J. Cell Set. 14, 215-223 (1974) 215

Printed in Great Britain

THE DENSITIES OF COLLOIDAL IRON

HYDROXIDE PARTICLES BOUND TO

MICROVILLI AND THE SPACES BETWEEN

THEM: STUDIES ON GLUTARALDEHYDE-

FIXED EHRLICH ASCITES TUMOUR CELLS

L.WEISS AND J. R. SUBJECK*Department of Experimental Pathology, Roswell Park Memorial Institute, Buffalo,N.Y., U.S.A.

SUMMARYThe densities of positively charged, colloidal iron hydroxide particles binding to the surfaces

of glutaraldehyde-fixed cultured Ehrlich ascites tumour cells, were determined on cells main-tained in suspension and monolayer cultures. In both cases, there was a higher density ofparticles bound to the surfaces of microvilli than to the spaces between them. Prior treatmentof the cells with ribonuclease did not significantly alter the proportions of particles binding tothe 2 different surface regions, indicating that the ribonuclease-susceptible binding sites aredistributed with similar mean densities over the microvilli and the regions between them. Priortreatment of the cells with neuraminidase reduced the densities of particles binding to microvillito those of the intermicrovillous spaces, indicating that the original higher density over themicrovilli is due to an increased number of sialic acid residues at their surfaces.

The findings are discussed in relationship to observed changes in the surface density ofmicrovilli when these cells convert from suspension to monolayer cultures.

INTRODUCTION

It seems likely that in a complex manner, the densities and distributions of ionogenicsites at their peripheries, help regulate some cell contact interactions (for review, seeWeiss & Harlos, 1972). Among the surface anionic moieties of interest in this respectare sialic acids (for review, see Weiss, 1973 a), and ribonuclease-susceptible groups(Weiss & Mayhew, 1969). In addition to these factors, it is a common observation thatmicrovilli often appear to play a role in promoting initial contacts between cells.

In this communication we report quantitative electron-micrographic studies of thebinding of positively charged colloidal iron hydroxide particles, which act as markersfor some anionic groups, to the surfaces of glutaraldehyde-fixed Ehrlich ascites tumour(EAT) cells. Some of the cells were incubated with either neuraminidase or ribonu-clease prior to fixation. We then use these data to compare the densities of particle-binding sites at the surfaces of microvilli and the spaces between them, in EAT cellscultured either in suspension or adherent to Epon surfaces.

• Present address: Biophysics Department, State University of New York at Buffalo,Buffalo, N.Y., U.S.A.

216 L. Weiss andj. R. Subjeck

MATERIALS AND METHODS

Cells

Ehrlich ascites tumour cells were grown in vitro in suspension culture in medium RPMINo. 1630 (Moore, Sandberg & Ulrich, 1966) supplemented with 5 % foetal calf serum. Suspen-sion cultures were divided in two; one half was allowed to continue to grow in suspension andthe other was cultured in monolayers on Epon substrates. After 7 h, by which time the cellsgrown in stationary cultures were adherent to and spread on the Epon, both types of culturewere fixed simultaneously.

Preparation of Epon substrate

The Epon is the same preparation used in this laboratory in embedding for electron micro-scopy (as described below). A layer of Epon was cured adherent to the bottoms of T-flasksor Petri dishes. The surfaces were then washed twice in distilled water, once with ethanol as asterilization procedure, and twice with Dulbecco's phosphate-buffered saline (PBS) at pH 7-2.The PBS contains 100 mg/1. CaCl,, and 100 mg/1. MgCls.6HsO. The vessels were filled withRPMI media 1630 + 5 % foetal calf serum and incubated at 37 °C for 24 h, then rinsed inPBS before use.

In test cultures prepared by this method, Ehrlich ascites tumour cells grew in monolayer forup to 7 days, in a manner indistinguishable from that seen on glass.

Enzyme treatment

Cells from suspension or in stationary culture were washed twice in PBS, incubated for30 min at 37 °C with either neuraminidase (Grand Island Biologicals, N.Y., U.S.A.; Vibriocholerae; 10 units/io7 cells) or ribonuclease A (Worthington Biochemical Corp., Freehold, N.J.,U.S.A.; 0-2 mg/107 cells) in total volumes of 3 ml in PBS at pH 72, and then washed twicein PBS before fixation. This and subsequent treatment did not detach the cells adherent tothe Epon-coated surfaces.

Fixation procedure

Glutaraldehyde, 0-3 % in phosphate-buffered solution, was prepared from a vacuum-distilled,nitrogen-packed stock (Polyscience Inc., Warrington, Penn.). Prior to use, the glutaraldehydewas warmed at 37 °C, and after 30 min at 37 °C in the fixative, the cells were left in it overnightat 4 °C.

Staining procedure

After fixation, the cells were washed twice with deionized, double-distilled water and reactedwith freshly prepared, colloidal iron hydroxide solution (Gasic, Berwick & Sorrentino, 1968).This reagent contained 0-15 g/1. of iron in the form of positively charged, 8-nm diametercolloidal particles of iron hydroxide (CIH) suspended in acetic acid at pH i-8. After exposureto the stain for 1 h at room temperature, the cells were washed once in 12% acetic acid, and oncein distilled water.

Preparation for electron microscopy

Cells were treated either directly on the Epon substrate (monolayer) or, in the case ofsuspended cells, as centrifuged pellets. Dehydration was accomplished in 70 %, 95 % and abso-lute ethanol, and substitution with 2-propenol. The material was embedded in a mixture ofEpon-Araldite, and 80-nm sections were cut, and counterstained on grids with 2 % alcoholicuranyl acetate.

Determination of particle density

Electron micrographs of sections perpendicular to the plane of the cell surface were taken at abase magnification of 20000 x and projected on to the screen of a profile projector (Nikon,

CIH-binding to mtcrovilli 217

Fig. i. End on views of a ioo-nm microvillus cut by an 8o-nm thin section. The variouspossibilities are discussed in the text.

Optical Comparator Model 6C) at a final magnification of 200000 x . Tracings of cell surfaceswere made directly on paper from the projection screen. The length of the traced cell surfacewas measured with a map-measuring wheel gauge, and the total number of particles wascounted. Oblique regions of surface were rejected. In the case of cells cultured on Epon, onlythe particles adsorbed to their free upper surfaces were enumerated.

Comparison of densities of surface-bound particles

A major interpretative problem lies in comparing the apparent densities of CIH-particlesbound to the curved surfaces of rod-like microvilli, with those on the comparatively flat surfacesof the IMV-spaces as seen in edge-on views. As equal lengths of sections of the 2 regions willcontain circumferential lengths of microvilli and the correspondingly shorter flat parts of IMV-spaces, even if the particle densities on the surfaces of both regions were equal, it would appearfrom the observed particle counts that their densities over the microvilli were higher.

The majority of apparent diameters of microvilli measured in longitudinal sections fell into therange 80—120 nm; the median of 100 nm is consistent with the reported observations of manyother workers. An additional interpretative problem therefore, is that only incomplete cross-sections of microvilli will appear in the 80-nm sections observed.

We have therefore developed a ' correction' factor to allow us to reduce the observed particledensities over microvilli to the corresponding value for flat surfaces, so that the observed den-sities may be compared with those for the IMV-spaces.

When a 100-nm diameter microvillus is cut in a section of 80 nm thickness, then part of itscircumference corresponding to a minimum of (100 — 80) = 20 nm of this diameter will beexcluded from the section. The various possibilities are shown diagrammatically in Fig. 1, inwhich end-on views of microvilli of 100 nm diameter, are located with different spatial relation-ships along a hypothetical X-axis between the 2 cut surfaces of the section. In each case theorigin of the X-axis is the centre of a microvillus.

Where C is the circumference, or arc length, of a microvillus of radius, r, enclosed in thesection:

(i) In the situations shown in Fig. 1 A, B, O < x < 30,

Ci(x) = 2r{n — cos"1 (x/r)}.

(ii) In the situations shown in Fig. 1 B-D, 30 < x < 50,

Cit(x) = 2r{n-cos-1 (x/r) -cos"1 ([8o-*]/r)}.

(iii) In the situations shown in Figs. 1 D, E, 50 < x ^ 80,

C4H(«) = 2,-OT-COS-1 ([8o-*]/r)}

The mean value of the arc length or circumference of the microvilli lying in the'thin sections

, jC(x)dxSdx '

C(x) = Ci(x) + Cu

(7 = 189 nm.

is

2i8 L. Weiss and J. R. Subjeck

Table i. Particle densities in tracings from electron micrographs of cellperipheries, calculated per 4000 nm2

Control RNase-treated NANase-treated

Suspension IMV spaces 2-261017(26) 1-9310-10(26) 0-16 ±0-03 (18)Suspension microvilli 4-2710-29(23) 3-3310-29(25) 0-2010-04(24)

Monolayer IMV spaces 2-1010-19(21) 1-6910-13(19) 0-1610-05(19)Monolayer microvilli 4-26 + 0-29(25) 3-5110-32(21) 0-1610-03(25)

Table 2. Particle density ratios — microvilli:IMV spaces

Control suspension 1-89 Control monolayer 2-03RNase suspension 1-73 RNase monolayer 2-08NANase suspension 1-25 NANase monolayer i-oo

Therefore, to compare the mean particle count over microvilli in 80-nm sections, with corre-sponding counts made over the IMV-spaces, the former should be multiplied by

80

789" = ° '4 2 3-RESULTS

Representative electron micrographs of cells reacted with CIH particles are shownin Figs. 2 (monolayer) and 3 (suspension), with or without prior enzyme treatment.

The mean particle densities (± standard error), given in particles per cm of tracingsfrom electron micrographs of cell peripheries, are shown in Table 1 for microvilli andIMV-spaces of cells cultured in monolayers and suspension. With the magnificationsused here, 1 cm of tracing corresponds to 50 nm of cell surface which, in 80-nm-thicksections, is equivalent to 4000 nm2 of surface. The values given for microvilli are thecorrected values as discussed above, and the numbers in parentheses are the numbersof samples. In both control and ribonuclease-treated (RNase-treated) cells, the particledensities for microvilli are approximately twice as great as those for correspondingIMV-spaces (Table 2). For both control and ribonuclease-treated groups, the differ-ences between particle densities on microvilli and IMV-spaces are statistically signi-ficant (P < o-ooi).

In the 2 sets of cells tested, the relative particle binding after neuraminidase(NANase) treatment yielded somewhat different results. As seen in Tables 1 and 2,particle binding to surface groups not removed by NANase is approximately the samein both the microvilli and the IMV-spaces.

Using projections of electron micrographs of thin sections in which 1 cm correspon-ded to 2630 nm of cell surface, the densities of microvilli per cm of projection weredetermined. The mean density at the surface of cells maintained in suspension was0-16 + O-O2 (s.E.) for 38 specimens, compared with o-io± o-oi (32) and 0-27 + 0-03 (29)at the free and ' adherent' surfaces respectively, of the cells maintained in monolayer.

CIH-binding to microvilli 219

DISCUSSION

Previous electrokinetic studies on EAT cells have revealed that although neur-aminidase- and ribonuclease-susceptible anionic groups together only account for36 % of their net surface negativity, removal of both species of groups results in a97% reduction in particle binding. The particles, therefore, are predominantlymarkers for the substrates of these enzymes at the surfaces of EAT cells. One explana-tion for this specificity, which has been discussed in detail, is that these particularcharged moieties occur in higher than average, surface charge density, so that eachparticle covers a cluster of very approximately 20-30 anionic groups (Weiss & Zeigel,1972).

The reaction with CIH is made at low pH and low ionic strength; the latter to pre-vent agglutination of the particles. Therefore, to avoid cytolysis, the cells were fixedbeforehand. Aldehydes cause cross-linking or coagulation of membrane proteins and'stabilization' of lipids (Zeiger, i960). They react with positively charged moietiesassociated with amino-, imino-, guanidino- and/or other surface groups (Weiss, Bello& Cudney, 1968) which constitute approximately 5 % of the ionogenic species at thesurfaces of the EAT cells used here (Weiss, 19736). Direct evidence in favour ofaldehyde-induced conformational changes in membrane proteins comes from studieson the circular dichroism of fixed erythrocyte membranes, which indicate partial lossof helicity; however, glutaraldehyde was the least traumatic of commonly usedfixatives studied (Lenard & Singer, 1968), and this prompted its use by us. Although thesurfaces of glutaraldehyde-fixed cells bear an undefined relationship to living ones, weconsider that our present comparisons of particle-binding densities within the 2experimental groups are valid since they have been made on different regions of cellsin the same sections, and comparison between experimental groups were made on cellsfrom the same parent cultures, treated with the same fixative. In our own work, thepresent results obtained using cells fixed in 0-3 % glutaraldehyde are quantitativelyvery close to those obtained on cells fixed with 3 % glutaraldehyde (Weiss, Jung &Zeigel, 1972). In addition, similar results to analogous work of our own on glutaralde-hyde-fixed erythrocytes (Weiss, Zeigel, Jung & Bross, 1972) have been obtained bylabelling with cationized ferritin under physiological conditions of ionic strength andpH, without prior fixation (Danon, Goldstein, Marikovsky & Skutelsky, 1972). Aftertreatment of erythrocytes with neuraminidase and fixation with glutaraldehyde, weobserved a 60 % loss in surface negativity associated with an 89 % decrease in the den-sity of bound CIH particles, whereas Danon et al. observed an 86-5 % decrease in thedensity of cationized ferritin particles bound to unfixed cells.

The results given in Table 1 indicate a higher density of particles bound to the sur-face of microvilli, than to the IMV-spaces for control and ribonuclease-treated groups.This 93 % increase cannot be accounted for in terms of the different sectional geo-metries of the microvilli and the IMV-spaces, since these considerations were takeninto account in formulating the correction factor (0-423).

After incubation with ribonuclease, most particle-binding is due to surface sialicacids, and following treatment with neuraminidase, most particle binding is due to

220 L. Weiss and J. R. Subjeck

surface groups associated with ribonuclease-susceptible sites (Weiss, Jung & Zeigel,1972). Therefore by comparing particle binding to microvilli and IMV-spaces beforeand after incubation with enzymes, some idea of the relative densities of the hypo-thetical charge clusters associated with these groups may be obtained.

The ratios given in Table 2 for cells grown both in suspension and in monolayercultures show that after ribonuclease treatment, the proportion of bound particles overthe microvilli and the IMV-spaces is approximately the same as in the controls. Adifference in distribution of ribonuclease-susceptible sites between these 2 regions istherefore not responsible for the observed increase in particle density over themicrovilli.

Following neuraminidase treatment, the particle density on the microvilli is reducedto approximately the same value as that of the IMV-spaces for both suspension andmonolayer cells (suspension, 0-4 > P > 0-5; monolayer, 07 > P > o-8). The higherdensity of bound particles over the microvilli is therefore due to higher density ofsuitable neuraminidase-susceptible sialic acids in these regions of the cell peripherythan in the spaces between them. There are at least two ' explanations' of this finding.Either microvilli are formed in pre-existing cell surface regions of high particle-bindingdensity or alternatively, such regions are concentrated on pre-existing microvilli.Our present data do not permit discrimination between these non-exclusive hypotheses.

The turnover time of proteoglycans in the cell periphery may be rapid (Warren &Glick, 1968), and in some cells the surface densities of ribonuclease-susceptible groupsmay increase with metabolism, which in turn may be modified by adhesion to solidsubstrata (Stoker, O'Neill, Berryman & Waxman, 1968). In addition to these examplesof physicochemical lability which may be associated with the related parameters ofadhesion and metabolism, morphological changes in the surfaces of adherent cells arealso well recognized. For example, Follett & Goldman (1970) have shown that whentrypsin-treated BHK21/C13 cells adhere to glass from suspension, there is a 70—80%decrease in the density of microvilli over their whole surfaces. It was therefore ofinterest to examine the surfaces of the EAT cells in the present experiments for dif-ferences associated with the change from suspension to monolayer culture. The countsof microvilli given earlier show that although in going from suspension to monolayerculture, there is a 44 % reduction in microvillus density at the free surface of the adher-ent cells, there is an increase of 75 % at their adherent undersurfaces. Preparationsof the type used here do not in themselves permit us to determine whether the densityof microvilli per whole cell changes on adhesion, or whether the increased densitieson their undersurfaces represent a redistribution or a de novo formation of these proces-ses in these regions. However, although fewer microvilli appear at the free surfaces ofthe adherent cells compared with those in suspension, both the numbers of adsorbedparticles (Table 1) and the ratios of particles adsorbed to the microvilli and the spacesbetween them (Table 2) remain substantially the same in the two situations. Thisfinding suggests that material at the surfaces of the ' disappearing' microvilli is notsimply redistributed over the whole free surface of the adherent cells.

Electrokinetic studies have shown that approximately 65 % of the negative surfacecharges are not associated with either ribonuclease- or neuraminidase-susceptible

CIH-binding to microvilli 221

sites. We cannot therefore extrapolate from the present observations on particle-binding, which occurs mainly to the 35 % of anionic groups sensitive to these enzymes,to total charge density at the microvilli and IMV-spaces. However, if the ratio of totalsurface charge density to ribonuclease- and neuraminidase-susceptible ionogenicgroups is the same for the 2 regions, then the surface charge density of the microvilliwill be approximately twice that of the IMV-spaces. Such regional differences in sur-face charge density on electrostatic considerations alone, would be expected to modifycontact interactions between microvilli and the charged regions of other cells.

We wish to thank Dr Robert Zeigel, Dr James Harlos and Ms Ohki Jung for their advice ondifferent aspects of this work. We also thank Mr R. Heinaman, Mr M. Reszka and Mr D. Huberfor valuable technical assistance.

This work was partially supported by Grant No. BC-87E from the American Cancer Society,Institutional Research, Grant IN-54 from the American Cancer Society, and National Institutesof Health Grant 5 Toi GM00718.

REFERENCES

DANON, D., GOLDSTEIN, L., MARIKOVSKY, Y. & SKUTELSKY, E. (1972). Use of cationized ferritinas a label of negative charges on cell surfaces. J. Ultrastruct. Res. 38, 500-510.

FOLLETT, E. A. C. & GOLDMAN, R. D. (1970). The occurrence of microvilli during spreadingand growth of BHK21/C13 fibroblasts. Expl Cell Res. 59, 124-136.

GASIC, G. J., BERWICK, L. & SORRENTINO, M. (1968). Positive and negative colloidal iron as cellsurface electron stains. Lab. Invest. 18, 63-71.

LENARD, J. & SINGER, S. J. (1968). Alteration of the conformation of proteins in red blood cellmembranes and in solution by fixatives used in electron microscopy. J. Cell Biol. 37, 117-121.

MOORE, G. E., SANDBERG, A. A. & ULRICH, K. (1966). Suspension cell culture and in vivo andin vitro chromosome constitution of mouse leukaemia L1210. J. natn. Cancer Inst. 36, 405-421.

STOKER, M., O'NEILL, C, BERRYMAN, S. & WAXMAN, V. (1968). Anchorage and growth regula-tion in normal and virus-transformed cells. Int. J. Cancer 3, 683-693.

WARREN, L. & GLICK, M. C. (1968). Membranes of animal cells. II. The metabolism andturnover of the surface membrane. J. Cell Biol. yj, 729—746.

WEISS, L. (1973 a). Neuraminidase, sialic acids and cell interactions. J. natn. Cancer Inst. 50,3-20.

WEISS, L. (19736). Studies on cellular adhesion in tissue culture. XIV. Positively chargedsurface groups and the rate of cell adhesion. Expl Cell Res. (submitted).

WEISS, L., BELLO, J. & CUDNEY, T. L. (1968). Positively charged groups at cell surfaces. Int.J.Cancer 3, 795-808.

WEISS, L. & HARLOS, J. P. (1972). Short-term interactions between cell surfaces. In Progress inSurface Science, vol. 1 (ed. S. G. Davison), pp. 355-405. Oxford: Pergamon.

WEISS, L., JUNG, O. S. & ZEIGEL, R. (1972). The topography of some anionic sites at the surfacesof fixed Ehrlich ascites tumor cells. Int. J. Cancer 9, 48-56.

WEISS, L. & MAYHEW, E. (1969). Ribonuclease-susceptible charged groups at the surface ofEhrlich ascites tumor cells. Int. J. Cancer 4, 626-635.

WEISS, L. & ZEIGEL, R. (1972). Heterogeneity of anionic sites at the electrokinetic surfaces offixed Ehrlich ascites tumor cells. J. theor. Biol. 34, 21-27.

WEISS, L., ZEIGEL, R., JUNG, O. S. & BROSS, I. D. J. (1972). Binding of positively chargedparticles to glutaraldehyde-fixed human erythrocytes. Expl Cell Res. 70, 57-64.

ZEIGER, K. (i960). Probleme der Fixation in Licht- und Elektronmikroskopie. 4th Int. Conf.Electron Microsc, Berlin, 1958, vol. 2 (ed. W. Bargmann), p. 17. Berlin: Springer Verlag.

{Received 31 May 1973)

2 2 2 L. Weiss and J. R. Subjeck

: « *

*

. - » , * *#

r

2 A 1

B

Fig. 2. Electron micrographs of segments of surfaces of cells in monolayers; controls (A)and after treatment with ribonuclease (B) and neuraminidase (c). x 66000.

CIH-binding to microvilli 223

3 A

•*' V-4

B

' .V.:

* • * !

Fig. 3. Electron micrographs of segments of surfaces of cells in suspension; controls(A) and after treatment with ribonuclease (B) and neuraminidase (c). x 66000.