J. Cell Sci. 48, 291-300 (1981) 2gi

Printed in Great Britain © Company nf Biologists Limited ig8i

ANALYSIS OF SKIN FIBROBLAST

AGGREGATION IN DUCHENNE

MUSCULAR DYSTROPHY

G. E. JONESDepartment of Biology, Queen Elizabeth College, Campden Hill Road, LondonWS 7AH, U.K.

AND J. A. WITKOWSKIJerry Lewis Muscle Research Centre (Department of Paediatrics), HammersmithHospital, London W12 oHS, U.K.

SUMMARY

Skin fibroblasts from patients with Duchenne muscular dystrophy have a low intercellularadhesiveness compared with normal cells when aggregated in a Couette viscometer (collisonefficiencies of 2-52 and 4-62, respectively). The pattern of aggregation was quantitated usinga digitizer system to measure the areas of particles (single cells and aggregates) formed after20 min aggregation. This size analysis showed that the majority of dystrophic cells remainedunaggregated but that a small number of very large aggregates was always formed. Normalcell suspensions only rarely contained large aggregates but contained many intermediate-sizeaggregates. These differences in intercellular adhesiveness and aggregate pattern formationindicate that there may be an alteration in the surface of dystrophic cells.

INTRODUCTION

Duchenne muscular dystrophy (DMD) is an X-linked disorder characterized bya progressive loss of muscle function accompanied by irreversible muscle degenerationleading to death of affected males in early adulthood (Rowland & Layzer, 1979).Although skeletal muscle is the primary tissue affected, there are also changes incardiac muscle in the later stages of the disease, and there is a small degree of intellec-tual impairment (Dubowitz, 1979). The pathogenesis of this disorder is unknownbut in recent years much research has been carried out exploring the possibility thatDMD is the result of an abnormality in the cell membrane (Rowland, 1976, 1980).For example, the very high serum level of muscle-specific creatine kinaae in patientswith DMD is believed to be the result of selective leakage of enzyme through thesarcolemma (Munsat, Baloh, Pearson & Fowler, 1973). A number of membraneabnormalities have been described in erythrocytes from patients with Duchennemuscular dystrophy, but many of these reports are contradictory (Rowland, 1980).For example, there have been reports of abnormal phospholipid composition inDuchenne erythrocytes, but other laboratories have found normal values and thedifferences may be due to autoxidation of phospholipids during membrane isolationprocedures (Plishker & Appel, 1980). Studies of DMD cells in culture have also been

292 G. E. Jones andj. A. Witkowski

controversial but it is clear that DMD muscle cells grown in culture do not showthe degenerative changes characteristic of muscle in vivo (Witkowski, 1977). Abnor-malities reflecting altered cell surface behaviour have been described in DMDlymphocytes that show reduced capping (Pickard et al. 1978) and in the behaviourof muscle cells prepared by enzymic dissociation of dystrophic muscle (Thompsonet al. 1977). The former finding is contentious (Stern, Kahan & Dubowitz, 1979)and the 'clustering' phenomenon described by Thompson et al. and believed to bea unique feature of dystrophic cells has now been described for cells other than frompatients with DMD (Ecob-Johnston & Brown, 1980).

We have been studying surface-mediated behaviour of dystrophic cells by measur-ing intercellular adhesiveness using the Couette viscometer (Curtis, 1969) and wehave recently shown that skin fibroblasts from patients with DMD are generally lessadhesive than normal cells (Jones & Witkowski, 1979). In the course of these experi-ments we observed that there was a small subpopulation of cells in the dystrophicpreparations that appeared to be very adhesive and formed a small number of largeaggregates although the majority of dystrophic cells did not aggregate.

In this communication we wish to report our further studies on the aggregationof normal and dystrophic cells, and in particular to describe in detail the anomolouspattern of aggregation of dystrophic cells.

MATERIALS AND METHODS

Cell cultureCultures of skin fibroblasts were established using a standard explant method from skin

biopsies taken from patients with DMD or from normal volunteers. Samples were age-matched as far as possible, although our previous results indicated that cell aggregation wasnot related to donor age (Jones & Witkowski, 1979). Cultures were established in Eagle'sminimal essential medium (MEM) supplemented with 10% foetal calf serum with 50 i.u. ofpenicillin and 50 /tg of streptomycin per ml. Cells were then grown routinely in the samemedium except that it was supplemented with 10% newborn calf serum and subcultureswere performed weekly. Cultures were used for experiments between passages 8 and 20, andwere matched for time following subculture and for degree of confluency. All tissue culturematerials were purchased from Gibco-Biocult (Paisley, Scotland).

Cell aggregation

Confluent monolayers of cells were washed once with 10 ml of calcium- and magnesium-freeHanks' saline, and then incubated at room temperature in 0-25 % trypsin in Ca'+- andMg!+-free saline for 20 min. The cells were suspended by gentle trituration, centrifuged at150 g and washed once in CaJ+-, Mgt+-free saline before resuspending at a concentration of05 x io" cells per ml in MEM containing 5 % (w/v) Ficoll 400 (Pharmacia). Ficoll is used toprevent sedimentation of cells during the course of an assay and has no effect on the measuredrates of aggregation of fibroblasts when compared with measurements made of aggregationin Eagle's Basal Medium (BME) supplemented with 10% foetal calf serum (see Results).The mean cell volume was obtained using a Coulter Counter (model ZBl) equipped with aCoulter Channalyser (model 1000) and the volume fractions of the suspension occupied bythe cells calculated from this and the cell concentration. Routine observations of suspensionswere performed to ensure that the single cell count was above 90 % of the total particleconcentration. The other 10% was invariably made up of doublets and triplets. Cell viability

Dystrophic cell aggregation 293

was assessed using trypan blue dye exclusion and found to be better than 85 % in all experi-ments. All aggregation experiments were carried out at 37 °C using Couette viscometers setat a shear rate of 10 s"1.

Determination of collision efficiency, a

Reaggregation of cells was performed under the conditions described above. Samples weretaken from the viscometers at 5-min intervals over a period of 25 min and total particle countsmade using a haemocytometer. The collision efficiency was calculated from these countsusing methods described elsewhere (Curtis, 1969; Jones, 1977).

Aggregate size distribution

Suspensions of single cells were aggregated for 20 min in Couette viscometers under thestandard conditions. At the end of that time, a IO-/J1 aliquot of cell suspension was taken,diluted into 1 ml of MEM and brought down on to a Metricel GA-i membrane filter of 5-/wnpore size (Gelman). The filter was attached to a clean glass slide and the cells fixed with 95 %ethanol for 30 min. The cells were then stained by a modified Papanicolou procedure, andafter clearing, the filter was mounted as a permanent preparation.

Fig. 1. Representative tracings of particles after 20 min aggregation. Cells wereprepared for area measurements as described in Materials and methods, and theirperimeters traced on the surface of the digitizer tablet. Scale bar, 50 /tm. (A, FANA,normal); (B, MUL, DMD). There are larger numbers of small and large particles inthe dystrophic preparation than in the normal.

Measurements of aggregate area were made using a Reichert MOP Digiplan. The specimenwas examined with a x 25 objective and the perimeters of the aggregates traced on thedigitizer surface using a drawing tube attachment projecting over the digitizer. Areas werecalculated directly in /im1. Two types of analysis were performed. In the first, all particles inrandom fields were measured regardless of whether the particles were single cells or aggregatesof any size; in the second analysis, measurements of aggregates only were made, single cellsbeing excluded. A total of 8 slides from 4 normal subjects and 8 slides from 4 patients withDMD were analysed. Examples of tracings made using this system are shown in Fig. 1.

Statistical analyses

Collison efficiencies of normal and dystrophic cells were compared using a nested analysisof variance (Sokal & Rohlf, 1969). Aggregate sizes were expressed as cumulative frequencies

294 G. E. Jones andj. A. Witkowski

(class intervals of 50 /im1 for randomly measured particles, and 1000 fim* for aggregatemeasurements) and normal and dystrophic frequencies compared with the Kolmogorov-Smirnov one-sample test (Siegel, 1956).

RESULTS

Aggregation kinetics

Table 1 gives the data from 18 further determinations of the intercellular adhesive-ness of normal and dystrophic skin fibroblasts and the pattern of aggregates formedis illustrated in Fig. 2. The results of these determinations performed in serum-freemedium agree very closely with our previous results using serum-supplementedmedium (Jones & Witkowski, 1979). These results suggest that serum componentsare not involved in the short-term aggregation we have measured, and that theinclusion of Ficoll 400 does not measurably alter the aggregation kinetics comparedwith those in serum-supplemented medium. The change in medium from BMEto MEM has had no effect on aggregation.

Table 1. Collision efficiencies of normal and dystrophic skin fibroblasts

Subject

ControlREDRAJFANA

DMDOSWIL

Age(y)

2

4

57

n

453

33

Mean

4-984-60446

2482-55

S.E.M.

O-27O'lO

cvoo

0-28o-io

Aggregate size distribution

A characteristic feature of suspensions of dystrophic cells after 20 min aggregationwas the presence of a small number of very large aggregates amongst a population ofsingle cells and the absence of intermediate si2e aggregates. Examination of dystrophiccell suspensions showed that these large aggregates were absent from the initialsuspensions and that they gradually appeared over the course of an aggregation assay.Although large aggregates were very occasionally seen in normal cell suspensions, theywere present in all dystrophic cell suspensions and contained very many cells (Fig. 3).Further analysis of the pattern of aggregation was carried out by measuring the areasof particles (single cells and aggregates) in cell suspensions. The areas of 50 normal'and 49 dystrophic single cells were measured on filters and the results show clearlythat the 2 types of cell had very similar areas (Table 2) when measured in this way.Measurement of cell volumes using the Coulter counter showed that the sizes ofnormal and dystrophic cells were very similar, and together with the present dataindicate that preparing cells for area measurements has no differential effect onnormal and dystrophic cells.

Dystrophic cell aggregation 295

2A

J JB

«#

Fig. 2. Human skin fibroblaats after ao min aggregation in a Couette viscometer.A, normal cell suspension with aggregates of 2 and 3 cells; B, DMD cell suspensioncontaining only single cells. Differential interference contrast, x 1000.

Fig. 3. A very large aggregate present in a suspension of DMD cells after 20 minaggregation. Differential interference contrast, x 1000.

In one series of experiments, area measurements were made on all particles foundin randomly distributed fields of duplicate filter preparations made from 4 normaland 4 dystrophic subjects, giving totals of 417 and 418 measurements, respectively(Fig. 4). For the purposes of analysis, the areas were expressed as a frequency distribu-tion with a class interval of 50 /tm2, and Fig. 4 shows the cumulative percentagefrequency of these measurements. It is clear from the curves that at 20 min the

296 G. E. Jones andj. A. Witkowski

Table 2. Area measurements (/on2) of normal and dystrophic single cells

Subject Sample Mean S.D.

FANA (control)

WIL (DMD)

AB

AB

2525

2425

350-2836280

36703600

71-616550

79-079-0

100 __

80

c

I

60

40

20

• •n..n--0"0-0-0"0"°"9-g"

200 600 1000 1400 2000 2400 2800 3200

Particle area, j im!

Fig. 4. Analysis of aggregate size at 20 min. Cumulative percentage distribution ofparticle areas (single cells and aggregates) measured on filter preparations using adigitizer. Normal cells (broken line), 4 subjects, measurements of 417 particles;dystrophic cells (continuous line), 4 subjects, measurements of 418 particles. Ordinatecumulative percentage frequency; abscissa: particle area, /ims.

dystrophic suspensions contain larger proportions of single cells (areas < 500 /tmJ)and very large aggregates than do the normal cell suspensions. Only 6 normal aggre-gates had areas greater than 3200 /im2 but there were 14 dystrophic aggregates largerthan this value. When the distributions were compared using the Kolmogorov-Smirnov test, they were significantly different (P < o-oi).

When aggregates alone were measured, there was a clear difference in the distribu-tion of aggregate size in cell suspensions from 3 normal and 3 dystrophic subjects(Table 3). After 20 min aggregation, approximately 91 % of normal aggregates hadareas of less than 3000 /im2, while only 53 % of aggregates in the dystrophic populationwere in that size range. Conversely there were more very large aggregates in thedystrophic suspensions: 4% of dystrophic aggregates were larger than 7000/fma

compared with 1 % of normal aggregates.

Dystrophic cell aggregation 297

Table 3. Cumulative percentage distribution of aggregate size

Subjects

Normal (3)DMD (3)

n

108154

< 10

242

1-9

1-3

Aggregate size (Jim* xA

3-5 5-7

909 973 98253-2 804 91-4

n = number of aggregates 1

7-9

99.1959

neasured

10-')

9-11

991973

n-13

100

993 1 0 0

DISCUSSION

Our present results confirm that dystrophic skin fibroblasts have a significantlyreduced intercellular adhesiveness determined by measurements of collision efficiencywhen compared with normal cells. The mean values of a for normal and dystrophiccells in the present study were 4-69 (S.E.M. = o n ; n = 12) and 2-52 (S.E.M. = 0-14;n = 6) respectively, compared with previous values of 4-53 (S.E.M. = 0-20; n = 24)and 2-53 (S.E.M. = 0-21; n = 17) (Jones & Witkowski, 1979). Pooled values of a were4-66 for normal and 2-52 for dystrophic cells, and an analysis of variance showed thisdifference to be significant at P = 0-07.

Aggregation kinetics have not been altered by changes in the composition of themedium used in the aggregation assays, and in particular by the omission of serum.Fibronectin is a component of serum that has been implicated in intercellularadhesion (Yamada, Yamada & Pastan, 1976; Zetter, Chen & Buchanan, 1976;Mautner & Hynes, 1977). If this is the case, from our results it would be necessaryto argue that residual fibronectin still bound to the cells after trypsinization wassufficient for short-term aggregation and/or that fibronectin is rapidly replaced froman intracellular store following trypsin treatment. It may also be that fibronectin isnot involved in cell-cell adhesion of fibroblasts, at least in the short-term assaysconducted here.

In carrying out this type of aggregation analysis, we would have preferred obtaininga frequency distribution of aggregate size by counting the numbers of single cells,doublets, triplets, etc., but this was not possible. The number of aggregates wassmall and it was not practically possible to accumulate sufficient numbers of the largeraggregates to make statistical analysis possible. It was also not possible to count theexact numbers of cells present in aggregates where they contained more than about5 cells. Using a digitizing system to measure the areas of cells and aggregates broughtdown on to filters enabled us to examine satisfactory numbers of cells and we believethe method will be of general application. It should be possible to detect distinct sizegroupings corresponding to the areas of single cells, doublets, triplets, etc., and whenthe data were plotted as a non-cumulative frequency distribution with class intervalof 100 fim2, it was possible to distinguish peaks at 350, 700, 1200 and 1600 /tm2

(Fig. 5). Measurements of single cells only gave a value of 350 fim2 and these peaksare approximate multiples of this value. The number of aggregates with areas greaterthan ]6oo//ms were too low to recognize individual peaks.

298 G. E. Jones andj. A. Witkowski

The cumulative frequency curves show very clearly that the dystrophic suspensionsat 20 min contain many more particles of area less than 400 /im2 (corresponding tosingle cells) than normal cell suspensions (56-2% dystrophic, 38% normal) (Fig. 4)and are what would be expected on the basis of the low values of collision efficiencydetermined for dystrophic cells. The reduced initial slope of the normal curveindicates that although normal aggregate suspensions contain fewer single cells thandystrophic suspensions, they contain larger numbers of particles in the range that

200-,

20-

2001 1 1 \ 1 1 1

600 1000 1400 1800

Fig. 5. Distribution frequency of particle areas to illustrate presence of single cellsand aggregates of 2, 3 and 4 cells (arrows). All values for normal and dystrophicparticle areas after 20 min aggregation have been pooled and log frequency (ordinate)plotted against area (abscissa, class interval 100 /tm!).

corresponds to aggregates of 2, 3 and 4 cells and smaller numbers of larger aggregatesso that only 1-7% of the normal population have areas greater than 3000 /*m! com-pared with 4-1 % of dystrophic particles.

In an attempt to characterize further this population of large dy9trophic aggregates,measurements of the areas of aggregates only were made, but these data must betreated with caution. In the course of making the measurements, it became clear thata bias was present towards measuring large aggregates in that small aggregates weremore likely to be ignored. Nevertheless, the data are valuable for comparative purposesbut should not be taken to be truly representative of the distribution of aggregatesizes. Although the numbers of large aggregates measured were small, the results

Dystropkic cell aggregation 299

indicate that dystrophic cell suspensions after 20 min aggregation contain more largeaggregates than do normal suspensions. For example, for areas greater than 7000 /tm2,there were only 2 normal aggregates but 13 dystrophic aggregates. It could be arguedthat dystrophic skin fibroblasts contain a small subpopulation of highly adhesive cellsbut this is difficult to explain in relation to the disease. As an X-linked disorder, allcells from a patient with Duchenne muscular dystrophy will be carrying the dystrophicgene and would be expected to behave similarly. A more acceptable explanation mightbe that all dystrophic cells have a common low adhesiveness but when an occasionalstable adhesion is made the resulting aggregate will increase its collision radius in thecell suspension; the rate at which the aggregate will collide with single cells will thusincrease. While the rate of formation of stable cell-to-cell adhesions remains low therewill be an increase in the recruitment of single cells into the aggregate because of thegreater number of successful collisions being made by the aggregate with other cells.

We can discount the possibility that the large aggregates are artefacts of heavytrypsin treatment as described by Whur, Koppel, Urquhart & Williams (1977) sinceinitial cell suspensions contain over 90% of single cells and appearance of largeaggregates does not occur by 5 min of incubation. Moreover, non-dystrophic cellsuspensions do not exhibit this phenomenon. This is not to say that we can ignorethe possibility that cells may differ in their susceptibility to trypsin attack. We areexploring this and other models to account for the behaviour of the dystrophic cells.

It is difficult to assess the significance of these results in relation to the pathogenesisof Duchenne muscular dystrophy. Intercellular aggregation is tacitly assumed to beclosely related to the strength of cell adhesions which may be modified by manyfactors associated with the structure and functions of the cell surface. Our resultsshowing altered intercellular adhesiveness of dystrophic cells for whatever causesupport the view that there may be some structural alteration in the plasma membraneof dystrophic cells, but how such a change might bring about a selective degenerationof muscle fibres is unknown.

We are grateful to the Muscular Dystrophy Group of Great Britain (J.A.W.) and theCentral Research Fund of the University of London (G.E. J.) for support, to Ms J. Carter fortechnical assistance, and to the Department of Human Metabolism, University CollegeHospital, for the use of the Reichert MOP Digiplan.

REFERENCES

CURTIS, A. S. G. (1969). The measurement of cell adhesiveness by an absolute method.J. Embryol. exp. Morph. 22, 305-325.

DUBOWITZ, V. (1979). Involvement of the nervous system in muscular dystrophies in man.Ann. N.Y. Acad. Set. 317, 431-439.

ECOB-JOHNSTON, M. & BROWN, A. (1980). Cluster formation is not a characteristic of Duchennemuscle in dissociated culture. In Proc. VHIth Syinp. on Current Research in MuscularDystrophy and Allied Neuromuscular Diseases, p. 21. London: Muscular Dystrophy Groupof Great Britain.

JONES, G. E. (1977). Cell disposition and adhesiveness in the developing chick neural retina.J. Embryol. exp. Morph. 40, 253-258.

JONES, G. E. & WITKOWSKI, J. A. (1979). Reduced adhesiveness between skin fibroblastsfrom patients with Duchenne muscular dystrophy. J. neurol. Sci. 43, 465-470.

300 G. E. Jones andj. A. Witkowski

MAUTNER, V. & HYNES, R. O. (1977). Surface distribution of LETS protein in relation to thecytoskeleton of normal and transformed cells. J. Cell Biol. 75, 743-768.

MUNSAT, T. L., BALOH, R., PEARSON, C. M. & FOWLER, W. (1973). Serum enzyme alterationsin neuromuscular disorders. J. Am. med. Ass. 226, 1536-1543.

PICKARD, N. A., GRUEMER, H. D., VERHILL, H. L., ISAACS, E. R., ROBINOW, M., NANCE, W. E.MYERS, E. C. & GOLDSMITH, B. (1978). Systemic membrane defect in the proximal musculardystrophies. New Engl. J. Med. 299, 841-846.

PLISHKEF, G. A. & APPEL, S. H. (1980). Red blood cell alterations in muscular dystrophy: therole of lipids. Muscle and Nerve 3, 70-81.

ROWLAND, L. P. (1976). Pathogenesis of muscular dystrophies. Arch. Neurol. 23, 315-321.ROWLAND, L. P. (1980). Biochemistry of muscle membranes in Duchenne muscular dystrophy.

Muscle and Nerve 3, 3-20.ROWLAND, L. P. & LAYZER, R. B. (1979). X-linked muscular dystrophies. In Handbook of

Clinical Neurology, vol. 40 (ed. P. J. Vinken & G. W. Bruyn), p. 349. Amsterdam: North-Holland.

SIEGEL, S. (1956). Non-parametric Statistics for the Behavioral Sciences. New York: McGraw-Hill.

SOKAL, R. R. & ROHLF, F. J. (1969). Biometry. San Francisco: Freeman.STERN, C. M. M., KAHAN, M. C. & DUBOWITZ, V. (1979). Lymphocyte capping in Duchenne

muscular dystrophy. Lancet i, 1300.THOMPSON, E. J., YASIN, R., BEERS, G. V., NURSE, K. & AL-ANI, S. (1977). Myogenic defect

in human muscular dystrophy. Nature, Lond. 268, 241-242.WHUR, P., KOPPEL, H., URQUHART, C. & WILLIAMS, D. C. (1977). Quantitative electronic

analysis of normal and transformed BHK-21 fibroblast aggregation. J. Cell Set. 23, 193-209.WITKOWSKI, J. A. (1977). Diseased muscle cells in culture. Biol. Rev. 52, 431-476.YAMADA, K. M., YAMADA, S. & PASTAN, I. (1976). Cell surface protein partially restores

morphology, adhesiveness and contact inhibition of movement to transformed fibroblasts.Proc. natn. Acad. Set. U.S.A. 73, 1217-1221.

ZETTER, B. R., CHEN, L. B. & BUCHANAN, J. M. (1976). Effects of protease treatment ongrowth, morphology, adhesion and cell surface proteins of secondary chick embryo fibro-blasts. Cell. 7, 407-412.

(Received 15 July 1980)