodium acetate-d₃ (pH 4.5), 2.5 mM TCEP and 25 mM DPC (20:80 protonated:deuterated). Data show intermolecular NOEs between the acyl chain of DPC and the indole ring of Trp11. Colored intermolecular NOE cross-peaks correspond to colored atoms on the DPC structure. Dashed lines in magenta and black indicate cross-peaks from DPC to Trp11 indole ring and aromatic rings of tyrosine or phenylalanine residues, respectively. Inset, signal corresponding to the indole NH group of Trp11.

To determine how Trp11 is inserted into membrane and to probe the depth of membrane insertion, we obtained 2D ¹H-¹H NOESY NMR data on Tat₈₆ in DPC micellar solution. Unambiguous intermolecular NOE cross-peaks between the acyl chain of DPC and the indole ring of Trp11 were observed (Figs. 7B). Of note, the methylene resonances of the DPC chain exhibited strong intermolecular NOE correlations to the indole ring. In addition, intermediate intermolecular NOE cross-peaks between the terminal methyl group of DPC and the indole ring of Trp11 were also observed, indicating a significant penetration of Trp11 in the interior of the micelles. Altogether, our data indicate that Tat₈₆ binding to membrane induces a conformational change allowing for Trp11 penetration in the interior of the membrane.

Discussion

In addition to its prominent role in HIV-1 transcription in infected cells, extracellular Tat can also induce severe cell dysfunction ranging from cell activation to cell death [65–67]. The Tat protein is strongly linked to the development and progression of HAND [8–14]. Persistent and active secretion of Tat from infected microglia/macrophages is particularly toxic for neurons [9, 17–22]. Central to the deleterious effects of Tat in the CNS is its ability to cross membranes during secretion and uptake by healthy cells [68]. Several lines of evidence suggested that Tat secretion is mediated by interaction with PI(4,5)P₂ [29, 31–35]. In this report, we devised an approach that allowed for production of homogenous and monomeric Tat₈₆ protein, which enabled detailed characterization of its interactions with PI(4,5)P₂ and membrane mimetics by NMR and biophysical methods. We have shown that the RKK motif constitutes the PI(4,5)P₂ binding site, and that Tat₈₆ interacts with lipid NDs enriched with PI(4,5)P₂ and/or PS 10-fold tighter than that of the free IP₃ ligand. These results are consistent with the finding that mutation of the RKK motif led to severe reduction of Tat secretion (only ~1% secretion) [33]. We have shown specific binding of IP₃ to Tat₈₆ compared to I(1,3,4)P₃ and I(1,3,5)P₃, indicating that the position of the phosphate groups is important for binding. This result is further supported by the finding that I(1,3,4,5)P₄ binds to Tat₈₆ with a similar affinity to IP₃. Our data also indicated that the Arg-rich motif appears to preferentially interact with PS, and that Trp11 is a critical residue for Tat penetration into membrane. The requirement of PS for efficient Tat₈₆ binding to membranes is reflected by the finding that Tat₈₆ is not able to bind to NDs made of 95:5 PC:PI(4,5)P₂. Furthermore, we provided evidence that Tat₈₆ binding to NDs with natural lipid composition supplemented with PI(4,5)P₂ is only slightly more efficient than PC:PS:PI(4,5)P₂, indicating that other membrane components play minimal role in binding. It is likely though that possible clustering/packing of brain extract lipids in NDs are responsible for binding of two Tat₈₆ molecules. Altogether, our data support a mechanism by which HIV-1 Tat secretion from the PM is mediated by a tripartite signal consisting of binding of the RKK motif to PI(4,5)P₂, Arg-rich motif to PS, and penetration of Trp11 in the membrane bilayer (Fig. 8).

Figure 8. Schematic model of Tat₈₆ bound to membrane bilayer.

Tat₈₆–membrane interactions are mediated by binding of the RKK motif to PI(4,5)P₂, the Arg-rich motif to PS, and Trp11 penetration through the membrane bilayer.

An interesting observation in this study is the finding that NMR signals corresponding to residues in the N-terminal loop are perturbed by IP₃ binding (Fig. 3A). These results suggest that, although the Tat₈₆ protein lacks secondary structure motifs, it may fold in an energetically favorable state that brings the N-terminal residues in close proximity to the HBR. Of note, this conformational arrangement was observed in the structure of Tat₈₆ determined by 2D NMR methods and MD calculations (Fig. 3C) [43] and other Tat₈₆ structures calculated by similar methods (Fig. S7) [69, 70]. In these structures, the indole ring of Trp11 is sequestered in the core of the protein. Our NMR data suggest that IP₃ binding alone does not induce major conformational changes in the Tat₈₆ protein. Therefore, upon binding to membranes it is likely that the Tat protein undergoes conformational change to enable penetration of the indole ring of Trp11 into the membrane bilayer. The requirement of an aromatic ring to facilitate efficient membrane penetration and therefore secretion is consistent with the finding that substitution of Trp11 with phenylalanine and tyrosine residues supports WT-like membrane penetration [35, 38]. Trp has unique physical-chemical properties and is often encountered in membrane proteins, especially at the level of the water/bilayer interface (reviewed in [71, 72]). It plays a role in membrane protein stabilization, anchoring, and orientation in lipid bilayers. The indole group of Trp is often involved in various types of interactions, such as π–cation or hydrogen bonds [71, 72]. Consistent with its characteristic as a membrane anchoring resides, Trp11 appears to stabilize binding of Tat₈₆ to membranes by penetrating deeply in the bilayer.

Perhaps the most surprising result is the finding that the RKK and Arg-rich motifs preferentially bind to PI(4,5)P₂ and PS, respectively. Preferential binding of PI(4,5)P₂ to lysine-rich motifs has been observed in numerous cases such as retroviral MA proteins [49–51, 61, 73–76], C2 [77–80], pleckstrin homology (PH) [81], Phox (PX) [77], and epsin N-terminal homology [82] domains as well as myristoylated alanine rich C-kinase substrate (MARCKS) [83]. Lysine- and Arg-rich motifs specificity to PI(4,5)P₂ and PS is similar to our recent findings on the human T-cell leukemia virus type 1 (HTLV-1) MA protein binding to membranes [50, 51]. We have shown that a lysine-rich motif (Lys47, Lys48, and Lys51) in HTLV-1 MA constitutes the PI(4,5)P₂ binding site, whereas an arginine rich motif (Arg3, Arg7, Arg14 and Arg17) located in the unstructured N-terminus binds to PS [50, 51]. These interactions proved to be critical for virus production and release.

Arg-rich motif selectivity to PS has been observed for many systems [84–88]. For example, cellular localization and membrane interaction of influenza A virus MA protein 1, an essential component involved in the structural stability of the virus and in the budding of new virions from infected cells, is mediated by an Arg-triplet motif and PS clusters [86–88]. PS has also been shown to be essential for retrograde membrane traffic at recycling endosomes (REs) [84], in which PS is most concentrated in REs among intracellular organelles, and evectin-2 protein, was targeted to REs by the binding of its PH to PS. Structural data confirmed the specificity of the binding of PS to the PH domain, which also appears to be mediated by two arginine residues. Taken together, we speculate that specificity of PI(4,5)P₂ and PS binding to Lys- and Arg-rich motifs is more common than previously thought.

In this study, we provided structural insights into the mechanism of Tat association with membranes and identified motifs that are critical for the interaction. However, the exact mechanism of Tat secretion through the PM has yet to be elucidated. A few models of Tat secretion have been proposed. In one report, it was suggested that a saddle-splay membrane curvature model is formed to allow entry of Tat to cells through an induced pore [89]. Other studies suggested that Tat deforms membrane shape and creates pores [35, 90]. It was also proposed that Tat secretion may proceed via oligomerization-mediated pore formation, spontaneous translocation, or incorporation into exosomes [29]. In the oligomerization-mediated pore formation model, multiple Tat protein molecules are recruited to the PI(4,5)P₂-enriched sites, resulting in oligomerization on the cytoplasmic side of the PM. Interaction with PI(4,5)P₂ has been shown to be required for Tat oligomerization and formation of pores [31, 35]. Accumulation of Tat on the PM was suggested to cause thinning of the PM bilayer, leading to arginine and lysine penetration into the bilayer [90–93]. As a result, a pore is formed allowing secretion of Tat molecules. In another model, it was proposed that Tat translocation can be initiated subsequent to PI(4,5)P₂ binding to the HBR [29, 31, 35, 92]. This interaction is further stabilized upon Trp11 insertion into the PM bilayer, leading to subsequent extracellular translocation of Tat. Tight Tat association with the PM may lead to destabilization of the bilayer caused by the Arg-rich motif, followed by passive translocation.

Previous studies have shown that short Tat-derived Arg-rich peptides not only penetrate membranes directly in a passive manner but also form physical pores, largely influenced by lipid topology and anionic charge of the membranes [90]. The finding that the RKK motif of Tat₈₆ recognizes PI(4,5)P₂ and the Arg-rich motif binds to PS, followed by Trp11 insertion into the bilayer may suggest that this synergy is required for efficient membrane penetration. It was reported that many membrane-active peptides and proteins, such as the Tat-derived peptides, display a stronger cell-membrane interaction with arginine over lysine [94, 95]. It was suggested that one of the advantages of arginine over lysine is its ability to form stable bidentate hydrogen bonds with phosphate and sulfate groups [96]. Because of the bidentate H-bonding ability and the planar Y-shape of the guanidinium groups, arginine side chains lay on the bilayer instead of being perpendicular as in the case of lysine, which creates a steric hindrance on the membrane surface and a lipid head crowding that generates a negative Gaussian curvature.

Recent reports have shown that a potent Tat inhibitor, didehydro-cortistatin (dCA) and derivatives, bind directly to Tat inhibiting its function by locking HIV-1 in persistent latency, leading to blockage of viral rebound [97–99]. It was shown that dCA crosses the blood-brain barrier, cross-neutralizes Tat activity from several HIV-1 clades and decreases Tat uptake by glial cell lines [99, 100]. Molecular modeling studies of dCA with Tat₈₆ revealed that dCA and analogs bind to the HBR [98]. An interesting future direction would then be to examine if dCA potently reduces Tat binding to the PM membrane and therefore inhibit secretion in vivo. Elucidation of the structural basis for Tat secretion and uptake may aid in the advancement of therapeutic intervention aimed at the inhibition of Tat’s activity not only in transcription and activation but also in neurotoxicity and HAND.

Experimental methods

Plasmid construction

The DNA gene encoding for HIV-1 Tat₈₆ was amplified from the pNL4–3 isolate (GenBank: AF324493.2) with NdeI and BaMHI restriction sites on the 5’- and 3’-end, respectively. The Tat₈₆ gene was then ligated into pET11a vector. Tat₈₆ (49–51)A, (55–57)A, and W11A mutant constructs were obtained using the QuickChange Lightning Site-Directed Mutagenesis kit (Agilent Technologies). Genes encoding for the (49–53)A, (52–57)A, and Δ49–57 mutants with NdeI and BaMHI restriction sites were purchased from Twist Bioscience, digested, and ligated into a pET11a vector. Positive clones were verified at the Heflin Genomics Core at the University of Alabama at Birmingham.

Protein expression and purification

The Tat₈₆ protein was expressed in E. Coli BL21(DE3) pLysS cells (Agilent Technologies). Cells were grown at 37 °C in Terrific Broth medium containing ampicillin (100 μg/mL) and chloramphenicol (34 μg/mL). When A₆₀₀ reached ~1.0, cells were induced with 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG; Gold Biotechnology). Cells were then grown at 37 °C for 18 h, spun down, and stored at −80 °C. The cell pellet was resuspended in a buffer containing 50 mM phosphates (pH 8.0), 10 mM Tris·HCl, 300 mM NaCl, and 10 mM tris(2-carboxyethyl)phosphine hydrochloride (TCEP). Cells were then sonicated, and lysate was spun down at 18,000g for 30 min at 4°C. The Tat₈₆ protein was isolated from inclusion bodies using an extraction buffer containing 50 mM phosphates (pH 7.2), 10 mM Tris·HCl, 300 mM NaCl, 10 mM TCEP, and 6M guanidine hydrochloride. The extract was sonicated, rocked overnight at 4°C, and spun down at 35,000g at 4 °C for 30 mins. The supernatant was loaded on cobalt affinity resin column equilibrated with extraction buffer. The resin was washed with 10x column volume using a buffer containing 50 mM phosphates (pH 7.3), 10 mM TCEP, and 6 M guanidine hydrochloride. The protein was eluted with a buffer containing 50 mM sodium acetate (pH 4), 10 mM TCEP, and 6 M guanidine hydrochloride. The elution fractions were renatured by overnight dialysis in 100 mM sodium acetate (pH 3.0) followed by 10 mM sodium acetate (pH 3.0). The Tat₈₆ protein was then concentrated using Amicon Ultra-15 centrifugal filter with 3 kDa cutoff (Millipore-Sigma) and loaded on Superdex-75 size exclusion column (Cytiva) using a buffer containing 50 mM sodium acetate (pH 4.0), 150 mM NaCl and 10 mM TCEP. The fractions containing the Tat₈₆ protein were pooled, concentrated, and stored at −20°C. Protein expression and purification of Tat₈₆ mutants were achieved as described for the WT protein. Uniformly ¹⁵N- and ¹⁵N,¹³C-labeled Tat₈₆ samples were prepared by growing cells in M9 minimal media containing ¹⁵NH₄Cl or ¹³C₆-D-glucose as the sole nitrogen and carbon sources, respectively. Protein purification was performed as described above.

Preparation of lipid nanodiscs.

Membrane scaffold protein 1 (MSP1 plasmid; Addgene) was expressed and purified as described [56, 60]. The MSP1 protein was stored at −20 °C in ND assembly buffer (20 mM Tris·HCl, pH 7.4, 100 mM NaCl, 0.5 mM EDTA, and 0.01% sodium azide). 1-palmitoyl-2-oleoylglycero-3-phosphocholine (PC), 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine (PS), L-α-phosphatidylinositol-4,5-bisphosphate (Porcine brain PI(4,5)P₂), and brain polar lipid extract (Porcine) containing 12.6% PC, 33.1% PE, 18.5% PS, 4.1% PI, 0.8% PA and 30.9% unknown lipids, were used as purchased (Avanti Polar Lipids) and stored as stock chloroform solutions at −20 °C. NDs were prepared by mixing the appropriate ratios (depending on the desired lipid composition in NDs) of PC, PS, brain lipid extract, and PI(4,5)P₂ chloroform stock solutions in a glass vial. Total lipid concentration was 50 mM. Chloroform was evaporated under a gentle airflow until a dry film was observed followed by lyophilization for 1 h. Lipids were re-dissolved by multiple freeze/thaw cycles in a buffer containing 20 mM Tris·HCl (pH 7.4), 100 mM NaCl, 100 mM sodium cholate, 0.5 mM EDTA, and 0.01% sodium azide. MSP1 was then added to the lipid mixture at a final molar ratio of 1:65 MSP1:lipids. Sample was rotated on an orbital shaker at room temperature for 1 h. Sodium cholate was removed by incubation of the sample mixture with Bio-beads SM adsorbents (Bio-Rad) on an orbital shaker overnight at 4 °C. Lipid NDs were then passed through a 0.2 μm filter to remove the Bio-beads and run on Superdex 200 Increase 10/300 GL column (Cytiva). ND fractions were pooled, concentrated to the desired concentration, and used immediately or stored at −80 °C.

NMR Spectroscopy

NMR data were collected at 35 °C on a Bruker Avance II (700 MHz ¹H) or Avance III (600 MHz ¹H) spectrometers equipped with cryogenic triple-resonance probes, processed with NMRPIPE [101], and analyzed with NMRVIEW [102] or CCPN Analysis [103]. The backbone resonances were assigned using standard triple resonance data (HNCA, HN(CO)CA, HN(CO)CACB, and HNCACB) collected on 300 μM sample of ¹³C,¹⁵N-labeled Tat₈₆ in a buffer containing 50 mM sodium acetate-d₃ (pH 4.2) and 10 mM TCEP. The triple-resonance experiments were collected in conventional or non-uniformly sampled (NUS) sparse mode (20% sampling density in indirect dimensions) according to schemes generated using hmsIST [104]. Two-dimensional ¹H-¹H NOESY spectra were collected at 35 °C with a 300 ms NOE mixing time. The Tat₈₆ sample used for the 2D NOESY data was prepared at 300 μM in 50 mM sodium acetate-d₃ (pH 4.2), 2.5 mM TCEP, and 25 mM DPC (20:80 protonated:deuterated) in 100% D₂O.

NMR titrations of Tat₈₆ with lipids

¹H-¹⁵N HSQC NMR titrations with IP₃ were conducted with 100 μM samples of ¹⁵N-labeled Tat₈₆ in 50 mM sodium acetate-d₃ (pH 4.5) and 10 mM TCEP using a 10 mM stock solution of IP₃ (Cayman Chemical) prepared in water. Chemical shift perturbations were calculated as $\Delta {\delta }_{HN}=\sqrt{\Delta {\delta }_H^2+0.04(\Delta {\delta }_N^2)}$, where Δδ_H and Δδ_N are ¹H and ¹⁵N chemical shift changes, respectively.

Isothermal Titration Calorimetry

Stock solutions of IP₃, I(1,3,4)P₃, I(1,3,5)P₃ and I(1,3,4,5)P₄ (Cayman Chemical) were prepared at 10 mM in water. Thermodynamic parameters of IP₃ or lipid NDs binding to WT or mutant Tat₈₆ were determined using a MicroCal PEAQ-ITC (Malvern Instruments). ITC experiments were conducted in a buffer containing 20 mM MES (pH 6.5) and 1 mM TCEP. IP₃ prepared at ~500 μM in the same buffer was titrated into ~25 μM Tat₈₆. Heat of reaction was measured at 25 °C for 19 injections. Heat of dilution was measured by titrating IP₃ into buffer and was subtracted from the heat of binding. Lipid NDs were prepared at a concentration of 8 μM and Tat₈₆ at 160–180 μM. Heat of dilution was measured by titrating Tat₈₆ into buffer and was subtracted from the heat of binding. Data analysis was performed using PEAQ analysis software. Thermodynamic parameters were determined by fitting baseline-corrected data with a single-site binding model. ITC experiments were collected in a set of three to ensure data reproducibility. Mean values and standard deviations of the thermodynamic parameters were calculated based on the three independent experiments.

Mass spectrometry

Approximately, 30 pmol of Tat₈₆ in 0.1% formic acid was trapped, desalted, and subsequently eluted using a Waters Vanguard C-18 trap column and Waters BEH C-18 column with an increasing gradient of acetonitrile. ESI-TOF spectrum was collected on a Waters Synapt G(2)-S(i) mass spectrometer in positive ion resolution mode using MassLynx 4.1. Raw spectrum was processed using the MaxEnt 1 function within MassLynx.

SEC-MALS

To assess whether the Tat₈₆ protein is monodisperse, 50 μg protein sample in a buffer containing 50 mM sodium acetate (pH 4), 100 mM NaCl, and 5 mM TCEP was injected onto a SEC analytical column (5 mm, 300Å, 4.6 mm; Wyatt Technology) on a Shimadzu HPLC system coupled to a DAWN 8 Ambient detector equipped with a light scattering module and an Optilab refractometer (Wyatt Technology). Data were analyzed using the ASTRA 8 software (Wyatt Technology).

CD Spectroscopy

CD spectra were acquired on a Jasco J815 spectropolarimeter from 260 to 185 nm at 25 °C. Scanning rate was set to 50 nm/min. Far-UV CD spectra were obtained for 20 μM Tat₈₆ in the free state or in 20 mM DPC micelles at 25 °C in a buffer containing 50 mM sodium acetate (pH 4.5) and 2 mM TCEP. The background signal from the buffer solution was subtracted from the spectrum.

Highlights:

• Molecular determinants of HIV-1 Tat binding to membrane are defined

• The Arg49-Lys50-Lys51 (RKK) motif is the PI(4,5)P₂ binding site

• The arginine-rich motif preferentially binds to phosphatidylserine

• Trp11 is a critical residue for membrane penetration

Abbreviations:

Tat

trans-activator of transcription

PM

plasma membrane

PI(4,5)P₂

phosphatidylinositol 4,5-bisphosphate

IP₃

inositol 1,4,5-trisphosphate

PC

1-palmitoyl-2-oleoylsn-glycero-3-phosphocholine

PS

1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine

DPC

dodecylphosphocholine

HBR

highly basic region

ND

nanodisc

NMR

nuclear magnetic resonance

HSQC

heteronuclear single quantum coherence

NOESY

nuclear Overhauser effect spectroscopy

CSP

chemical shift perturbation

ITC

isothermal titration calorimetry

Data Availability

The NMR chemical shift data for HIV-1 Tat₈₆ are available from the Biological Magnetic Resonance Data Bank under BMRB accession number 50973.