OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 ·...

20
J. Cell Sci. 18, 385-403 (i975) 385 Printed in Great Britain OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE M. S. STEINBERG AND D. R. GARROD* Department of Biology, Princeton University, Princeton, N.jf.oSs^o, U.S.A. SUMMARY Two problems are raised concerning the movement of cells during tissue-specific sorting-out of chick embryo cells in mixed aggregates, (i) A possible expectation from the hypothesis of 'contact inhibition' is that cells which are entirely surrounded by other cells in monolayer should be held stationary. Cells within solid aggregates, being totally surrounded by others, might also not be expected to move. How is it then that cell movement takes place within solid aggregates during sorting-out ? (ii) Are the movements of cells within sorting aggregates 'passive', being driven by adhesive differentials, or 'active', being merely guided by such differentials ? In order to study these questions, sorting-out experiments with chick embryonic limb bud mesenchyme and liver cells were carried out in monolayer culture, permitting direct observation of cell movements. Cell behaviour was observed by time-lapse cinematography. Sorting-out of these cells in monolayer began before and continued after the cells had spread to confluency. During sorting, liver cells showed ruffling activity even when they appeared to be totally surrounded by other cells. Both cell types showed contact inhibition as judged by the criterion of monolayering, for they did not move over each other but remained attached to the substratum. Yet the cells in the confluent monolayer were not immobilized. Because of this, we suggest that the observed restraint against overlapping did not result from an inhibition of movement. Several considerations, detailed in the text, suggest that cell movement during sorting-out involves active locomotion. Previous work suggests that sorting-out configurations are determined by the relative inten- sities of intercellular adhesive strengths, the more cohesive of 2 cell populations tending to adopt the internal position. While limb bud cells form internal islands surrounded by liver cells in solid aggregates, the reverse was found to be the case in these monolayers. This suggests that, in the monolayer, limb bud cohesiveness is depressed relative to liver cell cohesiveness. This is consistent with the observation that the limb bud cells flattened themselves markedly against the substratum, significantly decreasing their area of mutual apposition. INTRODUCTION This investigation was undertaken in the hope that some light would be shed on 2 problems concerning the mechanism of cell movements involved in the sorting-out process. The first problem was the following apparent paradox. When cells from 2 different vertebrate embryonic tissues are randomly intermixed in an aggregate, the cells of the 2 tissues sort out from one another (reviewed in Steinberg, 1964, 1970; Aber- • Present address: Department of Biology, The University, Southampton SO9 3TU, U.K.

Transcript of OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 ·...

Page 1: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

J. Cell Sci. 18, 385-403 (i975) 385Printed in Great Britain

OBSERVATIONS ON THE SORTING-OUT OF

EMBRYONIC CELLS IN MONOLAYER

CULTURE

M. S. STEINBERG AND D. R. GARROD*

Department of Biology, Princeton University, Princeton, N.jf.oSs^o, U.S.A.

SUMMARY

Two problems are raised concerning the movement of cells during tissue-specific sorting-outof chick embryo cells in mixed aggregates, (i) A possible expectation from the hypothesis of'contact inhibition' is that cells which are entirely surrounded by other cells in monolayershould be held stationary. Cells within solid aggregates, being totally surrounded by others,might also not be expected to move. How is it then that cell movement takes place within solidaggregates during sorting-out ? (ii) Are the movements of cells within sorting aggregates'passive', being driven by adhesive differentials, or 'active', being merely guided by suchdifferentials ?

In order to study these questions, sorting-out experiments with chick embryonic limb budmesenchyme and liver cells were carried out in monolayer culture, permitting direct observationof cell movements. Cell behaviour was observed by time-lapse cinematography. Sorting-outof these cells in monolayer began before and continued after the cells had spread to confluency.During sorting, liver cells showed ruffling activity even when they appeared to be totallysurrounded by other cells.

Both cell types showed contact inhibition as judged by the criterion of monolayering, forthey did not move over each other but remained attached to the substratum. Yet the cells in theconfluent monolayer were not immobilized. Because of this, we suggest that the observedrestraint against overlapping did not result from an inhibition of movement.

Several considerations, detailed in the text, suggest that cell movement during sorting-outinvolves active locomotion.

Previous work suggests that sorting-out configurations are determined by the relative inten-sities of intercellular adhesive strengths, the more cohesive of 2 cell populations tending toadopt the internal position. While limb bud cells form internal islands surrounded by livercells in solid aggregates, the reverse was found to be the case in these monolayers. This suggeststhat, in the monolayer, limb bud cohesiveness is depressed relative to liver cell cohesiveness.This is consistent with the observation that the limb bud cells flattened themselves markedlyagainst the substratum, significantly decreasing their area of mutual apposition.

INTRODUCTION

This investigation was undertaken in the hope that some light would be shed on2 problems concerning the mechanism of cell movements involved in the sorting-outprocess.

The first problem was the following apparent paradox. When cells from 2 differentvertebrate embryonic tissues are randomly intermixed in an aggregate, the cells ofthe 2 tissues sort out from one another (reviewed in Steinberg, 1964, 1970; Aber-

• Present address: Department of Biology, The University, Southampton SO9 3TU, U.K.

Page 2: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

386 M. S. Steinberg and D. R. Garrod

crombie, 1964; Trinkaus, 1965, 1967; Curtis, 1967). This process involves the move-ment of cells over considerable distances within the aggregate. On the other hand,normal tissue cells in monolayer cultures are usually stopped in their forward motionwhen they run into other cells. In the words of Abercrombie, Heaysman & Karthauser(1957), 'Movement of a cell in the direction of a contact tends to be prevented, aphenomenon termed "contact inhibition"'.

It has been argued that a logical consequence of contact inhibition is 'to hold afibroblast at a standstill when it is so closely surrounded by other fibroblasts that theonly surfaces available for its movement are those of its neighbours' (Abercrombie etal. 1957). Moreover, Weston & Abercrombie (1967) point out that 'An extrapolationof the hypothesis of "contact inhibition" of cell movement, from solid substrates invitro to intact tissues, suggests . . . that . . . random movement [of cells within intacttissues] would be slight. The movement of each cell within a tissue should be inhibitedby its neighbours.' The question of whether individual cells move within non-sortingtissue fragments was investigated experimentally by Weston & Abercrombie (1967).Using chick embryonic heart and liver fragments, they found that cells rarely movedacross the boundaries between fused heart or liver fragments. They concluded that'individual cells are probably not freely motile within intact tissues'. Nevertheless,cells undeniably do move within solid aggregates during the sorting-out process, sothat we are faced with the problem of how to reconcile the cell displacements duringsorting-out in aggregates with the inhibition of cell movements by mutual contact inmonolayer cultures.

The second problem is that of whether cell movements involved in sorting-out are'active', requiring intrinsic cellular motile activity, or 'passive'. Previous work hasled to the tentative conclusion that the sorting-out of embryonic chick cells is guidedby adhesive differentials, motility of cells being a requisite condition for rearrangementto occur (see Steinberg, 1964, 1970). The cause of this motility is of no consequence ifone is concerned only with identifying the conditions necessary for cell sorting. How-ever, adhesive differentials are, in principle, able to do more than give direction toactive movements that would take place in an unguided way in a uniform environment.Such differentials are also able to produce forces that could cause cell movements suchas those seen when cells sort out from one another. Carter (1967a, b) has pointed thisout and has coined the term 'haptotaxis' to refer to cell movements caused in thismanner. Movements propelled exclusively by interfacial forces of adhesion could beregarded as passive, requiring no special motile activity on the part of the cells.

Unfortunately, one cannot at present look inside a 3-dimensional cell aggregate inorder to observe directly the motile activity of individual cells. However, this problemmight be overcome if cell sorting experiments could be conducted in a 2-dimensionalsituation. Firstly, it might be possible to see whether cells from different tissue originscombined in mixed monolayer would be immobilized and prevented from sorting-out,a result which might be predicted from the hypothesis of contact inhibition. Secondly,direct observations of individual and small-group cell behaviour could be made whichmight make possible a distinction between a mechanism for cell rearrangement drivenby adhesive differentials and one dependent upon intrinsic cellular motile mechanisms.

Page 3: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

Sorting-out in monolayer 387

We have begun our approach to these questions by culturing chick embryonic 7-dayliver cells and 4-day limb bud mesoblast cells together on Falcon plastic surfaces, andobserving their motile behaviour (Garrod & Steinberg, 1973). It is known that in3-dimensional aggregates, intermixed limb bud mesoblast and liver cells sort out, thelimb bud cell population taking up the internal position (Moscona, 1957; Steinberg,1963, 1964, 1970).

MATERIALS AND METHODS

Liver cells were obtained from 7-day White Leghorn embryos as follows. Livers were dis-sected from the embryos in modified Hanks' Balanced Salt Solution* (HBSS) at 23 °C. Thelarger of the 2 liver lobes was separated fiom the smaller which was discarded. The large lobeswere then washed three or four times with modified calcium- and magnesium-free (CMF)Hanks' solutionf and minced into small fragments with iris knives. Forelimb buds were cutfrom 3p75- to 4-day embryos in HBSS. These were placed in 1 % crude trypsin (Difco 1-250)in HBSS and incubated at 4 °C for 60-90 min in order to loosen the epidermal outer layerfrom the mesenchymal cores (Szabo, 1955). Next the epidermis was removed from the limbbuds in HBSS at 23 CC, and the mesenchymal cores were stored in fresh HBSS, again at23 °C, until the dissociation procedure was begun.

For dissociation, the tissues were placed in o-i % crude trypsin (Difco 1:250) in CMF Hanks'solution at 37 °C, fragments from a maximum of 12 livers or 20 limb buds being added to10 ml of trypsin solution. Incubation in this solution was for 15-20 min at 37 °C in rollertubes. These tubes were placed on a test-tube rotator (Multipurpose Rotator, model 150V,Scientific Industries Inc., Springfield, Mass.) which revolved at 60 rev/min while holding thetubes at a constant angle of about 10—150 to the horizontal. Following incubation, the cells andtissue fragments were centrifuged down at about 400 g for 4 min, and the trypsin solution wasremoved and replaced with Eagle's Minimum Essential Medium (Grand Island BiologicalCompany) supplemented with 10 % horse serum plus penicillin, streptomycin and Fungizone(Amphotericin B) at respective concentrations of 100 units/ml, 100/Jg/ml and 025/tg/ml.Dissociation and resuspension of the cells was achieved by shearing the mixture for 15 s usinga Vortex mixer. The cells were again centrifuged down, this time at about 200 g for 3 min,and the medium replaced. After resuspension using the Vortex mixer, the suspension wascentrifuged at about 40 g for 1 min in order to remove any large, undissociated tissue fragments.

Suspensions of dissociated liver and limb bud cells were mixed to give a ratio by numberof 3:1 limb bud:liver, except where otherwise specified. The mixed suspension was put into35-mm tissue culture dishes (Falcon Plastics) at a concentration of approximately 8 x io9 cellsper dish, each dish containing 2 ml of medium. Dishes were incubated at 37 °C in a moistatmosphere of 5 % CO2-95 % air until required for time-lapse filming. The medium wasalways renewed immediately before filming.

Filming was carried out with a Nikon Model M inverted phase-contrast microscope equippedwith a Bolex 16-mm movie camera. The 10 x or 20 x objective was focused on the cells throughthe bottom of the culture dish. A Perspex chamber resting on the microscope stage, with ahole cut in the top to accommodate the condenser snugly, was gassed with 5 % COj-gs % airmixture. The filming temperature, 37 °C, was maintained by a thermostatically controlled hairdryer which discharged its effluent into a polyethylene bag completely surrounding the micro-scope stage.

All films were made with Kodak Plus-X negative film using an exposure of 0-5 s. The filmingintervals were between 6 and 20 s, and will be given where appropriate for specific films.

• The modified Hanks' Balanced Salt Solution used in these experiments contained, ingrammes per litre: NaCl, 8-oo; KC1, 040 ; Na,HPO4, 0 0 8 ; KH2PO4, 0 0 5 ; dextrose, 100;NaHCO,, 020 ; CaCl2.2H2O, 019 ; M g S O ^ H j O , 010 ; MgCl, .6H a 0, 010. The pH wasadjusted to 7-4.

f CMF Hanks' Solution was the same as HBSS except that the calcium and magnesiumsalts were omitted and 0-20 g/1. of phenol red was added as a pH indicator.

Page 4: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

388 M. S. Steinberg and D. R. Garrod

For histological preparations, monolayers were fixed in Bouin's or Zenker's fluid and thensoaked in 70 % ethanol for at least 48 h. They were then removed from the plastic surface withchloroform, as described by House & Stoker (1966). The ethanol treatment following fixationprevented the monolayers from curling. They were then either embedded for sectioning ormounted on slides. The mounting procedure was as follows. After removal from the plasticwith chloroform, the monolayers were washed with several changes of fresh chloroform. Theywere then transferred through 50:50 chloroform: ethanol (20 min) and absolute ethanol (20min) to 90 % ethanol. The monolayers were then allowed to settle on the surfaces of coverslipsin 90 % ethanol. Next, the coverslips were removed from the 90 % ethanol, the monolayersbeing held in place with a needle, and allowed to dry overnight at room temperature. Thefollowing day the preparations were stained with haematoxylin and eosin and mounted on slides.

RESULTS

Demonstration of sorting-out in 2 dimensions

When mixed cultures of liver and limb bud cells, prepared as described above, wereexamined after 48 h of incubation, 3 things were apparent (Fig. 3): (i) the cells hadspread to cover the bottom of the culture dish completely; (ii) the liver parenchymacells were grouped together in islands, just as occurs when they are cultured alone ata subconfluent density (see Garrod & Steinberg, 1975); and (iii) the space betweenthe liver islands was completely filled with limb bud cells. (Fibroblastic liver cellswere probably intermixed with the limb bud cells and could not be distinguished fromthem, but since a ratio of 3:1 limb bud: liver was normally used, the great majority ofcells between the liver islands would have been limb bud cells.)

These results were interesting in 2 respects. First, they showed that sorting-outhad taken place in 2 dimensions, though as yet they revealed nothing concerning themechanism by which this sorting-out had occurred. Secondly, it was evident that in2 dimensions liver cells took up the 'internal' position, forming a discontinuous phaseof liver islands surrounded by a continuous phase of limb bud cells. This is the reverseof the configuration taken up by these tissues in 3-dimensional aggregates, wherelimb bud adopts the internal position (Steinberg, 1964, 1970).

Further observations showed that the cells spread to form a confluent monolayerover the bottom of the culture dish between 10 and 20 h after the beginning of incuba-tion. It was therefore necessary to determine whether the observed sorting-out wascompleted before the monolayer became confluent or whether the process continuedafter confluency was achieved. Identical cultures were prepared and incubated for16, 24 and 48 h, after which time they were fixed. The monolayers were then mountedon coverslips and stained. These preparations were used to count the number of cellsper liver group, and the number of liver groups per given area, at the times given.

The area chosen was the field given by a 16 x /o-4O Zeiss Neofluar objective. It wasfound that the number of cells per liver group increased progressively with time, from16 through 24 to 48 h, and that the number of such groups per field decreased corre-spondingly (Table 1). There were 1-59 times as many cells per liver group at 48 asat 24 h, and 1-57 times as many groups per field at 24 as at 48 h. This experiment wasrepeated with similar results. Thus it seemed evident that sorting-out in 2 dimensionscontinued after the monolayer had become confluent, and that this was brought about

Page 5: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

Sorting-out in monolayer 389

by the fusion of small liver groups to form larger ones. The results of a time-lapsestudy of some aspects of this behaviour are presented in the next section.

Next, an experiment was performed in order to determine whether the formation ofa discontinuous phase by liver rather than limb bud cells was due to the use of a highlimb bud:liver ratio. Limb bud and liver cells were mixed in culture in the followingratios (limb bud:liver): 1:0-33, I : I> I:I'S> I : 3 ar>d 1:6-5. At all ratios it was foundthat the liver parenchyma cells tended to form islands surrounded by limb bud cells(Figs. 3-6). That is, junctions between liver and limb bud tended to be concavetowards the liver. It was thus evident that regardless of their relative numbers, in2 dimensions on a Falcon tissue culture surface, liver cells preferentially sort outinternally to limb bud cells.

Table 1. Results of liver cell counts

AverageAverage

number of livernumber of liver

• Ratios: 48f Ratios: 48

groups per field*cells per groupf

:24 = 1:1-57:48::24 = 1 59:1; 48:

16 =16 =

16

3918

2-45:1

h

00 0

0

[ ;

24:24:

1616

2 4

3528

= 1 :

= r;

h

•2

•8

1 1 3 .

48

22

45

h

•5

9

Examination of transverse sections of mixed monolayers showed a marked differ-ence in the degree of flattening of the 2 cell types. Whereas limb bud cells were extremelyflattened upon the substratum, liver cells were cuboidal. The average ratio of theheight of liver cells to the height of limb bud cells was 2-7:1. Under these conditions,therefore, the area of mutual apposition between liver cells was substantially greaterthan that between limb bud cells.

Time-lapse films of sorting-out in 2 dimensions

The sorting-out of liver and limb bud cells in 2 dimensions was recorded in 2 time-lapse films. Film 1, with a frame interval of 15 s, covered a period from about 4 h toabout 22 h after the beginning of culture. Film 2 began at about 23 h and continuedfor 16 h 40 min with an interval of 20 s. Thus film 1 covered the culture period pre-ceding and immediately following the attainment of confluency and showed theaggregation of liver cells into islands surrounded by a confluent limb bud monolayer(Figs. 7-12). Throughout film 2 the tissue culture surface was completely covered bya confluent monolayer of cells. The film showed the fusion of two liver islands whichwere separated by limb bud cells at the beginning of the film (Figs. 13, 14).

Four hours after setting up the culture (Fig. 7), considerable spreading of theindividual fibroblastic limb bud cells had taken place. Also there was already somecollection of liver cells into islands, although many liver cells were still single. Singleliver cells were brightly refractile and approximately circular. Through Figs. 7-10the movements of 7 liver cells have been traced; these are marked a to g in Fig. 7.Fig. 8 shows that cells c, d and e came together to form a small group, e having relin-quished its attachment to the larger liver group at the right-hand side of the picture.

Page 6: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

39° M. S. Steinberg and D. R. Garrod

Cells/and £ were also in contact. Sixty-two and a half minutes later (Fig. 9) there were2 cell groups - X, consisting of cells a to e, and Y, consisting of cells f and g - whichwere joined by pseudopodia. (At this stage, group Y may have contained more than2 cells but, if so, these could not be traced during repeated examinations of the film.)Groups X and Y then coalesced to form a single group (Fig. 10). Contact between

Fig. 1. Tracings from a time-lapse film showing changes in shape and fusion of liverislands within a confluent liver—limb bud monolayer. The interval between successivepictures is approximately 5-5 h. A and D are equivalent to Figs. 12 and 13, respectively,c shows the formation of contact between the 2 islands on the right-hand side. Thearea of mutual contact between them was progressively increased throughout theremainder of the film.

the 2 groups was formed first by pseudopodia, which then seemed to draw the groupstogether. Finally, this group of at least 7 liver cells joined the larger liver island at theright-hand side of the picture. Comparison of Figs. 11 and 12 shows that when thesmaller group joined the larger, the appearance of the cells in the smaller groupchanged: they became less refractile and their nuclei became visible. During theirmovement to join larger liver groups, the liver cells formed contacts with limb budcells from time to time. Such contacts appeared to be easily broken, unlike liver-livercontacts which, in general, were not broken once they had formed. (The contacts

Page 7: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

Sorting-out in monolayer 391

between limb bud cells could not be resolved well enough for accurate determinationof their behaviour.)

Ten hours after the beginning of culture (Fig. 11), the cells had spread almost toconfluency and the pattern of cell distribution had changed strikingly from that 6 hearlier (Fig. 7). The liver cells had collected into islands which were surrounded bythe limb bud monolayer. The remainder of the film showed the liver islands graduallyadopting more circular configurations (compare islands/) and q in Figs. 11 and 12),thus increasing liver-liver contacts at the expense of liver-limb bud contacts. Duringthis phase gaps occasionally appeared, separating cells at the edges of liver islandsfrom those of the surrounding limb bud monolayer. Liver cells extended pseudopodiainto these gaps.

Figs. 13 and 14 show that 2 liver islands, each initially surrounded by limb budcells, joined to form a single island during the course of film 2. The outlines of 3 liverislands, including the 2 which fused, were traced from the film at intervals of approxi-mately 5 h and 33 min and are shown in Fig. 1 A-D. From these tracings it is evidentthat the liver islands underwent changes of shape, even though they were surroundedby limb bud cells. The 2 islands on the right fused first over a small area of their edges;then the area of contact between them increased.

In this film, as in the previous one, gaps appeared at the junctions between limbbud and liver cells, into which the latter extended pseudopodia. A brief period ofpseudopodal activity by cells at the lower edge of the upper liver island in the figurepreceded the establishment of contact between this island and the lower island withwhich it fused.

In both films cells within liver groups showed ruffling activity even though theyappeared to be entirely surrounded by other cells. Ruffles passed over the nuclei ofliver cells, suggesting that ruffling occurred on the cells' upper surfaces. Also, ruffleswere observed on liver cells at the edges of liver islands, where they were in contactwith liver cells on one side and limb bud cells on the other.

DISCUSSION

The role of intrinsic motile activity in cell sorting

Of the various possible outcomes of these experiments, that one was realized whichcould in principle shed light upon how intermixed cells move in the process of sorting-out. Liver and limb bud cells did sort out from one another in 2 dimensions on aFalcon tissue culture plastic surface. The sorting process began before confluencywas achieved, but continued afterwards. We had expected, from previous observationsof Abercrombie & Ambrose (1958) among others, that ruffling activity of the cellmembranes would be paralysed at each point where cells made contact with oneanother, and that consequently all ruffling would cease when full confluency wasachieved. The exact role of ruffling activity in cell translocation is unknown (Aber-crombie, 1961), but the association of translational movement with the presence of aruffling membrane at the forward end of the cell has led to the widespread belief thatruffling is at least a symptom of intrinsic cellular motile activity capable of producing

Page 8: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

392 M. S. Steinberg and D. R. Garrod

forward motion, if not actually the 'cause' of the forward motion itself. Had rufflingindeed been suppressed at confluency, weight would have been added to the possi-bility that the sorting-out which continued to occur beyond that point was directlypropelled by the exchange of weaker adhesions for stronger ones. Contrary to ourexpectation, liver cells continued to ruffle after confluency was achieved. This rufflingwas observed both on the surfaces of liver cells surrounded partly by liver and partlyby limb bud cells, and on the surfaces of liver cells entirely surrounded by other livercells. We were also surprised to observe that the ruffles passed over the entire (pre-sumably upper) surfaces of these cells. Thus, while we succeeded in our goal ofbringing about cell sorting in 2 dimensions under conditions of full confluency, wefailed in our primary purpose of inhibiting, through these conditions, the action ofintrinsic cellular motile mechanisms in this thermodynamically unstable cell mixture.Consequently, our results shed less light than we had hoped upon the mechanism bywhich the cell translocations of sorting-out are propelled.

The observations reported here do not reveal whether the motive force for sorting-out is capable of being provided directly and exclusively by cellular adhesive differ-entials that favour de-mixing, or whether intrinsic cellular motile activity is requiredin addition to provide a selection of new contacts, which are then maintained orreplaced on the basis of their relative thermodynamic stabilities.

Other recent studies, however, indicate that interfacial forces and intrinsic cellularmotile activity both contribute to the cell displacements accompanying cell sorting.Steinberg & Wiseman (1972) have reported that cytochalasin B (CB), at concentrationsthat prevented cell locomotion on a plastic substratum, reversibly inhibited thesorting-out of heart and liver cells in mixed aggregates, suggesting that active cellmovements are required for cell sorting in this instance. Armstrong & Parenti (1972)obtained the same results in mixed heart-pigmented retina aggregates, but found thatsorting-out of neural retina and pigmented retina cells did proceed in the presence ofconcentrations of CB that completely abolished ruffled membrane activity in mono-layer cultures. In control aggregates, the pigmented retina cells came together toform many internal cell islands, which fused with one another repeatedly until asingle, internal mass of pigmented retinal tissue had been reconstituted, envelopedby an outer shell of neural retinal tissue. In the presence of CB, sorting-out ceasedat the multiple-internal-island stage. This suggested that active cell movementsrequired for these pigmented retinal islands to 'find' one another were indeed in-hibited, and that the sorting-out that had proceeded up to that time had been drivenby the interfacial forces of adhesion that in the other cell combinations could do nomore than guide the progress of such rearrangements. Steinberg & Wiseman haveintroduced the term ' cooperative cell locomotion' to describe circumstances in whichinterfacial forces act together with forces of internal origin in causing cell locomotion.

What the observations reported here do clearly establish is that the sorting-outwhich continued past confluency did involve the coming together and fusion of liverislands originally separated (at least to the eye of the observer) by limb bud cells.While certain encounters between tissue islands can occur as a consequence of shapechanges accompanying the rounding-up of originally irregular tissue boundaries, not

Page 9: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

Sorting-out in monolayer 393

all of the form-changes observed gave the appearance of being explainable in this way.One gains the impression that active movements of tissue islands were taking placealso. However, it must be recognized that cells sometimes extend long processeswhich may escape detection under the conditions of our phase- contrast observations,and that cellular extensions may in fact connect tissue islands which appear separateto the eye. The extension (and contraction) of such cellular processes would of coursequalify as one form of intrinsic cell motility, but the ' zippering up' of an original smallcontact mediated by such an extension would not. In the latter case, the first stagesin the rounding-up of a single barbell-shaped tissue island would appear to the eyeas the drawing together of a pair of separate islands.

Contact inliibition, contact paralysis and cell translocation

The observations discussed above raise the question of whether or not the cells inthese experiments have displayed contact inhibition. The term contact inhibition (ofmovement) was introduced by Abercrombie & Heaysman (1954) to denote thehypothesis that when a cell moving upon a substratum collides with another cell, itscontinued locomotion in the direction of the point of contact will usually tend to beinhibited. One of the consequences of contact inhibition would be that cells exhibitingit would tend not to move over each other'ssurfaces. Instead, they would tend to arrangethemselves to form a monolayer. We had anticipated that in mixed cultures of limbbud and liver cells, one type of cell might exclude the other from the plastic surfaceto form a bilayer rather than a monolayer; or that a mixed monolayer might form,but that translational movement of cells within it might cease, once confluency wasachieved. Neither of these possibilities was realized. Instead, the cells from bothtissues cooperated in forming a confluent monolayer all of whose members evidentlytended to maintain contact with the solid substratum while proceeding to sort out in2 dimensions. While we cannot be certain that cells from either source never movedover other cells of whatever kind, in general this did not seem to happen. Histologicalexamination of transverse sections of monolayers confirmed the impression gainedfrom phase-contrast observations that all of the cells appeared to be in contact withthe culture substratum.

Because the cells in these mixed cultures did not in general move over one another,it must be concluded that they have displayed contact inhibition. (The degree of cellor nuclear overlap is used as a quantitative measure of the degree of contact inhibition;see reviews in Abercrombie, 1970; Martz & Steinberg, 1973). How does this con-clusion accord with the observations that they continued both to ruffle and to trans-locate after confluency was achieved ? Careful consideration reveals that in fact thereis nothing whatever about the hypothesis of contact inhibition that requires thecessation either of ruffling or of cellular translocation at confluency. While cessationof ruffling at the point of cell-cell contact is commonly observed in cell populationsdisplaying contact inhibition, other cell populations have been observed to show 'anextremely strong tendency to monolayer, but . . . no paralysis of membrane' (seeDiscussion on pp. 270—271 in Abercrombie, 1967). Contact inhibition does not implycontact paralysis. Moreover, even if it were to be actually observed that in the fibro-

Page 10: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

394 M. S. Steinberg and I). R. Garrod

blast populations discussed by Abercrombie et al. (1957) one of the effects of contactinhibition ' is to hold a fibroblast at a standstill when it is so closely surrounded byother fibroblasts that the only surfaces available for its movements are those of itsneighbors' {op. cit. p. 276), such immobilization does not follow as a logically necessaryconsequence of a prohibition against overlapping. Cells in some other confluentpopulation could steadfastly refuse to cross over one another's surface, while simul-taneously milling about and exchanging positions with one another upon the substra-tum. In the present experiments, conditions favouring contact inhibition were designedto coexist with conditions favouring the translational cell movements of sorting-out,and the cells displayed both modalities of population behaviour simultaneously. Theindividual behaviour of liver cells in a pure, confluent monolayer is reported in thefollowing paper (Garrod & Steinberg, 1975), while that of 3T3 cells is reported else-where (Martz & Steinberg, 1974).

Culture morphology and adhesive differentials

Finally, we wish to address ourselves to the reversal of liver and limb bud phasesas seen in 2 as compared with 3 dimensions. When cells from these 2 organs sort outfrom one another in 3-dimensional aggregates, limb bud (chondrogenic) cells forman internal population of discontinuous, coalescing islands within a continuous massof liver cells (Steinberg, 1964, 1970). We were therefore surprised to find that in ourmixed monolayers not the limb bud but the liver cells formed the discontinuous phase.Much evidence has previously been presented to indicate that in 3-dimensional,2-phase, spontaneously sorting cell aggregates, as in 3-dimensional, 2-phase bodies ofimmiscible liquids, it is the more cohesive population that tends to take up a moreinternal position (Steinberg, 1958, pp. 74-75; 1962a-*:, 1963, 1970; Phillips, 1969;Phillips & Steinberg, 1969). The same principles should operate in 2-dimensionalsystems (see, for example, Gocletal. 1970; Gordon, Goel, Steinberg & Wiseman, 1972,Appendix A). This 'phase reversal' seemingly implies that limb bud mesoblast cellsare more cohesive than liver cells in aggregates, but less cohesive than the latter whenplated out on a Falcon plastic surface. Has the monolayer environment so quickly(less than 4 h) caused a change in the chemistry of the cell surfaces ? Or do individualcells have differentially adhesive surface regions which orient in a selective way onthe plastic, exposing only selected and specialized surface areas for lateral, cell-to-celladhesions ? While such events are in no way excluded, measurements of the relativeheights of liver and limb bud areas in the sorted-out monolayers offer a simpler andmore attractive explanation.

In the interior of a 3-dimensional aggregate, the whole of each cell's surface isaccessible for adhesion with other cells. This is not the case in a monolayer, becausepart of each cell's surface adheres to the substratum and another part remains exposedto the culture medium. Cells in a monolayer adhere to one another only in a regionforming a belt around the periphery of each cell. Furthermore, the breadth of thisbelt, and therefore the area of mutual apposition, depends upon the degree of flatteningof the particular cell on to the substratum. Cells of different kinds often flatten todifferent extents. Examination of cross-sections of monolayers showed in the present

Page 11: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

Sorting-out in monolayer 395

case that limb bud cells were more flattened on to the substratum than liver cells were:the height of liver cells averaged 2-7 times that of limb bud cells. Thus limb bud cellshad a substantially smaller area of mutual surface apposition than did liver cells(Fig. 2). The absolute strength of an adhesion between 2 surfaces depends upon (a) itsstrength per unit area, and (b) the area of contact between the two surfaces. Conse-quently, even though adhesions between limb bud cells may have had a greater strengthper unit of interfacial area than adhesions between liver cells, the substantially greaterheight of the interfaces between the latter may well have been sufficient to turn thisbalance and render liver cells the greater in absolute cohesiveness/>er unit of interfaciallength (as observed from above) in this situation.

LB

Fig. 2. Diagram to show appearance of monolayer in cross-section. The liver cells (L)are less flattened on to the substratum than the limb bud cells (LB), and have agreater area of mutual contact.

We have noted earlier that contacts among liver cells in our films were generallynot broken once they had formed, but that gaps commonly appeared between liverislands and the surrounding continuum of limb bud cells. It is interesting to considerthis observation in relation to the relative strengths of adhesion that the various kindsof cell-cell interface existing in these cultures may reasonably be deduced to possess.As has already been mentioned, much evidence exists to show that in 3-dimensionalcell aggregates containing cells of 2 types that tend to sort out, as in 2-phase liquidbodies that spontaneously tend to de-mix, it is that population whose elements coherewith the greater energy per unit of interfacial area which should tend to be envelopedby the other population. It must now be added that to favour sorting-out over inter-mixing, it is necessary that the adhesions between the 2 different kinds of elementsbe lower by a certain amount in adhesive energy per unit of interfacial area than thosebetween elements of the more cohesive population (see Steinberg, 1964, pp. 326-329).From the observation that limb bud (chondrogenic) cells sort out internally to livercells within 3-dimensional aggregates (taken together with the fact that other aspectsof the behaviour of this system point toward its control by adhesive differentials; seeSteinberg, 1962c, 1963, 1964, 1970), the deduction follows that limb bud-liver(LB-L) adhesions are weaker (per unit of interfacial area) than LB-LB adhesions.In the present experiments with monolayer cultures, the greater flattening of limbbud cells gives evidence of having diminished their area of mutual contact sufficientlyto bring their absolute cohesiveness (i.e. their energy of cohesion per unit of interfaciallength) below that of the less-flattened liver cells. Since the greater flattening of limbbud cells reduces their area of apposition to each other and to liver cells in exactly thesame degree (see Fig. 2), the ratio of LB-LB/L-L absolute adhesiveness would beexpected to remain constant. Thus LB-L adhesions would be expected to be weakerthan LB-LB adhesions in these monolayers just as in 3-dimensional aggregates. In

25 CE L 18

Page 12: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

396 M. S. Steinberg and D. R. Garrod

short, the evidence points toward a particular ranking of adhesive strengths at thevarious interfaces among limb bud and liver cells in these monolayer cultures. Thisranking is L-L > LB-LB > L-LB. The frequent gaps appearing at liver-limb budboundaries in our films may be significant in this connexion.

We thank Dr Eric Martz for technical suggestions, and Professor Lewis Wolpert, Dr Martz,Dr Morton Burdick and Dr Herbert Phillips for suggesting improvements in the manuscript.This research was supported by grant GB 5759X from the National Science Foundation, andPublic Health Service Research Grant no. CA-13605 from the National Cancer Institute.M.S.S. was Visiting Fellow at the Salk Institute during completion of the manuscript.

REFERENCES

ABERCROMBIE, M. (1961). The bases of the locomotory behaviour of fibroblasts. Expl Cell Res.,Suppl. 8, 188-198.

ABERCROMBIE, M. (1964). Cell contacts in morphogenesis. Archs. Biol., Liege 75, 351-367.ABERCROMBIE, M. (1967). Contact inhibition: the phenomenon and its biological implications.

Natn. Cancer Inst. Monogr. 26, 249-277.ABERCROMBIE, M. (1970). Contact inhibition in tissue culture. In Vitro 6, 128-142.ABERCROMBIE, M. & AMBROSE, E. J. (1958). Interference microscope studies of cell contacts in

tissue culture. Expl Cell Res. 15, 332-345.ABERCROMBIE, M. & HEAYSMAN, J. E. M. (1954). Observations on the social behaviour of cells

in tissue culture. II. ' Monolayering' of fibroblasts. Expl Cell Res. 6, 293-306.ABERCROMBIE, M., HEAYSMAN, J. E. M. & KARTHAUSER, H. M. (1957). Social behaviour of cells

in tissue culture. III . Mutual influences of sarcoma cells and fibroblasts. Expl Cell Res. 13,276-291.

ARMSTRONG, P. B. & PARENTI, D. (1972). Cell sorting in the presence of cytochalasin B. J. CellBiol. 55, 542-553-

CARTER, S. B. (1967a). Haptotaxis and the mechanism of cell motility. Nature, Land. 213,256—260.

CARTER, S. B. (19676). Effects of cytochalasins on mammalian cells. Nature, Lond. 213, 261-264.CURTIS, A. S. G. (1967). The Cell Surface: Its Molecular Role in Morplwgenesis. London and

New York: Logos/Academic Press.GARROD, D. R. & STEINBERG, M. S. (1973). Tissue-specific sorting-out in two dimensions in

relation to contact inhibition of cell movement. Nature, Lond. 244, 568-569.GARROD, D. R. & STEINBERG, M. S. (1975). Cell locomotion within a contact-inhibited mono-

layer of chick embryonic liver parenchyma cells, jf. Cell Sci. 18, 405-425.GOEL, N., CAMPBELL, R. D., GORDON, R., ROSEN, R., MARTINEZ, H. & YCAS, M. (1970). Self-

sorting of isotropic cells. J. theor. Biol. 28, 423-468.GORDON, R., GOEL, N. S., STEINBERG, M. S. & WISEMAN, L. L. (1972). A rheological mech-

anism sufficient to explain the kinetics of cell sorting. J. theor. Biol. 37, 43-73.HOUSE, W. & STOKER, M. G. P. (1966). Structure of normal and polyoma virus-transformed

hamster cell cultures. J. Cell Sci. 1, 169-173.MARTZ, E. & STEINBERG, M. S. (1973). Contact inhibition of what? An analytical review.

J. cell. Physiol. 81, 25-37.MARTZ, E. & STEINBERG, M. S. (1974). Movement in a confluent 3T3 monolayer and the

causes of contact inhibition of overlapping. J . Cell Sci. 15, 201-216.MOSCONA, A. A. (1957). The development in vitro of chimaeric aggregates of dissociated

embryonic chick and mouse cells. Proc. natn. Acad. Sci. U.S.A. 43, 184-194.PHILLIPS, H. M. (1969). Equilibrium Measurements of Embryonic Chick Cell Adhesiveness:

Physical Formulation and Testing of the Differential Adhesion Hypothesis. Ph.D. Thesis. TheJohns Hopkins University, Baltimore, Md.

PHILLIPS, H. M. & STEINBERG, M. S. (1969). Equilibrium measurements of embryonic chickcell adhesiveness. I. Shape equilibrium in centrifugal fields. Proc. natn. Acad. Sci. U.S.A.64, 121-127.

Page 13: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

Sorting-out in monolayer 397

STEINBERG, M. S. (1958). On the chemical bonds between animal cells. A mechanism for type-specific association. Am. Nat. 92, 65-82.

STEINBERG, M. S. (1962a). On the mechanism of tissue reconstruction by dissociated cells.I. Population kinetics, differential adhesiveness, and the absence of directed migration.Proc. natn. Acad. Set. U.S.A. 48, 1577-1582.

STEINBERG, M. S. (19626). Mechanism of tissue reconstruction by dissociated cells. II. Timecourse of events. Science, N.Y. 137, 762-763.

STEINBERG, M. S. (1962c). On the mechanism of tissue reconstruction by dissociated cells.III. Free energy relationships and the reorganization of fused, heteronomic tissue fragments.Proc. natn. Acad. Set. U.S.A. 48, 1769-1776.

STEINBERG, M. S. (1963). Reconstruction of tissues by dissociated cells. Science, N.Y. 141,401-408.

STEINBERG, M. S. (1964). The problem of adhesive selectivity in cellular interactions. InCellular Membranes in Development (ed. M. Locke), pp. 321-366. New York: Academic Press.

STEINBERG, M. S. (1970). Does differential adhesion govern self-assembly processes in histo-genesis ? Equilibrium configurations and the emergence of a hierarchy among populations ofembryonic cells. J. exp.Zool. 173, 395-434.

STEINBERG, M. S. & WISEMAN, L. L. (1972). Do morphogenetic tissue rearrangements requireactive cell movements ? The reversible inhibition of cell sorting and tissue spreading bycytochalasin B. J. Cell Biol. 55, 606-615.

SZABO, G. (1955). A modification of the technique of 'skin splitting' with trypsin. J. Path.Bad. 70, 545.

TRINKAUS, J. P. (1965). Mechanisms of morphogenetic movements. In Organogenesis (ed.R. L. De Haan & H. Ursprung), pp. 55-104. New York: Holt, Rinehart and Winston.

TRINKAUS, J. P. (1967). Morphogenetic cell movements. In Major Problems in DevelopmentalBiology (ed. M. Locke), pp. 125-176. New York: Academic Press.

WESTON, J. A. & ABERCROMBIE, M. (1967). Cell mobility in fused homo- and heteronomictissue fragments. J. exp. Zool. 164, 317-324.

{Received 1 February 1975)

25-2

Page 14: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

398 M. S. Steinberg and D. R. Garrod

Figs, 3-6. Photographs of 48-h monolayer cultures which were set up using differentinitial ratios of dissociated liver and limb bud cells. The limb bud:liver ratios were1 :o#33, Fig. 3; 1: i'S, Fig. 4; 1:3, Fig. 5 and 1:6'5, Fig. 6. The inset to Fig. 6 roughlyshows the boundaries of the liver islands.

Page 15: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

Sorting-out in monolayer

Page 16: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

400 M. S. Steinberg and D. R. Garrod

Figs. 7-10. Prints from a time-lapse film showing development of a mixed monolayerculture of liver and limb bud cells. Fig. 7 was taken approximately 4 h after settingup the culture, Fig. 8 after 5 h, Fig. 9 after 6'i h and Fig. 10 after 8'2 h. Some indi-vidual liver cells have been labelled a to g in Figs. 7 and 8. These coalesced to formgroups X and Y in Fig. 9 and group XY in Fig. 10. The latter group then joined thelarger liver group on the right-hand side of the photograph (see Fig. 11).

Page 17: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

Sorting-out in monolayer 401

Page 18: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

402 M. S. Steinberg and D. R. Garrod

Figs, i i , 12. Prints from the same time-lapse film as Fig. 7-10 but taken 103 and14-5 h respectively after setting up the culture. Note the rounding up of the liverislands p and q with time.Figs. 13, 14. Prints from a time-lapse film of a mixed monolayer similar to that shownin Figs. 7-12 but beginning 23 h after setting up the culture (Fig. 13). Approximately16-5 h later (Fig. 14) 2 liver islands (arrowed) which were initially separate hadjoined to form a single island. Rounding up of the other liver islands in the field hadtaken place. (See also Fig. 1.)

Page 19: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick

Sorting-out in monolayer

Page 20: OBSERVATIONS ON THE SORTING-OUT OF EMBRYONIC CELLS IN MONOLAYER CULTURE · 2005-08-21 · Sorting-out in monolayer 387 We have begun our approach to these questions by culturing chick