Lateral mobility of individual integrin nanoclustersorchestrates the onset for leukocyte adhesionGert Jan Bakkera, Christina Eichb, Juan A. Torreno-Pinac, Ruth Diez-Ahedoa, Gemma Perez-Sampera,Thomas S. van Zantenc, Carl G. Figdorb, Alessandra Cambib, and Maria F. Garcia-Parajoc,d,1

aBioNanophotonics group, Institute for Bioengineering of Catalonia and Centro de Investigación Biomédica en Red en Bioingeniería, Biomateriales yNanomedicina, Baldiri Reixac 15-21, 08028 Barcelona, Spain; bDepartment of Tumor Immunology, Nijmegen Center for Molecular Life Sciences, RadboudUniversity Nijmegen Medical Center, 6500 HB Nijmegen, The Netherlands; cICFO-Institut de Ciencies Fotoniques, Mediterranean Technology Park,08860 Castelldefels, Barcelona, Spain; dICREA-Institució Catalana de Recerca i Estudis Avançats, 08010 Barcelona, Spain

Edited by Thomas D. Pollard, Yale University, New Haven, CT, and approved January 31, 2012 (received for review October 6, 2011)

Integrins are cell membrane adhesion receptors involved in mor-phogenesis, immunity, tissue healing, andmetastasis. A central, yetunresolved question regarding the function of integrins is howthese receptors regulate both their conformation and dynamic na-noscale organization on the membrane to generate adhesion-com-petent microclusters upon ligand binding. Here we exploit the highspatial (nanometer) accuracy and temporal resolution of single-dye tracking to dissect the relationship between conformationalstate, lateral mobility, and microclustering of the integrin receptorlymphocyte function-associated antigen 1 (LFA-1) expressed onimmune cells. We recently showed that in quiescent monocytes,LFA-1 preorganizes in nanoclusters proximal to nanoscale raftcomponents. We now show that these nanoclusters are primarilymobile on the cell surface with a small (ca. 5%) subset of conforma-tional-active LFA-1 nanoclusters preanchored to the cytoskeleton.Lateral mobility resulted crucial for the formation of microclustersupon ligand binding and for stable adhesion under shear flow.Activation of high-affinity LFA-1 by extracellular Ca2þ resulted inan eightfold increase on the percentage of immobile nanoclustersand cytoskeleton anchorage. Although having the ability to bindto their ligands, these active nanoclusters failed to support firmadhesion in static and low shear-flow conditions because mobilityand clustering capacity were highly compromised. Altogether, ourwork demonstrates an intricate coupling between conformationand lateral diffusion of LFA-1 and further underscores the crucial roleof mobility for the onset of LFA-1 mediated leukocyte adhesion.

intercellular adhesion molecule ∣ single molecule detection ∣ cumulativeprobability distribution ∣ integrin lymphocyte function-associated antigen 1

Integrins are transmembrane α/β heterodimeric cell adhesionmolecules that play critical roles in cell–cell and cell–extracel-

lular matrix interactions. Most integrins dynamically modulatetheir adhesiveness through global conformational changes thatdefine their affinity state for ligands (1–3), and by their mem-brane lateral organization in clusters, also known as avidity (4–6).Although the mechanisms by which conformational changes ofintegrins leading to enhanced-binding affinity have been well es-tablished (1–3, 7, 8), there is no consensus yet on how integrinscontrol their spatial rearrangement on the cell membrane and therole of lateral mobility in cluster formation.

Integrin clustering alone can lead to increased avidity (4) dueto cooperative effects on resistance to bond breakage (9).Because bonds between individual integrins and their ligandscan be broken with relatively small forces (10), clustering appearsas an important mechanism to reinforce adhesion via cooperativebinding (10, 11). On leukocytes, rearrangement of integrins intoclusters upon activation occurs by diffusion, after release ofcytoskeleton constrains on integrin motion (12–14). These obser-vations led to the common belief that inactive integrins areanchored to the cytoskeleton and released when needed to rein-force binding (12, 15, 16), suggesting that integrin activation pre-cedes clustering. However, this model is hard to reconcile with

the fact that high-affinity (extend) integrins are more prone tointeract with the cytoskeleton via their cytoplasmic tails (2), lead-ing to immobile integrins that will compromise their capacityfor lateral diffusion; a process that is imperatively required forclustering (6).

A major integrin involved in the immune system is lymphocytefunction-associated antigen 1 (LFA-1). LFA-1 mediates leukocytemigration across the endothelium and within tissues and theformation of the immunological synapse by binding to its majorligand intercellular adhesion molecule 1 (ICAM-1) (17–19). Werecently showed that, on resting monocytes, LFA-1 function isassociated with its nanoclustering (20) forming nonintermixednanoscale supramolecular complexes together with lipid raftcomponents (21, 22). Importantly, a subset of these nanoclustersalready contained primed—i.e., extended LFA-1 molecules—prior to ligand binding (20, 23). Interestingly, a subset of high-affinity LFA-1 clusters has been also observed upon chemokinestimulation at the immediate contact between crawling T cellsand endothelial cells (24). Yet, dynamic insight on the spatial or-ganization of LFA-1 nanoclusters is lacking. Moreover, how thesedifferent subsets diffuse on the cell membrane and cooperatewith each other to generate adhesion-competent microclustersremains unknown. Here we applied single-molecule approachestogether with reporters of LFA-1 conformational state to dissectthe relationship between conformation, lateral mobility, andligand-induced microclustering of LFA-1.

ResultsLFA-1 Nanoclusters Are Mobile on Resting Monocytes. We previouslyshowed that LFA-1 forms distinct nanoclusters on resting mono-cytes prior to ligand binding (20, 21). Although this nanoscaleorganization resulted crucial for LFA-1 function (20), their dy-namic behavior has not been investigated yet. We applied single-dye tracking (SDT) to follow the lateral diffusion of individualsublabeled LFA-1 nanoclusters (Fig. S1 A–D and SI Text) on rest-ing monocytes. At our sublabeling conditions, individual intensitytrajectories on fixed cells showed discrete photobleaching stepsconsistent with single-dye detection (Fig. S1 C and D). Trajec-tories of LFA-1 nanoclusters labeled with the neutral mAbTS2/4 were generated from multiple movies (Fig. 1A) using asingle-molecule detection sensitive epifluorescence microscope(Fig. 1 B and C). At physiological cation conditions (0.4 mMCa2þ, Mg2þ), LFA-1 nanocluster mobility did not exhibit uniform

Author contributions: G.J.B., C.G.F., A.C., and M.F.G.-P. designed research; G.J.B., C.E.,J.A.T.-P., R.D.-A., G.P.-S., and T.S.v.Z. performed research; G.J.B., C.E., and J.A.T.-P.contributed new reagents/analytic tools; G.J.B., C.E., J.A.T.-P., G.P.-S., T.S.v.Z., and M.F.G.-P.analyzed data; and A.C. and M.F.G.-P. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. E-mail: maria.garcia-parajo@icfo.es.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1116425109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1116425109 PNAS ∣ March 27, 2012 ∣ vol. 109 ∣ no. 13 ∣ 4869–4874

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 3,

202

1

behavior. Whereas some nanoclusters were highly mobile, othersshowed a more restricted diffusion (Fig. 1D). Individual trajec-tories were analyzed by generating mean-square displacement(MSD) curves to obtain the diffusion coefficient D at short-time lags. The resultant D distribution varied from 10−3 to10−1 μm2∕s (Fig. 1E), evidencing a large heterogeneity in lateralmobility with a small percentage (ca. 5%) of stationary nanoclus-ters (Fig. S2 and SI Text).

To enquire on the type of diffusion exhibited by the mobileLFA-1 nanoclusters over longer time intervals (ca. 1.5 s) weapplied cumulative probability distribution (CPD) analysis (25).Using this approach, we could separate the mobile nanoclustersin two groups: (i) slow diffusing with Ds and αs, and (ii) fast dif-fusing with Df and αf , where D represents the short-time lag dif-fusion coefficients and α indicates the type of motion (α ¼ 1 for

Brownian diffusion; α < 1 for hindered, anomalous diffusion) (SIText). The analysis also rendered the relative fractions of slow andfast diffusing nanoclusters. At physiological cation conditions, thefast fraction (ca. 64%) of LFA-1 nanoclusters shows Df ¼ ð5.6�0.2Þ × 10−2 μm2∕s and αf ¼ 0.94� 0.06 (i.e., Brownian) (Fig. 1Fand Table S1). The slow fraction (ca. 32%) shows a diffusion coef-ficient of Ds ¼ ð1.4� 0.1Þ × 10−2 μm2∕s and αs ¼ 0.9� 0.2.Thus, in contrast to earlier reports about immobile LFA-1 in rest-ing Tcells (12, 15, 16) our data on resting monocytes demonstratethat, at physiological cation conditions, LFA-1 nanoclusters areprimarily mobile on the plasma membrane.

To confirm that we are tracking nanoclusters on monocytes,we also determined the diffusion of individual LFA-1 moleculeson monocyte-derived immature dendritic cells (imDC), whereLFA-1 is organized in a random (monomeric) fashion (20)(Fig. S3 A and B). As expected, the resultant D distribution ofLFA-1 on imDCs showed a clear shift toward larger D valuescompared to that on monocytes (Fig. S3C). To relate the changesin D to the size of the moving particles, we further appliedhydrodynamic theory (SI Text). Both the estimated values of Dfor individual LFA-1 molecules and nanoclusters agree well withthose measured on imDCs and monocytes, respectively, support-ing our statement that we are tracking individual nanoclusterson monocytes (see SI Text for further details). Additional controlexperiments at higher labeling conditions (Fig. S3 D and E) andhigh-density dual-color quantum dot tracking (Fig. S4) furthersupport our assignment (see SI Text).

Primed LFA-1 Nanoclusters Exhibit Multiple Diffusion Profiles on Rest-ing Monocytes. We previously showed that, on resting monocytes,approximately 25% of the LFA-1 nanoclusters are in a primedstate (20). To investigate the lateral mobility of these primednanoclusters, we used the conformation-dependent epitopeNKI-L16 (L16), which exclusively labels the extended conforma-tion of the αL subunit (26, 27) (Fig. 1A). Individual trajectoriesof L16þ-LFA-1 nanoclusters were generated (Fig. 2A) togetherwith the D histogram (Fig. 2B). A considerable higher fraction

A

B C

D

E F

Fig. 1. LFA-1 nanoclusters diffuse randomly on resting monocytes. (A) Sche-matic description of the SDT experiments. Individual LFA-1 molecules insidenanoclusters were labeled with TS2/4-ATTO520 or L16-ATTO647N using sub-labeling conditions. (B) Selected frame from a movie recorded at 100 ms perframe. Bright spots correspond to individual TS2/4-LFA-1 nanoclusters. Whitedots indicate the perimeter of the cell. (Scale bar: 5 μm.) (C) Selected LFA-1nanoclusters at different times to illustrate different mobility behavior: fast(Top), stationary (Middle), and slow (Bottom). (D) Representative LFA-1 na-nocluster trajectories displaying different lateral mobility, pseudocolor codedaccording to their apparent mobility: fast (orange), slow (blue), and station-ary (gray). (E) Normalized semilog distribution of D values at short-time lags.The vertical arrow indicates the threshold value of diffusion Dth above whichthe mobile population for CPD analysis has been selected (370 trajectoriesfrom 128 cells in multiple experiments). (F) Square displacement plots of theTS2/4 mobile fractions at different time lags as obtained from CPD analysis(157 trajectories).

A B

C D

Fig. 2. Primed LFA-1 nanoclusters exhibit multiple diffusion profiles on rest-ing monocytes. (A) Representative trajectories of primed LFA-1 nanoclusterslabeled with L16-ATTO647N. (B) Normalized semilog distribution of D valuesat short-time lags of L16þ-LFA-1 nanoclusters (bars) compared to that of thetotal (TS2/4) LFA-1 population (dashes) (669 trajectories from 49 cells onmultiple experiments). (C) Square displacement plots by fitting the CPD ofthe L16 mobile fractions at different time lags (380 trajectories). (D) Normal-ized fractions of stationary (gray), slow (blue), and fast (orange) mobile sub-populations for the total and primed nanoclusters. Error bars represent thestandard deviation.

4870 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1116425109 Bakker et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 3,

202

1

(ca. 20%) of L16þ nanoclusters was stationary, whereas the re-maining approximately 80% exhibited D values similar to thoserecovered with the TS2/4 reporter consistent with the fact thatprimed nanoclusters constitute a subset of the overall LFA-1population. Furthermore, CPD analysis revealed two differentdiffusion profiles (Fig. 2C): one fraction of L16þ nanoclusters(ca. 53%) exhibited fast, nearly Brownian diffusion (αf ¼0.91� 0.02), whereas the second fraction (ca. 28%) showed slowand anomalous diffusion (αs ¼ 0.49� 0.06) (Table S1). The over-all fractions of stationary, slow, and fast diffusing LFA-1nanoclusters for the two different reporters used are summarizedin Fig. 2D. These results indicate that primed nanoclusters displaymultiple diffusion behavior with stationary, anomalous, andfreely diffusing nanoclusters.

Reduction of Extracellular Ca2þ Restricts the Mobility of LFA-1Nanoclusters. Low extracellular Ca2þ or a combination of Mg2þ∕EGTA has been extensively used to promote high-affinity LFA-1(14, 28). We thus performed SDT experiments as a function ofCa2þ reduction to selectively induce extended nanoclusters andmonitor their lateral mobility on monocytes. Reduction of extra-cellular Ca2þ from 0.4 to 0.1 mM modestly reduced the TS2/4-LFA-1 mobile fraction (Fig. 3 A and G). In contrast, the diffu-sion of L16þ nanoclusters remained unaltered (Fig. 3 B and H).Strikingly, further reduction of Ca2þ down to 0.04 mM increasedthe number of stationary LFA-1 nanoclusters, regardless of thereporter used (Fig. 3 C, D, G, and H). Hence, the fraction of sta-tionary LFA-1 nanoclusters increased from 5% at 0.4 mM Ca2þto 43% at 0.04 mM Ca2þ for the TS2/4 reporter (Fig. 3G andTable S1) and from approximately 20% to 36% for the primed(L16þ) nanoclusters (Fig. 3H and Table S1). Consistent with thisincrease of stationary nanoclusters, the binned histograms ofthe diffusion coefficients of the total and primed nanoclustersrevealed an approximately twofold increase in the occurrenceof lowest diffusion values when extracellular Ca2þ was reducedto 0.04 mM (red arrows in Fig. 3 E and F). Moreover, theD curvedistributions of the total and primed nanoclusters decayedsharper at 0.04 mM, implying a severe slowing down of mobility(Fig. 3 E and F). CPD analysis further showed a significant de-crease on the fast mobile fractions and an anomalous behavior at0.04 mM Ca2þ regardless of the reporter used (Fig. 3 G and H,and Table S1). To rule out any effect of Mg2þ, we repeated ex-periments at low extracellular Ca2þ levels in the presence ofMg2þ and observed no differences on LFA-1 mobility (Fig. S5),consistent with the fact that the effect of Mg2þ is locally confinedto the headpiece of the integrin (2, 28). Collectively, these resultsreveal that reduction of extracellular Ca2þ dramatically decreasesthe mobility of LFA-1 on monocytes by immobilizing a large num-ber of nanoclusters and severely restricting the diffusion of theremaining mobile ones.

Remarkably, the slopes of the binned D distribution for thetotal LFA-1 nanoclusters (Fig. 3E) became progressively steeper,indicating a gradual slowing down of mobility as a function ofCa2þ reduction. Because Ca2þ depletion favors the extendedform of the integrin (2, 28), the changes in diffusion of the wholeLFA-1 population might reflect a shift in the conformation equi-librium from mainly inactive to the primed form. Indeed, thebinned diffusion histograms of the total and primed nanoclusters(Fig. 3I) showed an evolution profile in which both distributionsbecame more like each other as the concentration of Ca2þ wasreduced, until they remarkably matched each other at 0.04 mMCa2þ. Moreover, at 0.04 mM Ca2þ, the diffusion behavior of thetotal and primed nanoclusters became quite similar to each otherimplying full conversion from inactive to primed LFA-1 nanoclus-ters (Fig. 3 G–I and Table S1). To further confirm that the immo-bilization of LFA-1 is due to priming, we also performed SDTexperiments using Mn2þ, a strong integrin activator (3, 28), andthe activating antibody KIM185 (29). In both cases, the stationary

A B

C D

E F

G

I

H

Fig. 3. The lateral mobility of LFA-1 nanoclusters is affected by extracellularCa2þ. (A, C, and E) Diffusion coefficients for the total (TS2/4) LFA-1 populationat (A) 0.1 mM Ca2þ (bars) compared to 0.4 mM Ca2þ, Mg2þ (dashes); (C)0.04 mM Ca2þ (bars) compared to 0.4 mM Ca2þ, Mg2þ (dashes); (E) binneddistribution of D values for different extracellular Ca2þ conditions. The ver-tical arrow points to the lowest binned D value, and the curved arrow high-lights the slopes of the D values. (B, D, and F) D values for L16þ nanoclustersat (B) 0.1 mM Ca2þ (bars) compared to 0.4 mM Ca2þ, Mg2þ (dashes); (D)0.04 mM Ca2þ (bars) compared to 0.4 mM Ca2þ, Mg2þ (dashes); (F) binneddistribution of D values for different extracellular Ca2þ levels. (G and H) Nor-malized fractions of stationary, slow, and fast mobile subpopulations for (G)the total (TS2/4), and (H) primed (L16þ) nanoclusters at different Ca2þ levels.Error bars represent the standard deviation (p values are compared to therespective populations at 0.4 mM Ca2þ, Mg2þ: � ¼ p < 0.001 compared tothe stationary fraction; # ¼ p < 0.00003 compared to the fast fraction;† ¼ p < 0.03 compared to the stationary fraction; ‡ ¼ p < 0.001 comparedto the fast fraction). (I) Binned distribution of D values for TS2/4 (blacksquares) and L16 (green circles) at (Left) 0.4 mM Ca2þ, Mg2þ; (Center)0.1 mM Ca2þ; (Right) 0.04 mM Ca2þ. TS2/4 trajectories: 369 (0.4 mM Ca2þ,Mg2þ), 318 (0.1 mM Ca2þ), and 129 (0.04 mM Ca2þ). L16 trajectories: 669(0.4 mM Ca2þ, Mg2þ), 656 (0.1 mM Ca2þ), and 102 (0.04 mM Ca2þ).

Bakker et al. PNAS ∣ March 27, 2012 ∣ vol. 109 ∣ no. 13 ∣ 4871

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 3,

202

1

population increased considerably, similarly to the results ob-tained at low-Ca2þ conditions (Fig. S6).

The Actin Cytoskeleton Regulates the Diffusion of LFA-1 Nanoclustersat Low Extracellular Ca2þ Conditions. The immobilization of LFA-1nanoclusters on the cell membrane at low-Ca2þ levels suggeststheir anchoring to the cytoskeleton. We thus used cytochalasinD (CytoD) to test the effect of disrupting cytoskeletal interactionson LFA-1 diffusion. Treatment with CytoD released a significantfraction of stationary L16þ nanoclusters at 0.4 mM Ca2þ, Mg2þ(Fig. S7), and importantly released most of the L16þ nanoclus-ters at 0.04 mM Ca2þ (Fig. 4 A and C). This effect was extendedto the total LFA-1 nanoclusters at 0.04 mMCa2þ (Fig. 4 B andC).These data demonstrate that, at low extracellular Ca2þ levels,LFA-1 nanoclusters anchor to the cytoskeleton and suggest thatextracellular activators such as cations have a potent role onglobal integrin conformation by coupling extracellular stimulito intracellular signals that affect integrin lateral mobility.

The Decreased Mobility of LFA-1 Nanoclusters at Low-Ca2þ Levels Cor-relates with Restricted Microclustering Compromising Cell Adhesionand Spreading. Because several studies suggest that LFA-1 mobi-lity is associated with its ability to form microclusters (12, 30), weinvestigated the effect of extracellular Ca2þ on ligand-inducedLFA-1 clustering. For this purpose, dense ICAM-1 patterned sur-faces (5-μm squares) were prepared using microcontact printing(23). Clear accumulation of LFA-1 to the ICAM-1 regions wasobserved at physiological conditions compared to that at low-Ca2þ levels (Fig. 5 A and B). To quantify the degree of clusteringwe separated the data in two categories: (i) patches larger thanthe diffraction limit of our microscope (ca. 350-nm diameter)and defined them as microclusters and, (ii) smaller spots witha size limited by diffraction and defined them as nanoclusters(SI Text).

From 20 to 30 cells studied at each condition, the number ofL16þ microclusters on ICAM-1 regions was significantly differ-ent, with approximately 45% of the cells displaying L16þ micro-clusters at 0.4 mM Ca2þ, Mg2þ, and approximately 20% at0.04 mM Ca2þ (Fig. 5C). Interference reflection microscopyfurther showed extensive areas of close cell-ligand substratecontacts at 0.4 mM Ca2þ, Mg2þ, whereas reduced contact areasand finger-like extensions (filopodia) were observed at low-Ca2þlevels (Fig. 5 E–G). In addition, the number of cells that firmly

A B C

Fig. 4. The actin cytoskeleton regulates the diffusion of LFA-1 nanoclustersat low Ca2þ levels. (A and B) Normalized D distribution for (A) primed (L16þ)and (B) total (TS2/4) nanoclusters at 0.04 mM Ca2þ, with cells treated withCytoD (1 μg∕mL) (bars), compared to the control (DMSO, 1%) (dashes).(C) Normalized fractions of stationary (gray), slow (blue), and fast (orange)mobile subpopulations for L16 (Left) at 0.04 mM Ca2þ in DMSO and afterCytoD treatment. For this particular dataset, Dth ¼ 0.002 μm2∕s. Due tolimited statistics, only the stationary fractions of TS2/4 (Right) in DMSO andafter CytoD treatment were estimated. Error bars represent the standarddeviation (� ¼ p < 0.02 compared to stationary in DMSO; # ¼ p < 0.006 com-pared to stationary in DMSO). TS2/4 trajectories: 55 (DMSO) and 35 (CytoD).L16 trajectories: 94 (DMSO) and 72 (CytoD).

A B

C D

E F

G H

Fig. 5. LFA-1 microclustering and cell adhesion under static conditionsdepends on extracellular Ca2þ. (A and B) Representative frames of moviesrecorded in TIRF mode of L16þ-LFA-1 nanoclusters on monocytes seeded onICAM-1/BSA micropatterns at (A) 0.4 mM Ca2þ, Mg2þ and (B) 0.04 mM Ca2þ.Microclusters are surrounded in red, and nanoclusters are highlighted bywhite arrows. (Scale bars: 5 μm.) (C) Percentage of cells displaying microclus-ters at different cation conditions. Error bars are the standard deviationfrom three independent experiments. (D) Intensity distribution of nanoclus-ters on ICAM-1 regions at 0.04 mM Ca2þ (bars) and 0.4 mM Ca2þ, Mg2þ

(dashes). (E and F) Interference reflection microscopy of two representativemonocytes seeded for 20 min on ICAM-1 substrates at (E) 0.4 mM Ca2þ,Mg2þ and (F) 0.04 mM Ca2þ. Attachment areas were ð153� 100Þ μm2 inE and ð115� 44Þ μm2 in F, thus a reduction of 75% in firm contact area(55 cells inspected in each condition, over 2 separate experiments). (Scalebars: 5 μm.) (G) Histogram of firm cell attachment ratio for monocytesseeded for 20 min on ICAM-1 substrates at 0.04 mM Ca2þ (bars) and0.4 mM Ca2þ, Mg2þ (dashes). The attachment ratio was estimated by divid-ing the main contact area as measured by interference reflection micro-scopy by the corresponding bright field image. Mean attachment ratio is0.70 (σ ¼ 0.19) at 0.4 mM Ca2þ, Mg2þ and 0.55 (σ ¼ 0.16) at 0.04 mMCa2þ, thus a reduction to 79% compared to the total cell size at0.4 mM Ca2þ. (H) Relative cell adhesion at 0.4 mM Ca2þ, Mg2þ (white) and0.04 mM Ca2þ (black) at different seeding times on ICAM-1 substrates.Results from two independent experiments from 15 different bright fieldimages at each condition.

4872 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1116425109 Bakker et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 3,

202

1

adhered at low-Ca2þ levels reduced substantially compared tophysiological Ca2þ levels (Fig. 5H).

In the case of L16þ nanoclusters, we did not observe signifi-cant differences in the intensity distributions (Fig. 5D), indicatingthat low extracellular Ca2þ has no affect on LFA-1 nanocluster-ing. These results were further confirmed by transmission elec-tron microscopy of whole-mount monocytes (Fig. S8). Thus,although LFA-1 nanoclustering remains unperturbed at low-Ca2þ

levels, ligand-induced microclustering is greatly restricted anddirectly correlated with its decreased mobility on the cell mem-brane. Furthermore, these results demonstrate that extracellularCa2þ is not only an important regulator of ligand-binding affinitybut it is also required to couple ligand binding to global con-formational changes and downstream signals that trigger cellspreading and adhesion on ligand bearing substrates under staticconditions.

LFA-1 Mediated Adhesion of Monocytes Under Shear-Flow. Becauseextracellular Ca2þ affected conformational state, lateral mobilityand microclustering of LFA-1 on ICAM-1 bearing patterns,we studied the effect of extracellular Ca2þ on cell adhesion toICAM-1 substrates under shear flow. At low shear-flow condi-tions (0.2 dyn∕cm2) highest adhesion was found at 1 mM (Mn2þ,Ca2þ, Mg2þ), followed by 0.4 mM (Ca2þ, Mg2þ) (Fig. 6A), con-ditions where LFA-1 is laterally mobile and able to increase itsavidity state. Remarkably, adhesion was lowest at 0.04 mMCa2þ and slightly higher at 0.04 mMCa2þ, 1 mMMg2þ (Fig. 6A),a situation where high-affinity nanoclusters are generated butlateral mobility is compromised. Thus, despite the fact that lowextracellular Ca2þ levels promote primed LFA-1, the integrin isnot able to support firm adhesion.

At higher shear-flow conditions (0.5 dyn∕cm2), firm cell adhe-sion was equally maintained with 1 mM (Mn2þ, Ca2þ, Mg2þ)(Fig. 6B). However, cell adhesion was highly dependent on thepresence of Mg2þ, regardless of the Ca2þ levels used (Fig. 6B).Because low extracellular Ca2þ favors the extended form of theintegrin and Mg2þ increases the binding strength of LFA-1 toICAM-1 without affecting its lateral mobility (Fig. S5), these re-sults indicate that at higher shear forces affinity of LFA-1 to itsligand plays a more prominent role than its lateral mobility.

DiscussionEarlier models of LFA-1 regulation proposed that the moleculeis inactive and confined to the cytoskeleton in resting cells andreleased from its cytoskeletal constrains upon cell activation,resulting in increased mobility and clustering (12, 13, 15, 16).However, recent work on resting T cells demonstrated a morecomplex mobility pattern of LFA-1 with distinct diffusion profilesand different subsets of conformational states (31). Here weshowed that at physiological cation conditions, LFA-1 nanoclus-ters are predominantly mobile, with a subset of nanoclustersbeing in the extended, primed form. These primed nanoclustersshowed stationary, anomalous, and free diffusion behavior. Asthe L16 epitope recognizes full extension of the αL subunit(26, 27), regardless of whether the ligand-binding headpiece isclosed (intermediate form) or opened (high-affinity form), thediverse mobility profiles recovered in our studies might corre-spond to the different binding states of LFA-1. More comprehen-sive models describing different conformational states of LFA-1already hypothesized the existence of multiple intermediatestates depending on their transient interaction with the cytoske-leton (6). Our results now demonstrate this complexity. We pro-pose that this repertoire of multiple possibilities offered by LFA-1is beneficial for its diverse adhesive processes dynamically regu-lating migration, rolling, and firm arrest.

Our results also provide evidence for the strong effects ofextracellular Ca2þ on conformational state, lateral mobility,and ligand-induced microclustering of LFA-1. Although previousworks demonstrated the importance of extracellular Ca2þ regu-lating integrin activation (2, 28, 32), our data prove that extracel-lular Ca2þ also couples ligand binding to global conformationalchanges and downstream signaling that trigger cell spreading andfirm adhesion in the presence of its ligand. The mechanismby which extracellular Ca2þ couples conformational changes tolateral mobility on the cell membrane must be related to the pro-found changes that Ca2þ transmits throughout the global struc-ture of the integrin toward the cytoplasmic tails, by acting on itshybrid domain (33). Swinging out of the hybrid domain uponlowering of extracellular Ca2þ results in integrin extension andtransmembrane domain separation, so that both αL and β2 cyto-plasmic tails become available for interaction with other proteinson the inner side of the membrane that will regulate the lateralmobility of the receptor and its interaction with the cytoskele-ton (6).

Although our SDT experiments have been performed in theabsence of mechanical stimuli, recent simulations predicted thatforce accelerates the swinging out of the hybrid domain, facilitat-ing the transition into a high-affinity open headpiece state (33).Low shear forces were also seen to induce conformationalchanges on the α-I domain of LFA-1, increasing adhesive inter-actions with its ligand by stabilizing the open conformation of thisdomain (34). Interestingly, it has been recently suggested thatpreformed integrin anchorage to the cytoskeleton could be keyfor the integrin to load low forces and undergo instantaneous ac-tivation by surface-bound ligands (35). Our data demonstratethat, in the resting state, a significant fraction of LFA-1 is inprimed form and preanchored to the cytoskeleton. Activationof the integrin (by Ca2þ removal, Mn2þ, or activating antibodies)leads to an increase of cytoskeleton anchorage forming nascentsites for adhesion after ligand engagement. It will be then ex-pected that this preformed cytoskeleton anchorage create theperfect scenario for tensile forces to facilitate/accelerate the pro-cess of leukocyte adhesion on blood vessels. Low-affinity, mobileLFA-1 nanoclusters, as identified here, might contribute by dif-fusing to the adhesion sites initiated by the anchored extendedLFA-1 subsets.

A B

Fig. 6. Extracellular Ca2þ plays a differential role on LFA-1 adhesiveness un-der shear-flow conditions. Cell binding to ICAM-Fc coated surfaces quantifiedas a percentage of maximum binding observed at 1 mM Ca2þ, Mg2þ, Mn2þ at(A) 0.2 dyn∕cm2 and (B) 0.5 dyn∕cm2. Error bars are the standard deviationover average binding calculated over five different areas. One out of twoexperiments are presented (� ¼ p < 0.007 compared to 1 mM Ca2þ, Mg2þ,Mn2þ; # ¼ p < 0.005 compared to 1 mM Ca2þ, Mg2þ, Mn2þ; † ¼ p < 0.01compared to 0.04 mM Ca2þ; ‡ ¼ p < 0.007 compared to 0.4 mM Ca2þ.

Bakker et al. PNAS ∣ March 27, 2012 ∣ vol. 109 ∣ no. 13 ∣ 4873

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 3,

202

1

Materials and MethodsCell Culture and Labeling for Single-Dye Tracking. THP-1 monocytes werecultured in RPMI medium 1640 Dutch modification medium supplementedwith 10% fetal calf serum and antibioticantimycotic from Gibco. Cells werelabeled with 2 μg∕mL TS2/4-ATTO520, or 2 μg∕mL L16-ATTO647N to allow forsingle-dye tracking experiments (SI Text and Fig. S1). Further details on sub-strate preparation, micropatterns, labeling conditions under different cationconditions, and CytoD treatment are found in SI Text.

Shear-Flow Chamber Assay. Details of the experiments are described in SI Text.

Single-Dye Tracking. Experiments were performed using a homemade single-molecule Epi/total internal reflection fluorescence (TIRF) microscope. Sampleswere illuminated either in Epi or TIRF modes. Excitation was provided by aHe:Ne laser (4 ms at 633 nm, 1 kW∕cm2) or an Arþ-Krþ laser (2 ms at514.5 nm, ca. 2 kW∕cm2). Fluorescence was collected with a 1.45 N.A. oil

immersion objective and guided into an intensified CCD camera. Individualframes were retrieved at a frame rate of 10–20 Hz. Experiments wereperformed at 37 °C. MSD analysis was used to analyze trajectory data andto derive the short-range D values. CPD analysis was performed accordingto Schütz et al. (25) to derive the long-term D values of mobile trajectories(see SI Text for further details).

ACKNOWLEDGMENTS. We thank M. Rivas, M. Somunyudan and B. Joostenfor technical assistance, and C. Manzo for fruitful discussions. This workwas supported by grants from the Stichting voor Fundamenteel Onderzoekder Materie, IMMUNANOMAP EC-Research Training Network, SpanishMinistry of Science and Technology (MAT2010-19898), and Generalitat deCatalunya (2009 SGR 597) (to M.F.G.-P.); the Netherlands Organization forScientific Research (NWO) Veni Grant 916.66.028 and Human FrontiersScience Program Young Investigator Grant (A.C.); and partially by a Fondode cooperacion internacional en ciencia y tecnologia Grant (C.G.F.). A.C. isthe recipient of an NWO Meervoud subsidy.

1. Takagi J, et al. (2002) Global conformational rearrangements in integrin extracellulardomains in outside-in and inside-out signaling. Cell 110:599–611.

2. Luo BH, Carman CV, Springer TA (2007) Structural basis of integrin regulation andsignaling. Annu Rev Immunol 25:619–647.

3. Takagi J, Springer TA (2002) Integrin activation and structural rearrangement. Immu-nol Rev 186:141–163.

4. van Kooyk Y, van Vliet SJ, Figdor CG (1999) The actin cytoskeleton regulatesLFA-1 ligand binding through avidity rather than affinity changes. J Biol Chem274:26869–26877.

5. Katagiri K, et al. (2003) RAPL, a Rap1-binding molecule that mediates Rap1-inducedadhesion through spatial regulation of LFA-1. Nat Immunol 4:741–748.

6. Dustin ML, Bivona TG, Philips MR (2004) Membranes as messengers in T cell adhesionsignaling. Nat Immunol 5:363–372.

7. Shimaoka M, et al. (2003) Structures of the alpha LI domain and its complex withICAM-1 reveal a shape-shifting pathway for integrin regulation. Cell 112:99–111.

8. Carman CV, Springer TA (2003) Integrin avidity regulation: Are changes in affinity andconformation underemphasized? Opin Cell Biol 15:547–556.

9. Ward MD, Dembo M, Hammer DA (1994) Kinetics of cell detachment: peeling ofdiscrete receptor clusters. Biophys J 67:2522–2534.

10. Chen A,Moy VT (2000) Cross-linking of cell surface receptors enhances cooperativity ofmolecular adhesion. Biophys J 78:2814–1820.

11. Selhuber-Unkel C, et al. (2008) Cooperativity in adhesion cluster formation duringinitial cell adhesion. Biophys J 95:5424–5431.

12. Kucik DF, et al. (1996) Adhesion-activating phorbol ester increases the mobility ofleukocyte integrin LFA-1 in cultured lymphocytes. J Clin Invest 97:2139–2144.

13. Stewart MP, McDowall A, Hogg N (1998) LFA-1-mediated adhesion is regulated bycytoskeletal restraint and by a Ca2þ-dependent protease, calpain. J Cell Biol140:699–707.

14. Sigal A, et al. (2000) The LFA-1 integrin supports rolling adhesions on ICAM-1 underphysiological shear flow in a permissive cellular environment. J Immunol 165:442–452.

15. Jin T, Li J (2002) Dynamitin controls 2 integrin avidity by modulating cytoskeletalconstraint on integrin molecules. J Biol Chem 277:32963–32969.

16. Zhou X, Li J, Kucik DF (2001) The microtubule cytoskeleton participates in control of β2integrin avidity. J Biol Chem 276:44762–44769.

17. van Kooyk Y, et al. (1989) Enhancement of LFA-1-mediated cell adhesion by triggeringthrough CD2 or CD3 on T lymphocytes. Nature 342:811–813.

18. Dustin ML, Springer TA (1989) T-cell receptor cross-linking transiently stimulatesadhesiveness through LFA-1. Nature 341:619–624.

19. Marlin SD, Springer TA (1987) Purified intercellular adhesion molecule-1 (ICAM-1) is aligand for lymphocyte function-associated antigen 1 (LFA-1). Cell 51:813–819.

20. Cambi A, et al. (2006) Organization of the integrin LFA-1 in nanoclusters regulates itsactivity. Mol Biol Cell 17:4270–4281.

21. van Zanten TS, et al. (2009) Hotspots of GPI-anchored proteins and integrin nano-clusters function as nucleation sites for cell adhesion. Proc Natl Acad Sci USA106:18557–18562.

22. van Zanten TS, et al. (2010) Directmapping of nanoscale compositional connectivity onintact cell membranes. Proc Natl Acad Sci USA 107:15437–15442.

23. Diez-Ahedo R, et al. (2009) Dynamic re-organization of individual adhesion nanoclus-ters in living cells by ligand-patterned surfaces. Small 5:1258–1263.

24. Shulman Z, et al. (2009) Lymphocyte crawling on and through activated endotheliumis mediated by millipede-like contacts generated by chemokine triggered high affinityLFA-1 primed by shear forces. Immunity 30:384–396.

25. Schütz GJ, Schindler H, Schmidt T (1997) Single molecule microscopy in model mem-branes reveals anomalous diffusion. Biophys J 73:1073–1080.

26. Keizer GD, et al. (1988) A monoclonal antibody (NKI-L16) directed against a uniqueepitope on the alpha-chain of human leukocyte function-associated antigen 1 induceshomotypic cell–cell interactions. J Immunol 140:1393–1400.

27. Xie C, et al. (2004) The integrin-subunit leg extends at a Ca2þ-dependent epitope inthe thigh/genu interface upon activation. Proc Natl Acad Sci USA 101:15422–15427.

28. Chen JF, Salas A, Springer TA (2003) Bistable regulation of integrin adhesiveness by abipolar metal ion cluster. Nature Struct Biol 10:995–1001.

29. Andrew D, et al. (1993) KIM185, a monoclonal antibody to CD18 which induces achange in the conformation of CD18 and promotes both LFA-1- and CR3-dependentadhesion. Eur J Immunol 23:2217–2222.

30. Lub M, et al. (1997) Dual role of the actin cytoskeleton in regulating cell adhesionmediated by the integrin lymphocyte function-associated molecule-1. Mol Biol Cell8:341–351.

31. Cairo CW, Mirchev R, Golan DE (2006) Cytoskeletal regulation couples LFA-1 confor-mational changes to receptor lateral mobility and clustering. Immunity 25:297–308.

32. Salas A, et al. (2002) Transition from rolling to firm adhesion is regulated by the con-formation of the I domain of the integrin lymphocyte function-associated antigen-1.J Biol Chem 277:50255–50262.

33. Puklin-Faucher E, et al. (2006) How the headpiece hinge angle is opened: New insightsinto the dynamics of integrin activation. J Cell Biol 175:349–360.

34. Astrof NS, et al. (2006) Importance of force linkage in mechanochemistry of adhesionreceptors. Biochemistry 45:15020–15028.

35. Alon R, Dustin ML (2007) Force as a facilitator of integrin conformational changes dur-ing leukocyte arrest on blood vessels and antigen-presenting cells. Immunity 26:17–27.

4874 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1116425109 Bakker et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 3,

202

1