A Suite of Modular Fluorescence Assays Interrogate the Human Immunodeficiency Virus Glycoprotein-41 Coiled Coil and Assist in Determining Binding Mechanism of Low Molecular Weight Fusion Inhibitors

Abstract

Several different segments of the gp41 N-heptad repeat coiled coil have been constructed using N-terminal bipyridyl modification of composite peptides and inducing trimerization by adding ferrous ions. These metallopeptides act as receptors in fluorescence-binding assays with corresponding fluorescently labeled C-peptide probes. The Fe^II coordination complex quenches C-peptide fluorescence upon binding, and reversal of quenching by a small molecule inhibitor can be used to obtain the inhibitor-binding constant. A total of 10 peptide pairs targeting 25–46 residue segments of the coiled coil were constructed, with C-peptide probes of different lengths and binding affinities. The result is a suite of assays for exploring binding in the mM to nM range to any desired region of the coiled coil, including the hydrophobic pocket (HP), extended regions on either side of the pocket, or a region associated with T20 resistance mutations. These assays are high-throughput ready, and could be used to discover novel compounds binding along various regions of the gp41 coiled coil groove. They were used to evaluate a sub-μM low molecular weight fusion inhibitor, resulting in the finding that the molecule bound specifically to the HP and attained its potency from a low off-rate.

Introduction

Entry inhibitors are a relatively new class of approved human immunodeficiency virus type 1 (HIV-1) drugs, and have promising properties for combating resistance to current treatments and for prophylaxis.^1,2 They can act in both immune cell protection and disease prevention by targeting the steps of the viral life cycle that occur before infection of host cells. Fusion inhibitors are a class of entry inhibitors that act on the HIV transmembrane protein gp41, which is universally required by all strains for initial infection or cell-to-cell transmission.^3
–5 A large conformational change occurs in gp41 during fusion, involving exposure of a homotrimeric N-heptad repeat (NHR) coiled coil and subsequent folding of antiparallel C-heptad repeat (CHR) helices down the grooves of the coiled coil, to form a trimer of hairpins, a prerequisite for effective fusion pore formation.²

The gp41 coiled coil has been considered an important target for fusion inhibitors, based on peptides studies, which show that CHR peptides are highly effective fusion inhibitors.^6
–8 These include T20, currently the only FDA-approved fusion inhibitor.⁹ Small molecule fusion inhibitors have been sought for many years, but with limited success, having only μM potency against fusion.^10,11 They have chiefly targeted a known hydrophobic pocket (HP) on the coiled coil. However, in many cases, there is little experimental evidence confirming the exact binding site and no way to test inhibitors against different sites along the coil. Finding two inhibitors that bind in adjacent sites could enable a fragment-based drug design approach, in which a linker connecting two adjacent-binding fragments can have a multiplicative effect on their binding affinity.¹²

In previous work, we have demonstrated the use of metal–bipyridyl coordination to stabilize a segment of the gp41 coiled coil representing the HP,¹³ as well as an extended receptor targeting adjacent sites and high-affinity interactions.^14,15 This study expands on the analysis of inhibitors detected with these assays, and introduces a new assay that focuses on the N-terminal end of the coiled coil, a site prone to development of resistance mutations in viruses exposed to T20.¹⁶ C-peptide probes have been designed for fluorescence intensity assays that interrogate the NHR subsections. Adjusting C-peptide length tuned both assay selectivity and the range of detectable affinities. Small molecule inhibitors could easily be detected by competitive inhibition, as we have shown previously. Comparison of the results with the known biological activity of peptide segments of the CHR¹⁷ and small molecules¹⁸ confirmed the relevance of these biochemical screening assays for detection of fusion inhibitors. The ease and rapidity of the assays lends them to implementation in a high volume format. Utilization of the assays could result in identification of novel fragments or small molecules binding at specific subsites, with the possibility of linking them to generate high-affinity nonpeptide inhibitors. This concept has been applied for the derivation of C-peptide inhibitors of HIV-1 fusion containing multifunctional domains.^2,17

Materials and Methods

Materials

Peptides were prepared by solid-phase synthesis (Biosynthesis, Inc.). Small molecules were prepared according to previously published procedures.¹⁸ Fluorescein (FL) and Lucifer yellow (LY) were purchased from Molecular Probes (Invitrogen) as iodoacetamides and covalently attached through standard cysteine disulfide chemistry according to the manufacturer's instructions. Labeled peptides were purified by high pressure liquid chromatography using acetonitrile: H₂O with 0.05% trifluoracetic acid.

Circular Dichroism Experiments

Circular dichroism (CD) studies were performed on a DSM20 CD spectrophotometer from On-Line Instruments Systems, Inc., using 20 μM solutions of peptides in a 15 mM Tris-acetate buffer, pH 7.0, at 25°C.

Fluorescence Intensity Measurements

Fluorescence intensity was measured in 384-well low-volume plates (Greiner Bio-one) on a Spectramax M5 plate reader (Molecular Devices). Env# concentration was determined in 6 M Gdn.HCl from the absorbance at 291 nm, using the extinction coefficients of bipyridine (ɛ₂₉₁=12,000) plus the calculated contribution of any Trp and Tyr residues in the peptide sequence (www.expasy.org). FL-labeled peptide concentration was determined using ɛ₄₉₅=93,000 M⁻¹ cm⁻¹ in 6 M Gdn.HCl at pH 9; LY-labeled peptide concentration was determined using ɛ₄₂₅=10,800 M⁻¹ cm⁻¹ in 6 M Gdn.HCl. Fe(env#)₃ complexes were prepared by the addition of 1/3 stoichiometry of a freshly prepared ferrous solution to peptides in a 25 mM Tris-acetate buffer, pH 7.0. Fluorescence assays were carried out using a constant concentration of probe peptide, typically 15 nM of FL-labeled peptide or 200 nM of LY-labeled peptide. Typical receptor concentrations used in inhibition experiments were 3.6 μM [Fe(env2.0)₃ with C18-e2.0FL] 75 nM [Fe(env5.0)₃ with C32-e5.0FL], 1.5 μM [Fe(env5.0)₃ with C28-e5.1FL/LY], 3 μM [Fe(env5.0)₃ with C21-e6.0FL/LY], 3 μM [Fe(env6.0)₃ or Fe(env5.1)₃ with C28-e5.1LY/C21-e6.0LY], and 0.75 μM Fe(env6.0)₃ with C21-e6.0FL. Receptor concentrations are always reported as concentration of monomer env# peptide, which should be equal to the concentration of binding sites on three equivalent faces of the coiled coil. Fractional fluorescence values are reported relative to a control in which the Fe(env#)₃ receptor was omitted. The detailed protocol is provided in Tables 1 and 2.

Table 1.

Assay Protocol Table (C18-e2.0LY/FL, C21-e6.0LY)

Step	Parameter	Value	Description
1	Control in odd wells	9.6 μL	1:1 P=2X Cn-e#FL/LY and B=buffer
2	Assay in even wells	9.6 μL	1:1 P and R=2X Fe(env#)₃
3	Assay readout	Ex 485 (FL) or Ex 425 (LY) Em 538	Fluorescence intensity plate reader
4	Library compounds	0.4 μL	In DMSO, all wells
5	Incubation time	0–30 min	Sample equilibration
6	Assay readout	Ex 485 (FL) or Ex 425 (LY) Em 538	Fluorescence intensity plate reader

Step Notes

1. Greiner low volume tapered black/clear 384-well plates; P=30 nM Cn-#FL or 400 nM Cn-#LY in 0.02% Tween-20, 25 mM Tris, 25 mM acetic acid, pH 7.0, B=buffer (no Tween).

2. R=7.2 μM Fe(env2.0)₃ or 6 μM Fe(env5.0)₃ or 6 μM Fe(env6.0)₃ with C21-e6.0LY in a buffer (concentration of monomer peptide).

3. Determine assay bounds F _max, F _min and Z′.

4. To determine K_I, use serial dilution.

5. Measure immediately and at 15 min intervals for up to 1 h, to allow for equilibration.

6. Fractional fluorescence=(F_s±σ_s)/<F_c>, where<F_c>is the average value of the control fluorescence in odd wells, and F_s is the fluorescence of the sample. Large increase in control values with compound concentration indicates compound interacts with probe, for example, by micelle formation.

DMSO, dimethyl sulfoxide; LY, lucifer yellow; FL, fluorescein.

Table 2.

Assay Protocol Table (C32-e5.0FL, C28-e5.1LY/FL, C21-e6.0FL)

Step	Parameter	Value	Description
1	Probe P in all wells	4.8 μL	P=2X Cn-e#FL/LY
2	Library compounds	0.4 μL	In DMSO, all wells
3	Buffer B in odd wells	4.8 μL	Buffer B
4	Receptor R in even wells	4.8 μL	R=2X Fe(env#)₃
5	Incubation time	0–30 min	Sample equilibration
6	Assay readout	Ex 485 (FL) or Ex 425 (LY) Em 538	Fluorescence intensity plate reader

Step Notes

1. Greiner low volume tapered black/clear 384-well plates; P=30 nM Cn-#FL or 400 nM Cn-#LY in 0.02% Tween-20 in buffer.

2. To determine K_I, use serial dilution.

3. B=25 mM Tris, 25 mM acetic acid, pH 7.0.

4. R=150 nM Fe(env5.0)₃ with C32-e5.0FL, 3 μM Fe(env5.0)₃ with C28-e5.1FL/LY, 2 μM Fe(env6.0)₃ with C28-e5.1FL/LY, 1.5 μM Fe(env6.0)₃ with C21-e6.0FL in buffer (concentration of monomer peptide).

5. Measure immediately and at 15 min intervals for up to 1 h, to allow for equilibrium.

Results and Discussion

Peptide Design

The array of overlapping segments used to scan the full-length coiled coil is shown in Figure 1. NHR peptide env2.0 contained the residues ₅₆₅LLQLTVWGIKQLQARIL₅₈₁ of the HP, as previously described,¹³ and env3.0¹⁴ and env4.0 were extended to include a region either C-terminal or N-terminal to the pocket, respectively. Env5.0 and env5.1 extended the N-terminal region even further,¹⁵ and env6.0 contained just the N-terminal end of the coiled coil, without the HP. This region binds to the N-terminal two-thirds of T20. The env# peptides were modified with an N-terminal 5-carboxybipyridine (bpy) group and a linker sequence GQAV. Three bidentate bpy groups formed an octahedral complex with transition metal ions, stabilizing the trimeric form of the env# receptors.

Fig. 1.

Helical NHR and CHR domains of the gp41 extracellular region, showing the location of env# constructs and assay C-peptide probes listed in Table 3. For simplicity, only one NHR and CHR helix is shown from the trimer. The model shown is from protein data bank structures 2X7R²² (light gray) and 1IF3³⁹ (gray). Residues at i, i+4, and i+7 positions of the CHR are indicated in yellow (hydrophobic) or orange (polar). Env# peptides are bipyridylated and studied complexed to 1/3 stoichiometry Fe(II) (shown in purple and light purple for env6.0). C-peptides are labeled with a fluorophore at a C-terminal cysteine (fluorescein shown for C21-e6.0 in red). Env peptides contain the wild-type sequence, while C-peptides are mutated to increase helicity and affinity (see text). A known hydrophobic pocket binding small molecule, 14g, is shown in green (binding pose calculated using AutoDock4.2). NHR, N-heptad repeat; CHR, C-heptad repeat.

Figure 1 also depicts CHR peptides that were synthesized, aligned antiparallel to the NHR peptides, and labeled according to the number of residues and the cognate receptor. C18-e2.0 covered the HP and N-terminal adjacent region. C28-e5.1 and C21-e6.0 mimicked portions of T20 and did not contain HP-binding residues. C32-e5.0 and C39-e5.1 spanned close to the full length CHR. A fluorescent marker, LY or FL, was attached to each CHR peptide probe through a C-terminal cysteine, so that NHR–CHR binding could be followed by fluorescence quenching. The sensitivity of this experiment depended on the alignment of the fluorophore and metal complex, and C-peptides were designed with the fluorophore <18 Å from the metal. It was found that C-peptides designed for the receptors env5.1 and env6.0 were also quenched by env5.0, increasing the versatility of assay pairing.

Sequences of the peptides are indicated in Table 3. The wild-type (HXB2) sequence was used for all NHR peptides, to maintain receptor integrity for inhibitor detection, but C-peptides contained several helix-stabilizing modifications on the solvent-exposed face, based on studies demonstrating their effect on enhancing affinity.^8,19,20 Several C-peptide sequence variations were tested, with the sequences shown providing a good combination of high helicity and enhanced solubility.

Table 3.

NHR (env#) and CHR (Cn-e#) Assay Peptides

		% Helix^a		θ₂₀₈/θ₂₂₂
Name	NHR peptides	−Fe²⁺	+Fe²⁺	−Fe²⁺	+Fe²⁺
env2.0	bpy-GQAVEAQQHLLQLTVWGIKQLQARILAVEKK-amide	24	59	1.6	1.0
env3.0	bpy-GQAVEAQQHLLQLTVWGIKQLQARILAVERYLKDQQKK-amide	70	72	1.0	1.0
env4.0	bpy-GQAVQAIEQQQNNLLRAIEAQQHLLQLTEKK-amide	26	34	1.6	1.3
env5.0	bpy-GQAVSGIVQQQNNLLRAIEAQQHLLQLTVWGIKQLQARILAVEKK-amide	54	69	1.0	1.0
env5.1	bpy-GQAVEAARQLLSGIVQQQNNLLRAIEAQQHLLQLTVWGIKQLQARILAVEKK-amide	120	134	0.9	1.0
env6.0	bpy-GQAVEQARQLLSGIVQQQNNLLRAIEAQQHLLQLTEKK-amide	28	50	1.2	1.0

	CHR peptides	% Helix^a	% Change in complex^b
C18-e2.0	Ac-MTWBEWDREIBNYTSLIC-amide	17	+14
C27-e3.0	Ac-KSLEQBWNHTTWBEWDREIBNYTSLIC-amide	12	+29
C32-e5.0	Ac-TTWEAWDRAIAEYAARIEALIRAAQEQQEKNC-amide	68	+13
C39-e5.1	Ac-TTWEAWDRAIAEYAARIEALIRAAQEQQEKNEAELRELC-amide	45	+23
C28-e5.1	Ac-NYAALIEALIRAAQEQQEKNEAELRELC-amide	27	−9
C21-e6.0	Ac-ALIRAAQEQQEKNEAELRELC-amide	34	+19
T20	638-YTSLIHSLIEESQNQQEKNEQELLELDKWASLWNWF-673	n.d.	n.d.

# is a model number that refers to a particular construct of the NHR coiled coil; C-peptides are labeled by the number of residues n and a matching env#. C-peptides are fluorophore labeled at a C-terminal cysteine. Amino acids forming key interactions at a and d positions of the heptad repeat are indicated in bold. B is α-aminoisobutyric acid. T20 is included for reference, but is not an assay peptide.

^aDetermined by CD, using molar residue ellipticity of −33,500 degree·cm⁻²·dmol⁻¹ for 100% helix.

^bIncrease in helix content for the complex of C-peptide with cognate env# compared to the sum of component spectra.

NHR, N-heptad repeat; CHR C-heptad repeat; n.d., not determined.

NHR Trimer Formation

Metal binding to bipyridylated NHR peptides (env#) generally resulted in stabilization of the coiled coil, evidenced by an increase in helical content to 50% or higher in CD spectra and a 208 nm:222 nM peak intensity ratio of 1.0, which indicated α-helix bundle formation.²¹ An exception was the peptide env4.0, which did not give a well-structured helical coiled coil upon metal binding and subsequently performed poorly in fluorescence experiments. An apparent helical content >100% was observed for the peptide env5.1 both in the absence and presence of Fe²⁺. This was an indication of aggregation that was not ameliorated by metal binding, demonstrating a size limit of the self-assembly process in generating discreet trimers above a certain concentration. The CD results were confirmed by fluorescence studies (see below).

The receptor Fe(env6.0)₃ defined the region of gp41 at the base of the coiled coil, immediately adjacent to the fusion peptide proximal region that interacts with lipid membrane. A recent crystal structure of an extended segment of the gp41 ectodomain²² revealed a loss of helical periodicity in the N-terminal region that is included in Fe(env6.0)₃ (Fig. 1). This did not appear to affect the formation of a well-formed trimer, although it may explain the lower helical content observed compared to other well-formed receptors. Fe(env6.0)₃ is the segment of the gp41 coiled coil to which T20 binds, including the residues ₅₄₄LLSGIVQQQ₅₅₂ that are known to make important energetic interactions with T20.¹⁶ Mutations within these residues are strongly associated with resistance to T20,^23
–25 and a mutation of the T20 residue N656→L that binds to this segment abrogates fusion completely.^26,27 This receptor target could be used to interrogate small molecule replacements of T20. It is also straightforward to modify the receptor for small molecule discovery to include common T20 resistance mutations.

CD Studies of Six-Helix Bundle Formation

A change in the CD signature for a mixture of receptor and C-peptide compared to the sum of individual spectra reflected complex formation. Example CD spectra are shown in Figure 2 for the peptide pairs Fe(env5.1)₃/C39-e5.1, Fe(env5.1)₃/C28-e5.1, and Fe(env6.0)₃/C21-e6.0. An increase in helicity accompanied formation of the complex between cognate peptides Fe(env5.1)₃/C39-e5.1 and Fe(env6.0)₃/C21-e6.0, with the large negative molar residue ellipticity at both 222 and 208 nm indicative of six-helix bundle (6HB) formation.^21,28,29 Thus aggregation of Fe(env5.1)₃ did not preclude C-peptide binding. In Figure 2B, association of a truncated peptide C28-e5.1 with Fe(env5.1)₃ resulted in a distorted CD curve relative to that of the typical helix bundle. C28-e5.1 is missing the residues that bind in the HP, which is likely to lead to a combination of aggregation and an unstable six-helix structure. This effect has been observed in previous studies using T20 as the C-peptide.^17,30

Fig. 2.

Circular dichroism spectra for assay pairs using the long receptor Fe(env5.1)₃ (A, B) and the T20-binding site Fe(env6.0)₃ (C). The open triangles (▵) represent the sum of the C-peptide and Fe(env#)₃ spectra, the expected spectrum of the mixture if the peptides did not interact. All spectra were obtained at 20 μM peptide in a buffer at pH 7.

Fluorescence Studies of 6HB Formation

Dissociation constants (K_D) between peptide pairs were determined by direct titration of the Fe(env#)₃ receptor into a constant concentration of fluorescently labeled probe peptide.^13,15,29 Peptides were labeled with two different fluorescent probes, LY and FL. Figure 3 demonstrates fluorescence quenching for several receptor/peptide pairs, and the fit to an equilibrium 1:1 binding model. The fluorescence quenching experiment with Fe(env5.1)₃ confirmed that it aggregated above 0.1 μM, with the concomitant loss in binding sites reflected in an increase in fluorescence at receptor concentrations exceeding 0.1 μM. The K_D's for all assay pairs are reported in Table 4. The assay pair Fe(env3.0)₃/C27-e3.0 (K_D=1.2±0.2 μM) has been discussed elsewhere¹⁴ and is not evaluated further here.

Fig. 3.

Fluorescence-detected binding curves for env#/Cn-e# pairs depicted in Figure 1. Receptors used were env5.0 (black), env5.1 (red, left), or env6.0 (blue, right) and C-peptides used were C32-e5.0FL or C39-e5.1FL (•), C28-e5.1LY (▴) or C21-e6.0LY (○). Solid lines indicate fitting to a 1:1 binding model; the dashed line is a fit to IC₅₀. LY, lucifer yellow; FL, fluorescein. Color images available online at www.liebertpub.com/adt

Table 4.

Dissociation Constants (in μM) of Receptor–Probe Pairs Used in Fluorescence Assays

Probe	\documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes {10} {9} {7} {6} \begin{document}$${\rm Fe}({\rm env}2.0)_3^{2+}$$\end{document}	\documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes {10} {9} {7} {6} \begin{document}$${\rm Fe}({\rm env}5.0)_3^{2+}$$\end{document}	\documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes {10} {9} {7} {6} \begin{document}$${\rm Fe}({\rm env}5.1)_3^{2+}$$\end{document}	\documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes {10} {9} {7} {6} \begin{document}$${\rm Fe}({\rm env}6.0)_3^{2+}$$\end{document}
C18-e2.0LY	0.8±0.2
C18-e2.0FL	0.6±0.2
C32-e5.0FL		0.004±0.002
C39-e5.1FL		0.003±0.002	∼0.013
C28-e5.1LY		0.23±0.02	0.09±0.2	0.10±0.02
C28-e5.1FL		0.18±0.05	0.02±0.004	0.03±0.01
C21-e6.0LY		2.2±0.33	1.1±0.1	1.1±0.2
C21-e6.0FL		∼2.5^a	0.09±0.02	0.33±0.08

Receptor peptides are named env followed by a version number, probe peptides are labeled Cn followed by a version number. Experiments were repeated in quadruplicate and on at least two different days for each assay pair. The standard deviation is shown. Empty cells correspond to noncompatible pairs.

^aEstimated from the IC₅₀ (see text).

n, number of residues.

Inhibition constants were determined for unlabeled peptides, with the expectation that the observed K_I for a peptide inhibitor should equal the K_D of that peptide as a probe, barring any interference from the fluorescent group. This was found to be the case with one exception (see below). K_I's are reported in Table 5. Fe(env5.1)₃ was not generally used as a receptor in inhibition experiments due to aggregation at the required concentrations. In the typical experiment, a probe and an inhibitor were premixed before adding a receptor, although in moderate affinity assays (K_D ∼ 1 μM), rapid equilibrium was also achieved by adding the inhibitor to the premixed receptor and probe. K_D and K_I data enabled us to determine the energetic contribution of various subsites to 6HB formation, stability, and inhibition. K_D, or K_I in self-inhibition, is expected to reflect both the enthalpic contribution to binding energy from the length and characteristics of the helix–helix interface and the entropic penalty associated with C-peptide helical content changes (Table 3).¹⁴ K_I in heterologous assays, on the other hand, could provide information about the contribution of a subsite in stabilizing the longer ectodomain.

Table 5.

Inhibition Constants (in μM) for C-Peptide and Small Molecule Inhibitors Using Several Assay Pairs

					Fe(env6.0)₃ C21-e6.0FL/LY
Inhibitor	Fe(env2.0)₃ C18-e2.0FL/LY	Fe(env5.0)₃ C32-e5.0FL	Fe(env5.0)₃ C28-e5.1FL/LY	Fe(env5.0)₃ C21-e6.0FL/LY	LY	FL
C18-e2.0	0.8 (0.2)	1.1 (0.3)
C28-e5.1		1.4 (0.2)	0.3 (0.1)	0.2 (0.1)	0.45 (0.15)
C21-e6.0		>500	>150	4.9 (0.5)	0.8 (0.1)	∼60
C21-e6.0 N656L		>500	>150	126 (11)	5.6 (1)	∼150
T20	1.4 (0.3)	0.3 (0.1)	0.2 (0.04)	0.05 (0.01)	0.1 (0.03)
3a (MW 251)	3 (1)	8.5 (1.6)	69 (14)	139 (24)	184 (24)
14c (MW 357)	2.3 (0.3)	22 (1.5)	112 (16)	nsb, IC₅₀=63 ^a	nsb, IC₅₀=281 ^a
14g (MW 487)	1.1 (0.2)	1.8 (0.5)	1.7 (0.3)	4.1 (0.6)	nsb, IC₅₀=55 ^a

Inhibition constants obtained by competitive inhibition using the indicated metal-chelated NHR peptide Env# and fluorescently labeled CHR-peptide probe. In brackets is the standard deviation. Except where indicated, both probes gave the same K_I. Empty cells indicate no or poor overlap of the peptide inhibitor with an assay probe.

^aNonspecific binding, IC₅₀ in italics, sharp rise (Hill coefficient >2.5).

The HP Plus About Five Residues Immediately Upstream was the Only Significant Hotspot Mediating the Helix–Helix Interaction

C32-e5.0 had 50-fold increased affinity over C28-e5.1 for Fe(env5.0)₃, highlighting the importance of the HP-binding region, in agreement with previous binding and mutagenesis studies.^31
–33 C21-e6.0, directed to the region of the NHR that readily mutates upon exposure of the intact virus to T20, had 10-fold reduced affinity compared to C28-e5.1, indicating that important interactions with the NHR involved the segment ₆₃₇NYTSLI ₆₄₂ (wild-type sequence, residues in bold interact in the NHR groove). Truncated T20 variants missing this segment demonstrated a dramatic loss of antiviral potency.³⁴ These residues were also essential for the 0.8 μM affinity of the HP-binding peptide C18-e2.0 (unpublished observation and reference³⁵), although they are not considered to be part of the pocket-binding domain.

Inhibition data in heterologous assays confirmed the expanded HP interaction with C18-e2.0 as the most significant hotspot mediating 6HB formation. Thus, C18-e2.0 was able to inhibit the much longer Fe(env5.0)₃–C32-e5.0FL interface at about the same K_I as for self-inhibition, while peptides corresponding to other segments (C28-e5.1, C21-e6.0) had significantly reduced ability to compete with C32-e5.0FL compared to inhibition of self.

Interactions Between NHR and CHR at the Base of the Coiled Coil were Energetically Significant, but Could not Inhibit 6HB Formation

The peptide pair Fe(env6.0)₃ and C21-e6.0 representing the base of the 6HB formed a complex with a 1 μM K_D/K_I, a reasonably strong association, involving both hydrophobic and polar interactions (Fig. 1). The NHR segment ₅₄₀QARQLL₅₄₅ provided for a two-fold increase in binding potency with peptides C21-e6.0LY or C28-e5.1LY, as could be seen by comparing K_D's obtained with Fe(env6.0)₃ or Fe(env5.1)₃ with those obtained with Fe(env5.0)₃. The viral fusion-inactivating N656L mutation on the CHR²⁶ reduced affinity of C21-e6.0 by a factor of 7 for receptor Fe(env6.0)₃, 7.6 for receptor Fe(env5.1)₃ (data not shown), and 25 for receptor Fe(env5.0)₃. Differences in the effect of the mutation appeared to be due to compensatory interactions of C21-e6.0 with ₅₄₀QARQLL₅₄₅.

The reduction of potency of C21-e6.0 in heterologous assays was particularly striking, since it was unable to displace either C32-e5.0FL or C28-e5.1FL/LY from Fe(env5.0)₃ at concentrations up to 200 μM. It displaced C28-e5.1FL/LY from Fe(env6.0)₃ with six-fold reduced K_I compared to inhibition of self, and was unable to displace C21-e6.0FL (Fig. 4). There is an overwhelming evidence from structural and mutagenesis studies that the NHR–CHR interaction spanned by Fe(env6.0)₃–C21-e6.0 forms part of the 6HB core, and plays a critical role in gp41 function.^22,27 Thus, while there is clear evidence for NHR–CHR stabilizing interactions in this region, our data and other studies³⁶ suggest they do not appear to play a central role in disrupting or preventing 6HB formation.

Fig. 4.

Dissociation and inhibition constant determination for the assay pair Fe(env6.0)₃/C21-e6.0. (A) Binding of receptor Fe(env6.0)₃ to 0.2 μM C21-e6.0LY (●) or 15 nM C21-e6.0FL (○). (B) Inhibition curves using 3 μM Fe(env6.0)₃ and 0.2 μM C21-e6.0LY (●) or 0.75 μM Fe(env6.0)₃ and 15 nM C21-e6.0FL (○). The inhibitor C21-e6.0 was mixed with probe peptide before adding the receptor.

The Base of the Coiled Coil Showed Conformational Variability and Flexibility

Effects including nonspecific hydrophobic interactions and deviations from 1:1 binding in constructs containing residues at the base of the coiled coil were suggestive of conformational flexibility in this region. The segment ₅₄₀QARQLL₅₄₅ was implicated in nonspecific binding of the hydrophobic C-terminal FL group, which interfered in assays with Fe(env5.1)₃ or Fe(env6.0)₃, resulting in an observed K_D of FL-labeled peptides three-fold lower than the K_D for the corresponding LY-labeled peptides (Fig. 4A; Table 4) or the K_I of unlabeled peptides (Table 5). No association with free FL was observed. It was further found that unlabeled C21-e6.0 could not displace C21-e6.0FL from Fe(env6.0)₃, while readily displacing C21-e6.0LY (Fig. 4B), or either probe peptide from the receptor Fe(env5.0)₃. Interestingly, FL-labeling resulted in a K_D closer to the 100 nM K_I observed for T20 binding to Fe(env6.0)₃. T20 showed mid-nM inhibition in all assays except Fe(env2.0)₃–C18FL, which was to be expected, since T20 does not contain HP-binding residues. The hydrophobic tail of T20 binds to the fusion-peptide proximal region, which is not included in env6.0. Therefore some nonspecific hydrophobic effect was occurring with T20 as well.

The nonspecific binding resulted in deviations from normal 1:1 binding with C21-e6.0FL and to a lesser extent C21-e6.0LY. Binding curves displayed sharp transitions with a Hill coefficient >1.5 (e.g., Fig. 4A). The fluorescent label appeared to be a principle factor in inducing the effect, since unambiguous 1:1 binding was observed in the inhibition data for C21-e6.0 (Fig. 4B). The effect was observed only for the 21-residue peptide, where nonspecific interactions could likely contribute a significant proportion of the affinity. Some type of distortion is necessary to explain the fluorophore interaction, since it was designed to point away from the NHR–CHR interface in a purported six-helix structure. The observed cooperative binding transition may be associated with conformational variation from the expected six-helix structure. A recent crystal structure of full length ectodomain, including the membrane proximal external region (MPER), indicated that splaying and loss of the NHR helical structure occurred in this region.²² It is possible that FL could partially mimic the MPER in inducing a structural change.

High-Affinity Small Molecule Binding was Specific to the HP and Suggested a Dependence of Potency on Off-Rate

Assay performance was evaluated for potency and specificity with several confirmed low to sub-μM HP-binding inhibitors.¹⁸ Three of these, 3a, 14c, and 14g, are listed in Table 5, together with K_i's determined in all assays. They were selected because they have similar affinities in the HP-binding assay Fe(env2.0)₃/C18-e2.0, but differ in their ability to impede fusion.¹⁸ IC₅₀s for fusion inhibition were 5, 8, and 0.8 μM, respectively, for 3a, 14c, and 14g. 14g (plus some similar unpublished derivatives) is the only verified sub-μM fusion inhibitor in this scaffold class and is also the only small molecule tested in our laboratory to efficiently block the extended Fe(env5.0)₃–C32-e5.0FL helical interaction. To investigate whether this property could contribute to its improved biological activity, the assay suite was used to investigate the mechanism.

3a and 14c were weak or nonspecific inhibitors in all assays other than the HP-binding assay Fe(env2.0)₃–C18-e2.0. The inhibition curves for 14g in Figure 5A demonstrated low μM inhibition of the interactions Fe(env2.0)₃–C18-e2.0, and Fe(env5.0)₃–C32-e5.0. Minimal inhibition in the Fe(env6.0)₃–C28-e5.1 assay implied that the molecule was located specifically within the confines of the HP.¹⁵ Additional nonspecific inhibition occurred at the base of the coiled coil, as evidenced by a sharp rise in signal in assays with Fe(env6.0)₃ and C21-e6.0FL/LY at high inhibitor concentrations. However, unexpectedly, 14g inhibited C28-e5.1 and C21-e6.0 binding to Fe(env5.0)₃, with an apparent 1:1 binding mechanism, contrary to the results obtained with Fe(env6.0)₃. Further experimentation revealed that 14g did not inhibit C28-e5.1 or C21-e6.0 binding to Fe(env5.1)₃ (except nonspecifically at high concentrations). Rather, it increased the amount of bound C-peptide as evidenced by a reduction in baseline fluorescence upon addition of up to 25 μM 14g (Fig. 5B). The increased availability of binding sites for the probe peptide is presumed due to reduction of aggregation of the receptor when 14g occluded the HP. At concentrations above 25 μM, the familiar nonspecific binding of 14g to the base of the 6HB occurred. For comparison, 1:1 binding of C21-e6.0 due to competitive inhibition of Fe(env5.1)₃–C21-e6.0LY is shown in Figure 5B.

Fig. 5.

(A) Inhibition curves for 14g in various assays. Receptors used were Fe(env2.0)₃ (green), Fe(env5.0)₃ (black), or Fe(env6.0)₃ (blue, pink). C-peptide probes were C18-e2.0FL (green circles), C32-e5.0FL (black circles), C28-e5.1LY (triangles), C21-e6.0LY/FL (open circles). Solid lines are a fit to a 1:1 binding model, providing the K_I. Dashed lines are a fit to the IC₅₀ and are colored blue and pink for probes C21-e6.0LY and C21-e6.0FL, respectively. The y-axis is scaled for a minimum fluorescence of 0. (B) Inhibition of the interaction of 1.8 μM Fe(env5.1)₃ and 0.2 μM C21-e6.0LY by C21-e6.0 (blue) and 14g (black). (C) Fluorescence intensity as a function of time after addition of C32-e5.0FL to a mixture of Fe(env5.0)₃ (75 nM) and 3a (6.25 μM, closed circles), 14c (6.25 μM, open circles), and 14g (3.13 μM, closed triangles). Results from dimethyl sulfoxide only controls have been subtracted from the data. Early data points (<2,000 s) are omitted due to artifacts of physical mixing. Color images available online at www.liebertpub.com/adt

The apparent inhibition by 14g of C28-e5.1LY and C21-e6.0LY binding to Fe(env5.0)₃ appeared to be spurious in that it did not represent competitive inhibition of the NHR–CHR interaction. The only difference between this receptor and Fe(env5.1)₃ is the presence in the latter of the additional segment ₅₄₀QARQLL₅₄₅, which has been associated with nonspecific binding of both fluorescently labeled C-peptide and 14g. Indeed, nonspecific binding of 14g to Fe(env5.0)₃ was not observed. It is likely that Fe(env5.0)₃ is not an ideal assay partner for C21-e6.0FL/LY and C28-e5.1FL/LY, despite quenching the fluorescence of these peptides upon binding. It appeared that the C-terminal heptad overhang of the probe peptides, and the metal binding group of the receptor (Fig. 1) generated a non-natural site for 14g binding.

Having ruled out multiple-site binding for 14g as a mechanism for its increased potency in the high-affinity assay Fe(env5.0)₃–C32-e5.0FL and in fusion inhibition, off-rate was examined as a possible contributor to potency. Off-rate can be slowed by a protein conformational change associated with ligand binding.³⁷ Order of addition assays were used to obtain a rough estimate of the on and off-rates for C32-e5.0 with the receptor Fe(env5.0)₃, yielding k_on=10⁵ M⁻¹ s⁻¹ and k_off=0.0004 s⁻¹, which is within an order of magnitude of off-rates determined for a slightly longer peptide C37 in situ³⁸ (Supplementary Fig. S1; Supplementary Data are available online at www.liebertpub.com/adt). The slow kinetics permitted us to evaluate the effect of a small molecule ligand bound to Fe(env5.0)₃ by premixing the receptor with a low concentration of the ligand, and observing the time to equilibrium after adding C32-e5.0FL. This experiment is shown in Figure 5C for the ligands 3a, 14c, and 14g, and suggested three to five times slower off-rate for 14g than for 3a and 14c (see also Supplementary Fig. S1). Thus, the increased potency of 14g over earlier derivatives 3a and 14c could be attributed to differences in binding kinetics.

Conclusions

This study demonstrated a suite of assays for probing binding and segmental activity of the coiled coil domain of HIV-1 gp41. The assays encompass HP or T20 subsite detection, as well as full-length 6HB coverage. Dissociation and inhibition constants for peptides and small molecules accurately reflected both affinity and specificity provided the CHR probe peptide had a complete cognate-binding site on the Fe(env#)₃ receptor. The assay Fe(env6.0)₃/C21-e6.0LY, simulating the T20-binding site where resistance mutations are common, is the newest addition to our repertoire for examining gp41. The 1 μM K_D indicates that it can be used to detect inhibitors in the low μM to mM range that have a unique binding site apart from the HP. Experiments here demonstrate that use of the probe peptide in its LY-labeled form will be preferential over FL-labeling.

Importantly, many of the biophysical results that were obtained matched very well with observed biological data. The assays revealed that the HP was the sole hotspot mediating the NHR–CHR interaction, conferring significant stability to the 6HB. The first five residues of T20, which are critical for antiviral activity, were also necessary for high-affinity binding to the coiled coil. Peptides containing a known inactivating mutation N656L displayed a dramatic loss of potency. Evidence for a break in the rigid helical structure toward the base of the coiled coil was obtained, both from the inability of a C-peptide from this region to disrupt 6HB formation and from indication of conformational flexibility upon probe binding. The assay representing the 6HB segment at the base of the coiled coil showed unique features compared to the other assays, including interference of the FL group in binding, resulting in a cooperative binding transition, inability of the unlabeled peptide to compete with its FL-labeled counterpart, and nonspecific binding of hydrophobic compounds. The overall agreement between known biological properties and observed binding properties of the C-peptides confirmed the accuracy of the metallopeptide receptor models in representing the intact gp41 NHR trimer.

Evaluation of a sub-μM small molecule fusion inhibitor revealed that it was able to effectively inhibit the high-affinity interaction between a 40-residue NHR construct and a 32-residue CHR peptide. Small molecules that were not able to inhibit this interaction in the low to sub-μM range were found to have ≥10-fold weaker activity against cell–cell fusion, even if they had similar ability to disrupt an 18-residue CHR peptide binding in the HP. This property is demonstrated here with just three inhibitors, but recent results (unpublished) have indicated that it applies more broadly and may be a unique identifying property of effective low molecular weight fusion inhibitors. The mechanism of extended NHR–CHR inhibition appears to be associated with lower off-rate of the small molecule ligand, with binding specificity restricted to the HP. The assay suite could be used to aid in the discovery of alternative ligands, which bind adjacent to the pocket and which could be tethered to HP-binding inhibitors, leading to a significant increase in potency.

Footnotes

Acknowledgments

This work was supported by NIH grants NS066469 and GM087998. I thank Dr. Richard Shafer, UCSF, for use of the CD machine and Dr. G. Zhou for providing the low molecular weight compounds tested in the assays. Molecular graphics images were produced using Chimera, developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco, with support from the NIH (National Center for Research Resources grant 2P41RR001081, NIGMS 9P41GM103311).

Disclosure Statement

No competing financial interests exist.

Abbreviations:

References

Teissier

, Penin

, Pecheur

. Targeting cell entry of enveloped viruses as an antiviral strategy. Molecules, 2011; 16:221–250.

Naider

, Anglister

. Peptides in the treatment of AIDS. Curr Opin Struct Biol, 2009; 19:473–482.

Jacobs

, Garg

, Viard

, Raviv

, Puri

, Blumenthal

. HIV-1 envelope glycoprotein-mediated fusion and pathogenesis: implications for therapy and vaccine development. Vaccine, 2008; 26:3026–3035.

Groot

, Welsch

, Sattentau

. Efficient HIV-1 transmission from macrophages to T cells across transient virological synapses. Blood, 2008; 111:4660–4663.

Eckert

, Kim

. Mechanisms of viral membrane fusion and its inhibition. Annual Rev Biochem, 2001; 70:777–810.

Wild

, Oas

, McDanal

, Bolognesi

, Matthews

. A synthetic peptide inhibitor of human immunodeficiency virus replication: correlation between solution structure and viral inhibition. Proc Natl Acad Sci USA, 1992; 89:10537–10541.

Jiang

, Lin

, Strick

, Neurath

. HIV-1 inhibition by a peptide. Nature, 1993; 365:113.

Dwyer

, Wilson

, Davison

, Freel

, Seedorff

, Wring

, Tvermoes

, Matthews

, Greenberg

, Delmedico

. Design of helical, oligomeric HIV-1 fusion inhibitor peptides with potent activity against enfuvirtide-resistant virus. Proc Natl Acad Sci USA, 2007; 104:12772–12777.

Manfredi

, Sabbatani

. A novel antiretroviral class (fusion inhibitors) in the management of HIV infection. Present features and future perspectives of enfuvirtide (T-20) Curr Med Chem, 2006; 13:2369–2384.

10.

Gochin

, Zhou

. Amphipathic properties of HIV-1 gp41 fusion inhibitors. Curr Topics Med Chem, 2011; 11:3022–3032.

11.

Cai

, Jiang

. Development of peptide and small-molecule HIV-1 fusion inhibitors that target gp41. Chem Med Chem, 2010; 5:1813–1824.

12.

Hajduk

. SAR by NMR: putting the pieces together. Mol Interv, 2006; 6:266–272.

13.

Cai

, Gochin

. A novel fluorescence intensity screening assay identifies new low molecular weight inhibitors of the gp41 coiled coil domain of HIV-1. Antimicrob Agents Chemother, 2007; 51:2388–2395.

14.

Gochin

, Cai

. The role of amphiphilicity and negative charge in glycoprotein 41 interactions in the hydrophobic pocket. J Med Chem, 2009; 52:4338–4344.

15.

Cai

, Balogh

, Gochin

. Stable extended human immunodeficiency virus type 1 gp41 coiled coil as an effective target in an assay for high-affinity fusion inhibitors. Antimicrob Agents Chemother, 2009; 53:2444–2449.

16.

McGillick

, Balius

, Mukherjee

, Rizzo

. Origins of resistance to the HIVgp41 viral entry inhibitor T20. Biochemistry, 2010; 49:3575–3592.

17.

Liu

, Jing

, Cheung

, Lu

, Sun

, Yan

, Niu

, Farmar

, Wu

, Jiang

. HIV gp41 C-terminal heptad repeat contains multifunctional domains. Relation to mechanisms of action of anti-HIV peptides. J Biol Chem, 2007; 282:9612–9620.

18.

Zhou

, Wu

, Snyder

, Ptak

, Kaur

, Gochin

. Development of indole compounds as small molecule inhibitors of HIV-1 gp41. J Med Chem, 2011; 54:7220–7231.

19.

Oishi

, Ito

, Nishikawa

, Watanabe

, Tanaka

, Ohno

, Izumi

, Sakagami

, Kodama

, Matsuoka

, Fujii

. Design of a novel HIV-1 fusion inhibitor that displays a minimal interface for binding affinity. J Med Chem, 2008; 51:388–391.

20.

Otaka

, Nakamura

, Daisuke

, Kodama

, Uchiyama

, Nakamura

, Kobayashi

, Matsuoka

, Fujii

. Remodelling of gp41-C34 peptide leads to highly effective inhibitors of the fusion of HIV-1 with target cells. Angew Chem Int Ed, 2002; 41:2938–2939.

21.

Cohen

, Parry

. Alpha-helical coiled coils and bundles: how to design an alpha-helical protein. Proteins, 1990; 7:1–15.

22.

Buzon

, Natrajan

, Schibli

, Campelo

, Kozlov

, Weissenhorn

. Crystal structure of HIV-1 gp41 including both fusion peptide and membrane proximal external regions. PLoS Pathog, 2010; 6:e1000880.

23.

Greenberg

, Cammack

. Resistance to enfuvirtide, the first HIV fusion inhibitor. J Antimicrob Chemother, 2004; 54:333–340.

24.

Armand-Ugon

, Gutierrez

, Clotet

, Este

. HIV-1 resistance to the gp41-dependent fusion inhibitor C-34. Antiviral Res, 2003; 59:137–142.

25.

Rimsky

, Shugars

, Matthews

. Determinants of human immunodeficiency virus type 1 resistance to gp41-derived inhibitory peptides. J Virol, 1998; 72:986–993.

26.

Cao

, Bergeron

, Helseth

, Thali

, Repke

, Sodroski

. Effects of amino acid changes in the extracellular domain of the human immunodeficiency virus type 1 gp41 envelope glycoprotein. J Virol, 1993; 67:2747–2755.

27.

Chan

, Fass

, Berger

, Kim

. Core structure of gp41 from the HIV envelope glycoprotein. Cell, 1997; 89:263–273.

28.

Sackett

, Wexler-Cohen

, Shai

. Characterization of the HIV N-terminal fusion peptide-containing region in context of key gp41 fusion conformations. J Biol Chem, 2006; 281:21755–21762.

29.

Gochin

, Guy

, Case

. A metallopeptide assembly of the HIV-1 gp41 coiled coil is an ideal receptor in fluorescence detection of ligand binding. Angew Chem Int Ed, 2003; 42:5325–5328.

30.

Wild

, Greenwell

, Shugars

, Rimsky-Clarke

, Matthews

. The inhibitory activity of an HIV type 1 peptide correlates with its ability to interact with a leucine zipper structure. AIDS Res Hum Retroviruses, 1995; 11:323–325.

31.

, Konstantinidis

, Stewart

, Dekhtyar

, Ng

, Swift

, Matayoshi

, Kati

, Kohlbrenner

, Molla

. Conserved residues in the coiled-coil pocket of human immunodeficiency virus type 1 gp41 are essential for viral replication and interhelical interaction. Virology, 2004; 329:319–327.

32.

Chan

, Chutkowski

, Kim

. Evidence that a prominent cavity in the coiled coil of HIV type 1 gp41 is an attractive drug target. Proc Natl Acad Sci USA, 1998; 95:15613–15617.

33.

, Liu

, Jing

, Lu

, Cai

, Chin

, Debnath

, Kirchhoff

, Jiang

. Conserved residue Lys574 in the cavity of HIV-1 Gp41 coiled-coil domain is critical for six-helix bundle stability and virus entry. J Biol Chem, 2007; 282:25631–25639.

34.

Wild

, Shugars

, Greenwell

, McDanal

, Matthews

. Peptides corresponding to a predictive alpha-helical domain of human immunodeficiency virus type 1 gp41 are potent inhibitors of virus infection. Proc Natl Acad Sci USA, 1994; 91:9770–9774.

35.

Cole

, Garsky

. Thermodynamics of peptide inhibitor binding to HIV-1 gp41. Biochemistry, 2001; 40:5633–5641.

36.

Liu

, Lu

, Niu

, Xu

, Wu

, Jiang

. Different from the HIV fusion inhibitor C34, the anti-HIV drug Fuzeon (T-20) inhibits HIV-1 entry by targeting multiple sites in gp41 and gp120. J Biol Chem, 2005; 280:11259–11273.

37.

Cai

, Zhou

. Theory and simulation on the kinetics of protein-ligand binding coupled to conformational change. J Chem Phys, 2011; 134:105101.

38.

Kahle

, Steger

, Root

. Asymmetric deactivation of HIV-1 gp41 following fusion inhibitor binding. PLoS Pathog, 2009; 5:e1000674.

39.

Caffrey

. Model for the structure of the HIV-1 gp41 ectodomain: insight into the intermolecular interactions of the gp41 loop. Biochim Biophys Acta, 2001; 1536:116–122.