Influence of structured wafer surfaces on the characteristics of Caco-2 cells

Available online at www.sciencedirect.com

Acta Biomaterialia 5 (2009) 288–297

www.elsevier.com/locate/actabiomat

Influence of structured wafer surfaces on the characteristicsof Caco-2 cells

Iris Guell a, Heinz D. Wanzenboeck b, Sahar S. Forouzan b, Emmerich Bertagnolli b,Elisabeth Bogner a, Franz Gabor a, Michael Wirth a,*

a Department of Pharmaceutical Technology and Biopharmaceutics, Althanstrasse 14, Vienna, Austriab Institute of Solid State Electronics, Floragasse 7, Vienna, Austria

Received 31 January 2008; received in revised form 4 June 2008; accepted 17 July 2008Available online 6 August 2008

Abstract

Currently there is an increasing demand for high-throughput methods to identify and to verify the potential of new drug can-didates. Cell-based microelectronic biosensors might be powerful tools for rapid screening assays. However, reliable cultivation ofcells is influenced by the material characteristics and the surface topography of the wafers serving as growth supports. In order toinvestigate the influence of micropatterned structures on cell viability, Caco-2 cells were seeded on silicon wafers featuring trench/mesa patterns obtained by lithography and reactive ion etching. Besides determination of the cell growth pattern by electron micro-scopic inspection, the adherence of cells on different patterned silicon wafers and the formation of the tight junctional network wasinvestigated. Microstructured trench/mesa patterns, especially their lateral distances, remarkably influenced the adhesion and prolif-eration behavior of Caco-2 cells. Lateral distances below the average cell diameter were easily overgrown by the cells, whereasdimensions above the average cell diameter increasingly limited cell proliferation. Notably increased cell growth was observed usingtrenches with a width of 10–20 lm and a trench depth of around 35 lm. All in all, the results of this study might improve theproduction of microstructured biosensors and open up new perspectives concerning the combination of biosensors and microfluidicsystems.� 2008 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Caco-2; Structured surfaces; Cell adhesion; Trench; Mesa

1. Introduction

To date there is an increasing demand for rapid screeningof new drug candidates. Recently, research has recognizedcell-based assays as powerful tools not only for screeningthe efficacy but also for elucidation of the biopharmaceuticalcharacteristics of the drug of interest. That way, animalexperiments may be reduced in favor of cell-based assays,which are less time-consuming, less expensive, and withoutcontroversy regarding ethical considerations [1]. However,cell culture requires work-intensive handling, restricting itsroutine applicability for high-throughput screening.

1742-7061/$ - see front matter � 2008 Acta Materialia Inc. Published by Else

doi:10.1016/j.actbio.2008.07.029

* Corresponding author. Tel.: +43 1 42 77 55 407.E-mail address: [email protected] (M. Wirth).

This limitation may be overcome by adapting cell-based microelectronic biosensors for this purpose. Theversatility of bioelectronics for the screening of cellshas been demonstrated in several studies: the responseof cells to toxic substances was successfully investigatedby electrical impedance measurements [2], the meta-bolic activity during cultivation was monitored by anamperometric sensor system [3], and the influence ofgold-patterned silicon surfaces was investigated in termsof on-cell patterning [4].

Nevertheless, the microelectronic monitoring of livingcells has to meet specific requirements. In contrast to con-ventional tissue culture materials, microelectronic devicesare typically based on inorganic surfaces. Consequently,the biocompatibility of the microelectronic material used

vier Ltd. All rights reserved.

mailto:[email protected]

I. Guell et al. / Acta Biomaterialia 5 (2009) 288–297 289

plays an important role for successful cell cultivation. Atthis, gold, silicon, and silicon oxide revealed an adequatebiocompatibility, while several metals such as copper andtungsten have been proven less suitable [5].

Another crucial issue when working with adherentcells is to provide good attachment facilities on thewafers. It is known that cells can adhere to entities liketrenches or columns which are within the size of the cell[6]. Moreover, rat dermal fibroblasts were shown to fol-low the contours of microgrooves on polystyrene [7] andhuman fibroblasts were observed to alter their shape andorientation depending on the topography of micro-machined titanium surfaces [8]. However, there is lessinformation on the response of human cells to structuresin the range between 500 nm and 20 lm, which is awidely used dimension for microelectronic interfacesincluding electronic biochips.

Proper adhesion facilities are of great importance,particularly when working with microfluidic systemsdue to the induced shear stress. Microfluidic systemsprovide a well-established basis to miniaturize analyticaldevices for biological applications. Smaller analyticalplatforms mean low consumption of reagent whichresults in low cost and less expenditure of time [9].Moreover, the feasibility of microfluidic devices forreal-time monitoring of concentration-dependent biolog-ical responses was demonstrated [10].

Caco-2 cells are a widely accepted model for thehuman small intestinal epithelium to characterize intesti-nal drug absorption [11,12] and hence the Caco-2 cell lineis an approved model for the Biopharmaceutics Classifica-tion System (BCS) [13]. The Caco-2 cell model was estab-lished by J. Fogh in 1974 and derived from the humancolon adenocarcinoma of a 72-year-old Caucasian man[14]. Despite their colonic origin the cells form monolay-ers that undergo complete and terminal differentiationforming a columnar absorptive epithelium with tight junc-tions similar to that of the human small intestine [15].Due to their widespread application, new high-throughputscreening systems employing Caco-2 cell-based biosensorswould be of high interest and potential.

The aim of this work was to investigate the behaviorof cells on structured wafer surfaces and to identify theinfluence of surface patterns on cell attachment. For thispurpose, silicon wafers were patterned by lithography togenerate structures with dimensions above and below theaverage cell diameter. Caco-2 monolayers were grown onsuch trench-mesa patterns with lateral distance variationsin the range between 2 and 60 lm and characterizedmicroscopically concerning adhesion, cell growth andthe integrity of the cell layer. Furthermore, the influenceof the microstructures on the morphological differentia-tion was examined by immunofluorescent staining tech-niques. The results of this study should offer decisivebasic information for the design of microstructured inter-faces and perspectives for an expedient method to screennew drug candidates on human cells.

2. Materials and methods

2.1. Materials

2.1.1. Microstructured materials

As cell culture substrates p-doped (100) silicon wafers(single side polished) were used. The 600 wafers were brokeninto 20 � 20 mm pieces for individual processing in differ-ent batches. Prior to patterning the channel structures,the samples were cleaned with acetone and isopropanol.After processing the trench structures the 20 � 20 mmpieces were cut to identical 4.00 � 4.00 mm units with awafer saw with a 60 lm wide diamond cutting blade to pro-vide identical growth areas for cell culture experiments.Gases for etching of the trenches were SF6 (Linde, purity5.0) and oxygen (Air Liquide, purity 5.0).

2.1.2. Cell culture materials

The CyQuantTM cell proliferation assay kit wasobtained from Molecular Probes (Eugene, OR). CelLyticTM

Cell Lysis Reagent (Molecular Probes, Invitrogen), albu-min from bovine serum, and propidium iodide were pur-chased from Sigma (St. Louis, MO, USA). Purifiedmouse anti-ZO-1 antibody was obtained from BD Biosci-ences (San Jose, CA, USA). Goat-anti-mouse immuno-globulins/FITC were purchased from DAKO (Vienna,Austria). FluorSaveTM was obtained from Calbiochem(Darmstadt, Germany). All other chemicals were of analyt-ical grade and purchased from Merck (Darmstadt,Germany).

2.2. Processing of microstructured materials

2.2.1. Mask design

Masks were fabricated with an optical pattern generatorusing designs developed with a commercial computer-aided-design (CAD) program. Five-inch photomask blanks(Hoya) consisting of glass substrates with a thin, reflectivemetal coating were used. Trench/mesa designs of 6/12,8/14, 12/18, 16/22, 20/26, and 30/36 lm were chosen forphoto-resist patterning.

2.2.2. Pattern design

Microstructures were fabricated by optical lithographyusing a AZ 5214 photo-resist (Microchemicals, Ulm,Germany) which was deposited by spin coating at3500 rpm to form a resist layer of 1.5 lm in thickness.Exposure was performed with a MJB3 mask aligner (KarlSuess, Germany) equipped with a 200 W lamp. The resistwas developed with a metal ion free developer AZ 726(Microchemicals, Ulm, Germany). The photo-resist servedas a polymer mask for etching of the trench structures.

2.2.3. Trench fabricationThe three-dimensional trenches in the silicon substrates

were obtained by reactive ion etching (RIE) with a reactiveplasma etcher (Oxford Instruments, Plasmalab System 100)

290 I. Guell et al. / Acta Biomaterialia 5 (2009) 288–297

allowing a base pressure of 3 � 10�6 torr. The etch systemwas equipped with a radio-frequency (RF) plasma sourceand an inductively coupled plasma (ICP) source. Sampleswere mounted on a 600 Si wafer acting as carrier. For pre-conditioning the empty chamber was cleaned with pureoxygen plasma at high power setting for 30 min. Aniso-tropic etching of deep trenches was performed withsamples cooled down to 165 K by liquid nitrogen. Si etch-ing was performed at 14 mtorr with 13 W RF-power and100 W ICP-power. The etch gas SF6 was introduced with50 sccm and oxygen was introduced with 6.7 sccm. The fluxof etch gases was controlled with a mass flow controller.The duration of the etch process was used to control thedepth of the trench. Reactive ion etching in a SF6/O2 etchwas performed for a process time between 10 and 60 min.The actual depth was measured with a profilometer imme-diately after fabrication.

2.3. Surface geometry

The resulting trench structures were cut and thecross-sections were inspected by scanning electronmicroscopy. A LEO 1530 VP electron microscope witha Schottky emitter was operated at 10 keV acceleration.Secondary electron images were recorded from tiltedsamples and from cross-sections. The cross-sectionalview was used to measure the geometry of trenches pro-cessed. The real geometry was utilized for further dataevaluation.

2.4. Surface topography

The surface topography was measured by atomic forcemicroscopy (Dimension 3100, Digital Instruments, Veeco)in the tapping mode using a cantilever with a 300 kHz res-onance frequency and a tip apex radius below 10 nm. Mea-surement was performed at a scan speed of 3 lm s�1 (mesa)and 10 lm s�1 (trenches) for Si chips after deep reactive ionetching as well as after removal of the protective resist. Sur-faces were scanned over a 10 � 10 lm area, and RMS eval-uation was performed on a 2 � 2 lm field. The roughnesswas calculated as the standard deviation from the averagethickness.

Images designated with ‘‘trench” show the surface atthe bottom of a wide trench after 30 s etching with theparameters described above. Images designated as ‘‘sub-strate” show the surface of the unetched material afterremoval of the protective polymer layer after etchingand cutting.

2.5. Cell culture

Caco-2 cells (DSMZ, German collection of microorgan-isms and cell culture, Braunschweig, Germany) of passages33–40 were grown to confluence in 75 cm2 tissue cultureflasks (Costar, Corning, NY, USA) in RPMI-1640 cell cul-ture medium containing 10% fetal calf serum, 4 mM L-glu-

tamine and 150 mg ml�1 gentamycine in a humified 5%CO2/95% air atmosphere at 37 �C. The cells were subcul-tured by animal origin free Tryple Select (Gibco, Lofer,Austria). After the subcultivation the average cell diameterof the Caco-2 cells used amounted to 18.2 ± 1.4 lm asdetermined by microscopy and image analysis (Lucia Gsoftware). However, after adhesion and monolayer forma-tion, a change of Caco-2 cell dimensions can be observed.Hidalgo et al. [15] reported a decrease in cell width of42% for a 16-day culture.

2.6. Adhesion studies

For cell adhesion studies Caco-2 cells were seeded on4 � 4 mm structured substrates placed in the wells of a96-well microplate (IWAKI, Chiba, Japan) at a densityof 31,250 cells per cm2 and cultured for 3, 7, or 14 daysusing non-structured substrates (4 � 4 mm) as a control.The substrates were disinfected with 70% ethanol.

To determine the number of cells adhering to the struc-tured surfaces the DNA content of the cells was estimatedby the CyQuantTM test kit pursuing a modified protocol.After removal of the supernatant cells were lysed by incu-bation with 25 ll of CelLyticTM reagent diluted with thesame volume of 20 mM Hepes/NaOH buffer, pH 7.0, for15 min at room temperature. The DNA of the lysed cellswas stained by addition of 150 ll dye solution representinga mixture of 0.5 ll fluorescent dye with 7.5 ll buffer and142.5 ll water, according to the manufacturer’s instruc-tions. After storage for 20 min in the dark the fluorescenceintensity was determined at 485/535 nm (fluorescencemicroplate reader Spectra-Fluor SLT, Tecan, Groding,Austria). All assays were performed at least in triplicate.

2.7. Microscopic investigations

Cell growth and cell density were checked every secondday using an inverted optical microscope. Seven days post-seeding, the exact position of the cells was observed with ascanning electron microscope LEO 1530 VP operated at1 keV.

2.8. Morphological differentiation

To examine the morphology, Caco-2 cells were seededon 4 � 4 mm structured substrates as described above andcultured for 14 days. Caco-2 cells that adhered to struc-tured silicon were imaged by fluorescence microscopyafter double staining the cells to elucidate morphologicaldifferentiation. For this purpose cells were fixed for20 min with ice-cold methanol at �20�C and after wash-ing, the cells were rehydrated with PBS at room temper-ature. The tight junctions of the cell layer were stainedwith both the primary ZO-1 antibody and the secondarygoat anti-mouse IgG-FITC antibody supplemented with1 lg ml�1 propidium iodide for 1 h at 37�C. Finally, thecell layer was washed again and mounted for microscopy


using FluorSaveTM. Immunofluorescence images of labeledcells were obtained using a Nikon Eclipse 50i microscopeequipped with an EXFO X-Cite 120 fluorescence illumi-nation system. Excitation and emission filter blocks wereat 465–495/515–555 for green fluorescence and at 510–560/ > 590 for red fluorescence. All pictures were acquiredat 20 � magnification using Lucia Gv5.0 software forimage evaluation.

3. Statistics

All measurements were carried out in quadruplicate andrepeated at least 3 times. Statistical analysis was performedusing the Microsoft Excel� integrated analysis tool.Hypothesis tests among two data sets were made by com-parison of two means from independent (unpaired) sam-ples (t-test). A value of p < 0.05 was consideredsignificant. Descriptive statistical analyses were performedusing mean values and standard deviations.

4. Results

To estimate the quality of the processed microstructureddevices, patterns were investigated by scanning electronmicroscopy and atomic force microscopy in terms of sur-face geometry and surface topography.

4.1. Topography design

The intention of the design was to obtain an ideal pat-tern of equally distributed trench widths (Fig. 1-a) andmesa widths (Fig. 1-b) with a feature size of 6, 8, 12, 16,20, and 30 lm. In reality, however, fully anisotropic etch-ing was not observed for deep trench fabrication in the

Fig. 1. Schematic illustration of the trench/mesa design (a/b) during different prprocess.

available process range. To compensate for the lateral loss(Fig. 1-c) due to etching of the side-walls, a sacrificialperimeter of 3 lm on each side of the trench was allowed,resulting in a mesa design width that was 6 lm larger thanthe trench design width.

A perfectly anisotropic etch process would yieldtrenches with 90� side walls and a mesa/trench ratio asgiven by the lithographic mask design. Due to imperfectanisotropy during the etch process also isotropic etchingprocesses occur, leading to undesired etching of the sidewalls and resulting in an etch steepness of around 70�.The extent of isotropic etching may vary for differentgeometries and different processing times, resulting ininconstant temperatures, difficulties in the gas exchangein deep trenches, and in slight variations of the plasma gen-eration. Due to the etching of the side walls, sloped sidewalls and geometries with wider trenches and narrowermesas than expected were formed. For narrow trenchesthe etch species diffusion is hindered, leading to a reducedtrench depth in comparison to broad trenches. Due to theseprocess-related variations, the real structures as opposed tothe design structures are the basis for the cell cultivationexperiments.

4.2. Surface geometry

For imaging of the surface geometry the samples weremounted in the vacuum chamber of a scanning electronmicroscope either with no tilt (surface perpendicular tothe electron beam) or with a 45� tilt. The trenches wereevaluated with cross-section samples placed with a 90� tiltto allow frontal imaging of the trench structure. Surfaceareas were scanned with a 10 keV focused electron beamwith a magnification between 100� and 200� for details.

ocess steps: (a) trench width, (b) mesa width, (c) lateral loss due to the etch

Table 1Deviation of the real amplitudes from the mask design

Mask design Real design

Mesa (lm) Trench (lm) Mesa (lm) Trench (lm; top) Depth (lm)

10 8 5.9 12.2 35.312 8 6.6 13.8 36.114 8 8 14 3412 12 12.3 12 35.514 12 11.8 13.8 37.616 12 10.4 16.6 3722 16 5.8 31 15.422 16 2.5 35.5 5026 20 20 24.5 2936 30 22 43.6 31.8

Fig. 3. Top view of the AFM scan of the surface of an unetched mesa.


The chamber pressure during imaging was in the range of5 � 10�6 mbar. Images were recorded from the secondaryelectron signal generated by the scanning electron beam.

As exemplified in Table 1 the real design diverges fromthe mask design. Due to the etch process trenches are1.8–19.5 lm wider than expected, while mesas are scaleddown between 2.2 and 19.5 lm. In general, the deviationof the real amplitudes from the mask design appears tobe consistent for both the trenches and the mesas. A maskdesign of 10 lm mesa and 8 lm trench arises in 5.9 lm realmesa and 12.2 lm real trench, resulting in a lateral differ-ence of around ±4 lm. The same phenomenon wasobserved with major masks, where a design of 22 lm mesaand 16 lm trench arises in 2.5 lm real mesa and 35.5 lmreal trench, comprising a lateral difference of around±19.5 lm. All in all the deviations are given by the mask,following the rule: the wider the trench the narrower themesa. A linear coherence with the depth of the trenchwas not observed.

4.3. Surface topography

As the surface topography is supposed to influence thecell adhesion and proliferation, the surface of deposited

Fig. 2. Top view of the AFM scan of the surface of a trench etched in Si.

material layers was investigated by atomic force micros-copy. Generally the surface roughness was very low withina small bandwidth and in all cases an irregularly corru-gated surface was observed with a RMS (root meansquare) roughness below 3 nm. Fig. 2 shows the surfacetopography of a trench, where the roughness has a RMSof 0.060 nm. In contrast, the AFM scan of an unetchedmesa revealed a surface roughness given by RMS of2.4 nm (Fig. 3).

4.4. Influence of surface structures on cell adhesion

In order to identify the influence of three-dimensionallystructured microstructured substrates on the growthbehavior of cells, Caco-2 cells were seeded on the patternedsilicon (Si) wafers and the cell distribution and cell prolifer-ation was examined on surfaces with different lateral spac-ing as well as on structures with different depth. A detaileddescription of the dimensions of representative samplesused in this study is given in Table 2.

To estimate the influence of pre-structured surfaces oncell adhesion the number of adhered cells was determined3, 7 and 14 days post-seeding by determining the DNAcontent on the structured patterns using Caco-2 cells grownon non-structured p-doped Si substrates as a control.

As compared to the unpatterned reference, wafers with atrench depth of more than 40 lm and a mesa width of lessthan 3 lm (II–III; Table 2) showed a significant decrease(p = 0.045) of cell adhesion and proliferation independentof the trench width and the cultivation period.71,531 ± 317 adherent cells were found on substrates witha trench depth of 50 lm and a trench width of 2.5 lm onday 7 post-seeding as compared to 78,403 ± 1426 adherentcells on the unpatterned reference. A strong increase of pro-liferation rates was observed on wafers with a trench widthbetween 10 and 20 lm and a depth between 34 and 37 lm(IV, VI–XI; Table 2) irrespective of the duration of the cul-ture. Between 99,257 and 153,488 cells adhered per substrate

Table 2Dimensions of the microstructured substrates used for cell cultivation

Mesa (lm) Trench (lm; top) Depth (lm)

I 0.5 18 2II 2.3 30.3 46.5III 2.5 35.5 50IV 5.8 10.2 34.7V 5.8 31 15.4VI 5.9 12.2 35.3VII 6.6 13.8 36.1VIII 8 14 34IX 10.4 16.6 37X 11.7 13.8 37.6XI 12.3 12.1 35.5XII 12.5 17.5 38.8XIII 14.6 23.3 8XIV 15.3 17.1 47XV 17.5 17.5 28.7XVI 18.8 22.8 33.7XVII 19.2 21.2 33.8XVIII 20 19.8 29XIX 20 24.5 29XX 20.8 13.5 41XXI 22 43.6 31.8

Fig. 4. Mesh plot of the mean cell count of Caco-2 cells grown ondifferently structured Si wafers on day 14 post-seeding (n = 3).

Table 3Mean cell count of Caco-2 cells grown on differently structured Si waferswith a trench width between 12 and 14 lm and a mesa width of around6 lm as compared to a mesa width of around 12 lm (mean ± SD, n = 3)

Width of trench(lm)/width ofmesa (lm)

Mean cell count

Day 3 post-seeding

Day 7 post-seeding

Day 14 post-seeding

12.2/5.9 (VI) 50,540 ± 3144 110,079 ± 721 137,894 ± 310613.8/6.6 (VII) 60,434 ± 744 117,756 ± 1816 124,405 ± 381112.1/12.3 (XI) 54,002 ± 283 97,778 ± 1381 116,919 ± 73713.8/11.7 (X) 62,512 ± 1479 91,420 ± 2634 116,871 ± 1269


on day 14 post-seeding as compared to 85,052 adhered cellson the unpatterned reference. For all other evaluated trench/mesa patterns (Table 2) no significant impact of the underly-ing structure on the adhesion of Caco-2 cells was observed.To give an overview about the interaction between trenchdepth and trench width with the number of adherent cellson day 14 post-seeding the mean cell count on all substratesis shown in Fig. 4 as a 3D mesh plot.

In order to estimate the influence of the mesa dimen-sions on cell growth, the width of the mesa was correlatedto the width of the trench. At this, 10–20 lm trenches werechosen to elucidate the influence of small and wide mesas.Whereas on day 3 post-seeding no noteworthy influence

was observed, 7 days as well as 14 days post-seeding thecellular proliferation was significantly higher (day 7:p = 0.001; day 14: p = 0.017) on substrates with a mesaof around 6 lm in width as compared to substrates witha mesa of around 12 lm in width (Table 3).

4.5. Electron microscopic investigations

The impact of patterned wafers on the growth behaviorof cells as well as on the formation of connected cell layerswas elucidated by scanning electron microscopy. For thatpurpose cells were seeded on the patterned wafers andimaged after 7 days of culture. Although the cells are flatlypressed on the wafers and disruption or holes in the celllayer may emerge due to the evaporation process, thismethod provides for an exact determination of the positionof attached cells via high-resolution imaging.

SEM revealed that Caco-2 cells are able to bridge thetrenches and form a layer on top of the structures. This phe-nomenon was observed on trench/mesa patterned struc-tures with a lateral distance between the mesas in therange of 10–20 lm and a trench depth of 29 lm up to35 lm (Fig. 5A). Using substrates with trench widths inthe range of 20–60 lm (Fig. 5B) the Caco-2 cells were foundat the bottom of the trenches and, if the mesa were wideenough, on top of the mesa. Narrower trenches were simplyovergrown by the cells, because the dimensions were toosmall that Caco-2 cells could reach the base of the trenchesat the moment of seeding or enter during cell reproduction.

4.6. Tight junctional network of Caco-2 cells on structured

wafers

The tight junctional network that confirms the forma-tion of a continuous cell layer was monitored by stainingthe ZO-1 protein found in intercellular junctions of epithe-lial cells. Using immunofluorescent staining techniques, 14days post-seeding, a dense network of tight junctions wasassessed on structures with a trench width of 10–20 lmand a depth between 34 and 37 lm (IV, VI–XI; Table 2).Hence, these lateral microstructures had no major impacton the formation of tight cell layers of Caco-2 cells. Obvi-ously due to the narrow distance between the mesa, thedyeing solution seemed to remain in some trenches despite

Fig. 5. Electron microscopic images of Caco-2 cells on patterned silicon wafers, 7 days post-seeding, mounted with 45� tilt. (A) Trench width 15.3 lm,mesa width 16.5 lm, trench depth 29 lm, 1000� magnification. (B) Trench width 60.0 lm, mesa width 58.0 lm, trench depth 32.9 lm 500� magnification;1 keV acceleration voltage. Trenches (T) and mesas (M) are indicated by corresponding bars displaying the width of the trench/mesa structures.


rigorous washing. Thus, the trenches appear higlighted insome areas. In general, ZO-1 images revealed that the tightjunctional network of cells with an enhanced proliferationrate caused by the underlying patterned substrates was in-plane all over the structure. This confirms that Caco-2 cellsadhere to mesa and simply bridge the trenches. This phe-nomenon was observed on structures with a trench widthof 10–20 lm and a trench depth of 34–37 lm (IV, VI–XI;Table 2). Interestingly, no major differences were observedbetween small and wide mesas (Fig. 6A and B). Counterstaining of the cell nuclei confirmed a totally planar celllayer (Fig. 6C and D). In contrast, using trenches with awidth of >20 lm and a depth of >40 lm (II–III; Table 2)resulted in an uneven tight junctional network indicatingdirect attachment of Caco-2 cells to the bottom of thetrenches or the top of the mesa to form independent cellpatches without direct contact (Fig. 6E).

5. Discussion

The aim of our study was to investigate the influence ofmicro-patterned trench/mesa surfaces on growth and dif-ferentiation of Caco-2 cells to provide suitable surfacesfor a cell-based microelectronic biosensor. Even in naturethe topography of supports is known to modify cell growthalthough being not recognized due to the complexity ofthese surfaces. In contrast, man-made microstructures fab-ricated by lithography and reactive ion etching offer theadvantage that single structural parameters can be variedindependently, which should allow a clear correlationbetween the observed cell behavior and the individual fea-tures of the surface topography. To identify structures sup-pressing or promoting cell growth, patterned surfaces withstructure scales in the range of the cell diameter and slightlybelow or above were investigated regarding their influenceon cell attachment, cell viability and proliferative activity.

The development of micropatterned Si substrates byreactive ion etching delivered only an edge steepness of

around 70� at the best, so that a lateral loss on the realstructures compared to the designed structures had to beaccepted. Consequently, scanning electron microscopyrevealed that mesa were narrower and trenches were wideras given by the lithographic mask. The surface roughnessof etched and non-etched areas of the supports wasassessed by AFM-scanning the surface. The statisticalevaluation of the individual material layers given by theRMS displayed a roughness below 3 nm in all cases. Asthe measured surface roughness is very low and within anarrow range, the topographic differences of the samplesurface are assumed to exert no significant influence on cellgrowth. According to the literature regular features in therange of >30 nm influence cell adhesion [16], but no effectson cell adhesion and proliferation have been reported forirregular structures and for structures smaller than 3 nm.As the surface roughness in trenches (etched) and on mesas(unetched) is in a similar range, it is most likely that thesurface is perceived as smooth by the cell layer and exertsno influence on cell growth.

Furthermore, adhesion and proliferation of humanCaco-2 cells were investigated. Adhesion studies revealeda reduced adherence and proliferation on trench/mesastructures with >40 lm in depth and mesa of <13 lm inwidth. This decrease might be explained by the featuresof the trench/mesa structure: a mesa smaller than the celldiameter and a rather smooth edge steepness accumulateseeded cells at the bottom of the trenches. A mesa widthof 612 lm is below the cell diameter and offers less spaceto adhere, whereas a trench with a width above the celldiameter and a depth of more than 30 lm can neither bebridged by the cells nor perceived as flat substrate. Hence,on those patterns the cells can only grow within thetrenches and the lack of cell–cell contacts between the dis-tinct cell islets reduces cell growth and suppresses forma-tion of a continuous cell layer.

Interestingly, the number of adherent cells and cell growthwas increased by wafers with a trench width of 10–20 lm and

Fig. 6. Immunofluorescent images of the ZO-1 staining (A–C, E) or the cell nuclei (D). Trench width/mesa width/trench depth (lm): (A) 16.6/10.4/37(IX); (B) 14/8/34 (VIII); (C, D) 13.8/11.7/37.6 (X); (E) 13.5/20.8/41 (XX). Images (C) and (D) show the same area of the trench /mesa pattern. Trenches(T) and mesas (M) are indicated by corresponding bars displaying the width of the trench/mesa structures.


a depth of around 35 lm. This result may be due to a trenchwidth within the range of the cell diameter triggering Caco-2cells to simply bridge the gap formed by the underlyingtrenches. That way the culture medium can freely flowthrough the channels built up by the trenches and simulatethe physiological transport of proteins and ions at the baso-lateral side. Consequently such a device is proposed as a pro-totype for a microfluidic cell culture system that improves theperformance of the cells by continuously supplying them

with fresh culture medium. Microfluidic systems with con-centration gradients provide new tools for biological andchemical analysis such as monitoring of cellular responseand screening of drugs [17]. It is known that concentrationgradients of chemokines and growth factors play an impor-tant role in many biological processes such as immuneresponse [18] or cancer metastasis [19].

Moreover, microstructured landscapes with mesa ofaround 6 lm enhanced the cell proliferation more than


mesa of around 12 lm. A mesa of 12 lm in width providesan adhesion area for the major part of the cell to adhereand only a small part of the cell bridges the gap. In con-trast, for the smaller mesa it is likely that only a minor partof the Caco-2 cell adheres to the mesa and a major part ofthe cell body bridges the trench. Hence, the basolateral faceof the cells is accessible from the lower compartment andthe resulting transport of ions and proteins mimics physio-logical conditions, thus approaching in vivo conditionsmore accurately.

A general overview about the growing behavior andthe exact position of Caco-2 cells cultured on microstruc-tured Si substrates was acquired by microscopic inspec-tion. When the lateral distance between the patterns wasfar below the average cell diameter (<10 lm), the Caco-2 cells simply overgrew the wafers since the lateral dis-tance between two mesas is too narrow for the cells toreach the bottom of the trenches at the moment of seed-ing or during proliferation. On patterns with broader lat-eral distances a different cell distribution was observed:trenches with a width around the cell diameter andslightly below (10–20 lm) were simply bridged by thecells. However, if the lateral space of the trenches was>20 lm small islets of Caco-2 cells were observed. Thesecell patches predominantly accumulated in the trenches,indicating that the cells slide down the gently inclinededges and adhere at the bottom of the structure. At this,it has to be kept in mind that increasing the width of thetrenches is associated with decreasing edge steepness.Additionally, few cell islets were found on top of the mesawhen their width was higher than the cell diameter. Theseislets had no direct contact among each other so that nocontinuous cell layer was formed, which resulted in lowerproliferative activity.

Since Caco-2 cells are not electrically active such asnerve cells or muscle cells and the electrical conductivityof the monolayer is almost limited to the paracellularion flux [20], the development of tight junctions inducesthe so-called transepithelial electrical resistance [21,22].Thus, the expression of the ZO-1 protein initiates thedevelopment of the barrier function as well as the electri-cal resistance [23]. In view of the demands of high-throughput assays based on microelectronic biosensors,ZO-1 proteins of Caco-2 cells were stained. The immuno-fluorescent staining of ZO-1 proteins confirmed the for-mer morphological observations. On micropatternedstructures with trenches of 10–20 lm in width and around35 lm in depth a dense net of tight junctions was found atthe level of the mesa tops. Counter staining of the nucleiwith propidium iodide ascertained not only the planarityof the cell layer but also the adherence of the cells tothe mesa as well as the bridging of the trenches. The dif-ferent effects of mesa of 6 or 12 lm in width could not beobserved that clearly by immunofluorescent staining tech-niques. As already shown by the adherence studies, thetight junctional network on structured substrates with atrench depth of P40 lm (II, III, XIV, XX) was separated,

indicating direct attachment of the Caco-2 to the bottomof the trenches or to the tops of the mesa. Thus, indepen-dent islets of cells lacking any direct contact wereobserved.

6. Conclusions

The presented results indicate that careful characteriza-tion of substrate materials is the first and basic step for thedevelopment of cell-based microelectronic biosensors andmay significantly impact the future design of appropriatesensor surfaces. All in all, the results of our study revealthat microstructuring of the growth support has an enor-mous impact on the adhesion and proliferation of Caco-2cells. Whereas cell growth was reduced on structures withdeep trenches (>40 lm) and narrow mesas (<13 lm), atrench width of 10–20 lm and a trench depth of around35 lm enhanced cell proliferation due to the fact thatCaco-2 cells are able to bridge trenches not wider than theirdiameter provided that the mesa is wide enough to allowproper adherence. Furthermore, a trench depth of around35 lm provides the possibility to flow culture mediumthrough the trenches below the bridging cells which contin-uously supplies the cells with fresh medium from the baso-lateral side. That way, new vistas for the design ofinnovative biosensors are opened up by combining micro-fluidic systems with microelectronic detection for screeningthe toxicological and biopharmaceutical characteristics ofnew drug candidates.

Acknowledgements

We acknowledge B. Basnar for the atomic force micros-copy measurement of the sample surfaces. Fabrication ofmicrostructured surfaces was performed at the center formicro- and nanostructuring (CMNS) of the ViennaUniversity of Technology that is acknowledged for provid-ing this facility. Part of this paper and part of the work itconcerns were generated in the context of the CellPROMproject, funded by the European Community, contractNo. NMP4-CT-2004-500039 under the 6th FrameworkProgramme for Research and Technological Developmentin the thematic area of ‘‘Nanotechnologies and nano-sci-ences, knowledge-based multifunctional materials andnew production processes and devices”. The work and re-sults reported reflect the author’s views and the communityis not liable for any use that may be made of the informa-tion contained therein.

References

[1] Tavelin S, Grasjo J, Taipalensuu J, Ocklind G, Artursson P.Applications of epithelial cell culture in studies of drug transport.Methods Mol Biol 2002;188:233–72.

[2] Xing JZ, Zhu L, Jackson JA, Gabos S, Sun XJ, Wang XB, et al.Dynamic monitoring of cytotoxicity on microelectronic sensors.Chem Res Toxicol 2005;18:154–61.


[3] Pescheck M, Schrader J, Sell D. Novel electrochemical sensor systemfor monitoring metabolic activity during the growth and cultivationof prokaryotic and eukaryotic cells. Bioelectrochemistry2005;67:47–55.

[4] Lan S, Veiseh M, Zhang M. Surface modification of silicon and gold-patterned silicon surfaces for improved biocompatibility and cellpatterning selectivity. Biosens Bioelectron 2005;20:1697–708.

[5] Bogner E, Dominizi K, Hagl P, Bertagnolli E, Wirth M, Gabor F,et al. Bridging the gap – biocompatibility of microelectronic mate-rials. Acta Biomater 2006;2:229–37.

[6] Mata A, Boehm C, Fleischman AJ, Muschler G, Roy S. Growth ofconnective tissue progenitor cells on microtextured poly-dimethylsiloxane surfaces. J Biomed Mater Res 2002;62:499–506.

[7] Walboomers XF, Monaghan W, Curtis ASG, Jansen JA. Attachmentof fibroblasts on smooth and microgrooved polystyrene. J BiomedMater Res 1999;46:212–20.

[8] Oakley C, Jaeger NA, Brunette DM. Sensitivity of fibroblasts andtheir cytoskeletons to substratum topographies: topographic guid-ance and topographic compensation by micromachined grooves ofdifferent dimensions. Exp Cell Res 1997;234:413–24.

[9] Ong SE, Zhang S, Du HJ, Fu Y. Fundamental principles andapplications of microfluidic systems. Front Biosci 2008;13:2757–73.

[10] Yang M, Li CW, Yang J. Cell docking and on-chip monitoring ofcellular reactions with a controlled concentration gradient on amicrofluidic device. Anal Chem 2002;74(16):3991–4001.

[11] Miret S, Abrahamse L, de Groene EM. Comparison of in vitromodels for the prediction of compound absorption across the humanintestinal mucosa. J Biomol Screen 2004;9:598–606.

[12] He X, Sugawara M, Kobayashi M, Takekuma Y, Miyazaki K. Anin vitro system for prediction of oral absorption of relatively water-soluble drugs and ester prodrugs. Int J Pharm 2003;263:35–44.

[13] Kim JS, Mitchell S, Kijek P, Tsume Y, Hilfinger J, Amidon GL. Thesuitability of an in situ perfusion model for permeability determina-

tions: utility for BCS Class Ibiowaiver requests. Mol Pharm2006;3(6):686–94.

[14] Fogh J, Fogh JM, Orfeo T. One hundred and twenty-seven culturedhuman tumor cell lines producing tumors in nude mice. J Natl CancerInst 1977;59:221–6.

[15] Hidalgo IJ, Raub TJ, Borchardt RT. Characterization of thehuman colon carcinoma cell line (Caco-2) as a model systemfor intestinal epithelial permeability. Gastroenterology 1989;96:736–49.

[16] Martines E, McGhee K, Wilkinson C, Curtis A. A parallel-plate flowchamber to study initial cell adhesion on a nanofeatured surfaceIEEE Trans. Nanobioscience 2004;3:90–5.

[17] Yang M, Yang J, Li CW, Zhao J. Generation of concentrationgradient by controlled flow distribution and diffusive mixing in amicrofluidic chip. Lab Chip 2002;2:158–63.

[18] Zigmond SH. Ability of polymorphonuclear leukocytes to orient ingradients of chemotactic factors. J Cell Biol 1977;75:606–16.

[19] Condeelis J, Singer RH, Segall JE. The great escape: when cancer cellshijack the genes for chemotaxis and motility. Annu Rev Cell Dev Biol2005;21:695–718.

[20] Grasset E, Pinto M, Dussaulx E, Zweibaum A, Desjeux JF. Epithelialproperties of human colonic carcinoma cell line Caco-2: electricalparameters. Am J Physiol 1984;247:C260.

[21] Isoda H, Talorete TP, Han J, Nakamura K. Expressions of galectin-3,glutathione S-transferase A2 and peroxiredoxin-1 by nonylphenol-incubated Caco-2 cells and reduction in transepithelial electricalresistance by nonylphenol. Toxicol In Vitro 2006;20:63–70.

[22] Mathieu F, Galmier MJ, Pognat JF, Lartigue C. Influence of differentcalcic antagonists on the Caco-2 cell monolayer integrity or ‘‘TEER, ameasurement of toxicity”? Eur J Drug Metab Pharmacokinet2005;30:85–90.

[23] Delie F, Rubas W. A human colonic cell line sharing similarities withenterocytes as a model to examine oral absorption: advantages andlimitations of the Caco-2 model. Crit Rev Ther Drug Carrier Syst1997;14(3):221–86.

Influence of structured wafer surfaces on the characteristics of Caco-2 cells

Documents

Transcript of Influence of structured wafer surfaces on the characteristics of Caco-2 cells