ARTICLE IN PRESS

0142-9612/$ - se

doi:10.1016/j.bi

�Correspondfax: +1919 684

E-mail addr

Biomaterials 26 (2005) 6887–6896

www.elsevier.com/locate/biomaterials

MylarTM and Teflon-AFTM as cell culture substrates for studyingendothelial cell adhesion

Charles C. Anamelechi, George A. Truskey, W. Monty Reichert�

Department of Biomedical Engineering, Duke University, Durham, NC 27708-0281, USA

Received 29 November 2004; accepted 10 April 2005

Available online 28 June 2005

Abstract

The textured and opaque nature of DacronTM and ePTFE has prevented the use of these fabrics in conventional cell culture

techniques normally employed to optimize cell attachment and retention. This lack of optimization has led, in part, to the poor

performance of endothelialization strategies for improving vascular graft patency. Here we show that thin, transparent films of

MylarTM and Teflon-AFTM are viable in vitro cell culture mimics of DacronTM and ePTFE vascular graft materials, particularly for

the study of protein mediated endothelial cell (EC) attachment, spreading and adhesion. Glass substrates were used as controls. X-

ray photoelectron spectroscopy (XPS) and contact angle analysis showed that MylarTM and Teflon-AFTM have surface

chemistries that closely match DacronTM and ePTFE. 125I radiolabeling was used to quantify fibronectin (FN) adsorption,

and FN and biotinylated-BSA ‘‘dual ligand’’ co-adsorption onto glass, MylarTM and Teflon-AFTM substrates. Native human

umbilical vein endothelial cells (HUVEC) and streptavidin-incubated biotinylated-HUVEC (SA-b-HUVEC) spreading was

measured using phase contrast microscopy. Cell retention and adhesion was determined using phase contrast microscopy

under laminar flow. All surfaces lacking protein pre-treatment, regardless of surface type, showed the lowest degree of cell

spreading and retention. Dual ligand treated MylarTM films showed significantly greater SA-b-HUVEC spreading up to 2 h, but

were similar to HUVEC on FN treated MylarTM at longer times; whereas SA-b-HUVEC spreading on dual ligand treated Teflon-

AF was never significantly different from HUVEC on FN treated Teflon-AFTM at any time point. SA-b-HUVEC retention was

significantly greater on dual ligand treated MylarTM compared to HUVEC on FN treated MylarTM over the entire range of shear

stresses tested (3.54–28.3 dynes/cm2); whereas SA-b-HUVEC retention to dual ligand and HUVEC retention to FN treated Teflon-

AFTM gave similar results at each shear stress, with only the mid-range of stresses showing significant difference in cell retention to

Teflon-AFTM.

r 2005 Elsevier Ltd. All rights reserved.

Keywords: Teflon-AF; Mylar; Fibronectin; Avidin–biotin; Cell adhesion; Vascular grafts; Endothelialization

1. Introduction

Over half a million coronary artery bypass graft(CABG) operations are performed annually in the US ata health care cost of over $16 billion [1]. In spite of theprocedure’s prevalence, only 5% of bypasses are repeatoperations due in part because patients facing a secondCABG often lack transplantable autologous arteries or

e front matter r 2005 Elsevier Ltd. All rights reserved.

omaterials.2005.04.027

ing author. Tel.: +1919 660 5151;

4488.

ess: reichert@duke.edu (W.M. Reichert).

veins, thus necessitating the use of substitutes [2].Unfortunately, synthetic small diameter vascular graftshave unacceptable patency rates primarily due tolumenal thrombus formation and intimal thickening[3]. Lumenal endothelialization represents a clear,immediate, and practical solution to poor graft patency[4]. However, the promise of this approach has not beenrealized because of numerous unresolved issues con-cerning seeding efficiency, cell retention under flow, andachieving a quiescent and anti-thrombotic endothelialcell (EC) phenotype that resists thrombosis, intimalhyperplasia, and leukocyte adhesion [5–8].

ARTICLE IN PRESS

CC

OO

OO CH2 CH2 n

CF2 CF2* CF CF *

O O

C

CF3F3C

35 65

(a)

(b)

Fig. 1. Chemical structures of MylarTM (a) and Teflon-AFTM (b).

C.C. Anamelechi et al. / Biomaterials 26 (2005) 6887–68966888

The majority of approaches used to improve vasculargraft endothelialization have employed integrin-depen-dent surface conditioning to strengthen lumenal ECadhesion and retention [9,10]. All of these approacheshave been unsuccessful owing to the lack of rapid andfirm initial cell attachment necessary to nucleate theformation of stable, adherent, and quiescent endothe-lium in the graft lumen. Recently, we reported a dualligand approach to enhance EC attachment andspreading on vascular graft materials [11]. This techni-que combines high affinity integrin-independent ligandswith low affinity integrin-dependent ligands to promoterapid EC attachment and spreading. The high affinityavidin–biotin linkages bring the cell membrane in closeapposition to the pre-coated surface, thus facilitatingfaster integrin–fibronectin (FN) linkages that lead todownstream signaling that promotes faster cell spread-ing and firm attachment. The resulting formation of firmfocal contacts is intended to increase EC retention onvascular grafts and lead to improved patency.To date, we have shown that the dual ligand

treatment has a profound effect on in vitro ECattachment to glass surfaces tested under flow. Addi-tional in vitro testing showed that dual ligand treatedcells exhibited enhanced expression of vasoactive agentsNO and PGI2 [12]. Despite the pronounced differencesdemonstrated in vitro, in vivo testing of endothelializedsmall diameter ePTFE grafts showed that dual ligandand FN treated grafts performed equivalently in short-term patency testing in a rat femoral artery model [13].Clearly, in vitro testing on modified glass was a poorpredictor of the in vivo ePTFE results. Unfortunately, invitro testing of EC adhesion to ePTFE and Dacronusing standard cell culture testing methodology has beendifficult because both materials are textured and opaquefabrics.This study describes an in vitro system for character-

izing EC spreading and adhesion to thin transparentmimics of vascular graft materials [14]: MylarTM

(DuPont, Wilmington, DE), a film version of poly-ethylene terephthalate (PET); spun cast Teflon-AFTM

(DuPont, Wilmington, DE), a copolymer of per-fluoro(2,2-dimethyl-1,3-dioxole) (PDD) and tetrafluor-oethylene (TFE). MylarTM and Teflon-AFTM havewettabilities representative of that reported for PET[15] and PTFE [16]. MylarTM and Dacron possess thesame surface chemical composition because both arePET; whereas the surface chemical compositions ofTeflon-AFTM and ePTFE differ by the oxygen contentof the PDD comonomer. MylarTM and Teflon-AFTM

films are readily amenable to transmission microscopycharacterization of cell spreading, and are easilyincorporated into a two-dimensional flow model fortesting EC adhesion strength without the surfaceturbulence caused by textured materials. The currentmanuscript demonstrates the utility of these materials as

model substrates for in vitro studies of cell adhesion andretention by comparing the spreading and adhesionstrength of HUVECs subjected to dual ligand and FNsurface treatments.

2. Materials and methods

2.1. Substrates

The chemical structures of MylarTM and Teflon-AFTM are

shown in Fig. 1. Teflon-AFTM (DuPont, Wilmington, DE) is a

non-crystalline copolymer of 65mol% 2-bistrifluoromethyl-

4,5-difluoro-1,3-dioxole (PDD) and 35mol% TFE purchased

as a 6wt% solution in Fluorinert FC-40 solution (3M, St.

Paul, MN). MylarTM (DuPont, Wilmington, DE) is a non-

crystalline form of PET purchased as 37mm thick films. ePTFE

cardiovascular patch (Impra, Tempe, AZ) and donated

Dacron samples of woven PET were used as 1 cm2 samples.

Cell adhesion and retention studies used standard

100 � 300 � 1mm soda lime glass slides (Gold Seal, Portsmouth,

NH), 24mm� 60mm cover glass (Corning Labware and

Equipment, Corning, NY), 37.7mm MylarTM films cut to

shape, and Teflon-AFTM films solution spun-cast onto glass

microscope slides. Bare cover glass, cover glass coated with

Teflon-AFTM, and MylarTM films were used for contact angle

analysis, XPS, and 125I radiolabeling. Dacron and ePTFE

cardiovascular patch samples were cut to shape (1 cm2) and

used for XPS analysis.

Three cell adhesion systems were examined in this study:

‘‘No Ligand,’’ i.e. untreated HUVECs seeded onto untreated

substrates; ‘‘FN,’’ i.e. untreated HUVECs seeded onto

substrates pre-adsorbed with FN for 1 h; and ‘‘Dual Ligand,’’

i.e. SA incubated biotinylated-HUVECs (SA-b-HUVEC)

seeded onto substrates pre-adsorbed with a mixture of FN

and biotinylated BSA (b-BSA). All references in association

with dual ligand treatment imply the use of SA-b-HUVEC;

whereas HUVEC in association with FN and untreated

surfaces are native, untreated HUVEC. Seeding of untreated

HUVEC to surfaces with no treatment served as controls.

Additional controls used SA-b-HUVEC seeded onto untreated

surfaces and surfaces adsorbed with only FN and also

ARTICLE IN PRESSC.C. Anamelechi et al. / Biomaterials 26 (2005) 6887–6896 6889

HUVEC seeded on surfaces treated with co-adsorbed FN:b-

BSA (data not shown).

2.2. Sample cleaning and preparation

Glass samples (slides and cover glass) were placed in

CoplinTM jars and sonicated for 10min in 2% PCC-54

detergent cleaning solution (Pierce, Rockford, IL), rinsed

3� with tap water and then rinsed 2� with DI water. Washed

slides were placed in another CoplinTM jar and sonicated for

10min in a 1:1 mixture of MeOH:HCl. The slides were allowed

to sit in the CoplinTM jar for an additional 10min and then

sonicated for another 10min. Methanol and hydrochloric acid

were decanted and slides were rinsed 3� with tap water.

Rinsed slides were sonicated in DI water for 10min, rinsed 3�

with DI water, and rinsed with 100% ethanol in the same

CoplinTM jar. Clean slides were dried overnight at 65 1C. Slides

were stored in a square plastic PetriTM dish sealed with

parafilm until used.

MylarTM samples were cut to standard glass slide dimen-

sions, sonicated in 70% ethanol for 10min, rinsed with DI

water 3� , and dried under UV overnight at room tempera-

ture. Samples were stored in a sealed square plastic PetriTM

dish until used.

Using the manufacturer recommended procedure [17], clean

glass slides and cover glass were dip coated in a 2% solution of

1H, 1H, 2H, 2H-perfluorodecytriethoxy silane (Lancaster

Synthesis, Windham, NH) in 95% ethanol/5% water. The

slides were baked at 110 1C for 10min, coated with Teflon-

AFTM (1600s, DuPont, Wilmington, DE), and spin-cast on a

Headway Photoresist Spinner (Garland, TX) at 1000 rpm for

90 s. After spinning, samples were air dried for 10min, baked

at 112 1C for 10min to remove excess solvent, baked at 165 1C

for 5min to remove traces of solvent, and then baked at 330 1C

to anneal the film and promote adhesion to glass substrate [18].

All drying and baking was done in a high-temperature oven. A

film thickness of 61 nm was determined using a PROF1

DekTak 3ST profilometer (Veeco, Santa Barbara, CA).

Samples were stored in square plastic Petri dish slide holders

sealed with parafilm until used.

DacronTM and ePTFE cardiovascular patch samples used

for XPS analysis were cleaned by blowing with ultra-pure

nitrogen gas. Samples were stored in square plastic PetriTM

dishes sealed with parafilm until used.

2.3. Contact angle

Sessile drop, air–water contact angles were measured on

100 � 100 samples of glass, Teflon-AFTM coated glass, and

MylarTM using a NRL Goniometer-Model 100-00 (Rame-

Hart, Inc., Mountain Lakes, NJ). Six samples were analyzed in

each instance.

2.4. XPS analysis

XPS of all samples were conducted with 1 cm2 samples.

Spectra were collected on a Riber LAS2000 instrument (Riber

SA, Rueil Malmaison, France) equipped with a single pass,

cylindrical mirror (MAC2) analyzer, and a Mg Ka X-ray

source at the Analytical Instrumentation Facility (AIF) at

North Carolina State University. Samples were analyzed with

X-ray source at a 751 take-off angle, a spot size of 2mm, and

an analysis chamber pressure of �3� 10�10 Torr. The depth

probed under these conditions was 1–5 nm. Survey elemental

spectra from 1200 to 0.0 eV with a step size of 1.0 eV, dwell

time of 0.1 s, pass energy of 20 eV, and three scans were

conducted for all samples. All binding energies were calibrated

to the binding energy of the C 1s peak to 285 eV. High-

resolution spectra were obtained for F, O, and C. F 1s spectra

ranged from 669 to 689 eV with 10 scans, O 1s spectra ranged

from 520 to 540 eV with 10 scans, and C 1s spectra ranged

from 270 to 290 eV with 5 scans. All other conditions were the

same as those for the survey scans. High-resolution spectra

were resolved with CasaXPS (Casa Software Ltd., Cheshire,

United Kingdom). Software was used to determine percent

chemical bond composition of the surface.

2.5. Radiolabeling

FN, biotinylated bovine serum albumin (b-BSA), and

streptavidin (SA) were labeled with 1mCi Na 125Iodine

(DuPont NEN Research, Boston, MA) for 30min. Free 125I

was removed by passage over a PD-10 column (Pharmacia,

Peapack, NJ) equilibrated with PBS. The bound protein was

determined by precipitation with 10% (v/v) trichloroacetic

acid (TCA). 125I labeled proteins were diluted with unlabeled

proteins and specific activities were determined. Protein

adsorption for the dual ligand FN:b-BSA system was

examined by separately measuring the adsorption of 125I

labeled FN with unlabeled b-BSA, and the adsorption of 125I

labeled b-BSA with unlabeled FN. SA binding to b-BSA was

measured by titrating unlabeled FN:b-BSA surfaces with 125I

labeled SA. To quantify protein adsorption, 1 cm2 samples

were incubated with labeled proteins for as long as the

experimental incubation time. FN and b-BSA were incubated

for 1 h and SA was incubated for 40min. Samples were then

dipped 3� in a series of four 100ml volumes of PBS buffer

solutions. Samples were placed in 12� 75mm tubes containing

1ml buffer. 125I activity was determined by counting for 5min

in a Multi-Prias 1 gamma counter (Packard Instrument Co.,

Shelton, CT). On the following day, slides were rinsed,

transferred to new tubes, and counted for 5min. Adsorbed

protein amount was determined using the second day count.

2.6. Protein pre-treatment and cell culture

All cell culture reagents were obtained from Cambrex

(Walkersville, MD) unless otherwise specified. Human umbi-

lical vein endothelial cells (HUVEC), were grown to con-

fluence in gelatin-coated T25 polystyrene flasks (Corning Inc.,

Corning, NY) with endothelial basal media (EBM) supple-

mented with 0.5ml 10 g/ml human recombinant Epidermal

Growth Factor (hEGF), 0.5ml 1.0mg/ml Hydrocortisone,

0.5ml 50mg/ml Gentamicin and 50mg/ml 50mg/ml Ampho-

tericin-B mix, 3mg/ml bovine brain extract (BBE), and 10ml

fetal bovine serum (FBS). Cells were cultured in an incubator

with 95% air/5% CO2 at 37 1C. HUVECs from passages 2–4

were used for all experiments. For all experiments, confluent

HUVECs were trypsinized and centrifuged at 2200 rpm for

5min, re-suspended in fresh serum free media warmed to

ARTICLE IN PRESSC.C. Anamelechi et al. / Biomaterials 26 (2005) 6887–68966890

37 1C, and seeded onto glass, MylarTM and Teflon-AFTM

substrates at a concentration of 2� 105 cells/ml either with or

without protein pre-treatments in serum free media.

For ‘‘No Ligand’’ and ‘‘FN’’ cases, untreated HUVEC in

serum-free media were seeded onto substrates as described

above, except in the FN case where the substrates had been

pre-adsorbed with a 20mg/ml solution of FN in HEPES-BSS

for 1 h and washed gently with HEPES-BSS prior to seeding

with untreated cells. For the Dual Ligand case [19], substrates

were treated with a heterogeneous mixture of 1ml of 20 mg/mlFN (Sigma Aldrich, St. Louis, MO) and 200 ml of 2mg/ml b-BSA (Sigma Aldrich) in HEPES-BSS for 1 h and washed

gently with HEPES-BSS. Confluent HUVECs were trypsinized

and centrifuged at 2200 rpm for 5min, re-suspended in serum

free EBM, and biotinylated by treatment with 1mM sulfosuc-

cinimidyl-6-(biotinamido) hexaoate (NHS-LC-Biotin) (Pierce,

Rockford, IL) in HEPES-BSS for 20min. Biotinylated

HUVECs were treated with 50mg/ml SA for 40min, yielding

SA-incubated, biotinylated HUVEC (SA-b-HUVEC). SA-b-

HUVECs were seeded onto dual ligand substrates at a

concentration of 2� 105 cells/ml in serum free media.

2.7. Measurement of cell spread area

Cell spread area was measured at 0.5, 1, 2, 4, 8, 12, and 24 h

by phase contrast microscopy (Nikon Diapot, Tokyo, Japan)

with a 10� objective. Three fields were captured at each time

point for each surface and 100 cells were measured for each

sample at each time point. Cell images were captured by an

MTI camera (Dage-MTI, Michigan City, IN) connected to the

microscope and interfaced with a Macintosh computer. Images

were analyzed using NIH Scion image software. The projected

cell area was calculated by measuring the pixels enclosed by the

cell perimeter and converting to mm2 with the aid of a

micrometer slide.

Table 1

Percent elemental composition of polymers by XPS analysis

Material Carbon Oxygen Fluorine

2.8. Cell retention under flow

Slides with cells adherent for 1 h were placed in a variable

height flow chamber [20] and flow was induced for 2min with

Phenol Red Free EBM adjusted to a viscosity of 4 cP with

4.82 kD dextran (Sigma Aldrich, St. Louis, MO). A variation

in the method was used for MylarTM film samples where the

film was cut to the dimensions of the bottom flow plate and the

outer gasket mapped out the flow path. Fields of view along

the x-axis were counted pre- and post-flow. The percentage of

attached cells was calculated as a function of the applied stress.

The wall shear stress tw (dynes/cm2) was computed as

previously described [20]. The applied shear stress ranged

from 3.54 to 28 dynes/cm2. All experiments were performed at

37 1C. The critical shear stress (tc) was calculated using linear

regression fits of cell retention data. The critical shear stress

has been defined previously [11] as the stress at which 50% of

cells detached.

PET 77.36 22.64 —

Mylar 75.29 24.71 —

ePTFE 29.05 — 70.95

Teflon-AF

Uncured 31.36 7.90 60.74

Cured 28.43 9.04 62.53

2.9. Statistical analysis

StatView 5.0.1.0 (Cary, NC) was used to statistically

compare data to determine variability. One-way ANOVA plus

Fisher PLSD post hoc analysis was conducted to determine p

values. All data are reported as the mean7SEM.

3. Results

3.1. Contact angle

Sessile drop, air–water contact angle analysis wasconducted on 100 � 100 samples of MylarTM and Teflon-AFTM. MylarTM was determined to be moderatelywetting with a contact angle of 80.272.301, whileTeflon-AFTM was non-wetting with a contact angle of10872.61. Literature values for polar PET and non-polar PTFE are 801 and 1051, respectively [15,20].DuPont lists a contact angle of 1041 in their Teflon-AFTM materials specifications literature [21]. Theslightly less hydrophobic nature of Teflon-AFTM

compared to PTFE is likely due to the oxygens of thePDD comonomer.

3.2. XPS analysis

Table 1 contains the gross carbon, oxygen andfluorine surface compositions of DacronTM, MylarTM,ePTFE cardiovascular patch and Teflon-AFTM asdetermined by survey XPS scans of the polymersubstrates. The percent composition of carbon andoxygen measured for MylarTM and DacronTM closelyreflected the XPS-determined 72.03% carbon and27.97% oxygen content of PET reported elsewhere[21]. The percent composition of carbon and fluorinemeasured for ePTFE cardiovascular patch was slightlymore fluorine-rich than the XPS-determined 33.17%carbon and 66.83% fluorine content of PTFE reportedelsewhere [21]. While there were no literature referencesfor the XPS-determined composition of Teflon-AFTM,the measured composition of cured Teflon-AFTM filmclosely matched the stoichiometrically calculated com-position for Teflon–AFTM of 24.44% carbon, 9.30%oxygen and 66.27% fluorine.Tables 2 and 3 contain the results from high-

resolution XPS scans of MylarTM and Teflon-AFTM.Spectral splittings of carbon, oxygen and fluorine peaks

ARTICLE IN PRESS

Table 2

Resolved peaks of MylarTM high-resolution scans

C 1s O 1s

Bond C–C C–O CQO CQO C–O–CQO

Binding energy (eV) 284.7 286.0 288.6 532.0 533.5

Area 10205.5 2245.7 1291.1 15025.7 1805.6

Percent contribution 74.0 16.0 10.0 89.0 11.0

Table 3

Resolved peaks of Teflon-AFTM high-resolution scans

C 1s O 1s F 1s

Bond C–C C–O –CF2–CF2– –CF3 C–O C–F

Binding energy (eV) 284.8 288.6 289.9 292.2 533.5 687.6

Area 867.6 1301.4 5592.9 979.6 10394.7 130629.3

Percent contribution 10.0 15.0 64.0 11.0 100 100

Table 4

Surface densities (7SEM) of adsorbed fibronectin (FN), biotinylated BSA (b-BSA), and streptavidin (SA) on glass, MylarTM, and Teflon-AFTM

Glass (1010molecules/cm2) MylarTM (1010molecules/cm2) Teflon-AFTM (1010molecules/cm2)

FN 13.470.4# 12.370.5 9.8670.4

FN/b-BSA 63.971.8 56.673.9 54.873.6

FN/b-BSA 9.970.4* 8.670.4 8.670.3

FN/b-BSA/SA 3971.3* 3070.6 2970.3

Protein name in boldface indicates species labeled with 125I.

n ¼ 3.#Amount of protein adsorbed to glass significantly greater (po0:05) relative to Teflon-AF.*Amount of protein adsorbed to glass significantly greater (po0:05) relative to both Mylar and Teflon-AF Teflon-AF.

C.C. Anamelechi et al. / Biomaterials 26 (2005) 6887–6896 6891

were resolved by numerical fitting to Gaussian functionsin CasaXPS.The C 1s spectrum for MylarTM (Table 2) resolved

into peaks indicating the three binding states for carbonin PET: C–C (284 eV), C–O (286 eV), and CQO(288.6 eV). The dominant features in the C 1s spectrumwere aliphatic carbon (74.3%) with smaller contribu-tions from ether (16.3%) and carbonyl (9.4%) carbons.The O 1s spectrum in MylarTM resolved into bindingstates: O–CQO (532.0 eV) and C–O–CQO (533.5 eV).The major O 1s feature arose from carbonyl oxygen(89.27%) with a minor contribution from the etheroxygen (10.73%).The C 1s peak for Teflon-AFTM (Table 3) resolved

into four binding states: C–C (284.8 eV), C–O(288.6 eV), �CF2–CF2– (289.9 eV), –, and –CF3

(292.2 eV). The major component of the C 1s spectrumwas from the polymer fluorocarbon backbone (64.0%),followed by ether (14.9%), trifluoro (11.2%), andaliphatic (9.9%) bonded carbon. Singlet peaks wereobserved for high-resolution scans of O 1s (533.5 eV)and F 1s (687.6 eV) for Teflon-AFTM.

3.3. Protein adsorption

125I radiolabeling was used to determine surfacedensities of adsorbed FN, b-BSA, and SA as describedelsewhere [20]. Four different protein formulations wereincubated with each substrate: (1) 125I labeled FN, (2)125I labeled b-BSA mixed with unlabeled FN, (3) 125Ilabeled FN mixed with unlabeled b-BSA, and (4) 125Ilabeled SA binding to a pre-adsorbed mixed layer ofunlabeled FN and b-BSA. Table 4 lists the measuredsurface densities of protein adsorbed to glass, MylarTM,and Teflon-AFTM substrates. The general trend inadsorbed amount for all four formulations on all threesubstrates was glass4MylarTM�Teflon-AFTM. In fiveof eight cases the amount of protein adsorbed to glasswas significantly different (po0:05): FN alone relativeto Teflon-AFTM, FN co-adsorbed with b-BSA relativeto both Mylar and Teflon-AF, and SA bound to FN/b-BSA relative to both MylarTM and Teflon-AFTM. Therewere no significant differences in the amount of proteinadsorbed to either MylarTM or Teflon-AFTM for any ofthe four formulations employed.

ARTICLE IN PRESS

Cell Spread on Glass

0

250

500

750

1000

1250

1500

1750

2000

0.5 1 2 4 8 12 24Time (Hr)

Cel

l Sp

read

Are

a (u

m2 ) No Ligand

FNDual Ligand

# $ #

# $

# #

# #

# #

# $

Fig. 3. Cell spreading over a 24 h period on glass (n ¼ 6). #represents

significantly (po0:05) greater spreading than the ‘‘No ligand’’ system.$represents significantly (po0:05) greater spreading than the FN

system.

Cell Spread on Mylar

1750

2000

um

2 ) No LigandFN

C.C. Anamelechi et al. / Biomaterials 26 (2005) 6887–68966892

3.4. Cell spread area

Cell spreading was measured under static conditionsas described elsewhere [11]. Fig. 2 shows phase contrastimages of SA-b-HUVEC bound to dual ligand treatedMylarTM and Teflon-AFTM films indicating theiramenability to optical measurements of cell spreading.Figs. 3–5 are bar graphs of the average spread area ofcells seeded onto the glass, MylarTM and Teflon-AFTM

substrates subjected to no ligand, FN and dual ligandtreatments. For all substrates and surface treatments,the degree of cell spreading increased monotonicallywith adhesion time, ranging from greater than 1600 mm2

for dual ligand treated Teflon-AFTM at 24 h, to less than400 mm2 for untreated Teflon-AFTM at 30min. Regard-less of surface type, and for any given time point, alluntreated surfaces showed the lowest average cell spreadarea. With only two exceptions (glass at 24 h andMylarTM at 12 h), dual ligand treated surfaces showedthe highest average cell spread area for all surface types

Fig. 2. Phase contrast images of SA-b-HUVEC spread on MylarTM

(a) and Teflon-AFTM (b) treated with the dual ligand system after 4 h

of attachment (10� ). Bar ¼ 100mm.

0

250

500

750

1000

1250

1500

0.5 1 2 4 8 12 24Time (Hr)

Cel

l Sp

read

Are

a ( Dual Ligand

# $ # $

# $ #

# #

Fig. 4. Cell spreading over a 24 h period on MylarTM (n ¼ 6).#represents significantly (po0:05) greater spreading than the ‘‘No

Ligand’’ system. $represents significantly (po0:05) greater spreadingthan the FN system.

Cell Spread on Teflon-AF

0

250

500

750

1000

1250

1500

1750

2000

0.5 1 2 4 8 12 24Time (Hr)

Cel

l Sp

read

Are

a (u

m2 ) No Ligand

FNDual Ligand

# #

##

##

##

## #

# #

#

Fig. 5. Cell spreading over a 24 h period on Teflon-AFTM (n ¼ 6).#represents significantly (po0:05) greater spreading than the ‘‘No

Ligand’’ system. $represents significantly (po0:05) greater spreadingthan the FN system.

and time points. Dual ligand treated glass slides andMylarTM films showed significantly greater cell spread-ing within the critical time periods (1–2 h), but weresimilar to FN treated MylarTM at longer times; whereascell spreading on dual ligand treated Teflon-AFTM was

ARTICLE IN PRESS

Percent Cell Retention on Teflon-AF

0102030405060708090

100

3.54 7.07 14.15 17.7 19.5 28.3

Shear Stress (dynes/cm2)

Per

cen

t R

eten

tio

n

No LigandFibronectinDual Ligand

# #

#

#

# $

# $ # $

# #

# #

#

Fig. 8. Cell retention on Teflon-AFTM with various ligand treatments

as a function of shear stress (n ¼ 6) after 1 h attachment. #represents

significantly (po0:05) greater spreading than the ‘‘No Ligand’’ system.$represents significantly (po0:05) greater spreading than the FN

system.

C.C. Anamelechi et al. / Biomaterials 26 (2005) 6887–6896 6893

never significantly higher than on FN treated Teflon-AFTM at any time points.

3.5. Cell retention

Under static conditions, cells were seeded on thevarious substrates for 1 h at a concentration of 2� 105

cells/ml. The strength of adhesion was measured bysubjecting adherent cells to a range of physiologicallyrelevant shear stresses (3.54–28.3 dynes/cm2) in a vari-able height flow chamber [20]. The viscosity of flowmedia was adjusted to that of blood (4 cP) using 4.82 kDdextran to simulate in vivo conditions.Figs. 6–8 contain the results of cell retention to glass,

MylarTM and Teflon-AFTM. The values ranged from90% retention on dual ligand treated substratessubjected to 3.54 dynes/cm2, to a low of 0% retentionon untreated Teflon-AFTM subjected to X17.4 dynes/

Percent Cell Retention on Glass

0102030405060708090

100

3.54 7.07 14.15 17.7 19.5 28.3

Shear Stress (dynes/cm2)

Per

cen

t R

eten

tio

n

No LigandFibronectinDual Ligand

# $

# $ # $

# $ # $

# $

#

Fig. 6. Cell retention on glass with various ligand treatments as a

function of shear stress (n ¼ 6) after 1 h attachment. #represents

significantly (po0:05) greater retention than the ‘‘No Ligand’’ system.$represents significantly (po0:05) greater retention than the FN

system.

Percent Cell Retention on Mylar

0102030405060708090

100

3.54 7.07 14.15 17.7 19.5 28.3

Shear Stress (dynes/cm2)

Per

cen

t R

eten

tio

n

No LigandFibronectinDual Ligand

# $

# $ # $ # $

# $

# $

#

#

Fig. 7. Cell retention on MylarTM with various ligand treatments as a

function of shear stress (n ¼ 6) after 1 h attachment. #represents

significantly (po0:05) greater retention than the ‘‘No Ligand’’ system.$represents significantly (po0:05) greater retention than the FN

system.

Table 5

Critical shear stress (tc) of endothelial cells on treated and un-treated

substrates

No ligand Fibronectin Dual ligand

Glass 10.270.24 14.370.20# 24.070.14# $

MylarTM 4.270.22 12.470.21# 21.170.26# $*

Teflon-AFTM — 17.570.15# 21.770.21# $*

n ¼ 6.#tc is significantly (po0:001) greater than the ‘‘No Ligand’’ system for

the same substrate material.$tc is significantly (po0:001) greater than the FN system for the same

substrate material.*Not significantly (po0:05) different from each other.

cm2. For all substrates and surface treatments, thedegree of cell retention decreased steadily with increas-ing shear stress. Regardless of surface type, and for anygiven shear stress, all untreated surfaces showed thelowest average cell retention, and all dual ligand treatedsurfaces showed the highest average cell retention. SA-b-HUVEC retention was significantly greater on dualligand treated glass and MylarTM compared to HUVECon FN treated glass and MylarTM over the entire rangeof shear stresses tested; whereas SA-b-HUVEC reten-tion on dual ligand and HUVEC retention on FNtreated Teflon-AFTM gave similar results at each shearstress, with only the mid-range of shear stresses showingsignificantly higher cell adhesion to dual ligand treatedTeflon-AFTM.The critical shear stress defined as the shear stress at

which 50% of cells remain attached, is a quantitativemeasure of relative cell adhesion [11]. Table 5 lists themean tc for cell retention on glass, MylarTM, andTeflon-AFTM with and without FN and dual ligandtreatment. In general, FN and dual ligand treatmentsyielded similar values of tc on all three surfaces, whereas

ARTICLE IN PRESSC.C. Anamelechi et al. / Biomaterials 26 (2005) 6887–68966894

the untreated surfaces produced a range of values. Onlyfor untreated Teflon-AFTM, where the majority of cellsdetached under flow, was a tc value not calculable.Statistical significance was determined two ways: by

surface treatment and by substrate type. Looking atsurface treatment, all FN treated surfaces had signifi-cantly (po0:001) higher values of tc than did corre-sponding surfaces without protein treatment, and alldual ligand treated surfaces had values of tc that weresignificantly (po0:001) higher than did the correspond-ing FN treated surfaces; however, this difference was thesmallest between dual ligand and FN treated Teflon-AFTM. Looking by substrate type, dual ligand treatedMylarTM and Teflon-AFTM were the only conditionsthat yielded values of tc that were statistically (po0:05)indistinguishable.

4. Discussion

The endothelialization of synthetic small diametervascular grafts has been sought vigorously for at least 20years by a number of groups who have taken a variety ofapproaches [22–24]. Most attempts at graft endothelia-lization have involved, at least in part, the adsorption orcovalent coupling of extracellular matrix (ECM) pro-teins or cell adhesion peptides to graft surfaces toenhance the protein mediated attachment of EC to thelumenal surface [25]. Treatments to improve proteinadsorption to graft material include plasma treatment[26,27], chemical treatments [25,26], and fibrin glue [23].While mixtures of ECM proteins have been employed,such as Matrigel [28], it is more common to employsingle integrin-dependent ligands such FN [10] or RGDpeptide sequence [29]. However, the low affinity ofintegrin-dependent binding systems has limited theirsuccess. In order to increase the binding of EC underflow conditions, we developed a ‘‘dual ligand’’ adhesiontreatment that combines high affinity integrin-indepen-dent ligands to stabilize initial EC attachment, withlower affinity integrin-dependent ligands necessary forEC spreading and focal adhesion formation [11]. Thedual ligand approach involves the binding of SA-incubated ECs to surfaces pre-treated with a mixedmonolayer of FN and b-BSA.This study describes the use of smooth, transparent

films of MylarTM and Teflon-AFTM to address apersistent, albeit technical, problem that neither of themost common vascular graft materials, DacronTM andePTFE, are amenable to common in vitro cell adhesiontesting protocols because of their opacity and texture. Infact, cell adhesion to DacronTM and ePTFE fabrics canonly be observed using either SEM or fluorescentimaging [15,30] neither of which is suitable forcontinuous in situ cell monitoring. Consequently, mostEC adhesion strategies are optimized on modified glass

surfaces prior to testing on vascular graft materials. Aswe described elsewhere [13], this is a suboptimal strategybecause what works spectacularly well in vitro on glassmay be of little consequence in vivo using actualvascular grafts.Air–water contact angles measured for MylarTM and

Teflon-AFTM were consistent with the contact anglesreported elsewhere for PET [15] and PTFE [31]. Sincethe attachment of proteins to the polymer surfaces in thedual ligand treatment is a physical adsorption, ratherthan chemical coupling, the similarity of surface energiesis the minimum requirement for establishing MylarTM

and Teflon-AFTM as tissue culture mimics of DacronTM

and ePTFE.Survey and high-resolution XPS scans of MylarTM

and DacronTM yielded nearly identical surface chemicalsurface compositions; whereas the XPS of Teflon-AFTM

and the ePTFE cardiovascular patch differed, asexpected, by chemical shifts attributed to the oxygensof the PDD monomer. There were, however, a fewobserved discrepancies in the XPS measurements. Thecarbon composition measured for MylarTM and Da-cronTM exceeded the carbon XPS content reportedelsewhere for PET by a few percent [21], and the XPSmeasured for ePTFE was a few percent more fluorine-rich than that reported for XPS of PTFE [21]. A likelysource of excess carbon signal is contamination, possiblyfrom residual solvents used in the cleaning procedure[32]. The origin of the difference between ePTFE andPTFE fluorine content is not apparent. The intensities ofthe C 1s and O 1s peak intensities also differed betweenMylarTM and DacronTM. The difference in C 1s and O1s peak intensities between MylarTM and DacronTM

may have arisen from differences in electron captureafter X-ray excitation in thin film and textured speci-mens [33].Adsorption of the dual ligand proteins to glass,

MylarTM and Teflon-AFTM were tested using 125Iradiolabeling. There were no significant differences inthe amounts of FN, b-BSA and SA adsorbed toMylarTM and Teflon-AFTM; whereas glass bound more(11–26%) protein in all eight comparisons to MylarTM

and Teflon-AFTM, with five of these difference beingstatistically significant (po0:05). Although glass is veryhydrophilic it is well known to adsorb significantamounts of blood plasma proteins such as BSA andFN [34]; however, it was somewhat surprising thatMylarTM and Teflon-AFTM adsorbed such similaramounts given the difference in surface polarity. Never-theless, the amount adsorbed may be secondary to thereversibility and the conformational state of theadsorbed protein [35,36]. Finally, while adsorbedamounts showed some variability, the ratios of sur-face-bound proteins were remarkably consistent be-tween substrates: b-BSA to FN (6.5 on glass, 6.6 onMylarTM and 6.3 on Teflon-AFTM); SA to b-BSA (1.6

ARTICLE IN PRESSC.C. Anamelechi et al. / Biomaterials 26 (2005) 6887–6896 6895

on glass and 1.9 on MylarTM and Teflon-AFTM); andSA to FN (3.9 on glass, 3.5 on MylarTM and 3.0 onTeflon-AFTM).As expected, untreated surfaces showed the lowest

degree of cell spreading and retention compared to thecorresponding dual ligand and FN treated surfaces.Dual ligand treated MylarTM films showed significantlygreater SA-b-HUVEC spreading up to 2 h, but weresimilar to HUVEC spreading on FN treated MylarTM atlonger times; whereas SA-b-HUVEC spreading on dualligand treated Teflon-AFTM was never significantlydifferent than HUVEC spreading on FN treatedTeflon-AFTM at any time point. HUVEC retentionwas significantly greater on dual ligand treated MylarTM

compared to FN treated MylarTM over the entire rangeof shear stresses tested (3.54–28.3 dynes/cm2). OnTeflon-AFTM, however, HUVEC retention to dualligand and FN treatments gave similar results at eachshear stress, with only the mid-range of stresses showingany significant difference.The critical shear stresses determined for glass,

MylarTM and Teflon-AFTM showed a significant(po0:001) trend of untreatedoFN treatedodual ligandtreated substrates; however, the increase between FNand dual ligand treated Teflon-AFTM was less than halfof the increase in tc observed for MylarTM and glass. Itis also interesting that dual ligand treated glass yieldedsignificantly higher values of tc compared to MylarTM

and Teflon-AFTM. The critical shear stress result withTeflon-AFTM was particularly intriguing because it mayprovide important insight into why dual ligand treatedePTFE 2mm ID vascular grafts showed no advantageover FN treated grafts when tested in a rat femoralartery model [13]. The lack of difference between FNand dual ligand treatment with ePTFE in vivo wassurprising because dual ligand treatment dramaticallyout performed FN treatment when tested in vitro onglass [37]. The higher value of tc on glass compared toMylarTM and Teflon-AFTM in the current in vitro testsuggests that the dual ligand effect is less pronouncedwhen implemented on a vascular graft material surro-gate. Clearly, in vitro testing on glass suggested a morepronounced difference between FN and dual ligandsurfaces that was not evident with in vitro testing onTeflon-AFTM.Efforts are currently under way to determine (1) why

TeflonTM behaves in a manner that appears to mask theadvantage afforded by the high affinity avidin–biotinbinding of the dual ligand strategy, and (2) how toovercome this ambiguity. One possible explanation forthis observation comes from Grainger [38] who demon-strated that not only does the BSA out compete FN forbinding sites on TeflonTM, it also masks the tenth typeIII repeat unit within the cell binding RGDS recognitionsequence of FN. Thus, BSA masking the action of thecoadsorbed FN may have subverted the advantage

afforded by the high affinity avidin–biotin binding onTeflon-AFTM.

5. Conclusion

Here we demonstrated the utility of transparent filmsof MylarTM and Teflon-AFTM as tissue culture mimicsof DacronTM and ePTFE vascular graft materials. Thisstudy represents the necessary first step towards devel-oping an in vitro model that more accurately models ECadhesion to vascular graft materials, and that stillprovides for the use of conventional tissue culturemethods normally used to investigate cell attachment,spreading and adhesion. These materials were charac-terized by contact angle, XPS, and adsorption of thedual ligand proteins FN, b-BSA and SA. Measuring thespreading and retention of HUVEC without proteintreatment, with FN treatment, and with dual ligandtreatment, characterized cell adhesion. The observationthat cell retention to FN and dual ligand treated Teflon-AFTM was statistically indistinguishable, a result thatmirrors our in vivo observation with ePTFE, suggeststhat these materials are significantly better thanconventional in vitro testing on glass substrates. Currentefforts are directed at identifying the surface chemistrynecessary to optimize the impact of dual ligandtreatment for MylarTM and Teflon-AFTM films.

Acknowledgements

This research was supported by NIH grantHL044972. The authors gratefully acknowledge theassistance and technical advice of Dr. Bernard Chanand Mario Carson for help with the cell spreadingstudies. We would also like to thank Mr. Fred Steviewith the Analytical Instrumentation Facility (AIF) fromthe North Carolina State University. Comments ofanonymous reviews are acknowledged for useful sugges-tions regarding interpretation of the XPS data.

References

[1] Hlatky MA, Rogers WJ, Johnstone I, et al. Medical care costs

and quality of life after randomization to coronary angioplasty or

coronary bypass surgery. Bypass Angioplasty Revascularization

Investigation (BARI) Investigators. N Engl J Med

1997;336(2):92–9.

[2] http://www.americanheart.gov, 2004.

[3] Seifalian AM, Tiwari A, Hamilton G, Salacinski HJ. Improving

the clinical patency of prosthetic vascular and coronary bypass

grafts: the role of seeding and tissue engineering. Artif Organs

2002;26(4):307–20.

[4] Vinard E, Leseche G, Andreassian B, Costagliola D. In vitro

endothelialization of PTFE vascular grafts: a comparison of

ARTICLE IN PRESSC.C. Anamelechi et al. / Biomaterials 26 (2005) 6887–68966896

various substrates, cell densities, and incubation times. Ann Vasc

Surg 1999;13(2):141–50.

[5] Poole-Warren LA, Schindhelm K, Graham AR, Slowiaczek PR,

Noble KR. Performance of small diameter synthetic vascular

prostheses with confluent autologous endothelial cell linings. J

Biomed Mater Res 1996;30(2):221–9.

[6] Gumpenberger T, Heitz J, Bauerle D, et al. Adhesion and

proliferation of human endothelial cells on photochemically

modified polytetrafluoroethylene. Biomaterials 2003;24(28):5139–44.

[7] Hagerty RD, Salzmann DL, Kleinert LB, Williams SK. Cellular

proliferation and macrophage populations associated with

implanted expanded polytetrafluoroethylene and polyethyleneter-

ephthalate. J Biomed Mater Res 2000;49(4):489–97.

[8] Tiwari A, Salacinski H, Seifalian AM, Hamilton G. New

prostheses for use in bypass grafts with special emphasis on

polyurethanes. Cardiovasc Surg 2002;10(3):191–7.

[9] James NL, Schindhelm K, Slowiaczek P, et al. Endothelial cell

seeding of small diameter vascular grafts. Artif Organs

1990;14(5):355–60.

[10] Honduvilla NG, Bujan J, Lizarbe MA, Bellon JM, Olmo N,

Hernando A. Adhesion and stability of fibronectin on PTFE

before and after seeding with normal and synchronized endothe-

lial cells: in vitro study. Artif Organs 1995(2):144–53.

[11] Bhat VD, Truskey GA, Reichert WM. Fibronectin and avidin–-

biotin as a heterogeneous ligand system for enhanced endothelial

cell adhesion. J Biomed Mater Res 1998;41(3):377–85.

[12] Chan BP, Reichert WM, Truskey GA. Effect of streptavidin–-

biotin on endothelial vasoregulation and leukocyte adhesion.

Biomaterials 2004;25:3951–61.

[13] Chan BP, Liu WG, Klitzman B, Reichert WM, Truskey GA. In

vivo performance of dual ligand augmented endothelialized

expanded polytetrafluoroethylene vascular grafts. J Biomed

Mater Res Part B—Appl Biomater 2005;72B(1):52–63.

[14] Anamelechi CC, Reichert WM. Optimizing in vitro cellular

adhesion on polymer surfaces. Biomedical Engineering Society

Annual Meeting, 10-3-2003.

[15] Wang JCJ, Kwok SCH, et al. Characteristics and anticoagulation

behavior of polyethylene terephthalate modified by C2H2 plasma

immersion ion implantation-deposition. J Vac Sci TechnolA: Vac

Surf Films 2004;22(1):170–5.

[16] Brandup J, Immergut EH, Grulke EA. Polymer handbook. New

York: Wiley; 1999.

[17] DuPont Fluoroproducts. Teflon AF amorphous fluoropolymers:

processing and use. DuPont, 1998.

[18] www.DuPont.com. Teflon AF amorphous fluoropolymers: pro-

cessing and use. 2004.

[19] Mathur AB, Chan BP, Truskey GA, Reichert WM. High-affinity

augmentation of endothelial cell attachment: long-term effects on

focal contact and actin filament formation. J Biomed Mater Res

2003;66A(4):729–37.

[20] Truskey GA, Pirone JS. The effect of fluid shear stress upon cell

adhesion to fibronectin-treated surfaces. J Biomed Mater Res

1990;24(10):1333–53.

[21] Crist BV. Handbook of monochromatic XPS spectra. Chichester,

New York: Wiley; 2000.

[22] Jensen N, Lindblad B, Bergqvist D. In vitro attachment of

endothelial cells to different graft materials. Eur Surg Res

1996;28(1):49–54.

[23] Santhosh Kumar TR, Krishnan LK. Endothelial cell growth

factor (ECGF) enmeshed with fibrin matrix enhances prolifera-

tion of EC in vitro. Biomaterials 2001;22(20):2769–76.

[24] Salacinski HJ, Tiwari A, Hamilton G, Seifalian AM. Cellular

engineering of vascular bypass grafts: role of chemical coatings

for enhancing endothelial cell attachment. Med Biol Eng Comput

2001;39(6):609–18.

[25] Neff JA, Caldwell KD, Tresco PA. A novel method for surface

modification to promote cell attachment to hydrophobic sub-

strates. J Biomed Mater Res 1998;40(4):511–9.

[26] Lehle K, Buttstaedt J, Birnbaum DE. Expression of adhesion

molecules and cytokines in vitro by endothelial cells seeded on

various polymer surfaces coated with titaniumcarboxonitride. J

Biomed Mater Res 2003;65A(3):393–401.

[27] Dekker A, Reitsma K, Beugeling T, Bantjes A, Feijen J, van Aken

WG. Adhesion of endothelial cells and adsorption of serum

proteins on gas plasma-treated polytetrafluoroethylene. Bioma-

terials 1991;12(2):130–8.

[28] Patterson RB, Keller JD, Silberstein EB, Kempczinski RF. A

comparison between fibronectin and Matrigel pretreated ePTFE

vascular grafts. Ann Vasc Surg 1989;3(2):160–6.

[29] Walluscheck KP, Steinhoff G, Kelm S, Haverich A. Improved

endothelial cell attachment on ePTFE vascular grafts pretreated

with synthetic RGD-containing peptides. Eur J Vasc Endovasc

Surg 1996;12(3):321–30.

[30] Thomson GJ, Vohra RK, Carr MH, Walker MG. Adult human

endothelial cell seeding using expanded polytetrafluoroethylene

vascular grafts: a comparison of four substrates. Surgery

1991;109(1):20–7.

[31] Brandup J. Polymer handbook, 4th ed. New York: Wiley;

1999.

[32] Andrade J. Surface and interfacial aspects of biomedical

polymers. X-ray photoelectron spectroscopy (XPS) 1985;vol.

5:105–91.

[33] Walters M. Personal communication, 2005.

[34] Jenney CR, Anderson JM. Adsorbed serum proteins responsible

for surface dependent human macrophage behavior. J Biomed

Mater Res 2000;49(4):435–47.

[35] Baugh L, Vogel V. Structural changes of fibronectin adsorbed to

model surfaces probed by fluorescence resonance energy transfer.

J Biomed Mater Res Part A 2004;69A(3):525–34.

[36] Andrade JD, Hlady V. Plasma protein adsorption: the big twelve.

Ann NYAcad Sci 1987;516:158–72.

[37] Chan BP, Bhat VD, Yegnasubramanian S, Reichert WM,

Truskey GA. An equilibrium model of endothelial cell adhesion

via integrin-dependent and integrin-independent ligands. Bioma-

terials 1999;23–24:2395–403.

[38] Grainger DW, Pavon-Djavid G, Migonney V, Josefowicz M.

Assessment of fibronectin conformation adsorbed to polytetra-

fluoroethylene surfaces from serum protein mixtures and correla-

tion to support of cell attachment in culture. J Biomater Sci

Polym Ed 2003;14(9):973–88.