Single-Molecule Analysis of Human Immunodeficiency Virus Type 1 gp120-Receptor Interactions in Living Cells

A quantitative description of the binding interactions betweenhuman immunodeficiency virus (HIV) type 1 envelope glycoproteinsand their host cell surface receptors remains incomplete. Here,we introduce a single-molecule analysis that directly probesthe binding interactions between an individual viral subunitgp120 and a single receptor CD4 and/or chemokine coreceptorCCR5 in living cells. This analysis differentiates single-moleculebinding from multimolecule avidity and shows that, while thepresence of CD4 is required for gp120 binding to CCR5, the forcerequired to rupture a single gp120-coreceptor bond is significantlyhigher and its lifetime is much longer than those of a singlegp120-receptor bond. The lifetimes of these bonds are themselvesshorter than those of the P-selectin/PSGL-1 bond involved inleukocyte attachment to the endothelium bonds during an inflammationresponse. These results suggest an amended model of HIV entryin which, immediately after the association of gp120 to itsreceptor, gp120 seeks its coreceptor to rapidly form a new bond.This "bond transfer" occurs only if CCR5 is in close proximityto CD4 and CD4 is still attached to gp120. The analysis presentedhere may serve as a general framework to study mechanisms ofreceptor-mediated interactions between viral envelope proteinsand host cell receptors at the single-molecule level in livingcells.

INTRODUCTION

Top

Introduction

Current antivirial therapies for treatment of human immunodeficiencyvirus type I (HIV-1) infection have greatly reduced the mortalityand morbidity associated with HIV/AIDS. These treatments consistprimarily of protease inhibitors that prevent viral maturationand reverse transcriptase inhibitors that block synthesis ofproviral DNA. However, because of drug resistance (12), complementarystrategies targeting virus-cell interactions before HIV-1 entryare currently under intense examination (8). While a structure-basedmodel of interaction between HIV-1 envelope glycoproteins andhost cell receptors is relatively well established (6, 22),a quantitative description of how linkages between cell andvirus are formed to generate a functional junction remains incomplete.A quantitative description of the binding interactions betweenHIV-1 and cell surfaces is needed to further elucidate the molecularmechanisms of viral entry as well as potentially improve currentHIV-1 entry inhibitors (8, 30, 31, 36).

The current qualitative picture of HIV-1 entry consists of sequentialpairwise molecular interactions. HIV-1 targets primarily T lymphocytesand macrophages by initiating sequential binding interactionsbetween CD4 receptors on these host cells and a surface subunitof the HIV-1 exterior envelope (Env) glycoprotein, gp120 (5).Binding of gp120 to CD4 promotes a conformational change ingp120, which mediates its binding to a cellular chemokine coreceptor:CCR5 for R5-tropic viruses (26, 33) and CXCR4 for X4-tropicviruses (9). Here, we focus on CCR5, which is also targetedby drug developers as the most promising site because individualswho naturally lack CCR5 due to the ${Delta}$ 32 polymorphism are highlyresistant to HIV infection, with no obvious immune deficits(23, 35). The CXCR4 coreceptor does not become involved untiladvanced stages of the infection (7, 37); CXCR4 use occurs inabout half of individuals who progress to AIDS. Binding of gp120to its coreceptor triggers a further conformational change ofgp120, which exposes a previously buried hydrophobic fusiondomain located at the N terminus of the transmembrane subunitof Env, gp41. Finally, binding of gp41 to the host cell surfacepromotes fusion between viral and cell lipid membranes, whichmediates the entry of the genome-containing viral protein coreinto the host cell's cytoplasm to initiate the infection process(40, 43).

Crystallographic studies of gp120-receptor complexes have providedinvaluable details about the structure of the HIV-1/cell bindinginterface at the atomic level (22, 33, 41, 42), but the actualdetails of the binding kinetics for virus-cell attachment remainpoorly defined for living cells and at the single-molecule level.All kinetic measurements thus far have used of bulk methodsand do not provide any information about the micromechanicsof individual gp120-receptor bonds. For instance, the forcerequired to break a single gp120-CD4 bond, its adhesion force,is unknown. Whether coreceptor CCR5 mechanically reinforcesthe primary gp120-CD4 bond is also unknown. Here, we used single-moleculeforce spectroscopy to probe quantitatively the kinetics andmicromechanical properties of the principal bonds involved inthe binding interactions between HIV-1 and its host cell. Thissingle-molecule analysis does not require the use of solubleproteins or protein labeling, has exquisite sensitivity, accommodatesliving cells, and can unambiguously distinguish single-moleculebinding events from multiple-molecule adhesion (i.e., avidity).

Using single-molecule analysis, we probed directly the bindinginteractions between an individual recombinant HIV-1 envelopeglycoprotein subunit gp120 and an individual receptor CD4 inthe presence and absence of the coreceptor chemokine CCR5 inliving cells. We find that, while the presence of CD4 is indeedrequired for CCR5 to bind to gp120, the force required to rupturethe gp120-CCR5/CD4 bond and its lifetime are significantly higherthan those for the gp120-CD4 bond. Together these results suggesta revised model for the initial binding interactions betweenHIV-1 and the host plasma membrane.

MATERIALS AND METHODS

Top

Materials and Methods

Cell culture and reagents. GHOST cell lines (developed by V. KewalRamani and D. Littman)were grown in Dulbecco's modified Eagle's medium (American TypeCulture Collection, Manassas, VA) supplemented with 10% fetalcalf serum (American Type Culture Collection), 500 µg/mlG418 (Sigma, St. Louis, MO), 100 µg/ml penicillin-streptomycin(Sigma), 100 µg/ml hygromycin (Sigma), and, for the coreceptorencoding cell line, 1 µg/ml puromycin. HOS cell lines(developed by N. Landau) were grown in Dulbecco's modified Eagle'smedium with 10% fetal calf serum and 1.0 µg/ml puromycin.These cell lines were expanded on a 2- to 3-day cycle in a humidified5% CO₂-95% air incubator maintained at 37°C. For molecularforce probe (MFP) experiments, during the passage of cells,200 µl of a cell suspension of 1 x 10⁶ cells/ml was addedto each 60-mm tissue culture dish containing 5 ml of culturemedium and was incubated overnight with 5% CO₂ at 37°C toallow cell spreading and restoration of normal cell morphology.Immediately before the MFP experiment, the medium was changedto serum-free medium containing HEPES (Invitrogen, Carlsbad,CA) to stabilize the pH while outside the incubator environment.

The cell surface CD4 complex monoclonal B4 has preferentialrecognition for cell surface CD4 over recombinant soluble CD4(sCD4). It exerts apparent neutralizing activity against HIVby blocking access to the CD4 cell surface complex. Recombinanthuman sCD4 is the full-length extracellular domain of humanCD4 (amino acids 1 to 370). This protein is reactive with HIV-1gp120 and anti-CD4 monoclonal and polyclonal antibodies. Itsglycosylation pattern is identical to that of the natural humanprotein. Recombinant HIV-1 BaL gp120 (molecular mass, 116 kDa)is the Env protein subunit of the macrophage-tropic virus thatbinds to CD4. This protein was prepared from HEK293 cells andpurified by immunoaffinity chromatography using a monoclonalantibody specific for the conformational epitope of gp120. TheGHOST/HOS cell lines, cell surface CD4 complex monoclonal B4,recombinant soluble CD4, and BaL gp120 were obtained throughthe NIH AIDS Research and Reference Reagent Program (NIAID).

Attachment of gp120 onto the cantilever tip. The cantilever tip (Veeco Instruments, Santa Barbara, CA) wasfirst incubated in 1.0 mg/ml bovine serum albumin (BSA) (Sigma)overnight at 4°C. The cantilevers were then rinsed threetimes with phosphate buffer. BaL gp120 was then chemically cross-linkedat room temperature to the cantilever tip with the membrane-impermeativeagent bis(sulfosuccinimidyl)suberate (Pierce, Rockford, IL).Afterwards, the cantilevers were again washed three times withphosphate buffer and incubated overnight with 1.0 mg/ml BSAto block nonspecific binding. To verify that the functionalityof gp120 was preserved in the cross-linking attachment, thetips were incubated with 1.0 mg/ml sCD4 with 5% CO₂ at 37°C.The cantilever tip with the cross-linked protein was then allowedto interact with cells plated on the bottom of a tissue culturedish. Upon retraction, no binding between the cell surface andthe cantilever tip occurred (see more about this control experimentbelow).

MFP measurements. Single-molecule experiments were conducted using an MFP (AsylumResearch, Santa Barbara, CA) (Fig. 1A). This instrument measuresforces, with subpiconewton force resolution and millisecondtime resolution, as a function of cantilever-sample separationdistance, which is measured with subnanometer resolution. TheMFP uses technology similar to the more conventional atomicforce microscope by making use of a flexible cantilever tipthat deflects in response to forces from between tip and cellsurface.

View larger version (29K):
[in this window]
[in a new window]
FIG. 1. Schematic of the MFP used to measure binding interactions between individual HIV-1 gp120 molecules and receptors in a living host cell. (A) Schematic of the molecular force probe used in the measurements of gp120-receptor binding interactions. (B) A low density of recombinant BaL gp120 molecules are cross-linked to a cantilever, which is gently brought into contact with a cell. The cell is engineered to express either CD4 alone, coreceptor/chemokine CCR5 alone, or both CD4 and CCR5. The time of contact between cell and cantilever is kept to a minimum (<1 ms) to ensure mostly single-molecule interactions. The cantilever is subsequently pulled with a controlled reproach velocity. The time-dependent force applied on the bond between a gp120 molecule and a single cell receptor and the time-dependent deformation of the cell membrane-bound proteins are recorded simultaneously. The gp120-functionalized cantilever is shown in black when a gp120-receptor bond is subjected to a pulling force before rupture and in light gray after bond rupture.

Two outputs are generated from MFP experiments: the photodetectorsensor output (in volts) and the linear variable differentialtransformer (LVDT) output. The photodetector sensor output istranslated into cantilever deflection (in nanometers) by multiplyingthe sensor output by the inverse optical lever sensitivity,which is the inverse slope of the sensor output versus the LVDToutput curve in the constant-compliance regimen. The LVDT outputis translated into the tip-sample separation distance by takingthe difference between the total cantilever movement and thedisplacement due to the applied force. The force measured fromthe cantilever deflection is calculated based on Hooke's lawfor linear elastic springs, F = k ${Delta}$ x, where F is the force (expressedin piconewtons), k is the bending constant of the cantilever(in piconewtons per micrometer), and ${Delta}$ x is the deflection (innanometers). The bending constant of the cantilever tip used,which in this case is the largest and softest triangular cantilever(nominal spring constant of 10 pN/µm), was measured bythe nondestructive thermal oscillation method before each newexperiment. The time-dependent deflection of the cantileverwas determined by laser deflection onto a photodetector andoccurred at a rate of 1.0 kHz. Each cantilever was calibratedby the nondestructive thermal oscillation method before use(19); the cantilever had a mean bending constant of 10 pN/nm.

A 60-mm tissue culture dish containing receptor-expressing cellsat 60% confluence was placed on the MFP stage. The protein-tetheredcantilever was positioned directly above a single cell. Thedwell time between the cantilever tip and the cell was minimizedto reduce the occurrence of multiple bond rupture events. Asthe cantilever retracted at a preset velocity, the force wasrecorded as a function of vertical displacement. Igor Pro 4.09software (Wavemetrics, Inc., Lake Oswego, OR) was used to recordrupture forces and loading rates from the force-versus-distanceplots collected during experimentation. Rupture forces werebinned corresponding to increments of 100 pN/s for loading ratesof between 100 and 1,000 pN/s and 1,000 pN/s for loading ratesof between 1,000 and 10,000 pN/s. The loading rate applied toprotein-cell receptor interaction is equal to the product ofthe slope of the individual time-dependent force profiles (inpiconewtons per micrometer) prior to rupture and the reproachvelocity (in micrometers per second), also called retractionvelocity.

RESULTS

Top

Results

Single gp120-CD4 bond force spectroscopy in living cells. We used a single-molecule force spectroscopy assay to characterizethe binding interactions between a single recombinant HIV-1envelope glycoprotein BaL gp120 and its major receptor CD4 andcoreceptor CCR5 on a living host cell. This assay measures rigorouslykinetics and micromechanical parameters at the molecular level,thereby providing a wealth of biochemical and biophysical informationabout gp120-mediated bonds, which is not provided by traditionalbinding assays. Specifically, our assay measures the ruptureforce, dissociation rate, tensile strength, reactive compliance,and lifetime of the bond made by a single HIV-1 envelope glycoproteinsubunit and individual receptors in living cells.

This assay uses an MFP (15), which consists essentially of aflexible triangular cantilever with tethered recombinant gp120distributed at low density (Fig. 1). The cantilever is positionedover and gently brought into contact with a single living hostcell and then pulled at a prescribed reproach velocity (Fig.1A). The cell surface displays either the primary cell receptorfor HIV-1 binding, CD4 (CD4⁺ GHOST parental cell line), thecoreceptor/chemokine CCR5 (CD4^– HOS cell line), or bothCD4 and CCR5 (GHOST Hi-5 cell line). The progressive deflectionof the cantilever was recorded during the approach and reproachof the calibrated cantilever as a function of its vertical displacementuntil bond breakage(s) occurred (Fig. 2). These deflectionswere translated into force-displacement spectra such as thoseillustrated in Fig. 2. The spectrum was flat (constant force)when no rupture event occurred and displayed one (typically)or two (rarely) rupture events when bonds ruptured.

View larger version (29K):
[in this window]
[in a new window]
FIG. 2. Force-distance spectra for gp120-CD4 interactions in living cells. Typical profiles of the time-dependent force acting on a gp120-CD4 bond under tension as a function of the time-dependent deformation of that bond are shown. From these force-distance spectra, the k_off rates of the bonds between HIV-1 envelope proteins and host cell receptors can be readily extracted. The probability of binding events between gp120 and its cell receptors depends on the contact time between cantilever and cell surface, the density of gp120 on the cantilever, and the approach force used to bring the cantilever in contact with the cell. These parameters are controlled to ensure mostly single-molecule interactions, as shown by single arrows in profiles 6 and 10 (numbered from the top) to indicate single-bond breakage. The two arrows in profile 1 show a rare case of double-bond breakage. Rupture force (in piconewtons) and loading rate (in piconewtons per second) are extracted from the height of the peak (in piconewtons) at the rupture point and from the product of the slope right before the peak (in piconewtons per micrometer) and the reproach velocity (in micrometers per second), respectively. Here, the reproach velocity was 15 µm/s. Experiments were conducted using CD4⁺ GHOST cells.

For each cell receptor assayed (CD4 and CCR5) and at each testedreproach velocity (5 to 45 µm/s), hundreds of force-displacementspectra were collected. From these force-displacement spectra,large data sets of rupture forces (i.e., the heights of thediscrete force step at rupture [Fig. 2]) and loading rates wereextracted and pooled into histograms (see Fig. 4). As shownbelow, this molecular-binding assay is specific, it allows usto report kinetic and biophysical properties of single-moleculeinteractions, and it leads to the direct measurement of ratesof dissociation of a single gp120 molecule and individual CD4/coreceptorsexpressed in living cells at equilibrium (no applied force)and as a function of the applied force.

View larger version (21K):
[in this window]
[in a new window]
FIG. 4. Single-molecule analysis of gp120-CD4 binding interactions. (A) Rupture force (in piconewtons) required to break a single gp120-CD4 bond as a function of the loading rate (in piconewtons per second) to which the bond is subjected. The unstressed dissociation rate k⁰_off and the reactive compliance x_ß of the bond are calculated from this curve through Bell's model analysis (see text). This curve features a break at a characteristic loading rate of $~$ 350 pN/s, which separates a regimen where the bond lifetime is long at low loading rates from a regimen where the bond lifetime is shorter at high loading rates (see text). (B) Distributions of rupture forces required to break a single gp120-CD4 bond at different reproach velocities. (C) Schematic of the intermolecular potential of the gp120-CD4 interaction qualitatively based on data shown in panel A. Red represents the inner barrier potential of the gp120-CD4 interaction at high loading rates; blue represents the outer barrier potential at low loading rates. The width of each potential well is taken as the reactive compliance at low and high loading rates. Experiments were conducted using CD4⁺ GHOST cells.

Distinguishing single-molecule interactions from multiple-molecule interactions. For long contact times between cell and cantilever and for highdensities of gp120 molecules on the cantilever, gp120 moleculesformed multiple bonds with the surface of the host cell. However,we were able to promote single gp120-receptor bond formationas opposed to multiple bonds by taking the following measures.(i) Cantilever and cell were allowed to interact for only ashort time (<1 millisecond) to target the percentage of successfulcell-cell interactions to be 10 to 30%. Based on Poisson distributionstatistics, when 30% of cell-cantilever contacts result in gp120-CD4binding events, >80% of these binding events involve a singlebond, 15% involve double bonds, and <3% involve triple bonds(15, 16). Figure 2 illustrates force-displacement spectra whenno binding event occurred (spectra 2 to 5 and 7 to 9 [numberedfrom the top]) and when distinct bond rupture events (spectra1, 6, and 10) occurred. (ii) The force applied on the cell bythe cantilever upon initial contact was lowered to reduce thesurface area of contact between cantilever and cell surface(4, 34) and therefore reduce the probability of forming multiplebonds. (iii) The density of gp120 on the cantilever was madelow by using short incubation times during cantilever preparation(see Materials and Methods).

We note that simple visual inspection of the force-displacementspectra readily distinguishes single-molecule interactions from(rare) multiple-molecule interactions. The former involve asingle bond rupture event (marked by an abrupt force) (see,e.g., force-displacement spectra 6 and 10 in Fig. 2), whereasthe latter involve multiple bond rupture events (spectrum 1in Fig. 2). We also verified that the rupture force histogramobtained at each reproach velocity (see Fig. 4B) displayed aunique well-defined peak as opposed to multiple quantized peaksthat would occur if rupture force measurements for multiplebonds were inadvertently included in the data sets (4). Togetherthese precautions ensured that the adhesive interactions betweencell and cantilever recorded in our assay involved one bond,that nonspecific interactions between cantilever and cell wereinsignificant, and that only parameters characterizing singlegp120-receptor binding interactions are reported here.

Single-molecule force spectroscopy measurements are specific. We verified that the gp120-cell interactions probed by our instrumentwere specific to CD4 by treating the cells with saturating amountsof an anti-CD4 function-blocking monoclonal antibody. We foundthat this cell treatment completely abrogated interactions betweengp120 and CD4⁺ cells (n = 300 and 14 cells). A typical force-displacementspectrum for this case is shown in Fig. 3 (compare spectra 1and 2 [numbered from the top]). We also found that cantileverswithout cross-linked gp120 molecules and blocked with BSA elicitedno bond rupture events (n = 310 and 10 cells) (see illustrativeflat spectrum 3 in Fig. 3). Finally, we verified that, throughcompetitive binding, the presence of saturating concentrationsof sCD4 abrogated the binding interactions between gp120 moleculesand CD4 (n = 278 for 12 cells) (see flat spectrum 4 in Fig.3). Together, these controls ensure that the reported biochemicaland biophysical parameters are specific to gp120-CD4 bonds.

View larger version (29K):
[in this window]
[in a new window]
FIG. 3. The interactions between gp120-coated cantilevers and cell receptors are specific. Typical force-distance spectra for the binding association between individual gp120 and CD4 molecules in the absence (first spectrum) and presence (second spectrum) of an anti-CD4 function-blocking monoclonal antibody (mAb), for a cantilever without cross-linked gp120 (third spectrum), and for CD4⁺ cells in the presence of sCD4 (fourth spectrum) are shown. Experiments were conducted using CD4⁺ GHOST cells.

The kinetic and micromechanical properties of a single gp120-CD4 bond. Our assay measures kinetics and mechanical parameters, includingthe dissociation rate, tensile strength, and reactive compliance,which characterize the dynamics and micromechanics of a bondbetween gp120 and its host cell receptor and coreceptor expressedon the surface of living cells. These properties were evaluatedusing Bell’s model for the three main molecular pairs(gp120-CD4, gp120-CCR5, and gp120-CCR5/CD4) known to mediateearly R5 HIV-1 entry (3, 11). The Bell model relates the ruptureforce required to break a bond between two molecules to theapplied loading rate. The same model has been successfully usedby our group and others to characterize the binding interactionsbetween ligands and the surface of living cells (e.g., selectin/PSGL-1interaction) (1, 10, 15). The mean rupture force, f_m, is plottedas a function of the natural log of the loading rate, r_f, (Fig.4A):

(1)
where k_B = 1.38 x 10^–23J/K is the Boltzmann constant and T = 310 K is the absolutetemperature of the cell culture medium. By fitting the datawith equation 1, we extracted the unstressed dissociation rateconstant, k⁰_off, and the reactive compliance, x_ß,of the gp120-CD4 bond. We note that while MFP measurements werecollected under a nonzero loading rate, data extrapolation toa zero loading rate yields an unstressed (equilibrium) dissociationrate.

The experimental curve f_m versus r_f shows that the rupture forcerequired to break a single gp120-CD4 bond does indeed grow linearlywith the natural logarithm of the loading rate applied to thatbond. However, this force-loading rate curve had a remarkableprofile: the rupture force did not grow monotonically with loadingrate. Instead, the rupture force grew with two different slopesbefore and after a characteristic loading rate of $~$ 350 pN/s (Fig.4A) (see more in Discussion). For comparison, the rupture forcefor the selectin/PSGL-1 bond grows uniformly over the same rangeof loading rates (15, 16).

Our force-spectroscopy measurements show that a gp120-CD4 bondcould withstand mean forces of up to 26 pN before rupture, fora loading rate of 200 pN/s (Fig. 4A). The fitted f_m versus r_fcurve using equation 1 yielded unstressed dissociation rateconstants of 4.1 s^–1 at low loading rates (<350 pN/s)and 5.3 s^–1 at high loading rates (>350 pN/s), correspondingto bond lifetimes of 0.24 second and 0.19 second, respectively.This fitted data also determined that the reactive complianceof the gp120-CD4 bond was 0.14 nm at low loading rates and 0.11nm at high loading rates. Following Jarzynski's reconstructionof an equilibrium free energy from single-molecule pulling measurements(18), the presence of two distinct slopes in the rupture forcesignifies that the free energy of interaction between gp120and CD4 features two energy wells (Fig. 4C) (see more in Discussion).

The tensile strength, lifetime, and reactive compliance of the gp120-CCR5/CD4 bond are higher than those of the gp120-CD4 bond. Using living cells that expressed the coreceptor/chemokine CCR5alone (no CD4), we applied the single-molecule spectroscopymethod and the Bell model analysis described above to extractthe tensile strength, lifetime, and reactive compliance of thegp120-CCR5/CD4 bond. Importantly, we detected no bond ruptureevents between gp120 and CCR5 in the absence of sCD4 (Fig. 5).Therefore, in agreement with results from traditional bindingassays, our assay suggests that gp120 and CCR5 cannot bind directlyin the absence of CD4 in GHOST cells.

View larger version (41K):
[in this window]
[in a new window]
FIG. 5. Single-molecule analysis of gp120-CCR5 and gp120-CCR5/CD4 bonds. (A) Typical force-distance spectra for the gp120-CCR5 bond in the presence of sCD4, gp120-CCR5/CD4, and gp120-CCR5 in the absence of sCD4. (B) Distributions of rupture forces to break a single bond between a single gp120 molecule and the protein complex CCR5/sCD4 at different reproach velocities. (C) Rupture force for single gp120-CCR5 bonds (circles) in the presence of sCD4 and single gp120-CCR5/CD4 bonds (squares) as a function of the loading rate applied to those bonds. Here again, characteristic ruptures occur in the force-loading rate curves, at threshold loading rates of $~$ 450 pN/s for the gp120-CCR5/sCD4 complex and $~$ 650 pN/s for the gp120-CCR5/CD4 complex. All experiments were conducted using CD4⁺ CCR5⁺ GHOST Hi-5 cells.

In contrast, in the presence of sCD4, we found that a significantlystronger interaction mediated the binding between gp120 andCCR5 compared to that between gp120 and CD4. Like the gp120-CD4bond, the gp120-CCR5/sCD4 bond had a adhesion force that grewlinearly with the natural log of the loading rate, at two differentslopes before and after a higher characteristic loading rateof $~$ 450 pN/s. The dissociation rate of the gp120-CCR5/sCD4 bondwas significantly lower than that of the gp120-CD4 bond, i.e.,1.3 s^–1 in the low-loading-rate regimen (<450 pN/s)and 2.1 s^–1 in the high-loading-rate regimen (>450pN/s), corresponding to bond lifetimes of 0.77 second and 0.48second, respectively. The gp120-CCR5/CD4 bond could withstandmean forces of up to 29 pN before rupture, for a loading rateof 200 pN/s. Therefore, the force required to break a gp120-CCR5/sCD4bond was similar to that required to break the bond betweengp120 and CD4. Nevertheless, the gp120-CCR5/sCD4 bond was lessdynamic and lasted significantly longer than the gp120-CD4 bond(1/k⁰_off = 0.77 s versus 0.24 s).

The dynamic interactions between gp120 and CCR5 in the presence of CD4 in living cells. Using GHOST His-5 cells expressing both CD4 and CCR5, we measuredthe binding kinetics between gp120 and the coreceptor/chemokineCCR5 in the presence of CD4. The protein complex gp120-CCR5/CD4could withstand mean forces of up to 34 pN before rupture, fora loading rate of 200 pN/s. The rupture force measurements fittedto equation 1 yielded a k⁰_off rate of 0.7 s^–1 at low loadingrates (<450 pN/s). Therefore, at low physiological loadingrates, the tensile strength of the gp120-CCR5/CD4 bond was significantlyhigher and its lifetime (1.43 seconds) was significantly longerthan those of the gp120-CD4 bond (0.24 second). The treatmentof the cells with function-blocking monoclonal antibodies andsmall-molecule inhibitors against both CD4 and CCR5 eliminatedall specific interactions between cells and the gp120-coatedcantilever tip. Further, we verified that the measurements ofthe force required to break the gp120-CD4 bond obtained fromGHOST Hi-5 cells expressing both CD4 and CCR5 receptors in thepresence of a function-blocking antibody against CCR5 led tothe same kinetic and mechanical Bell parameters as those obtainedwith the parental GHOST cell line (CD4 only) (data not shown).

DISCUSSION

Top

Discussion

We have described a single-molecule analysis of the bindinginteractions between HIV-1 gp120 and its major cell receptorCD4 and coreceptor CCR5. Our approach allows us to obtain notonly familiar kinetic parameters, such as dissociation rateand bond lifetime (albeit at the single-molecule level), butalso fundamental micromechanical properties of molecular bonds,such as rupture force and reactive compliance, which describethe mechanical strength of gp120-receptor bonds.

Previous structural and biochemical studies suggest that thefunctional forms of CD4 and CCR5 are monomeric (38, 39), althoughthe state of aggregation of CD4 may be cell type dependent (24,25). One could imagine that, upon contact of gp120 with theplasma membrane, gp120 molecules could interact with a clusterof CD4 molecules at the same time. However, if CD4 receptorswere to fortuitously form an aggregate for a moment, the rateof reproach of the cantilever away from the cell remains sufficientlylow to detect the rupture of individual molecular pairs (see,e.g., Fig. 2). Multiple bonds would have to break simultaneouslyto be indistinguishable from single bond rupture event. Thelifetime of the bonds involving multiple copies of CD4 or gp120would be substantially longer, and the rupture force histogramswould display multiple quantized peaks (as opposed to the observedunique well-defined peak), if rupture forces for multiple bondswere inadvertently part of the data sets (4). We also note thatour assay shows that the lifetimes of the tested bonds, gp120-CD4,gp120-CCR5/sCD4, and gp120-CCR5/CD4, depend critically on theforce applied to those bonds, a property that is not capturedby traditional assays.

To place our gp120-CD4 interaction measurements in context,we compare the kinetic and micromechanical properties of thegp120-CD4 bond with those of important molecular bonds involvedin immunology and metastasis. Table 1 compares the dissociationrate (k⁰_off) and reactive compliance (x_ß), as wellas the mean rupture forces of the gp120-CD4, gp120-CCR5/CD4,and gp120-CCR5(sCD4) bonds, with those of P-selectin/PSGL-1and P-selectin/carcinoma ligand bonds. These bonds are involvedin leukocyte-endothelium interactions during inflammation andin carcinoma-endothelium interaction during metastasis, respectively(21). The kinetic and micromechanical parameters k⁰_off, x_ß,and rupture force were assayed at the single-molecule level,in living cells, using the same molecular force probe assay(15, 16). HIV infection occurs mostly in the lymph nodes, whereforces acting on the virus and loading rates, to which gp120-CCR5/CD4bonds are subjected, are expected to be low. In contrast, uponbinding to the endothelium, leukocytes and cancer cells arecontinuously subjected to hemodynamic flows, which subject P-selectin/ligandbonds to large forces (20). At low loading rates, the tensilestrength of all gp120-receptor/receptor bonds is much lower(P < 0.0001) than that of the P-selectin/PSGL-1 or P-selectin/carcinomaligand bonds: much smaller forces are required to break a gp120-receptor/receptorbond than to break P-selectin/ligand bonds. Moreever, the equilibriumlifetimes of gp120-receptor/receptor bonds were much shorterthan those of P-selectin/ligand bonds. This result indicatesthat in the presence of a low force, such as the random forcedue to the Brownian motion of the virus on the surface of itshost cell (of the order k_BT ${eta}$ / ${tau}$ , where ${tau}$ is bond lifetime, ${eta}$ is themedium viscosity, and k_BT is the thermal energy), the time fora gp120-receptor/receptor bond to dissociate is much shorter(P < 0.0001) than the time to dissociate the P-selectin/ligandbonds.

View this table:
[in this window]
[in a new window]
TABLE 1. Summary of the biochemical and biomechanical properties of gp120-CD4, gp120-CCR5/sCD4, and gp120-CCR5/CD4 compared to those of P-selectin/PSGL-1 and P-selectin/cancer cell LS174T ligand molecular bonds^a

The equilibrium interaction energy for the gp120-receptor canbe semiqualitatively obtained from nonequilibrium pulling experiments.Following Jarzynski's reconstruction (18), we find that theequilibrium free energy of interaction between gp120 and itscoreceptor features two wells. This is in sharp contrast tothe P-selectin/PSGL-1 bond involved in leukocyte attachmentto the endothelium, which features only one slope and, therefore,whose free energy displays only one well. This implies thatgp120 and its receptors have to overcome two energy wells inorder to detach spontaneously from each other. The sum of thereactive compliances for these two wells can be interpretedas the effective length of reaction path during mechanical retractionand the widths of the inner and outer potential barriers characterizingthe binding interaction between gp120 and receptors (Fig. 4C).The underlying structural origin of these two wells is unclearat this time. Nevertheless, we could speculate that it may stemfrom the considerable conformational flexibility within gp120(27). This test of this hypothesis will have to await mutationalstudies in combination with our MFP assay.

Using previous K_d measurements (2, 17), we can estimate theassociation rate (k_on) of gp120 for CCR5 by using our k⁰_offmeasurements: it is about 500-fold higher than the associationrate of gp120 for CD4 (k_on of $~$ 2 µM/s versus $~$ 0.004 µM/s).Together with previous crystallographic data, the results obtainedhere by single-molecule force spectroscopy suggest revisionsof the current model of the early steps leading to HIV-1 entry,at least in cultured GHOST cells. In those cells, CD4 is absolutelyrequired for the binding interaction between gp120 and its receptors(e.g., CCR5) (although the CD4 requirement depends on the isolateand the type of cells [13, 14]). Indeed, in the absence of CD4,interactions between gp120 and host cells are completely abrogated,and CCR5 alone cannot engage CD4. Nevertheless, the associationof gp120 with CCR5 (with the required presence of CD4) is muchmore favorable than the association of gp120 with CD4. Thissuggests that right after the association of gp120 with CD4,gp120 seeks CCR5 to rapidly form a new bond. This "bond transfer"occurs only if CCR5 is in close proximity to CD4 to allow, withina time equal to or smaller than the lifetime of the gp120-CD4bond, for CCR5 to form with CD4 a bond with gp120. Given theshort lifetime of the gp120-CD4 bond, if CCR5 is not in theclose vicinity of CCR5, gp120 will detach rapidly from CD4 andno bond can be established between gp120 and CCR5. CD4 and CCR5do not seem to interact directly (38), although this may dependon receptor density and the cell type-dependent organizationof these receptors on the cell surface (28, 29, 32). Therefore,the requirement of the close proximity of CCR5 to CD4 for effectivebinding of HIV to its receptors does not seem to be regulated.

The assay and analysis presented here may serve as a generalframework to study mechanisms of receptor-mediated interactionsbetween viral envelope proteins and host cell receptors at thesingle-molecule level. In particular, this approach could beused to study the single-molecule kinetics and micromechanicalproperties of the bonds that R5X4 gp120 makes with receptorsrelative to R5 and X4 gp120 molecules.

ACKNOWLEDGMENTS

We thank our colleague Robert F. Siliciano for advice and supportand Sean X. Sun for fruitful discussions. We also thank oneof the reviewers for insightful suggestions.

This work was funded by NASA grant NAG9-1563 and NIH grant GM065835.

FOOTNOTES

* Corresponding author. Mailing address: Department of Chemical and Biomolecular Engineering, The Johns Hopkins University, 3400 N. Charles Street, Baltimore, MD 21218. Phone: (410) 516-7006. Fax: (410) 516-5510. E-mail: wirtz{at}jhu.edu.

REFERENCES

Top