Appl.Phys.A 65,341–348
(1997)
C Springer-Verlag 1997
Invited paper
Fluorescence interference-contrast microscopy of cell adhesion
on oxidized silicon
D.Braun,P.Fromherz ∗
Department of Membrane and Neurophysics,Max-Planck-Institute for Biochemistry,D-82152Martinsried-München,Germany (Fax:+49-89/8578-2822,E-mail:fromherz@biochem.mpg.de)Rec
eived:26June 1997/Accepted:30June 1997
Abstract.Standing modes of light in front of the reflecting surface of silicon modulate the excitation and emission of fluorescent dyes.This effect was ud to determine the dis-tance of a biomembrane from an oxidized silicon chip.The membrane of a red blood cell (ghost)was stained with a cya-nine dye and attached with poly-lysine to a surface structured with microscopic steps of silicon dioxide on silicon.The sys-tem was illuminated in a microscope.The fluorescence inten-sity of the membrane depended on the height of the steps.The data were fitted by an optical theory which accounts both for the interference of the exciting light and for the interference of the emitted light at a finite aperture.The distance between the membrane and the silicon dioxide was determined to be 12nm .
PACS:33.50.D;68.35.G;78.66;87.22
Nerve cells and silicon devices can be joined by electrical in-duction when the cell membrane is attached cloly to an ox-idized chip [1].Recording as well as stimulation of neuronal activity from oxidized silicon have been reported [2,3].The strength of coupling depends on the width of the electrolyte which parates the insulating layers of the cell membrane and the silicon dioxide.This width was estimated to be in the range 10–100nm on the basis of the coupling experiments themlves [4].No direct measurements are available.
教师幸福感悟随笔
Usually the attachment of cells to surfaces is studied by reflection interference contrast microscopy (IRM /RICM)[5–9]or by total internal reflection fluorescence microscopy (TIRFM)[10–15].In both methods the cell is illuminated by visible light through a transparent support.In the RICM method the light is reflected from the substrate/electrolyte and electrolyte/cell interfaces and gives ri to an interference pattern.In the TIRFM method the cell is illuminated under the condition of total reflection;the membrane or electrolyte are labelled with a fluorescent dye which is excited by the evanescent wave.
∗Correspondence author
RIC-microscopy and TIRF-microscopy cannot be ud for cells on silicon,which is not transparent in the visible range.For that reason a new method of fluorescence inter-ference contrast (FLIC)microscopy was propod to map the distance between a membrane and oxidized silicon [16].FLIC-microscopy takes advantage of the Wiener effect [17–19],the interference of incident and reflected light above a mirror.The standing modes of the electromagnetic field above the surface of silicon modulate the excitation and the emission of a fluorescent dye which is disperd in an ad-jacent solvent or bound to a macromolecule or membrane.Some features of FLIC-microscopy were tested with a dry monomolecular film [16].
In this paper we consider FLIC-microscopy as a tool to study cell adhesion.It is applied to determine the distance be-tween a biomembrane and a silicon chip.As a test system we ud ghosts of human erythrocytes becau of their homo-geneous membrane.In the first part we describe the concept of the method and explain the experimental technique.In the cond part we prent the data and discuss the results.1Materials and methods
First we consider the principles of FLIC-microscopy as ap-plied to cell adhesion.Then we describe the fabrication of the chips,the preparation of stained erythrocyte membranes,the attachment of the cells to the chip,the photometric t-up,the optical theory and the evaluation of the data.1.1Principles of FLIC-microscopy
Silicon reflects visible light.The interference of the incident light with the reflected light gives ri to standing modes of the electromagnetic field.As a result the electronic excitation of a dye molecule depends on its position in front of the mir-ror.By analogy,the fluorescence emission of a dye molecule is affected due to the interference of light which is emitted with and without reflection,or in other words due to emis-sion into an unoccupied standing mode of the electromagnetic field.
342
Fig.2.Amphiphilic trimethin-indocarbocyanine dye DiIC12ud to stain
the erythrocyte membrane
21–23]with two dodecyl-chains(DiIC12,Molecular Probes)
(Fig.2).Three aspects were relevant for the choice of this
dye:(i)The transition dipole of excitation and emission is
parallel to the cell membrane[24];(ii)theflip-flop of the dye
from one side of the membrane to the other side is slow[25];
(iii)photobleaching of the dye is rather weak[26].
We made a5mM solution of the dye in ethanol.5µl were
added to15ml TRIS buffer such that a suspension of mi-
crocrystals was formed.It was ud to resuspend the upper
(liquid)part of the centrifuged pellet of ghosts.We isolated
the stained ghosts immediately afterwards by centrifugation
at3000g for10min.The supernatant with the microcrystals
of the dye were discarded and the ghosts were resuspend-
ed in15ml TRIS buffer.Most ghosts had an erythrocyte-like
shape.Some were spherical with a diameter of about6µm.
1.4Chips
We prepared chips from polished n-doped(4–8Ωcm)four-
inch silicon wafers(Freiberger,Freiberg,Germany).They
were cleaned by the standard RCA procedure[27].We pre-
pared a homogeneous layer of silicon dioxide with a thickness
of about135nm by thermal growth in an oven at1000◦C
(E1200Lab,Centrotherm,Blaubeuren).The wafer was cov-
ered with a photoresist by spin-coating and illuminated in
a mask-aligner through a metal mask with stripes of5µm
width,parated by5µm.After development we removed
about83nm of the oxide in the open areas by etching with
fluoric acid[27].Then the photoresist was stripped,the wafer
was cleaned and coated again with photoresist.In a cond
illumination we ud the same mask rotated by90◦.After de-
velopment we removed about42nm of the oxide in the open
areas.Then the photoresist was stripped and the wafer was cut
into chips(3.4cm×1.0cm).The surface exhibited a pattern
of squares with5µm edge length as illustrated in Fig.3.The
heights of the oxide were about11,53,92and135nm.
343
The chips were sonicated for10min at70◦C in an acidic
detergent(5%Ultrax102,KLN Heppenheim),rind with
Milli-Q water(Millipore Inc.),sonicated for2min at70◦C in an alkaline detergent(2%Tickopur RP100,Bandelin,Berlin)
and rind with Milli-Q again.Then the chips were sonicat-
ed four times in Milli-Q water for10min(twice at70◦C, twice at room temperature)and dried with nitrogen[16,28].
After cleaning,the thickness of the oxides was measured by
牛开头的四字成语an ellipsometer(SD2000,Plasmos,München)using a re-fractive index n ox=1.460at633nm.For that purpo we ud a quadruple of reference squares with a size500µm×
500µm which were arranged on the chip at a paration of 1cm.They were fabricated together with the microscopic
steps under identical conditions of etching.
1.5Adhesion
The chips were placed into a petri dish of35mm diameter
(Falcon3001,Becton Dickinson,Plymouth).The ghost mem-brane did not adhere to clean silicon dioxi
de as it is negatively charged[29].We ud poly-lysine to induce attachment[30]. The chips were incubated in a solution of poly-L-lysine(MW 10,000,Sigma,Heidelberg)in TRIS buffer at a concentration of0.5mg/ml for2h at room temperature.They were rind three times with TRIS buffer.(The layer of poly-L-lysine had a thickness of about0.7nm after rinsing with milli-Q water and drying,as estimated by ellipsometry.)3ml of the sus-pension of ghosts were added to the wet chip.The ghosts were allowed to diment for20min.We removed the liquid with non-adhering cells carefully and added3ml of TRIS buffer.
1.6Photometry
Thefluorescent ghosts were studied in the same t-up as ud for monomolecularfilms[16].A water immersion ob-jective(100×)with a numerical aperture1.0was ud in the microscope(Axioskop,Zeiss,Oberkochen).The cells were focud in red light(>630nm).Then the chip was illuminat-ed monochromatically at546nm by a high pressure mercury lamp(Zeiss)through a dichroic mirror(Q565LP,AHF Anal-yntechnik,Tübingen)and a bandpassfilter(546/10nm, 546FGS,Andover,Salem,NH,USA).Thefluorescence was detected around610nm through the dichroic mirror and a bandpassfilter(610/70nm,AHF Analyntechnik,Tübin-gen).
Afluorescence picture was taken by illumination for 40ms.We ud a CCD camera with752×582pixels(Sony chip ICX039AL,HRX,Theta System,München).The size of a single pixel corresponded to approximately90nm×90nm of the object.The signal of every cond line of the camera was transferred to a PC by a frame grabber(ITEX AFG,Stemmer,München)with8-bit resolution.We evalu-ated the storedfluorescence pictures in three steps:(i)the adhesion area of a single ghost was approximated by an el-lip and divided into four ctors on each oxide step;(ii) the number of pixels with a certain brightness was count-ed on each ctor to form four histograms of the intensi-ties;(iii)Gaussians werefitted to the histograms.The low-est components(<10%)of a histogram were not consid-ered.1.7Theory
Thefluorescence intensities(average with standard deviation) on the four oxides of different thickness werefitted by an op-tical theory of interference.We summarize here the crucial relations.For details e[16,31].
The probability per unit time P ex for excitation of a dye molecule is determined by the intensity of illumination I(λin) (quanta per area,time and wavelength interval),the extinc-tion coefficient of the dyeε(λin)and the relative strength F in of the electricalfield of incident light at the position of the dye projected onto the direction e ex of the transition dipole of excitation.We obtain(1)by averaging over
all directions and polarizations of the incident light within the aperture of the microscope,by averaging over all orientations of the dye in the plane of the membrane and by integration over all wavelengthsλin of the incident light1:
P ex∝
dλin I(λin)ε(λin) |F in·e ex|2 .(1)
The probability per unit time P em to detect an emitted quantum from an excited molecule is given by(2)below.It depends on the quantum yieldΦdet(λout)of the detection sys-tem,on thefluorescence spectrum(quanta per wavelength interval)of the dye,on the relative strength F out of the local electricalfield of that mode which accepts the emitted photon and on the direction e em of the transition dipole of emission. We average over the polarizations and directions of detected light within the aperture of the microscope and over the orien-tations of the dye in the plane of the membrane.Finally we integrate over the wavelengths of detectionλout:
P em∝
dλoutΦdet(λout)f(λout) |F out·e em|2 .(2)
The local relativefield strengths F in and F out depend on the optical properties of the asmbly and on the position of the dye molecule,in particular on the thickness d ox of the oxide and on the distance d cleft between membrane and support.They are computed by matrix methods[16,32].
Under stationary illumination we detect an averageflow Jflof quanta per unit time from a dye molecule.It depends on the probability of detected quanta per unit time P em from an excited molecule and on the probability that the molecule
1The various directions of exciting radiation may be integrated in the membrane(medium3)at the position of the absorbing dye and the di-rections of emitted radiation in water(medium4)at the entrance of the detecting objective.The energyflows per unit polar angle dΘin3
and dΘout
4春节剪纸
,respectively,depend on the direction,even if the illumina-tion through the objective and the emission from the dye are homoge-neous with respect to different directions within the angle of aperture. This effect is due to refraction and becomes important for high apertures as ud in the prent experiments.It can be described as a modula-tion of the intrinsic aperture functions–which are A in
Θin3
=1and A out
Θout4
=1within the angle of aperture and zero elwhere–as
n3cosΘin3
独领风潮n4cosΘin4
A in
Θin3
and
n4cosΘout
4
n3cosΘout
3
A out
Θout4
,
respectively,withΘin4expresd byΘin3andΘout
3
byΘout
4
according to Snellius’law.Appropriate substitutions are to be made when the integration of illumination is computed in water(medium4),too.
344
is in its excited state.We obtain(3)with the total transition probabilities offluorescence kfland of nonradiative decay k nr:
Jfl=
1
kfl+k nr
·P ex·P em.(3)
If the transition probabilities offluorescence and non-radiative decay are independent of the thickness of the oxide, we may u(4).This approximation was shown to be ad-equate for the cyanine dyes S9[16]and S27/DiIC18[31] in a dry monomolecular lipidfilm as well as in a wet lipid membrane:
Jfl∝P ex·P em.(4)
1.8Optical model
We describe the optics of an adhesion site by a model with five homogeneous and isotropic layers as sketched in Fig.4. It consists of bulk silicon,a layer of silicon dioxide(thickness d ox),a layer of the extracellular medium(thickness d cleft),the cell membrane(thickness d mem)and the intracellular medi-um.The complex refractive index of silicon(refractive index n Si,attenuation indexκSi)was taken from[33].We ud the table of refractive index of silica from[34]and matched the dispersion of the table to a refractive index n ox=1.460at 632.8nm of our thermally grown oxide.(The same value was ud to determine the thickness of the oxide by ellipso-metry.)We described the extracellular cleft by the refractive
cad字体安装index of water n cleft=1.333.The membrane was character-ized by a thickness d mem=4nm[7]and a refractive index n mem=1.450[7].For the cytoplasm we ud the refractive index of water n cyt=1.333.In the model the dye molecule was placed within the membrane layer clo to its lower sur-face.The transition moments of excitation and emission were aligned parallel to the membrane with random orientation in the plane.The angles of aperture in water were47.3◦for excitation and48.6◦for emission.The values correspond to numerical apertures of0.985and1.0of the objective,re-spectively.The lower aperture for excitation was assigned
on the basis of systematic measurements offluorescence in-terference with supported lipid membranes[31].The exci-tation was monochromatic withλin=546nm.An emission
345
高考作文题目
Fig.6.Photograph of fluorescent erythrocyte membranes (ghosts)stained with the dye DiIC 12on stepped silicon dioxide on silicon.Four areas of homogeneous fluorescence are en in each cell due to the membrane at-tached to four levels of oxide.In some cells this pattern is superpod by a quence of rings.The scale bar is 5µm
肯德基炸鸡height.On each step the intensity is rather homogeneous.In some cells a ring-shaped pattern is superpod.It is due to the upper part of the cell when it has the shape of a flat cupola being within the focus of the microscope.
First we evaluate the fluorescence intensities of the at-tached membrane of one lected cell.Then we consider a population of attached ghosts.Finally we evaluate the pro-file of the upper membrane from the ring-shaped pattern.2.1Single cell
A digitized image of a lected cell is depicted in Fig.7.The height of the steps of silicon dioxide is indicated.The fluores-cence is low on the thinnest oxide of 10.9nm ,modest on the oxide of 52.7nm ,bright on the oxide of 92.1nm and mod-est again on the oxide of 135.4nm .The histograms of the intensities within the four areas are shown in the figure.The data were fitted by Gaussians.The average intensities and the width of the distributions (±σ)are plotted versus the height of the steps in Fig.8.
The four data points in Fig.8were fitted with the op-tical theory of interference using (4)with (2)and (3).The wavelength of excitation,the spectrum of emission,the n-sitivity spectrum of detection and the apertures of excitation and emission were defined by the optical t-up.They deter-mine the peri
od of the wave.The five-layer model (Fig.4)with silicon,silicon dioxide,extracellular medium,mem-brane and intracellular medium determined the pha of the wave.The distance d cleft of the outer surface of the membrane from the silicon dioxide was a free parameter.The other free parameters were the amplitude of the wave and a constant background.As a result we obtained d cleft =12.1nm .2.2Accuracy
The total statistical errror of the distance d cleft of a single ghost had two sources:the error of the fit of the optical theory and the uncertainty of the thickness of the oxides.We found that the statistical deviation of the fit was ∆d cleft =±0.4nm .
The cond error was due to the different positions of ellip-sometry and ghost on the chip.The uncertainty of the oxide thickness within 1cm along the chip was ±0.1nm .Thus the total statistical error of the distance between membrane and oxide was ∆d cleft =±0.41nm .
There were three sources of systematic errors:(i)the measurement of thickness by the ellipsometer;(ii)the meas-urement of fluorescence intensity by the CCD camera and by digitization;and (iii)the uncertainty of the parameters which enter the optical model.We estimated the precision of the ellipsometer to be ∆d ox =±0.2nm .This corresponds to an error of d cleft =±0.22nm .The error of the
intensity due to the detection system was ±0.5units in a range of 0–255units.The resulting error was around d cleft =±0.15nm .The estimated errors of the parameters of the optical model are summarized in Table 1together with the resulting
errors
Fig.7a,b.Fluorescence of a single ghost.a Image taken by a CCD-camera.The thickness of the four oxide layers is indicated.The scale bar is 5µm .(The bright spots are dye crystals.)b Histogram of fluorescence intensity on the four regions.The number of pixels with a certain intensity is plotted versus the intensity in 8-bit resolution
346
marketing
Parameter Value Systematic Resulting
error error in d cleft n Si atλ=633nm3.87±0.05±0.00nm κSi atλ=633nm0.017±0.005∓0.04nm n cleft1.333+0.05−0.50nm n mem1.450±0.05∓0.06nm d mem4.0nm±0.5nm∓0.04nm n cyt1.333+0.05−0.02nm angle of transition dipole90◦−10◦−0.03nm aperture excitation47.3◦±1◦±0.42nm aperture emission48.6◦±1◦±0.29nm max.of spectrum emission565nm5nm±0.01nm 2.4Photobleaching
A problem with thefluorescence measurements is the -
lective photobleaching of the dye.The light intensity at the
position of the membrane is not identical on the four steps of different height.As a conquence the rate of irreversible
bleaching of the dye by photochemical reactions from the
excited state is not identical.The relative intensity offluores-cence on the four steps may change.An example is shown
in Fig.9when a cell was obrved for40ms before and
after10s of illumination.Thefirst data t led to a distance d cleft=12.2±0.3nm(stochastic error).The relative intensi-ties of the four data points were changed in the cond meas-
urement due to photobleaching.We obtained d cleft=16.0±3.5nm.The quality of thefit was lower,as the condition of
homogeneous staining was no longer fulfilled.
2.5Cell population
We evaluated thefluorometric data of25ghosts.The mean
distances d cleft of the membrane from silicon dioxide with their total stochastic error are shown in Fig.10.As an av-erage of the distances weighted with their stochastic
error
