effects of vitamin a on the behaviour of migratory neural...

J. Cell Sri. 57, 331-350 (1982) 331

Printed in Great Britain © Company of Biologists Limited 1982

EFFECTS OF VITAMIN A ON THE BEHAVIOUR

OF MIGRATORY NEURAL CREST CELLS

IN VITRO

PETER THOROGOOD1, LINDA SMITH1, ALASTAIR NICOL1,ROSE McGINTY1 AND DAVID GARROD1

1 Department of Biology, Medical and Biological Sciences Building, Bassett CrescentEast, Southampton SO9 3TU, U.K. and1 CRC Medical Oncology Unit, University of Southampton, Southampton GeneralHospital, Southampton SO9 $XY, U.K.

SUMMARYIt has been proposed elsewhere that the teratogenic effects of retinoids on craniofacial

morphogenesis are caused by a disturbance of the migration of cranial neural crest cells. Theeffects of 3-5 x io~6 M and 3-5 x io~* M-retinol on the migration of avian neural crest cellsin vitro have been investigated by monitoring cell morphology, locomotory behaviour, fibro-nectin distribution and actin-micronlament organization. Retinol retards migration by affectingcell-to-substratum adhesiveness. Cells exposed to medium containing retinol are less adherentto the substratum, and although the cell surface is very mobile, are unable to extend or maintainlamellipodia. As a consequence the cells do not actively translocate. Fibronectin distributionat the cell surface is sparse, possibly as a result of shedding, and actin distribution remainsdiffuse.

At the retinol molarities used all these effects are reversible. Thus cells allowed to recoverin normal medium flatten out, display lamellipodia and commence active translocation.Fibronectin becomes organized into a fibrillar array and actin micronlaments become organizedinto cables. The period needed for this recovery is directly related to the molarity of retinolduring the initial exposure; after recovery the retinol-treated cells are virtually indistinguish-able from control cells.

We propose that in vivo the effects of retinoids might be to impair cell-extracellular matrixinteraction, thus impeding a cell's ability to migrate through that matrix. Contrary to previoussuggestions, the in vivo effects are probably not in any way ' specific' to neural crest cells butare more accurately considered as ' selective', in that any cell undergoing migration would besimilarly affected.

INTRODUCTION

Vitamin A and its analogues, collectively termed the retinoids, not only have func-tional roles in a wide range of normal physiological systems but can also be potentteratogens. Their effects range from changes in the differentiation of integumentalstructures, such as scale primordia differentiating into feathers (Douailly, Hardy &Sengel, 1980), to congenital limb abnormalities such as phocomelia (Kochar, 1977).Exposure during early development has a dramatic effect on craniofacial morpho-genesis ; in the mammalian embryo, for example, exposure, in utero or in vitro, duringa critical period of development produces a cleft palate and a deficient facial skeletonarising from a disturbed pattern of skeletogenesis rather than from a failure of skeletal

332 P. Thorogood, L. Smith, A. Nicol. R. McGinty and D. Garrod

differentiation per se (e.g. see Morriss & Thorogood, 1978). In work on other verte-brate groups it has been demonstrated that the greater part of the connective and skele-tal tissues in the face are ec/omesenchymal, that is to say, derived from cranial neuralcrest cells migrating into the presumptive facial region during early head development(e.g. see Johnston, 1966; Le Lievre & Le Douarin, 1975). It has been proposed by anumber of authors that the craniofacial effect of hypervitaminosis A are due to retardedneural crest migration (Poswillo, 1975; Morriss & Thorogood, 1978). This results in adeficient facial mesenchyme and disturbs the synchrony of morphogenesis, therebydistorting the migration pathways and causing an abnormal distribution of the reducedmesenchyme.

These proposals have not been tested directly and the precise mode of action ofteratogenic retinoid levels has not yet been defined. For the avian embryo, whereparameters of migration rate, migration routes and the differentdative fate of cranialneural crest cells have been studied, two brief reports provide circumstantial evidencethat migration in vivo might be impeded by retinoids (Hassell, Greenberg & Johnston,1977; Keith, 1977). In this paper we describe the effects of one particular retinoid,retinol, on the locomotory and social behaviour of avian cranial neural crest cellsmigrating from primary explants in vitro. The experiments were designed in anattempt to answer the following questions: (i) does retinol affect the locomotoryability of neural crest cells ? (ii) if so, how is this effect implemented ? (iii) is impairmentof locomotion specific to neural crest cells as previous reports imply?

MATERIALS AND METHODS

Culture technique

Blastoderm of the Japanese quail (Coturnix coturnix japonica) at stages 9-9 + (Hamburger &Hamilton, 1951) were washed briefly in Dulbecco's phosphate-buffered saline and placed into'alpha' Eagle's minimal essential medium (otMEM) containing 10% foetal calf serum (FCS).The tips of the mesencephalic neural folds were dissected free (see fig. 3 of Thorogood, 1981),and placed as primary explants onto heat-sterilized glass coverslips in 30-mm plastic tissue-culture dishes (Sterilin) containing z ml of <xMEM supplemented with 10 % FCS, ioo unitg/mlpenicillin, 100 mg/ml streptomycin and 0-25 mg/ml Fungizone (GIBCO). Mesenchymal cellsmigrating from the explants were regarded as migratory neural crest cells; the differentiationof such cells into skeletal tissues, when grown in organ culture, has been described elsewhere(Bee & Thorogood, 1980). Trarei-retinol (Sigma) was dissolved in absolute ethanol and addedto experimental cultures to give final concentrations of i-o fig and io-o fig retinol/ml medium,that is3-5Xio~*M and 3-5 x io"5 M, respectively (dose levels of io~* M and above are thoughtto be cytotoxic to most cell types; for further details, see Lotan, 1981). Control cultures receivedan amount of ethanol (4 fiX) equal to that added to experimental cultures. Cultures were main-tained at 37 °C in s % CO, in air, in a humidified incubator. The duration of the culture periodand frequency of medium changes varied according to experiment (see below).

Analysis of cell locomotion

Each culture was assigned a code number and all measurements and photography werecarried out ' blind' by a second person. The rate of outgrowth from explants was monitoredby measuring the diameter of outgrowth at 24 h intervals; for each culture the average of twomeasurements made at 900 was recorded. The morphology of living cells was monitored usinginverted phase-contrast microscopy and recorded by either 35 mm photography using Ilford

Vitamin A and neural crest cells 333

Pan F or by time-lapse video recording (see below). Cultures were terminated by washingtwice in PBS, and then they were fixed in absolute methanol and stained by the May-Grun-wald-Giemsa technique (Paul, 1975). The locomotory and social behaviour of the cells wasmonitored by time-lapse video recording for periods up to 48 h. An Hitachi low-light intensitycamera (model HV-175K) was fitted to an Olympus inverted phase-contrast microscope andcoupled to a FOR-A video time generator (model VTG-377) and National time-lapse videorecorder (V'l'K NV-8030); the image was displayed on a Sanyo T.V. monitor (V.M. 4219).Recordings were made on Sony High Density V-62 Video tapes, at the 80 h setting on therecorder, and played back for analysis at normal speed; where necessary the field-by-fieldfacility was used for analysis. Cultures for video recording consisted of explants in 5 ml ofmedium in plastic tissue-culture flasks (Sterilin), which were gassed with 5 % COI( and theentire microscope was kept in a 37 CC environment. In all other respects flask cultures wereidentical to the dish cultures described earlier, and no differences in cell morphology and rateof outgrowth were detected between the plastic and glass (coverslip) substrates.

Immunofluorescent localization of fibronectin and actin

Fibronectin was prepared from citrated chicken plasma by affinity chromatography on agelatin-Sepharose 4B column according to Engvall & Ruoslahti (1977). Antibodies to fibro-nectin were raised in rabbits and then affinity-purified on a fibronectin-Sepharose 4B columnaccording to Chiquet, Puri & Turner (1979).

Actin for use as an antigen was prepared from chicken gizzards as follows: clean chickengizzards were minced and extracted overnight in 50 % glycerol containing 0-067 % phosphate-buffered saline (PBS) and o-oi mM-phenylmethylsulphonyl fluoride (PMSF) at 4 CC. Theywere then extracted with two 15-min washes with o-i M-KC1 to polymerize actomyosin in thetissue. Divalent cations and tropomyosin were then removed by washes with 0-05 M-NaHCOscontaining o-i mM-PMSF followed by C05 M-NaHCO, containing o-i mM-PMSF and ando-oi M-EDTA. The material was then washed several times with distilled water and then withacetone at o °C, and then blended with acetone before being air-dried overnight. Actin wasthen extracted from the acetone powder according to Spudich & Watt (1971). The actin pre-paration was then solubilized by boiling for 3 min in 1 % sodium dodecyl sulphate (SDS) and1 % /?-mercaptoethanol and electrophoresed on a preparative 15% polyacrylamide gel (L aemmli).The 46000 molecular weight band was then cut from the gel, eluted (Lazarides & Weber,1974) and used to immunize rabbits that had been screened previously to demonstrate theirfreedom from anti-actin autoantibodies.

For staining, cells were washed 3 times with PBS at room temperature, fixed in freshlyprepared 3-5 % paraformaldehyde in PBS for 1 h and then washed ino-i M-NH4C1 in PBSfor 30 min. Cells were then washed in PBS alone prior to staining. For intracellular localizationof actin the cells were made permeable by treatment with acetone (2 min in 50 % acetone at— 20 °C, 5 min in 100 % acetone at 4 °C and 2 min again in 50 %) acetone at — 20 °C and thenwashed in PBS. Cells were stained for 30 min with either anti-fibronectin or anti-actin antibody,washed thoroughly with many changes of PBS, stained with fluorescein-conjugated sheepanti-rabbit immunoglobulin G (IgG) (Wellcome), washed again and mounted in 90 % glycerol10 % PBS.

Fluorescence was observed on a Zeisa Photomicroscope II and photographs taken on HP5 orFP4 black and white film.

RESULTS

Effects on rates of outgrowth

Explant cultures of two different ages (0 h and 24 h) were exposed to retinol attwo concentrations, 3-5 x io"6 M and 35 x io~8 M, in three different series of experi-ments. In the first series, 24 h explants that had been grown in normal aMEM andalready consisted of sizeable monolayered outgrowths, were cultured in retinol-containing medium for a further 72 h (see Fig. 1). Cultures in 3-5 x io"6 M-retinol

334 P. Thorogood, L. Smith, A. Nicol, R. McGinty and D. Garrod

o

160

140

120

100

80

60

40

20

- (~JLo—-—°—c

¥r i i

24 48 72Time (h)

96

24 48Time (h)

Fig. i. Rate of outgrowth of cultures exposed to retinol 24 h after explantation.( • ) Controls; ( • ) 3-5 x io"8 M-retinol; (O) 3-5 x io~s M-retinol. Each pointrepresents the mean of between 10 and 20 individual measurements.Fig. 2. Rate of outgrowth of cultures exposed to retinol from the time of explantation.The broken line from 48 h onwards for the 3-5 x io"8 M-retinol cultures indicatesthat cell detachment occurred regularly and resulted in a subconfluent outgrowth(symbols as for Fig. 1.)Fig. 3. Recovery of outgrowth after 24 h treatment with retinol. Cultures weretreated with retinol at the time of explantation but this was replaced with normalmedium after 24 h (arrow) (symbols as for Fig. 1).

showed a rate of outgrowth approximately the same as that shown by the controls,in that over a 3-day period both types of culture attained a 160 % increase in outgrowthdiameter. Cultures exposed to the lower concentration of retinol were regularly foundto be slightly larger than those of the controls. Rate of outgrowth from explants in3-5 x io~8 M-retinol was considerably reduced; following 24 b exposure the rate of


outgrowth was approximately half that of the controls and the 3-5 x io"6 M cultures.This retardation in outgrowth rate continued and after 72 h exposure these culturesshowed only a 75% increase in outgrowth diameter compared with the 160%increase shown by the other two categories of culture. Outgrowths in the concentra-tion of retinol were at this stage difficult to measure accurately as the outgrowths werevery disorganized, with large areas denuded of cells.

In the second set of experiments explants were established at o h in mediumcontaining retinol, that is without the benefit of 24 h growth in normal aMEM. Theeffects seen over a 72 h period were similar to those observed in the previous experi-ments; there was a dramatic decrease in the rate of outgrowth in 3-5 x io~6 M-retinol,whereas cultures at the lower concentration, 3 -5x10"^ , showed an increase inoutgrowth diameter that paralleled that of control cultures (see Fig. 2). Whereasgrowth of the 3-5 x io~s M-retinol cultures began to slow down after 24 h, the size ofboth 3-5 x io"6 cultures and control cultures more than doubled in the 24 h to 72 hperiod of culture.

A third set of experiments revealed that the effects of 24 h exposure to retinol,which is sufficient to change cell morphology dramatically (see below), are reversible.Initial explantation was done in retinol-containing medium and, after 24 h, themedium was completely replaced with normal aMEM. Growth rates were unimpairedby this initial exposure to either concentration of retinol (see Fig. 3) and cell morpho-logy at 72 h was virtually identical to the controls in spite of having been abnormalduring and immediately following exposure to retinol (see next section).

Effects on cell morphology

Changes in cell morphology resulting from exposure to retinol were basicallysimilar in cultures treated at 24 h* and at o h. However, for reasons of brevity commentswill be made only on the latter. Altered cell morphology became apparent afterapproximately 10 h exposure, at which time early outgrowths in control cultureswere composed of confluent monolayers of flattened mesenchymal cells with a distinctedge to the outgrowth. Retinol-treated cultures consisted of less-flattened cells, oftenspindle-shaped, and 'criss-crossing' of cells could occasionally be detected; theedges of such cultures were usually looser and less well-defined.

These effects became more distinct with time so that at 24 h the majority of cellsin retinol cultures, at both concentrations, were either triangular, spindle-shaped orrounded, in contrast to the control outgrowths in which the multipolar, sometimesstellate, cells remained flattened and confluent (see Figs. 4A, B). Several observa-tions indicated that in retinol cultures at this stage the cells were less adherent to thesubstratum.

First, in control outgrowths criss-crossing of cells was rare, whereas in retinolcultures the incidence was high (see arrows in Fig. 4 A). Secondly, the refractile halosor edges of cells exposed to retinol, seen when viewed in phase-contrast, reinforcethe interpretation that these cells were not flattened against the substratum. These

• Cells that had grown out from the explant before exposure to retinol at 24 h of culturewere slightly more resistant to the effects of the retinoid.


effects of retinol became progressively more extreme over a 72 h period of exposure,although the effects were always more dramatic in the 3-5 x io~6 M-retinol cultures.After 72 h of exposure at this concentration most, if not all, of the cells were eitherrounded, spindle-shaped or had detached from the substratum, giving a very raggedand disorganized appearance to the outgrowth, which in some cases virtually dis-integrated into individual cells (not illustrated). At the lower concentration of retinolmany more cells remained attached to the substratum but retained the rounded orspindle-shaped morphology. Outgrowths of such cultures were much looser thancontrol outgrowths. (It was noted that there was a higher incidence of melanogenicdifferentiation than in controls, although this was not measured). Control cells, at leastat the periphery of the outgrowth, showed a greater degree of flattening by 72 h,consequently appearing much larger (cf. Figs. 4c with 6 c). This may be due to anincreased cell-substratum adhesiveness.

The transition to a relatively less adhesive, rounded or spindle-shaped morphologywas reversible. Cells that had acquired this morphology, following 24 h exposure toretinol (as seen in Fig. 4 A) recovered and displayed a phenotype virtually identicalto control cells of the same culture 'age', when placed in normal aMEM (see Figs.5 A-6B inclusive). The speed of resumption of normal morphology was related to theconcentration of retinol to which the cells were initially exposed. Outgrowths from3-5 x io~* M cultures at 48 h (i.e. a 24 h recovery period) consisted of a majority offlattened cells and relatively fewer rounded or spindle-shaped cells, when comparedwith outgrowths of 3-5 x io~6 M cultures (see Fig. 5 A). However, by 72 h (i.e. a 48 hrecovery) the majority of cells in both types of culture were flattened and relativelyfew rounded or spindle-shaped cells were seen even in 3-5 x io~5 M cultures (seeFig. 6A). Outgrowths in retinol cultures were sometimes subconfluent and this wasprobably the result of earlier detachment of some cells during or immediately afterexposure to retinol.

To test whether these effects were in any way specific to cells of neural crest origin,identical cultures were set up using explants of 7-day heart ventricle as a source offibroblasts. Although such explants were growing more slowly than neural crestexplants, the heart fibroblasts still showed the same response on exposure to retinolas neural crest cells. Over a 48 h exposure period the cells became less adherent to thesubstratum and displayed a rounded or spindle-shaped morphology. Outgrowths in

Fig. 4. Cell morphology at the edge of outgrowths 24 h after explantation in: A, 3-5 xio~* M-retinol; and B, control cultures. Arrows indicate point of cell cross-over. Phase-contrast optics. Bar, 50 /tmFig. 5. Cell morphology at the edge of outgrowths 48 h after explantation; A, 24 h in3-5 x io"5 M-retinol, containing medium followed by standard medium for 24 h;B, control cultures in standard medium throughout culture period. Phase-contrastoptics. Bar, 50 fim.Fig. 6. Cell morphology at the edge of outgrowths 72 h after explantation; A, 24 h in3-5 x io~5 M-retinol, containing medium followed by standard medium for 48 h;B, control cultures in standard medium throughout culture period. Phase-contrastoptics. Bar, 50 /tm.


control cultures were composed of cells with a typical fibroblastic morphology. Inaddition, the incidence of spontaneous rhythmic myogenic contraction of the heartexplants was reduced in a concentration-related fashion. Those cultures exposed to3-5 x io~5 M failed to exhibit any detectable contractile activity, whereas approximatelyhalf of those exposed to 3-5 x io"6 M-retinol did show some contractility; virtually allcontrol cultures displayed rhythmic contractions.

All the following results are based on ' recovery '-type experiments in which explantsat o h were exposed to retinol for 24 h and then allowed up to 72 h to recover.

Effects on cell behaviour

Time-lapse video recording was used to monitor the locomotory and contactbehaviour of outgrowing cells in control and 3-5 x io~5 M-retinol cultures, includingcultures in which an initial period of exposure to retinol for 24 h was followed by aperiod of recovery in normal aMEM. Data are presented here from recordings oftypical cell behaviour at the edge of explants, the analysis having been made at 50-60 min intervals over 3-4 h periods. In each case the culture was positioned so thatapproximately one-half to two-thirds of the monitor screen was occupied by out-growing cells. Total cell number was recorded, together with the numbers of cellsclassed as 'rounded', 'mitotic', 'dead', 'blebbing', or involved in cross-overs, all ofwhich were expressed as percentages.

Table 1. Cell behaviour in control cultures at 24 h

Time Total Rounded* Mitotic Dead Blebbing Crossovers(min + g) cells (%) (%) (%) (%)

0

57 + 43" 7 + 3177 + 5223+26

5i6 i

546 i

63

0

4 95 63 34 8

0

3 31 90

1 6

0 0 0

o 49 3-3o 5-6 74o 3-3 3'3i-6 3-2 3-2

• Rounded cells are subdivided into 'mitotic', 'dead' and 'blebbing'.•f The term cross-overs is used to denote the proportion of cells involved in over- or under-

lapping neighbouring cells.

At approximately 24 h control cultures were composed largely of flattened multi-polar cells of which a very low proportion were involved in cross-overs or were roundedand blebbing (Table 1). Cell movements tended to be slow, gradual lamellipodialextensions with occasional jerky movements of trailing processes. In contrast, out-growths that had been exposed to retinol for approximately 24 h consisted of up to40 % rounded cells, virtually all of which were alive and actively blebbing at a fastrate (Table 2) but rarely moving aross the substratum. Most of the cells not classifiedas rounded were either spindle-shaped or triangular, as found in previous experiments.Continuous exposure in retinol for a further 24 h - that is, 48 h exposure in all - ledto an increase in the proportion of rounded, blebbing cells and in the incidence of cellscrossed over each other (Table 3). The surface activity of the cells was intense;


Table 2. Cell behaviour in 3-5 x io~5 M-retinol cultures at 24 h

Time(min + s)

0

59 + 4919 + 1279 + 5342 + 45

• Rounded

Totalcells

5241465048

cells adherent

Rounded(%)

26-92 6 82 6 140-02 9 2

Mitotic(%)

0

0

0

0

0

to upper surface of other

Dead(%)

0

2-42-26 0O

cells not

Blebbing(%)

2 6 924-42 3 9

34'°2 9 2

Crossovers(%)*

3 97 34-44-08-3

counted as cross-overs.

Table 3. Cell behaviour in 3-5 x io~5 M-retinol culturesafter 48 h continuous exposure

Time Total Rounded Mitotic Dead Blebbing Cross-overs(min + s) cells (%) (%) (%) (%) (%)

0

59+40119 + 17179 +II

77*757877

3 9 0493487649

00

00

2 65'32 6i - 3

36-444-O46-2636

1 3 0

5 35'2

12-0

# Virtually all cells not classified as rounded are either spindle-shaped or triangular.


Time Total Rounded Mitotic Dead Blebbing Cross-overs(min + s) cells (%)

0

58 + 8118 + 13178 + 9

25*242729

0

0

0

10-3+

0000

0 0 0

o 0 00 0 0

35 o °• Fewer cells occupy an equivalent proportion of the field to that found in the 24 h analysis,

because at 48 h the cells are more flattened and therefore appear larger.•f Two out of these three cells rounded up prior to differentiating as neurones.

blebs were put out and retracted continuously and small transient ruffles appeared,but the cells were unable to extend or maintain any lamellipodia-like extensions.Control cells at approximately 48 h were, by contrast, much more flattened andbecause they formed a confluent monolayer, far fewer cells occupied the field of view(compare left-hand columns in Tables 1 and 4). There were virtually no roundedcells and cell cross-overs were not detected. All movements were slow and gradual bylamellipodial extensions from cells into cell-free spaces, either in an outward directionor, occasionally, backwards should space open up 'behind' a cell. In those culturesin which 24 h exposure to retinol had been followed by 24 h in normal aMEM apartial reversion to normal cell behaviour could be detected (Table 5). A largeproportion of the non-rounded cells were flattened and moving like the control cells.


Table 5. Cell behaviour in 3-5 x io~5 M-retinol cultures after 24 hexposure followed by a 24 A recovery period


0

59 + 57120 + 6179 + 16239 + 10

34#

3839364 i

8'85'37 78-37-3

0

0

2 60

0

2 92 62 62-8

2-4

5-92 62 6S-64 9

u - 85-30

S-64 9

• In all fields analysed, virtually all of the cells not classified as rounded were flattened andspread.



0

59 + 52H9±37181+7241+8

18191718

19

5-60

0

0

0

S-60

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

Table 7. Cell behaviour in 3-5 x io~5 M-retinol cultures after25 h exposure followed fry a 48 A recovery period

Time Total Rounded Mitotic Dead Blebbing Cross-overs(min + s) cells (%) (%)

o 21 o o o 4'8* o59 + 35 23 o o o 4-3 o114 + 40 24 o o o 4-2 o117 + 12 23 o o o 4 - 3 0237 + 55 22 O O O O O

• These percentages for blebbing cells represent one cell that blebbed during most of theperiod analysed.

Up to 9 % of cells remained round and a proportion of these were actively blebbing.Cross-overs still occurred, possibly as a result of incomplete recovery of normal cell-substratum adhesiveness.

In control cultures at 72 h there was little net movement of the cells, which weretypically very flattened against the substratum and consequently appeared large.Blebbing and cross-overs were rarely, if ever, seen and only limited ruffling wasobserved (Table 6). Cultures of equivalent age, but in which cells had been exposed toretinol for the first 24 h and then allowed 48 h to recover, presented a similar appear-ance (Table 7). Most peripheral cells were flattened, and rounded or blebbing cellswere very rare; the incidence of cross-overs was similarly very low. Thus the effects


of retinol noted after 24 h exposure (Table 2) and after partial recovery (Table 5)were no longer apparent.

Effects offibronectin distribution

In control cultures at 24 h the cells of the original explant fragment were enmeshedin a positively staining skein of fibronectin and this remained so throughout theperiod of culture (Fig. 7 A). However, most of the central cells within the outgrowthwere negative (not illustrated) and the peripheral cells stained weakly, with a faintpunctate distribution (Fig. JB). Positive staining of cells at the growing edge of controlcultures was stronger at 48 h and a fibrillar pattern was sometimes seen along withthe punctate type; more central cells still remained negative (Figs 8A, B). At 72 h theperipheral cells, by now considerably flattened and spread, stained positively, revealinga well developed fibrillar distribution offibronectin (Fig. 8 c). A fine punctate patternwas still seen on smaller (possibly less flattened) cells and a small proportion ofperipheral cells were regularly found to be negative. More central cells within theoutgrowth at 72 h still remained largely negative, although occasionally individualcells that were strongly positive for fibronectin were found. The rounded and fusiformcell morphologies in cultures exposed to retinol were associated with a very differentfibronectin distribution. Rounded cells were found to be negative for fibronectinwhether situated centrally or peripherally. Spindle-shaped and relatively unflattenedcells were largely negative, apart from a few small spots or stitches of positive stainingalong the edges of the cells. At 24 h these effects were seen clearly in the 3-5 x io~5 M-retinol cultures (Fig. 7c, D) and were also present, although in a less acute form, inthe 3-5 x io"6 M cultures.

If such cultures, resulting from initial exposure to retinol for 24 h, were placed intonormal aMEM, the fibronectin distribution began to recover and resemble that seenin controls, just as cell morphology and behaviour had done in earlier experiments.Cultures initially exposed to 3-5 x io~6 M-retinol for 24 h, and then given 24 h torecover, had a fibronectin distribution indistinguishable from controls at 48 h (notillustrated) and at 72 h (Fig. 8F). Cultures exposed to the higher concentration ofretinol took longer to recover and this parallels previous results on morphology. How-ever, by 72 h - that is, after 48 h recovery and when many of the cells had begun toflatten - punctate and fibrillar patterns of positive staining were seen, associatedparticularly with the peripheral cells (Fig. 8D, F). The intensity and pattern of stainingwas approximately the same as that seen in 48 h controls (cf. Fig. 8 B, D). One differencefrom controls was that more central cells within the outgrowth stained with faintpunctate patterns than did central cells in control cultures. This may be related to theearlier observation that control outgrowths are generally confluent, whereas out-growths in ' recovered' retinol cultures may often be subconfluent due to cell detach-ment during exposure.

Effects on actin distribution

In 24 h control cultures central cells had a diffuse actin distribution but in peri-pheral cells, while most of the actin was diffuse, some was organized into slender

342 P. Thorogood, L. Smith, A. Nicol, R. McGinty, and D. Garrod

Fig. 7. Immunofluorescent staining of surface fibronectin on cells in 24 h cultures,stained first with rabbit-anti-fibronectin followed by fluorescein-conjugated sheepanti-rabbit. A, Central cells in control cultures; B, peripheral cells in control cultures;C, central cells in 3-5 x IO"5 M-retinol cultures; D, peripheral cells in 3*5 x io~* M-retinol cultures. Fluorescence microscopy. Bar, 50 fim.

fibrils, which partly or entirely traversed the cell (Fig. 9 A, B). By 48 h most of theperipheral cells had flattened further and a distinct fibrillar array of actin could beseen clearly in some cells (Fig. 9c). The organization of actin in central cells at thisstage (Fig. 9D) was not unlike that seen in peripheral cells 24 h earlier (Fig. 9 A). Theprogressive flattening of the cells with time produced a typical 'cable' pattern of


Fig. 8. Immunofluorescent staining of surface fibronectin on cells in cultures initiallyexposed to retinol for 24 h and then allowed a further 48 h to recover in normalmedium. Staining as for Fig. 7. A, Central cells in control cultures at 48 h; B, peri-pheral cells in control cultures at 48 h; c, peripheral cells in control cultures at 72 h;D, peripheral cells in cultures initially exposed to 3-5 x io~° M-retinol for 24 hfollowed by 48 h recovery; E, as for D, but central cells; F, as for D but initial exposureto 3-5 x io - ' M. Fluorescence microscopy. Bar 50 /im.

344 P- Thorogood, L. Smith, A. Nicol, R. McGinty and D. Garrod

Vitamin A and neural crest cells

Fig. io. Immunofluorescent staining of cytoplasmic actin in cells of cultures exposedto 3-5 x io~° M-retinol for 24 h, and of cultures identically treated and allowed 48 hto recover. Staining as for Fig. 9. A, Peripheral cells at 24 h; B, central cells at 24 h;C, peripheral cells at 72 h, i.e. after 48 h recovery; D, as for c, but central cells.Fluorescence microscopy. Bar 50 fim.

Fig. 9. Immunofluorescent staining of cytoplasmic actin in cells of control culturesat various times. Cells permeabilized with acetone and stained first with rabbitanti-actin followed by fluorescein-conjugated sheep anti-rabbit antibodies. A, Peri-pheral cells at 24 h; B, central cells at 24 h; c, peripheral cells at 48 h; D, central cellsat 48 h; E, peripheral cells at 72 h; F, central cells at 72 h. Fluorescence microscopy.Bar, 50 /im.

12 CEL 57


actin staining in peripheral cells (Fig. O,E), although the less-spread central cells stillretained a largely diffuse actin distribution (Fig. O,F).

Exposure to retinol, causing the rounded or fusiform cell morphology, resulted inparallel changes in actin distribution. Such cells after 24 h exposure to either con-centration of retinol had a diffuse actin distribution and even in peripheral cells nofibrillar organization could be seen (cf. Figs, IOA, 9A). However, a period of recoveryin normal aMEM, to produce a normal cell morphology as described earlier, resultedin the actin becoming organized progressively into cables in synchrony with cellflattening. The rate of recovery was again dependent upon the concentration of retinolto which the cells had been exposed; 3-5 x IO~8M cultures were virtually indistin-guishable from controls after 24 h recovery, whereas the majority of cells in 3-5 x IO~5M

cultures began to show a cable pattern of actin staining only at 72 h, that is after48 h recovery (Fig. 10 c, D).

DISCUSSION

The locomotory abilities of cranial neural crest cells in vitro are clearly impairedby exposure to retinol at non-lethal levels. The rate of cellular outgrowth fromexplanted fragments of premigratory neural crest cells is reduced and time-lapsestudies show individual cell movement to be retarded. Cells in control cultures movein a smooth, gradual fashion effected principally by lamellipodia and with occasionalrapid movements resulting from the retraction of trailing processes, whereascells exposed to retinol move far less, extend and retract blebs and small rufflescontinuously, and appear unable to extend or maintain any lamellipodial exten-sions. This is paralleled by a change in the morphology of individual cells within theoutgrowths, especially at the periphery where most cell movement is occurring.Normally, neural crest cells flatten progressively and spread with time in culture andthis is especially marked by 72 h; in this respect they behave similarly to heart fibro-blasts in explant culture (Couchman & Rees, 1979). However, with exposure toretinol cells become less adhesive to, and may even detach from, the substratum. Thispresents a range of morphologies, from ' triangular' to fusiform or spindle-shaped torounded, as cell-substratum adhesiveness is lost and the cell gradually loses itslocomotory abilities. (The cells may not necessarily be less adhesive to each other, asrounded cells were sometimes seen in small clusters; retinoids may affect cell-substratum and cell-cell adhesions differentially.) The surface activity of retinol-affected cells is not less than that of control cells but may even be increased, and it isthe lack of substratum contact and inability to extend and maintain lamellipodia thatapparently impairs the cell's locomotory ability. This i9 coincident with a failure toorganize extracellular fibronectin, which in control cultures changed from a diffuse/punctate to a fibrillar distribution during the culture period. (See also Newgreen &Thiery (1980), who describe a similar ability to organize fibronectin in normal cranialneural crest cells of the chick embryo.) Furthermore, actin distribution, which incontrol cells changes from a diffuse distribution to a cable organization, remainsdiffuse. On present evidence it is not possible to determine whether retinol causes


changes in the cytoskeleton initially, and fibronectin distribution is affected secondarilyor vice versa.

However, all these effects, at the retinol concentrations employed, are apparentlyreversible, so that cells with dramatically altered morphology after 24 h exposure,given a period of recovery, ' catch up' with control cells. They resume normal ratesof outgrowth, develop normal cell phenotypes, becoming progressively flattened,display normal contact and locomotory behaviour, and express fibrillar patterns ofextracellular fibronectin and cable arrays of cytoskeletal actin. This ability to recoverwas directly related to the concentration of retinol to which the cells were initiallyexposed. Although not tested, it is probable that cells exposed for longer periods or tohigher concentrations of retinol would not recover so readily. Vitamin A-typecompounds or retinoids exhibit a wide range of well-described physiological effectsthat, with the exception of retinol in the visual system remain to be satisfactorilyexplained at the cellular and molecular levels. Retinoids influence growth and differen-tiation in many non-visual systems, notably skeletal tissues, the male and femalereproductive tracts and a variety of epithelia. Effects on epithelia are possibly relatedto the observation that retinoids can have an inhibitory effect on the incidence andgrowth of carcinogen-induced tumours, a large proportion of which are carcinomas,i.e. of epithelial origin (see a review by Lotan, 1981). From the plethora of experi-mental work on the activity of retinoids three possible modes of action have emerged.First, that retinoids as lipid-soluble compounds have a direct effect on cell membranesby entering the lipid phase of the membrane and there causing changes in viscosityand permeability, which may subsequently precipitate a wide range of physiologicalevents (Dingle & Lucy, 1965). Secondly, since cytoplasmic binding proteins forcertain retinoids have been identified recently, it has been proposed that retinoids mayact like steroid hormones, causing changes in patterns of gene expression (Chytil &Ong, 1978). Thirdly, it has been suggested that phosphorylated vitamin A maynormally function as a lipid intermediate in the glycosylation of membrane glyco-proteins (De Luca, 1977) thus influencing membrane function and possibly cell-matrix interactions. The precise mode of actin of retinoids in any non-visual systemstill remains to be established clearly and it is possible that in some systems retinoidsmay act simultaneously at several loci within the cell.

In the present system the simplest and possibly the most likely mode of action isby insertion into the plasmalemma, causing viscosity /fluidity changes. This correlateswith the observation that the plasmalemma of retinol-affected cells was very mobile,and with the possibility that loss of organization of membrane-bound fibronectinreceptors underlies the inability of such cells to develop a fibrillar fibronectin distri-bution. If so then the loss of actin organization or failure to organize actin into cablesmay be secondary to this and consequently actin may remain or become diffuse. Thusexposure to retinol produces a hypermobile membrane with impaired ability toorganize membrane receptors for a major extracellular glycoprotein known to mediatein cell-substratum adhesion. This, coupled with disorganization of actin micro-filaments severely impairs the cells' ability to adhere to a substratum to spread andto translocate actively. A similar relationship between fibronectin and the organiza-


tion of the actin cytoskeleton is seen in protease-treated cells, urea-treated cells andtransformed cells (see e.g. Hynes, Ah', Mautner & Destree, 1978).

How does this interpretation correlate with the apparently specific effects of retinoidson cranial neural crest or on the pattern of neural crest-derived tissues reported in theliterature? Preliminary results with heart fibroblasts revealed that they respond toretinol in a manner similar if not identical, to neural crest cells as judged by morpho-logical criteria in that they exhibit reduced cell-substratum adhesiveness. We propose,that the action of retinol is not specific for neural crest cells but that retinol mayaffect any moving or migrating cell, in which case the effects in vivo might be inter-preted more accurately as selective rather than specific. Thus retinol selectively affectsindividual cells undergoing large-scale movements or migrations and impairs theirability to interact with, and move through an extracellular matrix. A predictableconsequence of this would be that patterns and rates of migration would be disrupted.However, the majority of cells in an embryo do not display such movements and astheir motility is confined to small-scale 'associative' movements such cells remainlargely unaffected. Thus heart fibroblasts in vivo could only be pffected (if at all) in aminor fashion, which might escape attention. However, in vitro with constraintsupon movement eliminated, the heart fibroblasts migrate from the explant and areaffected in a parallel way. If this interpretation is correct then primordial germ cells,which also undergo extensive migration during embryogenesis, might also be affected,causing impaired locomotion and possibly resulting in deficient colonization of thepresumptive gonad by the primordial germ cells. However the true extent of anyimpairment of cell migration, but it primordial germ cells or neural crest may belessened by the embryo's ability to regulate for a disturbance.

Few other studies have been made of the effects of retinoids on the locomotoryability of embryonic cells. Exposure in utero or in vitro retards migration of explantedlimb mesenchyme from mouse embryos and cells from embryos exposed to retinoidsin utero will recover during the culture period (Kwasigroch & Kochar 1975); cell-surface effects were invoked as an explanation, although there was no apparent reduc-tion in cell-substratum adhesiveness (Kochar 1977). Morriss (1975) examinedretinoideffects on primary mesenchyme from primitive-streak stage rat embryos but theconcentrations used were possibly cytotoxic for these particular cells, as rapid celldetachment and blebbing were accompanied by lysis. In a screening programme forpotential teratogenic compounds using embryonic cells in culture, it was observedthat retinoids elicit a change in neural crest cell morphology in which the cells became'spindle-shaped or rounded rather than flattened', but no further comment was made(Wilk et al. 1980).

One of the novel and unpredicted aspects of the present results is that the neuralcrest reacts to retinoid exposure in a manner different from that exhibited by a numberof cultured cells lines. Biologically active retinoids generally cause an increase in cell-substratum adhesion or a reduced detachability of cultured fibroblastic cells, which iscoincident with an increase in surface-associated fibronectin (Jetten, Jetten, Shapiro &Poon, 1979), and is also dose-dependent and reversible (Adamo, De Luca, Akalovsky& Bhat, 1979). This contrast is not simply due to a fundamental difference between


embryonic cells and those from a cultured line, because mouse limb-bud mesenchymecells will also respond to retinoid exposure by increasing the binding of fibronectinat the cell surface and subsequently remaining undifferentiated (Lewis, Pratt,Pennypacker & Hassell, 1978). Furthermore, there is a report of a cell line respondingin a similar way to that found in the present results; i.e. cells of an intestinal epithelialline react to retinoic acid by an increase in detachability, which presumably reflects adecreased cell-substratum adhesiveness (Shapiro & Poon, 1979). Thus not all cells inculture response in similar or identical ways to exposure to retinoids; decreased orincreased cell-substratum adhesiveness may be due to different modes of action, forexample, differences in retinoid-binding proteins. However, within a limited range ofexamples, cells with a recent epithelial 'history' or origin react in a common wayin vitro and show reduced cell-substratum adhesiveness on exposure to retinoids.(For example, primary mesenchyme recently derived from 'epiblast', neural crestarising from ectoderm and a cell line derived from intestinal epithelium.)

The normal physiological role of vitamin A in control of epithelial growth anddifferentiation has frequently been invoked as a possible factor in the efficacy ofvitamin A in causing carcinoma regression. In vivo retinoids may possibly impair the(carcinoma) cells' ability to move through an extracellular matrix and thus reduce theinvasive and metastatic properties of the carcinoma cells. If so, this would parallelthe effects of retinoids on migration of the neural crest, a morphogenetic movementthat can be interpreted as an invasiveness of a temporary and 'controlled* character.

The work was supported by grants from the Medical Research Council and Cancer ResearchCampaign to P. V. T. and D. R. G., respectively. Drafts of the manuscript were typed by SueBroomfield and Sue Coxson, whom we thank.

REFERENCESADAMO, S., DE LUCA, L. M., AKALOVSKY, I. & BHAT, P. V. (1979). Retinoid-induced adhesion

in cultured transformed mouse fibroblasts. J. natn. Cancer. Jnst. 6a, 1473-1478.BEE, J.A. & THOROGOOD, P. V. (1980). The role of tissue interactions in the skeletogenic

differentiation of avian neural crest cells. Devi. Biol. 78, 47-62.CHIQUET, M., PURI, E. C. & TURNER, D. C. (1979). Fibronectin mediates attachment of

chicken myoblasts to a gelatin-coated substratum. J. biol. Chem. 354, 5475-5482.CHYTIL, F. & ONG, D. E. (1978). Cellular vitamin A binding proteins. In Vitamins and Hor-

mones, vol. 36 (ed. P. L. Munson, E. Diczfalusy, J. Glover & R. E. Olson), pp. 1-32. NewYork: Academic Press.

COUCHMAN, J. R. & REES, D. A. (1979). The behaviour of fibroblasts migrating from chickheart explants: changes in adhesion, locomotion and growth, and in the distribution ofactomyosin and fibronectin. J. Cell Sci. 39, 149-165.

DE LUCA, L. (1977). The direct involvement of vitamin A in glycosyl transfer reactions ofmammalian membranes. In Vitamins and Hormones, vol. 35 (ed. P. L. Munson, E. Dicz-falusy, J. Glover & R. E. Olson), pp. 1-57. New York: Academic Press.

DINGLE, J. T. & LUCY, J. A. (1965). Vitamin A, carotenoids and cell function. Biol. Rev. 40,422-461.

DOUAILLY, D., HARDY, M. H. & SENGEL, P. (1980). Formation of feathers on chick footscales: a stage-dependent morphogenetic response to retinoic acid. J. Embryol. exp. Morph.58, 63-78.

ENGVALL, E. & ROUSLAHTI, E. (1977). Binding of soluble forms of fibroblast surface protein,fibronectin, to collagen. Int. J. Cancer 20, 1-5.


HAMBURGER, V. & HAMILTON, H. L. (1951). A series of normal stages in the development ofthe chick embryo. J. Morph. 88, 49-92.

HASSELL, J. R., GREENBERG, J. H. & JOHNSTON, M. C. (1977). Inhibition of cranial neural crestdevelopment by vitamin A in the cultured chick embryo. J. Embryol. exp. Morph. 39,267-271.

HYNES, R. O.( ALI, I. U., MAUTNER, V. M. & DESTREE, A. (1978). LETS glycoprotein; arrange-ment and function at the cell surface. In The Molecular Basis of Cell-Cell Interaction (ed.R. A. Lerner & D. Bergsma). Birth Defects series, vol. 14, no. 2, pp. 139-153. New York:A. R. Liss.

JETTEN, A. M., JETTEN, M. E. R., SHAPIRO, S. S. & POON, J. P. (1979). Characterization of theaction of retinoids on mouse fibroblast cell lines. Exp Cell Res. 119, 289-299.

JOHNSTON, M. C. (1966). Cranial neural crest cells in the chick embryo. Anat. Rec. 156, 143-155-

KEITH, J. (1977). Effects of excess vitamin A on the cranial neural crest in the chick embryo.Ann. R. Coll. Surg. 59, 479-483.

KOCHAR, D. M. (1977). Cellular basis of congenital limb deformity induced in mice by vitaminA. In Morphogenesis and Malformations of the Limb (ed. D. Bergsma & W. Lenz), BirthDefects series, vol. 13, no. i, pp. m-154. New York: A. R. Liss.

KWASIGROCH, T. E. & KOCHAR, D. M. (1975). Locomotory behaviour of limb bud cells. ExplCell Res. 95, 269-278.

LAZARIDES, E. & WEBER, K. (1974). Actin antibody: the specific visualization of actin filamentsin non-muscle cells. Proc. natn. Acad. Sci. U.S.A. 71, 2268-2272.

LE LIEVRE, C. S. & LE DOUARIN, N. M. (1975). Mesenchymal derivatives of the neural crest:analysis of chimaeric quail and chick embryos. J. Embryol. exp. Morph. 34, 125-154.

LEWIS, C. A., PRATT, R. M., PENNYFACKER, J. P. & HASSELL, J. R. (1978). Inhibition of limbchondrogenesis in vitro by vitamin A: alterations in cell surface characteristics. Devi Biol. 64,3I-47-

LOTAN, R. (1981). Effects of vitamin A and its analogs (retinoids) on normal and neoplasticcells. Biochim. biophys. Acta 605, 33-91.

MORRISS, G. M. (1975). Abnormal cell migration as a possible factor in the genesis of vitaminA-induced craniofacial anomalies. In Neva Approaches to the Evaluation of Abnormal Mam-malian Embryonic Development (ed. D. Neubert), pp. 678-687. Berlin: Geo. Thieme Verlag.

MORRISS, G. M. & THOROGOOD, P. V. (1978). An approach to cranial neural crest cell migrationand differentiation in mammalian embryos. In Development in Mammals (ed. M. H. Johnson),vol. 3, pp. 363-412. Amsterdam: Elsevier/North-Holland.

NEWGREEN, D. & THIERY, J. P. (1980). Fibronectin in early avian embryos: synthesis anddistribution along the migration pathways of neural crest cells. Cell Tiss. Res. a n , 269-291.

PAUL, J. (1975). Cell and Tissue Culture, 5th edn. Edinburgh: Churchill Livingstone.POSWILLO, D. (1975). The pathogenesis of the Treacher Collins syndrome (mandibulofacial

dysostosis). Br. J. oral. Surg. 13, 1-26.SHAPIRO, S. & POON, J. P. (1979). Retinoic acid-induced alterations of growth and morphology

in an established epithelial line. Expl Cell Res. 119, 349-357.SPUDICH, J. A. & WATT, S. (1971). The regulation of rabbit skeletal muscle contraction. I.

Biochemical studies of the interaction of the tropomyosin-troponin complex with actin andthe proteolytic fragments of myosin. J. biol. Chem. 246, 4866-4871.

THOROGOOD, P. V. (1981). Neural crest cells and skeletogenesis in vertebrate embryos. Histo-chem.J. 13, 631-642.

WILK, A.L., GREENBERG, J. H., HORIGAN, E. A., PRATT, R. M. & MARTIN, G. R. (1980).Detection of teratogenic compounds using differentiating embryonic cells in culture. InVitro 16, 269-276.

(Received 28 January 1982 - Revised 5 May 1982)

effects of vitamin a on the behaviour of migratory neural...

Documents