Inhibition of cell adhesion by a synthetic polymer adsorbed to glass

shown under defined hydrodynamic stress

NORMAN F. OWENS, DAVID GINGELL

Department of Anatomy and Biology as Applied to Medicine, The Middlesex Hospital Medical School, Cleveland St, London W1P 6DB, L'K

and PAUL R. RUTTER

B.P. Research Laboratory, British Petroleum Pic, Snnbiin^ on Thames, Middlesex, UK

Summary

A co-polymer -with hydrophobic and hydrophilicsegments was allowed to adsorb from aqueoussolution onto glass previously made hydrophobicby derivatization with octadecyl dimethylchloro-silane. The polymer is thought to adsorb via itshydrophobic segments, leaving the hydrophilicsegments free to extend into the water. Afterallowing cells to settle on the treated surface, theshear stress at the chamber wall required toremove red blood cells, Dictyosteliutn discoid-eum amoebae and Escherichia coli was deter-mined in a calibrated laminar flow chamber. Onoctadecyl glass a shear stress of 2-3 Nm~2 evicts50% of adherent red cells and E. coli. No D.

discoideum amoebae could be removed at5Nm~2. In striking contrast, the lowest exper-imentally obtainable shear stress of 0*03 Nm~z

removes 97-0-99-5% of cells of all three typesfrom the polymer-treated surface, even after acell residence time of l h without flow in theabsence of free polymer. The minimum shearstress of 0-03 Nm~2 corresponds to only =20times the force of gravity on a red cell. Themechanism of action of the polymer and theimplications of the results are discussed.

Key words: cell adhesion, blood compatible syntheticpolymer.

Introduction

The prevention of cell adhesion to surfaces undercontrolled conditions has long been the goal for thedevelopment of surfaces that are non-adhesive to bloodcomponents, marine microorganisms and bacteria.The search has been for a material that would haveminimal attractive forces for cells and proteins, andthat would possess maximal repulsive forces. Thisimplies that the combined energies of van der Waals'attraction, entropic attraction (i.e. 'hydrophobic bond-ing') and hydrogen bonding should be smaller than thesum of electrostatic repulsion, hydration repulsion andentropic repulsion due to thermal motion of flexiblemolecular chains. It should, in addition, be mechan-ically and chemically stable. The search for such amaterial has largely concentrated on syntheticpolymers (see Andrade, 1985a; Nordeei al. 1986) andsince their surface properties in a biological milieu willbe modified by protein adsorption, the latter processhas been extensively studied (see Andrade, 19856).

Journal of Cell Science 87, 667-675 (1987)Printed in Great Britain © The Company of Biologists Limited 1987

The relationship between surface physicochemicalproperties and biocompatibility is not a trivial problemand is still much disputed after more than two decades(see the critical discussion by Ruckenstein & Gourisan-kar, 1984). In this period only limited progress hasbeen made towards the goal of a blood-compatibleinterface for vascular prostheses or kidney dialysisequipment (Brash, 1983).

Of the large number of polymers that have beenassessed for blood compatibility, polyethylene oxide(PEO), either alone or as a component of a co-polymer,has been found to have notable anti-adhesive proper-ties. George (1972) described a dramatic reduction inthe adhesion of platelets to glass in the presence ofPEO. Earlier, Hiatt et al. (1971) had found that PEOprevented the adsorption of rabies virus to glass.Whicher & Brash (1978) and Brash & Uniyal (1979)showed that little protein adsorption occurred on asolid co-polymer composed of PEO and polyurethane.Thrombin and platelet adsorption to similar co-polymers was studied by Sa da Costa et al. (1980,

667

1981), using thick layers of polymer on glass beads,produced by evaporating the non-polar solvent. Theyperformed XPS analyses on the particle surface layersand concluded, after comparing the results with XPSof polyurethanes alone, that lack of cell and proteinadsorption was a function of the PEO moiety andsuggested that an amorphous network of stronglyhydrated PEO would be maximally anti-adhesive.Merrill et al. (1982) further characterized the lowplatelet adhesiveness of related co-polymers based onpolyethylene glycol and polyurethane. Similarly, Moriet al. (1982) described the striking in vitro and in vivoanti-thrombogenic properties of PEO-methacrylate-PVC co-polymers, but detected substantial proteinadsorption by scanning electron microscopy (SEM)after one day under in vivo conditions. They madeand tested polymers with PEO side-chains (arrangedherringbone fashion on a backbone of PVC) of variouslengths (5-100 monomers) and found the longest to bethe most blood-compatible.

Against the background of these results, we describea relatively simple and well-defined water-solublepolymer consisting of two polyethylene oxide segmentsspaced with a relatively hydrophobic polypropylenesequence. The polymer adsorbs strongly to a hydro-phobic glass surface, where it almost totally preventsthree very different cell types from adhering, as judgedby the criterion of liquid shear in a calibrated flowsystem.

Materials and methods

Materials

The bifunctional polymer Pluronic (F108) was obtained fromICI Pic. Penicillin-streptomycin (5000IU) from GibcoChemicals, Middlesex, UK was stored below 0°C. Concan-avalin A (ConA) conjugated with tetramethylrhodamine Bisothiocyanate was purchased from Sigma Chemical Co.,Poole, UK. A stock solution O-Smgml"1 prepared in Trisbuffer at pH 7-4 was stored in 0-2 ml samples below 0°C untilrequired. Bacteriological peptone and yeast extract wereobtained from Oxoid Chemicals Ltd, Basingstoke, Hamp-shire, UK. Disodium hydrogen orthophosphate, sodiumdihydrogen orthophosphate, sodium chloride and D+-glucosewere all of AnalaR quality from BDH Chemical Co., Poole,UK. The procedure for obtaining surface chemically purewater by distillation from alkaline potassium permanganatewill be described elsewhere (Owens et al. 1987). Microscopecoverslips (40 mm X 22 mm X 0-17 mm) from Chance Prop-per Pic, Smethwick, UK were rendered hydrophobic byovernight immersion in a 2% solution of octadecyl dimeth-ylchlorosilane (Sigma, Southampton, UK) in chloroform.The treated coverslips were first rinsed with chloroform, thenin distilled water to remove HC1. If required for later use theywere stored in chloroform. Precision micrometer flow valves(Nupro Co., types S and M) were purchased from NorthLondon Valve & Fitting Co., London, UK.

Cultures

Vegetative cells of Dictvostelium discoideum (Ax2 strain)were grown to the fully differentiated spore fruiting stage at22°C on 2% nutrient agar inoculated with Escherichia coliB/r. The dormant spores were germinated in glucose-freeaxenic medium containing 0-5 ml penicillin-streptomycin for72 h in a rotary incubator at 22CC. This was followed by agrowth cycle in axenic medium supplemented with 86 mM-glucose. Ax2 amoebae were isolated from shaken culturesduring log-phase growth ( l x 106cells ml"1) by allowing thecells to settle onto a clean glass surface rather than the usualcentrifugation routine. In so doing we avoided any tendencyfor the cells to rupture during harvest. The adherent cellswere first washed with 20mM-NaCl and then resuspended inthe same electrolyte using gentle fluid flow from a Pasteurpipette to dislodge cells from the glass surface. Streamingflow experiments were done in unbuffered 20 mM-NaCl at thepH of our distilled water (pH 5-8-6-2). The work was donein a temperature-controlled laboratory at 20(±0-5)°C.

Human red blood cells were obtained by venipuncture intocitrate buffer, pH 6-8. The cells were washed three times bycentrifuging with Dulbecco A phosphate-buffered saline(PBS), pH7-4 (Oxoid, UK) at 400 £. A 10ft! sample of thepacked cells was diluted with similar PBS to give a celldensity of 8xl0 6 ml~ ' . This was kept at room temperatureuntil required. Streaming flow experiments were also done inPBS at pH 7-4.

E. coli B/r were incubated in nutrient broth no.2 (Oxoid)for 48 h at 37°C. A 30 ml suspension of bacteria was pelletedby centrifugation at 1000 £f for lOmin. Bacteria were washedby four successive 10-min centrifugations at 1000 g using30 ml samples of PBS at pH7-4. The washed cells wereresuspended and stored in PBS until required. Streamingflow experiments with E. coli were conducted in PBS,pH7-4.

Methods

A laminar flow of electrolyte from an elevated reservoir waspassed through a rectangular conduit in which a derivatizedglass coverslip formed the upper wall of a closed parallel platechannel 388 ftm wide. The wall shear TW exerted by the flowagainst the solid surface was calculated for varying rates offlow from the equation:

dP' d/ b , (1)

where P is the hydrostatic pressure drop across the conduit oflength / and wall separation 2b. The conduit design, flowsystem and calibration procedure are described in detailelsewhere (Owens et al. 1987) so only the salient features aregiven here.

The components of the conduit are illustrated in Fig. 1. Aglass coverslip C was held in a recess on the upper face of apolycarbonate block D by a stainless steel frame A. This wassecured by screws that passed through a silicone rubbergasket B into a stainless steel baseplate E. The flow conduitwas formed between the coverslip and a flat-bottomedchannel across the polycarbonate block D. Fluid entersthrough a cylindrical pipe integral with the end wall of blockD leading into a confusor of elliptical section and then intothe conduit. Fluid exit is via a symmetrical arrangement.

668 vV. F. Owens et al.

Fig. 1. The flow chamber.

The baseplate E, which has a central hole for transmittedillumination, locates the assembly on the microscope stage.

A Bowline of glass and silicone rubber tubing was connec-ted to the chamber and the whole system filled with electro-lyte, care being taken to exclude any air bubbles. Exchange ofelectrolyte for 0 1 % polymer solution followed, to ensurethat the polymer-coated glass did not become exposed at anystage to air. After 2 h the system was flushed with electrolyteand the cells were injected into the inverted chamber on aninverted microscope through a silicone rubber diaphragm in aside arm, using a hypodermic syringe with an 8 cm needle.Sedimentation of cells onto the derivatized glass wall wasessentially complete after 20min. Cells were observed with aZeiss 25 X oil-immersion objective lens under bright-field andIRM illumination. The flow rate through the conduit wasvaried by means of precision micrometer valves mounted inparallel in the flowline. On polymer-treated surfaces, cellswere exposed to a minimal flow rate of 001 mis"1 for 30 s incontrast to those on control surfaces where the flow rate wasextended to 2-0 ml s~' over the same period. Cell-substratuminteractions under hydrodynamic shear were recorded onclosed-circuit TV using a Chubb surveillance camera connec-ted to a video recorder (Sony U-Matic) and monitor. Laterexperiments utilized a sensitive 'Falcon' SIT camera (Cus-tom Camera Devices Ltd, Wells, UK).

Calibration of flow systemFor a flow chamber of the type used, the Reynold's number isgiven by R= \-32VnxM^b/rj, where Vmnx is the maximumvelocity of flow (along the central axis of the chamber), 2b isthe separation of the flat parallel walls and 7) is the absoluteviscosity. The critical Reynold's number Rc corresponds to acritical value of Vnax below which flow is laminar and abovewhich it is turbulent (Eskinazi, 1975, p. 384). For a rectangu-lar cuvette like ours where length > width *> depth,Rc = 2300. In our system at maximum flow R = 200 < Rc sothat flow is laminar under all our experimental conditions.

Although flow in the cuvette will always be laminar, theliquid must pass a certain distance along the channel before aconstant parabolic velocity profile is established. This estab-lishment length Lc is a function of chamber shape andReynold's number, Le = 0013W? (Sparrow, 1955). For oursystem Lc is less than 1 % of the conduit length so flow will befully developed over virtually the entire length of the conduit.This conclusion is consistent with the findings of VanWagenen & Andrade (1980) for the measurement of stream-ing potentials generated by a laminar flow of electrolytebetween parallel glass plates. Fig. 2 shows a linear relation-ship between the wall shear, calculated from equation (1),and the applied pressure difference across the chamber (P).

Results

Numerical results for the behaviour of red blood cells,D. discoideum and E. coli under hydrodynamic stresson control octadecyl glass and F108-treated octadecylglass are shown in Table 1. On F108-treated glassbetween 97-0 and 99-5 % of cells that were initially incontact with the surface were removed at a wall shearstress rw = 0-03 N m~2. On the control surface 100-0 %

Inhibition of cell adhesion 669

1(X) 600 700 800300 400 500Pressure p (Nm 2)

Fig. 2. Calculated shear stress at the glass wall of the flowchamber as a function of applied hydrostatic pressuredifference across the chamber. (# ) Low flow rate valve;(O) higher flow rate valve. Final point, both valves open.

Table 1. Behaviour of cells under hydrodynamicstress

A. Polymer-treated glass

RBCE. coliD. discoideum

B. Control

RBCE. coliD. discoideum

Number ofexpts

542

Beforeflow

13611070

137

octadecyl glass

Numberof Before .

cxpts flow

2 1861 3603 29

For 50% removal in 30 s:E.coli, 3-0 Nm-2.

After

0-03

18636029

After flow(0-03 Nm- 2 )

1754

% Cellsremoved

98-899-597-0

flow °h Cells removedr2) (Nm-2)

5 0-03

0 0-0126 0-029 0-0

red blood cells (RBC),

5

100-065-00-0

2-3 Nn-T2;

of all three cell types remained stuck at this shear stress.Increasing the flow rate to give rw = 5Nm~ causedthe removal of all red cells and 65-0% of E. coli butfailed to remove any D. discoideum amoebae. Nor wereany of the latter removed at 5-9Nm~2, which iscurrently our maximal obtainable shear stress. Wefound that half the red cell population and also halfthe bacteria were removed at 2-3 and 3'0Nm~2,respectively.

Fig. 3A,B,C shows transmitted light images of redcells, D. discoideum amoebae and E. coli on F108-

treated glass before flow, taken from videotape. Theuneven background is due to light transmitted by thenon-optical plastic base of the chamber. Fig. 3D,E,Fshows typical corresponding fields at minimum flow,giving O03Nm~2 wall shear stress. The 0-5 s photo-graphic exposure time shows cells that have beenremoved by flow as streaks or blurs, while the few staticcells are sharply focused. The behaviour of cells on thecontrol surface of octadecyl glass, before and after flow,is shown in Fig. 4. We also made two tests on thepolymer-treated glass to assess its durability. In thefirst, physiological saline was allowed to flow throughthe chamber at full rate for 1 h. After this the surfacewas as anti-adhesive to red blood cells as it was beforeexposure to prolonged flow. In the second test thepolymer-coated glass was allowed to air dry for 30 min,before being re-hydrated in physiological saline for30 min. The prevention of cell adhesion by this surfacewas indistinguishable from that of polymer-treatedglass kept rigorously out of contact with air.

We found that it was possible to define the boundaryzone between polymer-treated and untreated glassusing a fluorescent dye adsorption test. Rhodamine-conjugated ConA was shown to adsorb strongly tohydrophobic glass but failed to adsorb to glass treatedwith F108, thus both providing a convenient practicalassay for polymer adsorption and showing that F108vetoes protein adsorption as well as cell adhesion.

Discussion

The parallel plate hydrodynamic shearing techniquethat we have used for assessing cell-to-substratumadhesion is well known. It has been extensivelyemployed to investigate both the adsorption of cellsfrom flowing suspensions onto the chamber walls aswell as the removal of attached cells (Doroszewski et al.1977, 1979; Forrester & Lackie, 1984; Hochmuthe/ al.1973; Mohandasef al. 1973, 1974). The hydrodynamicremoval force on a cell attached to the chamber wall canbe determined from the wall shear stress rw and there isreason to believe that this force is not significantlydependent on the details of cell shape and the degree ofspreading (Hubbe, 1981) for any given cell type.

Our data for red cells on siliconized glass (Table 1)can be directly compared with the work of Mohandas etal. (1974) for the removal of red cells by hydrodynamicshear in a saline solution. These authors report thatcells are removed during 300 s flow only whenT W > 0 - 3 5 Nm~2. Cell removal was found to depend onboth Tw and time. Their fig. 8 shows that half the cellsare removed in 300 min on exposure to 1 N m~2 and allare gone in the same time at 1-5 N m~2. We found thatin 30 s half the initially adherent cells are removed fromoctadecyl glass at 3 N m~ and that all are removed at

670 N. F. Owens et al.

*r • . •'••* ''•%. w r003 Nm2

r i ' I • •s ••<•-!.."< . /

003 Nm2

Fig. 3. Cells on F108-treated glass before flow (left column) and after flow at the minimum rate (right column). A,D. Redcells; B,E, Dictyostelium; C,F, E. coli. Wall shear stress and scale inset.

5 N m 2. Since our flow times differ substantially fromthose of Mohandas et al. (1974) we can only state thatthere is no evidence that the two sets of results are

inconsistent. In view of our results with F108, it is ofinterest that Mohandas et al. (1974) found a largedecrease in percentage adhesion in the presence of

Inhibition of cell adhesion 671

fibrinogen. Unfortunately they did not measure pro-tein adsorption to siliconized glass and give no shearstress data. The analysis of red cell removal fromoctadeeyl glass is considered in more detail elsewhere(Owens et al. unpublished data).

Our data for bacterial adhesion can be comparedwith those of Fowler & MacKay (1980), who used aradial flow chamber to examine the growth of unident-ified cocci in protein-free media. Although cell countsare not given, cells are stated to be unable to attach andmultiply where the wall shear stress exceeds 2-6 N m~2

(units of Tw in their Tables 7.1 and 7.2 are not correctlyexpressed). This is very similar to our value of3-0 N m~2 for 50 % removal of E. coli, but owing to thevery different conditions employed, the similarity ofthe two sets of results may be fortuitous. Taking thedimension of E. coli as a cylinder of 0-5 jum radius and3 jxra long, the bacterium is equivalent to a sphere of0-8fim radius. Consequently, using E=32r~2

(Hubbe, 1981) we obtain a characteristic removal forceof 1X10"10N per cell at 3 N m " 2 .

In the case of F108-treated glass, the minimal wallshear stress used in our system, 0-03 Nm~2, removesbetween 97-0 and 99-5 % of attached cells, according tocell type. In view of the widely different surfacecompositions of these cells (prokaryote, eukaryote andslime mould) this result argues persuasively that F108is a very potent general anti-adhesive. Therefore itsmode of adsorption and mechanism of action are ofconsiderable interest.

Pluronic F108 consists of two hydrophilic poly-oxyethylene chains separated by a central hydrophobicpolyoxypropylene chain and has the general formula:

H-(OCH2CH2),,(OCH2CH(CH3)V-(OCH2CH2)nOH ,

where a is approximately 130 and b is approximately45. The molecule adsorbs on hydrophobic surfaces insuch a way that the polyoxyethylene chains are oriented

003

. •

Fig. 4. Cells on control hydrophobed glass before flow (left) and after flow (right). A,C. Red cells; B,D, Dictvostelium.

672 N. F. Owens et al.

CH,

CH,

CH.

CH,

CH,

Water

Polyethylene oxide)

Poly(propylene glycol)

0-CH-CH,-0-CH-CH,-0 0-CH-CH,-0\ ' A ' A A ' I

CH ACH \ A ^ J

Fig. 5. Diagram of a molecule of the bifunctional polymerF108 adsorbed at the interface between water andhydrophobed glass.

away from the surface and into the aqueous phase (seeFig. 5) to produce an adsorbed layer approximately26 nm thick (Kayes & Rawlins, 1979). The highsolubility of polyoxyethylene is usually attributed tothe helical structure that it adopts in water, resulting inan oxygen-oxygen distance of 0-288 nm, which is verysimilar to that between these atoms in water itself. Thecorresponding distance of 0-75 nm in polyoxypro-pylene is very different, rendering this material water-insoluble (Tronel-Peyroz et al. 1983).

Since the molecular composition of F108 includes noformal negative charges, Debye electrostatic repulsioncannot be the mechanism of non-adhesiveness. It isinteresting that charge repulsion is in any case oflimited effectiveness in preventing adhesion.Trommler et al. (1985) found that although the pro-portion of red cells adhering to glass was reduced at lowionic strength (when repulsion is substantiallyincreased) a significant proportion of cells neverthelessmade molecular contacts with the glass and becamestrongly adherent.

Thus we seek another explanation for the action ofF108. The fact that red cells can remain indefinitely incontact with a 0-01 % aqueous solution of the polymer,or on a surface treated with polymer, without haemoly-sis shows that it does not disrupt the cell membrane.There remain only two likely candidates for the repul-sion observed, steric (or entropic) repulsion and waterstructuring (hydration repulsion).

Steric repulsion is to be expected because of theextension of the hydrophilic arms into the water. If so,repulsion should begin when the glycocalyx of anapproaching cell begins to interdigitate with the hydro-philic segments of the adsorbed polymer, thus reducingthe available modes of gyration of the polymer chainsand decreasing entropy. A direct measurement ofintermolecular repulsion associated with PEO chainshas been obtained by Klein & Luckham (1982) usingthe Israelachvili technique. They measured the forcebetween mica plates with adsorbed PEO in 0-1 M-KNO3. Repulsion began at a mica-mica separation of6R (R = radius of gyration of the polymer) and rosemonotonically as the separation was reduced. Littleevidence of desorption over 72h was found. Thus theeffective length of the individual polymer chains whenthey began to repel mutually is 37?. This corresponds to18-20 nm for one polymer (iV/r = 40000; degree ofpolymerization, TZ~900) or 33-38 for another(Mr= 100000; « = 2300). Under conditions wherePEO can simultaneously adsorb to both mica plates,cross-bridging forces could be measured (Klein, 1986).These results suggest that repulsion in the case of thePEO co-polymer that we used (;/ ~ 130) might begin ata comparatively small effective extension distance, say3nm.

Evidence for thermal flexibility of PEO chains in amixed polymer was obtained by Mori et al. (1982) whorelated the width of the peak I3C nuclear magneticresonance (n.m.r.) signal to the degree of polymeriz-ation (ti) of the PEO moiety. They found that increas-ing n to 100 decreases the width of the signal peak,indicating high flexibility (no graph shown).

The other probable source of repulsion, water struc-turing, as an important component of interaction forcesis a relatively recent discovery (Le Neveu et al. 1977;Parsegian et al. 1979; Parsegian & Rau, 1984; see alsoIsraelachvili, 1985). The electric field at a surface (dueto adsorbed ions, ionogenic groups or possibly dipoles)alters the orientation and thus the structure of thenearby water. The result is that water molecules withinabout 2nm of a surface are to some extent associatedwith the surface. Although the individual bindingenergy per molecule is extremely small, the energyneeded to remove large numbers of them from anymacroscopic region between approaching surfaces isvery substantial. Between the lamellae of lecithin liquidcrystals the hydration pressure at 1 nm separationexceeds 107Nm~2 and rises to near 109Nm~2 or 104

atmospheres at 0-2 nm (Parsegian et al. 1979). How-ever, until the extent of water structuring associatedwith F108 is measured it will not be possible toappreciate fully how the PEO exerts its remarkableanti-adhesive effects.

Inhibition of cell adhesion 673

The authors thank British Petroleum Pic for financialsupport and for permission to publish this paper. D.G.thanks the Wellcome Trust and the SERC for financialsupport.

References

ANDRADE, J. D. (1985a). Surface and Interfacial Aspectsof Biomedical Polymers, vol. 1, Surface Chemistry andPhvsics (ed. J. D. Andrade). New York, London:Plenum Press.

ANDRADE, J. A. (19856). Surface and Interfacial Aspects ofRiomedical Polymers, vol. 2, Protein Adsorption (ed. J.D. Andrade). New York, London: Plenum Press.

BRASH, J. L. (1983). Hydrophobic polymer surfaces and

their interaction with blood. Ann. N. Y. Acad. Sci. 283,356-371.

BRASH, J. L. & UNIYAL, S. J. (1979). Dependence of

albumin-fibrinogen simple and competitive adsorptionon surface properties of biomaterials. jf. Polym. Sci.,Poly. Symp. 66, 377-389.

DOROSZEWSKI, J., GOLAB-MEYER, Z. & GURYN, W. (1979).

Adhesion of cells in flowing suspensions: effects ofshearing force and cell kinetic energy. Microvasc. Res.18, 421-433.

DOROSZEWSKI, J., SKIERSKI, J. S. & PRZADKA, L. (1977).

Interaction of neoplastic cells with glass surface underflow conditions. Expl Cell Res. 104, 335-343.

ESKINAZI, S. (1975). Principles of Fluid Mechanics.Boston: Allyn & Bacon.

FORRESTER, J. V. & LACKIE, J. M. (1984). Adhesion of

neutrophil leucocytes under conditions of flow. J. CellSa. 70, 93-110.

FOWLER, H. W. & MCKAY, A. J. (1980). The

measurement of microbial adhesion. In MicrobialAdhesion to Surfaces (ed. R. C. W. Berkley et al.), p.143. Chichester, UK: Ellis Horwood.

GEORGE, N. J. (1972). Direct assessment of plateletadhesion to glass: a study of the forces of interaction andthe effects of plasma and servan factors, platelet functionand modification of the glass surface. Blood 40, 862-874.

HIATT, C. W., SHELVKOV, A., ROSENTHAL, E. J. &

GALIMORE, J. N. (1971). Treatment of controlled poreglass with poly(ethylene oxide) to prevent adsorption ofrabies virus. J. Chromatogr. 56, 362-364.

HOCHMUTH, R. M., MOHANDAS, N. & BLACKSHEAR, P. L.

(1973). Measurement of the elastic modulus for red cellmembrane using a fluid mechanical technique. Biophys.J. 13, 747-762.

HUBBE, M. A. (1981). Adhesion and detachment ofbiological cells in vitro. Progr. Surface Sci. 11, 65-138.

ISRAELACHVILI, J. N. (1985). lntennolecular and SurfaceForces. London: Academic Press.

KAYES, J. B. & RAWLINS, D. A. (1979). Adsorption

characteristics of certain polyoxyethylene-polyoxypropylene block copolymers on polystyrene latex.Coll. Polymer Sci. 257, 622-629.

KLEIN, J. (1986). Surface interactions with adsorbedmacromolecules. J. Coll. Interface Sci. I l l , 305-313.

KLEIN, J. & LUCKHAM, P. (1982). Forces between twoadsorbed polyethylene oxide layers immersed in a goodaqueous solvent. Nature, Land. 300, 429-431.

LE NEVEU, D. M., RAND, R. P., PARSEGIAN, V. A. &

GINGELL, D. (1977). Measurement and modification offorces between lecithin bilayers. Biophys. J. 18, 209-230.

MERRILL, E. W., SALZMAN, E. W., WAN, S., MAHMUD,

N., KUSHNER, L., LINDON, J. N. & CURME, J. (1982).

Platelet-compatible hydrophilic segmented polyurethanesfrom polyethylene glycols and cyclohexane diisocyanate.Trans. Am. Soc. artif. int. Organs 28, 482-487.

MOHANDAS, N., HOCHMUTH, R. M. & SPAETH, E. E.

(1974). Adhesion of red cells to foreign surfaces in thepresence of flow. J . Biomed. mat. Res. 8, 119-136.

MOHANDAS, N., HOCHMUTH, R. M. & WILLIAMSON, J. R.

(1973). Deformation of blood cells adhering to a surface.Bibl.Anat. 11, 69-75.

MORI, Y., NAGASKA, S., TAKIUCHI, H., KIKUCHI, T.,

NOGUCHI, N., TANZAWA, H. & NOISHIKI, Y. (1982).

Trans. Am. Soc. artif. int. Organs 28, 459-462.NORDE, W., MACRITCHIE, F., NOWICKA, G. & LYKLEMA,

J. (1986). Protein adsorption at solid-liquid interfaces:reversibility and conformational aspects. J. Coll.Interface Sci. 112, 447-456.

OWENS, N. F., JOHNSTON, D., GINGELL, D. & CHAPMAN,

D. (1987). Surface properties of long chain 10:12diynoic acids at an air-water interface. Thin Solid Films(in press).

PARSEGIAN, V. A., RAND, R. P., FULLER, N. & RAU, D.

C. (1979). Osmotic stress for the direct measurement ofintermolecular forces. Meth. Enzym. 127, 400-416.

PARSEGIAN, V. A. & RAU, D. C. (1984). Water nearintracellular surfaces. J. Cell Biol. 99, 196s-200s.

RUCKENSTEIN, E. & GOURISANKAR, S. V. (1984). A surface

energetic criterion of blood compatibility to foreignsurfaces. J. Coll. Interface Sci. 101, 436-451.

SA DA COSTA, V., BRIER-RUSSELL, D., SALZMAN, E. W. &

MERRILL, E. W. (1981). ESCA studies on polyurethanes:blood platelet activation in relation to surfacecomposition. J. Coll. Interface Sci. 80, 445-452.

SA DA COSTA, V., BRIER-RUSSELL, D., TRUDEL, G.,

WAUGH, D. F., SALZMAN, E. W. & MERRILL, E. W.

(1980). Polyether-polyurethane surfaces: thrombinadsorption, platelet adsorption and ESCA scanning.J. Coll. Interface Sci. 76, 594-5% (letter).

SPARROW, E. M. (1955). Analysis of laminar forced-convection heat transfer in entrance region. NationalAdvisory Committee on Aeronautics. Technical News,3331. NAGA.

TROMMLER, A., GINGELL, D. & WOLF, H. (1985). Red

blood cells experience electrostatic repulsion but makemolecular adhesions with glass. Biophys. J. 48, 835-841.

TRONEL-PEYROZ, E., RAOUS, H. & SCHUMANN, D. (1983).

A study of the interfacial behaviour of polyoxyethylene

674 N. F. Owens et al.

at a mercury-aqueous solution interface: compact and WHICHER, S. J. & BRASH, J. L. (1978). Platelet-foreigndiffuse layers. J. Coll. Interface Sci. 92(1), 136—153. surface interactions: release of granule constituents from

VAN WAGENEN, R. A. & ANDRADE, J. D. (1980). Flat adherent platelets. J . Biomed. mat. Res. 12, 181-201.plate streaming investigations: hydrodynamics andelectrokinetic equivalency. J. Coll. Interface Sci. 76, {Received 23 December 1986 —Accepted, in revised form,305-314. 7 April 1987)

Inhibition of cell adhesion 675