Quantifying the Effects of Molecular Orientation and Length on Two-dimensional Receptor-Ligand Binding Kinetics*

Surface presentation of adhesion receptors influences cell adhesion, although the mechanisms underlying these effects are not well understood. We used a micropipette adhesion frequency assay to quantify how the molecular orientation and length of adhesion receptors on the cell membrane affected two-dimensional kinetic rates of interactions with surface ligands. Interactions of P-selectin, E-selectin, and CD16A with their respective ligands or antibody were used to demonstrate such effects. Randomizing the orientation of the adhesion receptor or lowering its ligand- and antibody-binding domain above the cell membrane lowered two-dimen-sional affinities of the molecular interactions by reducing the forward rates but not the reverse rates. In contrast, the soluble antibody bound with similar three-dimen-sional affinities to cell-bound P-selectin constructs regardless of their orientation and length. These results demonstrate that the orientation and length of an adhesion receptor influences its rate of encountering and binding a surface ligand but does not subsequently affect the stability of binding. Cell adhesion fundamental G1 subtracting determined 50 times excess unlabeled G1, from total binding. All assays performed in triplicate. statistical significance of the difference between three-dimensional affinities for different molec- ular orientations or performed using the Student’s t test.


INTRODUCTION
Cell adhesion is a fundamental biological process that is mediated by specific interactions between adhesion receptors and their ligands on other cell surfaces or in extracellular matrix (1,2).For example, interactions between selectins and glycoconjugates mediate leukocyte tethering to and rolling on vascular surfaces at sites of inflammation or injury (3)(4)(5).The selectin family of adhesion molecules has three known members: P-, E-, and L-selectin.Their common structure is an N-terminal, C-type lectin (Lec) 1 domain, followed by an epidermal growth factor (EGF)-like module, multiple copies of consensus repeat (CR) units characteristic of complement binding proteins, a transmembrane segment, and a short cytoplasmic domain (3)(4)(5).As another example, circulating immunoglobulin G (IgG) bind to foreign particles or damaged tissue through their dual antigen-binding fragments (Fab).The conserved Fc fragment of this bifunctional molecule is available for binding by Fc receptors (Fc Rs) on the immune cell surface.Such binding triggers a wide variety of immune responses (6).
To mediate cell adhesion, the interacting receptors and ligands must anchor apposing surfaces of two cells or of a cell and the substratum, i.e., two-dimensional (2D) binding, which differs from interactions in the fluid phase, i.e., three-dimensional (3D) binding.The binding affinity K a is the ratio of equilibrium concentration of bonds to those of free receptors and ligands.However, concentration is measured as number of molecules per volume in 3D but number of molecules per area (i.e., surface density) in 2D, resulting in different units for K a (M -1 in 3D and m 2 in 2D).The kinetic forward-rate k f also has different units in different dimensions (M -1 s -1 in 3D and m 2 s -1 in 2D).The 2D k f is the rate of bond formation between unit densities of receptors and ligands that are respectively anchored on two apposing surfaces of unit area.By comparison, the 3D k f is the rate of bond formation between unit concentrations of receptors and ligands at least one of which is in solution of unit volume.There has been increasing recognition that 2D binding parameters are not readily conversable from their 3D counterparts (7).One of the reasons for this is that 2D interactions are influenced by the surface environment, just as 3D interactions are affected by the solvent environment (8).One of the surface environmental factors is the distance that the binding pocket of the adhesion receptor extends outward from the cell surface.For example, extending the binding site of CD58 above the cell membrane by interposing four Ig-like domains (~15 nm) in the stalk enhances its binding to CD2 on T lymphocytes (9).Under static conditions, P-selectin glycoprotein ligand 1 (PSGL-1)-expressing neutrophils adhere equivalently to cells expressing wild-type P-selectin that contains 9 CR units and to P-selectin constructs with as few as two CRs.However, P-selectin requires at least 5 CRs to mediate optimal rolling of flowing neutrophils under dynamic shear conditions (10).K562 cells bearing 2-GSP-6, a glycosulfopeptide modeled after the binding domain of PSGL-1, which was attached to the membrane-distal region of a nonbinding molecule ~50 nm above the cell surface, roll more stably on P-selectin than cells bearing 2-GSP-6 randomly attached to cell surface proteins (11).HL-60 cells adhere to immobilized full-length E-selectin that contains all six CRs but not to shorter E-selectin constructs with two or less CRs (12).
Another surface environmental factor is the orientation of the binding pockets of adhesion receptors.For example, HL-60 cells adhere to the short E-selectin constructs with two or less CRs when they are captured by a nonblocking monoclonal antibody (mAb) adsorbed on the plastic surface (12).The ligand binding affinity and kinetics of Fc RIIIA (CD16A) are affected by its surface anchor (glycosylphosphatidylinositol or transmembrane) (13).Cell adhesion to fibronectin is modulated by surface chemistries that alter fibronectin adsorption (14).
Cells expressing an L-selectin construct that replaces its EGF domain with the EGF domain of P-selectin roll better on L-selectin ligands than cells expressing wild-type L-selectin, perhaps because an altered orientation of the lectin domain enhances the association rate with surface ligands (15).
The differential effects of shortening P-selectin on adhesion under flow and static conditions (10) suggest that altering molecular length may affect the kinetics of the receptor-ligand interaction.The EGF domain swapping study (15) suggests that orientation also influences kinetics.To directly test this possibility, we quantified the effects of molecular orientation and length on cell adhesion by comparing the 2D kinetics of the same molecules coupled to the surface of human red blood cells (RBCs) using different methods: coating via chromium chloride (CrCl 3 ) coupling, which led to random orientation, or capturing by a nonblocking mAb, which resulted in a more uniform orientation, or direct binding to the RBC surface via an anti-human RBC antibody, which yielded fully uniform orientation.2D kinetics was also compared between captured short and long constructs of the same molecule.In addition, 3D affinities and 2D bindings of a mAb to surface-bound P-selectin constructs of different orientations and lengths were compared.Our results indicate that the orientation and length influence the 2D forward-rate but not the 2D reverse-rate of the molecular interaction, whereas they have no effects on 3D affinity.These results provide insights into the biophysical mechanisms by which the molecular orientations and lengths of adhesion receptors affect cell adhesion.

{ Figure 1 }
Site density determination Site densities of proteins coated on RBCs or expressed on CHO cells were measured using flow cytometry and/or immunoradiometric assay (IRMA).To measure densities of randomly oriented selectins coated via CrCl 3 coupling, RBCs were incubated with anti-selectin blocking mAbs (G1 or ES1) at a concentration of 10 g/ml in 200 l of FACS buffer (RPMI/5 mM EDTA/1% BSA/0.02% sodium azide) on ice for 40 min, and then incubated with FITC-labeled goat anti-mouse secondary antibody.To measure densities of randomly or uniformly oriented RbIgG, RBCs were directly incubated with FITC-labeled goat anti-rabbit antibody.To measure CD16A expression, CHO cells were incubated first with anti-CD16 mAb CLBFcgran-1 and then with FITC-labeled goat anti-mouse secondary antibody.After washing, the cells were analyzed by flow cytometry.The site densities were then calculated by comparing the fluorescence intensities of the cells with those of standard beads (Bangs Labs, Fishers, IN) (24).To measure densities of uniformly oriented selectins (sPs, sEs, or PLE), one set of RBCs pre-coated with a range of densities of the relevant capturing mAb (five to seven densities for each selectin) were incubated with FITC-labeled goat anti-mouse antibody and the fluorescence intensities were measured using flow cytometry as above.Another set of RBCs pre-coated with the same range of densities of capturing mAb were incubated with the corresponding selectin and their site densities were measured by IRMA (21,26,28).A calibration curve was obtained for each selectin by plotting the selectin site density against the mean fluorescence intensity of the capturing mAb, thereby allowing calculation of the site densities of the uniformly oriented selectin from the mean fluorescence intensities of the capturing mAb.

2D kinetics measurements
The micropipette adhesion frequency assay for measuring 2D kinetics has previously been described (24)(25)(26).Briefly, a single selectin (or RbIgG) coated RBC and a single carbohydrate ligand-expressing HL-60 cell (or a CD16A-expressing CHO cell) was respectively aspirated by two micropipettes with respective diameters of ~1.5 and ~3 m via a suction pressure of 1-4 mmH 2 O (10-40 pN/ m 2 ).Adhesion between the RBC and the HL-60 (or CHO) cell was staged by placing them onto controlled contact via micromanipulation.The presence of adhesion at the end of a given contact period was detected mechanically by observing microscopically the deflection of the flexible RBC membrane upon retracting it away from the HL-60 (or CHO) cell.Such detection was reliable and unambiguous for >90% of the retractions, as clearly observable membrane deflections could be generated by a force as low as 2 pN at the RBC apex, far lower than the strength of a single bond (24,25).This contact-retraction cycle was repeated a hundred times to estimate the adhesion probability, P a , at that contact duration, t.For each surface presentation of each molecular species examined, ~100 pairs of cells were used to obtain several P a vs. t curves that correspond to different receptor and ligand densities, m r and m l .Each binding curve was fitted to a small system probabilistic kinetic model (24)(25)(26), to estimate a pair of parameters: the zero-force reverse rate, k r 0 and effective binding affinity, m l A c K a 0 (if m r was known) or A c K a 0 (if both m r and m l were known), where A c is the contact area, which was kept constant in all experiments.Multiple pairs of (k r 0 , m l A c K a 0 ) or (k r 0 , A c K a 0 ) values were obtained for each molecular orientation and length to allow evaluation of the mean and standard deviation.The statistical significance of (or the lack thereof) the difference between the 2D affinities (or reverse-rates) of the same pair of interacting molecules being presented on the cell surface with different orientation or distance above the membrane was assessed using the Student t-test.
In one set of experiments, adhesion frequencies between RBCs coated with G1 via CrCl 3 coupling and RBCs coated with P-selectin constructs of different orientations or lengths were measured at a single contact time of 4 s, and the results were expressed as -ln(1 -P a )/(m r m l ).5-6 pairs of cells were measured for each P-selectin surface presentation, and the statistical significance of the differences between results from different conditions were assessed by the Student t-test.
3D affinity measurements Scatchard analysis was used to determine 3D binding affinities of 125 I-labeled G1 in fluid phase for either long (sPs) or short (PLE) P-selectin constructs coated on RBC surfaces either uniformly by capturing mAbs (S12 or 1478) pre-coated via CrCl 3 coupling or randomly via direct CrCl 3 coupling (29,21).Briefly, 200 l of 0.1-6 g/ml 125 I-G1 was added to 5 10 6 P-selectin-coated RBCs.After incubation at 4 o C for 30 min, 500 l of a 1:9 ratio of apiezon oil:dibutyl phthalate (Sigma) oil mixture was added into each sample, which was then centrifuged to separate the free 125 I-labeled G1 from the cells.Radioactivity associated with the cell pellets was measured using a gamma counter.Specific binding of G1 was calculated by subtracting the nonspecific binding, which was determined by adding 50 times excess unlabeled G1, from total binding.All assays were performed in triplicate.The statistical significance of the difference between 3D affinities for different molecular orientations or lengths was performed using the Student t-test.

RESULTS
Binding is specific Binding was quantified by the adhesion frequency at sufficiently long contact time (t ) in control experiments.As exemplified for binding of P-selectin-coated RBCs to PSGL-1-expressing HL-60 cells (Fig. 2), adhesion frequencies measured from the micropipette assay were mediated by specific selectin-ligand interactions, because they were present when the RBCs were coated with appropriate nonblocking anti-selectin mAbs to capture the corresponding soluble selectin constructs but were abolished when mismatched selectin constructs not recognized by the coated capturing mAbs or isotype-matched irrelevant mAbs were used (Fig. 2).In addition, binding was blocked by mAbs against PSGL-1 (PL1), P-selectin (G1), and by the calcium chelator EDTA.Binding specificity between P-selectin-coated RBCs and G1-coated RBCs, E-selectin-coated RBCs and HL-60 cells, and between RbIgG-coated RBCs and CD16A-expressing CHO cells were confirmed by similar experiments (data not shown).

{ Figure 2 }
Binding curves follow simple receptor-ligand binding kinetics Dependence of adhesion frequency on contact duration was measured using the micropipette assay at contact times ranging from 0.25-10 s, as exemplified for P-selectin-coated RBCs interacting with HL-60 cells (Figs. 3a, 3b, 6a, and 6b).The adhesion probability, P a , was obtained by removing the nonspecific adhesion frequency, P n (dashed lines in Figs.3a, 3b, 6a, and 6b, obtained by fitting the directly measured nonspecific adhesion frequency to Eq. ( 1)), from the directly measured total adhesion frequency, P t , according to P a = (P t -P n )/(1 -P n ) (30).The data (points in Figs.3a, 3b, 6a, and 6b) exhibit the shape of typical binding curves, with a transient phase where P a increased with t and a steady phase where P a reached equilibrium.Eq. ( 1) was used to fit each binding curve to obtain two parameters: k r 0 and m l A c K a 0 (if m r was known) or k r 0 and A c K a 0 (if both m r and m l were known).For each set of interacting molecules presented on the RBC surface at each orientation and length, the mean reverse-rate and mean effective binding affinity were calculated from several pairs of k r 0 and m l A c K a 0 (or A c K a 0 ) values that respectively best fitted several P a vs. t curves obtained by varying the densities of the receptors and/or ligands.The mean k r 0 and m l A c K a 0 (or A c K a 0 ) values were then used, along with the corresponding m r and/or m l values measured from independent experiments, to predict each P a vs. t curve (solid lines in Figs.3a, 3b, 6a, and 6b).It is evident that the model fits the data well, imparting confidence in the estimated kinetic parameters.

Molecular orientation affects forward-rates but not reverse-rates
To alter molecular orientation, soluble selectin constructs were coated randomly via CrCl 3 or coupled more uniformly via capturing by a mAb pre-coated on RBC surfaces (cf.Fig. 1, a and b).For each coating method, densities of selectin binding sites were quantified using adhesion blocking mAbs, and P a vs. t curves were measured using the micropipette assay.To bind HL-60 cells with the same level of steady-state adhesion frequency, ~3.5-fold higher site density was required for RBCs coupled with randomly oriented sPs than for RBCs coupled with uniformly oriented sPs (e.g., compare curves labeled 2.4 m -2 in Fig. 3a and 8.7 m -2 in Fig. 3b).By comparison, the time required to reach half equilibrium binding level, t 1/2 , was indifferent to the coating methods.
Since the equilibrium binding P a ( ) is related to the effective binding affinity, m l A c K a 0 = -ln[1 -P a ( )]/m r (if m r was known) or A c K a 0 = -ln[1 -P a ( )]/(m r m l ) (if m r and m l were known), and the half-time t 1/2 is related to the reverse-rate, k r 0 0.5/t 1/2 (13,24), our data indicate that, when both interacting molecules were immobilized on cell surfaces, randomly oriented sPs bound ligands with a lower effective binding affinity but a similar reverse-rate compared to the more uniformly oriented sPs.These observations were confirmed by comparing the kinetic parameters obtained from fitting Eq. ( 1) to the P a vs. t curves (Fig. 3c).The zero-force reverse-rates for the randomly oriented and more uniformly oriented sPs were similar (k r 0 = 1.1 ± 0.1 and 0.9 ± 0.1 s -1 , respectively, P > 0.45).By contrast, the effective affinity for the randomly oriented sPs was 3.1-fold lower than that for the more uniformly oriented sPs (m l A c K a 0 = 0.08 ± 0.002 and 0.25 ± 0.01 m 2 , respectively, P < 0.015).The effective forward-rate (calculated from m l A c k f 0 = m l A c K a 0 k r 0 ) for the randomly oriented sPs was 2.6-fold lower than that for the more uniformly oriented sPs (m l A c k f 0 = 0.09 and 0.23 m 2 /s, respectively).This isolation of the orientation effects to the forward-rate suggests that better receptor orientation on the cell surface enhances effectiveness for ligand binding by providing easier access to ligands immobilized on the apposing cell surface.
It follows from this hypothesis that the dissociation of pre-formed receptor-ligand bond would not be affected by orientation, as was observed.
The effective forward-rate for the randomly oriented RbIgG was 8.3-fold lower than that for the uniformly oriented RbIgG (A c k f 0 = 0.44 × 10 -6 and 3.65 × 10 -6 m 4 /s, respectively).These data corroborate the previous results and provide additional support to the conclusion that uniformly oriented molecules bind their counter-molecules with higher effective affinity than randomly oriented molecules.

Molecular length affects forward-rates but not reverse-rates
Previous flow chamber experiment has shown that more neutrophils accumulated on a long sPs than on a short PLE coated on the chamber floor at matched densities, demonstrating the effects of length on transient adhesion (22).To quantify the effects of molecular length on 2D ligand binding kinetics, micropipette experiments were performed using the same P-selectin constructs: sPs was captured by mAb S12, which binds the 4 th CR from the membrane anchor (10), whereas PLE was captured by mAb 1478, which binds an epitope tag added to the C-terminus of the EGF domain.
Thus, both P-selectin constructs were oriented properly on their respective RBC surfaces but the sPs extends the lectin binding pocket 5 CRs, or 15 nm, further outward from the RBC surfaces than the PLE (Fig. 1d).For nearly identical site densities (4.6 and 4.8 m -2 , respectively), the steady-state adhesion frequencies were higher for the long sPs than the short PLE (0.68 and 0.42, respectively) (Fig. 6, a and b).By comparison, the time required to reach half steady-state level was similar for the two P-selectin constructs of different lengths.These observations were confirmed by comparing the kinetic parameters obtained from fitting Eq. ( 1) to the P a vs. t curves (Fig. 6c).The effective affinity for the long sPs was 2.3-fold higher than that for the short PLE (m l A c K a 0 = 0.25 ± 0.01 and 0.11 ± 0.003 m 2 , respectively, P < 0.03), while the zero-force reverse-rates were similar for the long and short constructs (k r 0 = 0.9 ± 0.1 and 1.1 ± 0.1 s -1 , respectively, P > 0.4).The effective forward-rate for long sPs was 1.9-fold higher than that for short PLE (m l A c k f 0 = 0.23 and 0.12 m 2 /s, respectively).The similarity in the effects of molecular length and orientation on the 2D forward-rates but not the reverse-rates suggests a similar underlying mechanism: sPs is more effective to bind ligands than PLE because it is easier for surface-bound ligands to access long than short receptors anchored on the apposing cell surfaces; however, the dissociation of pre-formed receptor-ligand bond would not be affected by the molecular length.

Lack of effects of molecular orientation and length on 3D affinity
The isolation of the effects of molecular orientation and length to the forward-rate suggests that randomizing the orientation and shortening the length of a membrane-bound molecule reduce its accessibility by the binding partner, especially when the partner is also anchored to the apposing cell membrane that restricts its ability to bind a molecule with suboptimal orientation and length.This hypothesis predicts that the effects of molecular orientation and length are associated primarily with 2D binding, and will diminish if the counter-molecules are no longer restricted by their surface anchor.solution G1 (Fig. 7b).Similarly striking cooperative effects were also observed in micropipette experiments using RBCs coated with randomly oriented short molecules, PLE, ELE, or Fc fragment of IgG interacting with HL-60 cells or CD16A-expressing CHO cells.In spite of the high molecular densities as assessed by flow cytometry, adhesion frequencies remained comparable to nonspecific adhesions, thereby preventing kinetic rates from being measured (data not shown).By comparison, much higher adhesion frequencies were detected using similar site densities of the same short but more uniformly oriented molecules (PLE and ELE, captured by mAbs 1478 and 1D6, respectively) (Fig. 6 and ( 26)) or of their long but randomly oriented counterpart molecules (sPs, sEs, and RbIgG) (Figs. 3-6), thereby allowing kinetic measurements.
These combined data demonstrate that uniform orientation and extended length cooperatively enhance 2D binding affinity.(13).Another example is that NIH 3T3 fibroblasts adhered better to fibronectin absorbed on OH-group expressing self-assembled monolayers than to fibronectin absorbed on COOH-, NH 2or CH 3 -group expressing monolayers.This effect was interpreted as substrate-dependent conformational changes in fibronectin, since differential binding was also observed for soluble anti-fibronectin mAb and for cell surface 5 1 integrin (32).
Our results indicate that altering the molecular orientations and lengths of adhesion receptors can affect their functions without altering the conformation of the ligand-binding domain.This conclusion is supported by multiple lines of evidence.Conformational changes of the binding site likely affect reverse-rate while orientation changes of the binding site likely affect forward-rate (33).Extending the binding site above the surface by adding a polymer chain spacer can generate a long range attractive force that may enhance the forward-rate but would not affect the reverse-rate (34).Randomizing orientation would reduce accessibility, thereby affecting forward-rate, but would have little effect on the stability of the bond once it is formed, thereby not affecting reverse-rate.These were exactly what were observed (Figs.3-5).
The fact that similar effects were observed for four molecular systems, two of which are structurally related (P-selectin and E-selectin) but the other two not (RbIgG and G1), also suggests that the different coating methods caused changes in orientation rather than in conformation.Like physiosorption, CrCl 3 coupling likely results in random orientation that renders some of the immobilized molecules inaccessible for binding by the counter molecules.
Coating with an anti-RBC antibody results in uniform orientation of the Fc portion of IgG for Fc R binding that is identical to the physiological situations.Capturing the molecules via nonblocking mAbs results in an intermediate degree of uniformity in orientation as the capturing mAbs themselves were randomly coated on RBC via CrCl 3 coupling.Consistent with this contention, the forward-rate changes produced by different coating methods were greater for the Fc R system than the two selectin systems (compare Figs. 3c and 4 with Fig. 5).
It follows from the above arguments that shortening the molecular length would produce similar effects, as this likewise would affect accessibility but not stability, which was also observed (Fig. 6).Furthermore, the same reasoning predicts that the effects of random orientation would be worsened by shortening the molecule; or conversely, that the diminishing binding due to random orientation could be partially restored by lengthening the molecule.
Indeed, for similar differences in the degree of uniformity, the effects for the shorter E-selectin seemed greater than for the longer P-selectin (compare Figs. 3c and 4) and the effects for the even shorter G1 seemed the greatest (Fig. 7c).In addition, while randomizing the orientation of the long molecules (sPs, sEs, and RbIgG) via CrCl 3 coupling reduced binding, doing the same using their short counterparts (PLE, ELE, and Fc fragment of IgG) diminished binding below the level detectable by our micropipette assay (data not shown).
Finally, it seems intuitive that altering molecular orientation and length would affect 2D binding but not 3D binding.The reason is that 2D binding places far greater demands on accessibility due to the linkage of the molecules to apposing surfaces, which restricts their motions in 2D.Releasing one of the molecules from the surface enables it to approach its binding partner via 3D motions, which likely allows it to gain access to suboptimally oriented and shortened molecules that are inaccessible to its surface-bound counterpart.Indeed, Scatchard

FIGURE 1 .
FIGURE 1. Schematics of long or short molecules coated on RBC surface with uniform or

FIGURE 3 .
FIGURE 3. Binding curves and kinetics of uniformly or randomly oriented sPs.Adhesion

FIGURE 4 .
FIGURE 4. Kinetics of uniformly or randomly oriented sEs.Micropipette experiments were

FIGURE 5 .
FIGURE 5. 2D kinetic reverse-rates and effective affinities of CD16A for uniformly or randomly

FIGURE 6 .
FIGURE 6. Binding curves and kinetics of long or short P-selectin constructs.Adhesion

FIGURES FIGURE 1
FIGURES