Corneal epithelial adhesion strength to tethered-protein/peptide modified hydrogel surfaces

Christopher Wallace,1,2 Jean T. Jacob,1 Albert Stoltz,1,2 Jingjing Bi,1 Kirk Bundy2

1LSU Eye Center, LSU Health Sciences Center, 2020 Gravier Street, Suite B, New Orleans, Louisiana 701122Department of Biomedical Engineering, Tulane University, Lindy Boggs Building, New Orleans, Louisiana 70118

Received 24 June 2004; accepted 5 August 2004Published online 8 November 2004 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jbm.a.30199

Abstract: In this study, we investigated the suitability ofmicrojet impingement for use on hydrogel materials to de-termine the cellular adhesion strength of corneal epithelialcells grown on novel hydrogels with extracellular matrixproteins (laminin and/or fibronectin) or a peptide sequence(fibronectin adhesion promoting peptide, FAP) tethered totheir surface with poly(ethylene glycol) chains. The defor-mation of the hydrogel surface in response to the force of themicrojet was analyzed both visually and mathematically.After the results of these experiments and calculations de-termined that no deformation occurred and that the pres-sure required for indentation (1.25 � 106 Pa) was threefactors of 10 greater than the maximum pressure of themicrojet, the relative mean adhesion strength of primaryrabbit corneal epithelial cells grown on the novel poly(2-hydroxyethyl methacrylate-co-methacrylic acid) hydrogelswas determined and compared with that of the same type ofcells grown on control glass surfaces. Only confluent celllayers were tested. Cells grown on control glass surfacesadhered with a mean relative adhesion strength of 488 � 28

dynes/cm2. Under identical conditions, cells grown on lami-nin- and FAP-tethered hydrogel surfaces were unable to beremoved, indicating an adhesion strength greater than 516dynes/cm2. Cells grown on fibronectin- and fibronectin/laminin (1:1)-tethered surfaces showed significantly lowerrelative adhesion strengths (201 � 50 and 189 � 11 dynes/cm2, respectively), compared with laminin- and FAP-teth-ered surfaces (p � 0.001). Our results demonstrate that themicrojet impingement method of cell adhesion analysis isapplicable to hydrogel substrates. Additionally, analysis ofour test surfaces indicates that fibronectin tethered to thishydrogel in the quantity and by the method used here doesnot induce stable ligand/receptor bonding to the epithelialcell membrane to the same degree as does laminin or FAP.© 2004 Wiley Periodicals, Inc. J Biomed Mater Res 72A:19–24, 2005

Key words: cell adhesion; hydrogels; epithelial cells; extra-cellular matrix (ECM) proteins; fibronectin adhesion pro-moting peptide (FAP)

INTRODUCTION

More than 120 million people today have visionproblems caused by damage to or irregularities of thecornea. Artificial materials to aid the keratoplastic pro-cedures used to treat these corneal defects have beenunder development for many years. Although muchprogress has been made recently, the optimal materialhas yet to be achieved.1 The development of a materialthat allows corneal epithelial cells to attach and form anormal epithelial structure is the ultimate goal. A keystep in developing such a material is to identify the

material surface that allows the strongest cell adhesionto form between the epithelial cell membrane and thesurface.

Cell surface adhesion phenomena are mediated bycell membrane receptor molecules, that is, integrins,which interact with complementary ligand proteins inthe membrane of another cell or in the extracellularmatrix (ECM). Integrin-mediated cell membrane ad-hesion to ECM proteins not only initiates focal adhe-sions to the underlying surface but also activates sig-naling pathways that control cell morphology,proliferation, and differentiation.2 The mechanicalstrength of specific cell adhesion has been related tothe receptor/ligand bond affinity and number.3

Most of the techniques used to measure the strengthof cellular adhesion to a surface, such as peel tests andsingle cell adhesion tests, have limitations that preventthem from being accurately applied to hydrogel orelastomeric surfaces. Microjet impingement has beenused previously to determine differences in the bioad-

Correspondence to: J. T. Jacob; e-mail: jjacob@lsuhsc.eduContract grant sponsor: National Eye Institute, National

Institutes of Health; contract grant numbers: EY012367,EY002377

Contract grant sponsor: Research to Prevent Blindness,Inc., New York, NY

© 2004 Wiley Periodicals, Inc.

hesion of fibroblasts and osteoblasts to various non-elastomeric surfaces.4,5 Beginning in 1956, Glauert6

and Bakke7 sequentially addressed the use of theforces produced by the wall jet to determine the ve-locity distributions at the surface. This technique wasfirst used in the biomedical field in 1982, when Desh-pande and Vaishnav,8,9 in an effort to measure theshear strength of the vascular endothelium in humans,modeled the effects of the wall jet using stress versusradial distance curves. Using the Navier-Stokes andcontinuity equations, they were able to model theshear stresses induced by microjet impingementgraphically and mathematically.

The microjet impingement technique involves asubmerged laminar microjet of fluid impinging per-pendicularly on a flat, rigid plane upon which a con-fluent layer of cells has been grown. The flat, cell-covered plane causes the fluid to flow radiallyoutward, resulting in shear stresses distributed radi-ally outward across the surface. If the shear stressesare strong enough, some of the attached cells erodefrom the surface, forming a lesion. By measuring theradii of the lesions and comparing them to the shearstress versus radial distance curves, it is possible todetermine the shear stresses required to erode the cellsfrom the surface,6–9 thus providing an indirect mea-surement of the strength by which the cells are adher-ing to the surface.

The microjet impingement technique is adaptablefor hydrogel surfaces as long as the hydrogel does notdeform under the stress of the impinging microjet. In

this study, we determined the suitability of microjetimpingement for use on hydrogel surfaces and thenmeasured the adhesion strength of primary rabbit cor-neal epithelial cells to specifically surface-modifiedhydrogels.

MATERIALS AND METHODS

Microjet impingement apparatus

The microjet impingement apparatus used was based onthe impingement technique described by Bundy et al.4 andHallab et al.10 (Fig. 1). To use the published stress versusradial distance calibration curves, a Reynolds number (Re)of 2000 was maintained during the impingement. Using anRe of 2000, a needle diameter, d, of 0.69 mm, a kinematicviscosity, v, of 0.75 � 10�6 m2/s for phosphate-bufferedsaline (PBS) at 37°C, and the equation Re � (u � d)/v,11 amean fluid velocity, u, of 2.17391 m/s was calculated. Thevolumetric flow rate, V, of the apparatus was calculatedusing the equation V � � � u (0.5 � d)2, where u and d arethe same as above and � is 3.14159.11 This equation gives aV of 48.7732 mL/min.

The Harvard ‘44’ Programmable syringe pump (HarvardApparatus, Holliston, MA) was programmed by enteringthe selected volumetric flow rate, the inner diameter of the100-mL custom syringe (Hamilton Company, Reno, NE),and the time of impingement (30 s). The microjet nozzleand/or needle (19-gauge, flat-tipped; Hamilton Company)were attached to the syringe via a LeurTight fitting (Up-

Figure 1. Photograph of microjet impingement apparatus with parts labeled.

20 WALLACE ET AL.

church Scientific, Oak Harbor, WA), which was attached toanother LeurTight fitting locked to the syringe. The verticalposition of the flow microjet nozzle was controlled using theDigi-Matic Precision Height Gauge (Mitutoyo Corporation,Japan).

Hydrogel deformation analysis

We analyzed the possibility of hydrogel deformation bothvisually and mathematically. Deformation of the hydrogelwas visually studied using digital photography of the mi-crojet impingement process. Specifically, surface-modifiedhydrogels were cut and fitted snugly into the impingementchamber with the front edge of the gel touching the clearfront of the chamber and held in place with a three-sided,U-shaped, poly(methyl methacrylate) stabilizer (open sidefacing the front of the chamber for visualization). The flowchamber was filled with water, the impingement syringefilled with PBS dyed with red food coloring, and the needleset to 1.38 mm above the surface of the gel. A digital videocamera was placed parallel to the top surface and verticallyeven with the hydrogel to capture images of the gel. Digitalstill captures from the camera were then analyzed to mea-sure deformation. The number of pixels per millimeter wascalculated from the known impingement needle height of1.38 mm. The digital camera yielded a maximum pixels permillimeter count of 21.0. This process was done with theneedle impinging 2.5 cm from the front of the flow chamber,which was determined to be the center of the hydrogel.Impingement was then repeated at 1.5 cm and 1 mm fromthe front of the flow chamber to decrease the chance ofvisual distortion between the microjet center and the cam-era.

The absence of visible deformation would be expected ifthe pressure force applied failed to stress the hydrogelenough to cause a measurable amount of strain. A mathe-matical estimate of the pressure force that could be sus-tained without causing unacceptable strain levels was cal-culated using the tensile strength values for the hydrogels.Using the modulus of elasticity, 1782966.83 Pa, we calcu-lated the pressure required to cause a 0.1-mm indentation orsurface strain of 7% to be 1.25 � 106 Pa. Based on a fluidvelocity, u, of 2.2 m/s, the approximate density of water, p,at 1000 kg/m3, and the equation F/A � p � u2, the maxi-mum downward pressure that could be generated by thewater microjet flow was determined.12 The maximumdownward pressure of the microjet and the pressure neededto deform the hydrogel were then compared.

Surface-modified polymers

The tethered-protein/peptide surface-modified hydrogelsused for testing were prepared in thin sheets. Completedetails of the polymerization and tethering chemistry, aswell as the analytical and cell growth results on the modifiedsurfaces, are reported elsewhere.13,14 Briefly, poly(ethyleneglycol) (MW 3400 D; Nektar Therapeutics, formerly Shear-water Polymers, Huntsville, AL)-tethered proteins—laminin

(MW 410 kDa, from basement membrane of Engelbreth-Holm-Swarm mouse sarcoma; Sigma-Aldrich, St. Louis,MO), fibronectin (MW 450 kDa, from human plasma; Sigma-Aldrich), a mixture of the two (1:1), or a fibronectin adhesionpromoting peptide (FAP, MW 1023 D; Bachem Biosciences,King of Prussia, PA)—were covalently added to the surfaceof poly(2-hydroxyethyl methacrylate-co-methacrylic acid)(Polysciences, Warrington, PA) using carbodiimide (Sigma-Aldrich) chemistry. Immediately before cell culture, discs ofthe hydrogel were cut using a 35-mm armed punch (SmallParts, Miami Lakes, FL) under sterile conditions from thesterile thin sheet and placed in sterile PBS.

Corneal epithelial cell culture

Primary rabbit corneal epithelial cells were isolated fromwhole eyes (Pelfreeze, Rogers, AZ) using methods adaptedfrom those published by Stocker et al.12 and Gipson andGrill.15 Isolated cells were seeded on four different types ofprotein and/or peptide surface-modified hydrogel samplesin the 35-mm-diameter wells of a six-well culture plate. Thesurface-modified hydrogel samples were held at the bottomof the wells using sterilized Teflon™ bushings. Twelve sam-ples of each surface-modified hydrogel were tested (twocultures with six samples each). Glass coverslips were usedfor the control growth surface because epithelial cells do notreach confluence on unmodified hydrogel surfaces.14 Twen-ty-four control samples were tested (two cultures with 12samples each). All seeded samples were incubated at 37°Cwith 5% CO2 until confluence was achieved.

Microjet impingement of surfaces

Upon removal from the incubator, confluent cell surfaceswithin the wells were placed into the impingement appara-tus and submerged in PBS at 37°C. The microjet nozzle waspositioned perpendicularly to the center of the well contain-ing the sample at a height of four times the radius of thenozzle (1.38 mm). Surfaces with nonconfluent layers werenot tested.

Cell staining and analysis

Immediately after impingement, samples and their cellswere fixed in 3.7% formaldehyde for 10 min. The fixed cellswere then stained with 0.2% Coomassie brilliant blue indeionized water/methanol/acetic acid (5:5:1 by volume) for10 min and immediately destained with deionized water/methanol/acetic acid (5:5:1) for 10 min to define the lesions.

Still images of each impinged and stained sample weretaken using a CCD camera in conjunction with an Eagle EyeII (Stratagene, La Jolla, CA) still video system. Each imagewas filtered and converted into a black and white image(Photoshop; Adobe Systems, San Jose, CA), identifying thelesion area (black; no cells) and the nonimpingement area

CELLULAR ADHESION TO MODIFIED HYDROGELS 21

(white; attached cells), respectively (Fig. 2). The diameters ofthe lesion were taken at 5° intervals for 180° using an in-house, customized pupil diameter measurement softwareprogram. The mean of the 36 resulting diameters for eachsample was taken and then made dimensionless by dividingthe resulting radius by the radius of the microjet nozzle(0.345 mm). Shear stresses generated by the flow required toerode the cells at the perimeter of the lesions were estimatedusing the radius of the lesions and published stress versusradial distance calibration curves.

Statistical analysis

Statistical analysis of the numerical data was performedusing an analysis of variance method in the Statistical Anal-ysis Systems (SAS, Cary, NC) program.16

RESULTS

Hydrogel deformation

The dye flow tests showed that the elastomericproperties of the hydrogel samples did not influencethe laminar flow of the impinging microjet stream. Ifsignificant hydrogel deformation occurred, it wouldbe accompanied by flow separation from the hydrogelas the amount of deformation and therefore the angleof deformation increased. Because the microjet streamwould follow a roughly linear path determined by the

angle of the surface it first strikes, an upward-facingcone would result if the hydrogel deformed signifi-cantly, as shown in Figure 3. Because the fluid velocitynear the hydrogel would be much less than the fluidvelocity seen at upward angles, the hydrogel would besubjected to significantly less shearing flow across itssurface than would be predicted. However, the dyedmicrojet stream did not exhibit this type of behavior.The visible cross-section of fluid motion, once thetransient turbulence cleared, was flat against the hy-drogel until very near the chamber wall, as is clearlyseen in Figure 4.

Figure 2. Lesion analysis. (A) Still tissue culture well image showing impinged and stained cell layer. (B) Filtered image. (C)Conversion of filtered image into black and white image provides a measurable lesion area.

Figure 3. Diagram of fluid flow when the sample surfacedeforms in response to the force of the impinging fluid.

22 WALLACE ET AL.

The maximum pressure generated by the fluid jetused throughout this experiment, based on the equa-tion given in the Materials and Methods section above,was calculated to be 4.84 � 103 Pa. Although uncon-trollable experimental conditions such as hydrogeldrying and water jet turbulence affect both the theo-retical maximum allowable pressure and the theoret-ical maximum experimentally observed pressure,there is no reason to expect extreme error. The actualvalues that these two pressure forces estimate shouldfall within one factor of 10 of the idealized values.Because the pressure required for indentation (1.25 �106 Pa) is three factors of 10 greater than the maximumpressure of the microjet, even the worst-case scenar-io—an unexpectedly high experimental pressure andan unexpectedly low maximum allowable pressure—would fall well within acceptable levels for no defor-mation.

Cellular adhesion strength

Five of the 24 control glass samples did not haveconfluent cell layers within 7 days and were excludedfrom further study. Epithelial cells reached confluenceon all of the surface-modified hydrogel samples. Mi-crojet impingement at a Reynolds number of 2000eroded cells from the glass surface, as well as fromsome of the fibronectin-grafted hydrogel surfaces, butdid not erode cells from any of the surfaces graftedwith laminin or FAP alone.

The calculated shear stresses required to erode cellsfrom a normalized area on the tested surfaces aregiven in Figure 5. All of the 19 control glass samples

with confluent cell layers formed lesions during im-pingement. The resulting shear stress required toerode the cells from the surfaces of these samples was488 � 28 dynes/cm2. Nine of the 12 samples graftedwith either fibronectin or fibronectin/laminin (1:1 byvolume) formed significant lesions (Table I). The re-sulting shear stresses required to erode the cells fromthe surfaces of these samples were 201 � 50 and 189 �11 dynes/cm2, respectively. Cells grown on surfacesgrafted with laminin alone (n � 12) or FAP alone (n �12) formed no lesions in any of the samples, exhibitinga significantly stronger adhesion strength than cellsgrown on surfaces grafted with fibronectin (p � 0.001).

DISCUSSION

The microjet impingement method assumes that thecell adhesion surface is rigid and does not deformunder the stress of the jet. Because the surface-modi-

Figure 4. Photograph shows a sample of laminin surface-modified hydrogel without cells being impinged with amicrojet stream of red dye-colored PBS. The dye allows thedirect observation of the impinging fluid flow. Note that thevisible cross-section of the impinging fluid remains flatagainst the surface-modified hydrogel until very near thechamber wall.

Figure 5. Relative mean adhesion strength of corneal epi-thelial cells to protein- and/or peptide-modified hydrogelsurfaces. *No lesions were formed in cells grown on hydro-gel surfaces grafted with laminin or FAP, indicating a rela-tive adhesion strength greater than 516 dynes/cm2. Valuesare expressed as means � SD.

TABLE IRelative Mean Adhesion Strength of Corneal Epithelial

Cells to Protein- and/or Peptide-ModifiedHydrogel Surfaces

Material

Number ofConfluent

CellSamples

Number ofSamples

WithLesion

Formed

Number ofSamplesWith NoLesion

Formed

Glass (control) 19 19 0Laminin 12 0 12FAP 12 0 12Fibronectin 12 9 3Laminin/fibronectin (1:1) 12 9 3

CELLULAR ADHESION TO MODIFIED HYDROGELS 23

fied hydrogels used in this study are elastomeric, itwas first important to determine if any deformation ofthe hydrogel would confound the results of our celladhesion study. The lack of deformation observedduring impingement, as well as the calculated level ofstrain on the hydrogel, strongly supports the assump-tion of rigidity in the hydrogel. Perhaps more telling,however, was the qualitative analysis made possibleby viewing the recording of the centered impingementexperiment. Because the freshly dyed PBS leaving theneedle could be distinguished from PBS that had beenin the chamber for longer amounts of time, the motionwithin the chamber was clearly visible. Any flow dis-turbance caused by surface deformation or irregular-ity in the microjet would have been observed. Becausethe fluid motion stayed flat against the hydrogel, therewas no deformation or irregularity present. This resultsupports not only the assumption of rigidity but alsothe assumption of laminarity and all other assump-tions made about the flow pattern and the forces as-sociated with it. Therefore, the shear stress imposedby microjet impingement on the cells on the peripheryof the lesion area is equivalent to the relative adhesionstrength of the cells to the underlying hydrogel sur-face.

In a confluent layer of cells, cell adhesion occursboth with the underlying substrate and with adjacentcells. Generally, initial cell adhesion is characterizedby cell membrane receptors binding with ligands onthe underlying substrate and subsequent depositionby the cell of its own ECM. As attached and spreadingcells encounter sister cells, intercellular adhesion be-tween adjacent cell membranes occurs. The integrityof the cell layer is dependent on the optimization ofboth of these phenomena. The microjet impingementmethod determines the total adhesive strength of theconfluent cell layer, both intercellular and cell/sub-strate.

Based on our microjet impingement results, the cor-neal epithelial cell layer adhered more strongly to thehydrogels grafted with laminin and FAP than to thecontrol glass surface. Because no significant lesion wasformed on any of the cell layers grown on the hydro-gels modified with laminin or FAP only, we wereunable to determine exactly to what degree the cellsadhered more strongly to these surfaces than to theglass. However, a shear force of 488 � 28 dynes/cm2

was not sufficient enough to break the chemical bondbetween corneal epithelial cells and the hydrogelsmodified with laminin and FAP only. There are sev-eral possible explanations for the significantly lesseradhesion strength to the fibronectin-modified surfac-es: 1. the modified surfaces presented too little fi-bronectin to provide the ligand surface concentrationnecessary for the formation of stable epithelial cellreceptor-ligand bonds; 2. the percentage of fibronectin

molecules bound to the surface with the appropriatereceptor binding sites exposed to the epithelial cellswas inadequate for strong cell adhesion; and/or 3. ourtethering technique presented competitive ligand siteson the fibronectin molecule that have a down-regulat-ing effect on epithelial cell adhesion, such as the RGDpeptide, too prominently for adequate cell/substrateadhesion. Further studies are needed to clarify theseissues.

Our current studies have shown that microjet im-pingement allows determination of the relative celladhesion strength of confluent cell layers on hydrogelsurfaces and present biomaterials researchers with anew tool to investigate the integrity of the cell/mate-rial interface.

References

1. Wang MX, Karp CC, Selkin RR, Azar DT. Corneal and con-junctival surgery. In: Yanoff M, Duker J, editors. Ophthalmol-ogy. London: Mosby; 1999. p 5.12.1–5.12.18.

2. Hynes RO. Integrins: versatility, modulation, and signaling incell adhesion. Cell 1992;69:11–25.

3. Kuo SC, Lauffenburger DA. Relationship between receptor/ligand binding affinity and adhesion strength. Biophys J 1993;65:2191–2200.

4. Bundy KJ, Roberts OC, O’Connor K, McLeod V, Rahn B. Quan-tification of fibroblast adhesion to biomaterials using a fluidmechanics approach. J Mater Sci Res Mater Med 1994;5:500–502.

5. Giliberti DC, Anderson KA, Dee KC. A jet impingement inves-tigation of osteoblastic cell adhesion. J Biomed Mater Res 2002;62:422–429.

6. Glauert MB. The laminar boundary layer on oscillating platesand cylinders. J Fluid Mech 1956;1:97–110.

7. Bakke P. An experimental investigation of a wall jet. J FluidMech 1957;2:467–472.

8. Deshpande M, Vaishnav R. Submerged laminar jet impinge-ment on a plane. J Fluid Mech 1982;114:213–236.

9. Vaishnav RN, Patel DJ, Atabek HB, Deshpande MD, PlowmanF, Vossoughi J. Determination of the local erosion stress of thecanine endothelium using a jet impingement method. J Bio-mech Eng 1983;105:77–83.

10. Hallab NJ, Bundy KJ, O’Connor K, Moses RL, Jacobs JJ. Eval-uation of metallic and polymeric biomaterial surface energyand surface roughness characteristics for directed cell adhe-sion. Tissue Eng 2001;7:55–71.

11. Kent W. Kent’s mechanical engineer’s handbook. 12th ed. NewYork: John Wiley & Sons; 1950. p 379–381.

12. Stocker FW, Eiring A, Georgiade R, Georgiade N. A tissueculture technique for growing corneal epithelial, stromal, andendothelial tissues separately. Am J Ophthalmol 1958;46:294–298.

13. Bi J, Downs JC, Jacob JT. Tethered protein/peptide surfacemodified hydrogels. J Biomater Sci Polym Ed 2004;15:905–916.

14. Jacob JT, Rochefort J, Bi J, Gebhardt BM. Corneal epithelial cellgrowth over tethered-protein/peptide surface-modified hy-drogels. J Biomed Mater Res 2004. Forthcoming.

15. Gipson IK, Grill SM. A technique for obtaining sheets of intactrabbit corneal epithelium. Invest Ophthalmol Vis Sci 1982;23:269–273.

16. Freund RH, Little RC, Spector PC. SAS system for linear mod-els. Cary, NC: SAS Institute; 1986.

24 WALLACE ET AL.