Development 1987 GURDON 285 306

Development 99, 285-306 (1987)Printed in Great Britain © The Company of Biologists Limited 1987

Review Article 285

Embryonic induction - molecular prospects

J. B. GURDON

Cancer Research Campaign Molecular Embryology Research Unit, Department of Zoology, University of Cambridge, Downing Street,Cambridge CB2 3EJ, UK

Synopsis

Section no. page

(1) Introduction 1(2) Molecular assays 2(3) Instructive and permissive inductions 5(4) The timing of the response 8(5) The localization of the response 9(6) The nature of an inducer 11(7) Early steps in the inductive response 14(8) Comparison with other cell signalling

systems 15

(9) Prospects for further analysis at themolecular level

(10) ConclusionReferences

Key words: induction, embryonic molecular markers,competence in induction, inducer, Xenopus, Triturus.

171818

(1) Introduction

Embryonic induction is an interaction between one(inducing) tissue and another (responding) tissue, asa result of which the responding tissue undergoes achange in its direction of differentiation. This isprobably the single most important mechanism invertebrate development leading to differences be-tween cells and to the organization of cells into tissuesand organs.

The phenomenon is considered to have been dis-covered by Spemann in 1901 and by Lewis (1904) whoboth established that in certain species of Rana theformation of a lens from ectoderm is dependent on aninfluence of the underlying optic lobe of the brain.Interest in induction was enhanced by the famousSpemann & Mangold (1924) experiment in which thedorsal lip of the unpigmented Triturus cristatus em-bryo was transplanted to the ventral region of apigmented Triturus taeniatus gastrula, where it in-duced the pigmented host cells to form a secondaryaxis (Spemann, 1938). Within the secondary axis isthe neural tube which is induced in ectoderm by theunderlying mesoderm, a process referred to as pri-mary induction. It was soon discovered that a varietyof easily obtainable materials could substitute formesoderm in inducing ectoderm to undergo neuraldifferentiation, and it was expected in the 1930s that

these could serve as sources for purifying a neuralinducing substance (excellent review by Witkowski,1985). This optimism disappeared when it was foundthat newt embryo ectoderm was so far predisposedto neural differentiation that almost any substancecould act as an inducer (Holtfreter, 1934, 1947;Waddington, Needham & Brachet, 1936; Barth,1968). Although it is now nearly a century sinceembryonic induction was discovered, the molecularbasis of inducers and the inductive response remainsalmost totally obscure.

Embryonic induction has been most fully analysedin Amphibia, on account of the large size and easymanipulation of their eggs and embryos, and amphib-ian work is emphasized in this review. Cell interac-tions having the characteristics of induction probablyconstitute the single most prevalent mechanism in thedevelopment of the vertebrates. Though much lessstudied, inductive cell interactions certainly takeplace in the development of invertebrates. Forexample, in sea urchins, vegetal pole micromerestransplanted to the animal half of an embryo inducesecondary vegetal differentiation (Horstadius, 1973).In the nematode C. elegans, a signal from an adjacentgonadal anchor cell induces hypodermal precursorcells to adopt one of two types of vulval cell lineage,as opposed to nonvulval development in the absenceof a gonadal cell (Sternberg & Horvitz, 1986).

Noenk_zuliyadi

Sticky Note

http://dev.biologists.org/content/99/3/285.full.pdf

286 /. B. Gurdon

The purpose of this review is to summarize somecommon properties of embryonic inductions and toconsider ways whereby current methods of molecularbiology, which are proving of such enormous value tofields as diverse as taxonomy and membrane physi-ology, may be usefully applied to the extraordinarilyrecalcitrant problem of embryonic induction.

(2) Molecular assays

For analysis at the molecular level, the single mostdesirable characteristic in any experimental inductionsystem is a rapid and quantitative assay for an earlymolecular event. The importance of having an assayfor an early inductive response should be stronglyemphasized. It has become increasingly clear thatmost, if not all, inductions consist of several separatesteps. For example, liver induction in the chickrequires first an induction by heart mesenchyme andthen by liver mesenchyme, neither alone being suf-ficient (Wessells, 1977). Triturus lens induction nor-mally involves three sequential inductions by endo-derm, heart mesoderm and retina (Jacobson, 1966).If an induction can only be recognized after severalsequential steps have taken place, this greatly compli-cates attempts to resolve the process into its com-ponent parts. Therefore to simplify the problem, it isalmost essential to use a direct assay for an earlyresponse. For example, the first synthesis of a singlegene's transcripts can be detected using sensitivebiochemical probes that depend on the in vitroformation of nuclease-resistant hybrids. However,such assays do not give localization at the cellularlevel. For a complete analysis, it is essential to be ableto apply either in situ hybridization with nucleic acidprobes or a specific antibody, to distinguish inducedfrom uninduced cells at the cellular level. Antibodieshave the great advantage of providing single cellresolution, a situation very hard to attain by in situhybridization in embryos.

The following paragraphs give examples of molecu-lar markers that can be used in some of the mostwidely studied inductions. In normal amphibian de-velopment, cells of the ectodermal lineage are sub-jected to at least three sequential inductions. As seenin Fig. 1, some of the animal (future ectodermal) cellsof a blastula are induced by vegetal cells to formmesoderm; this includes muscle for which severalexcellent markers are available (Figs 2-4), eitheras muscle actin-specific probes (Mohun, Brennan,Dathan, Fairman & Gurdon, 1984; Gurdon, Fair-man, Mohun & Brennan, 1985), or as antibodies(Kintner & Brockes, 1984; Smith, Dale & Slack,1985). An antibody that recognizes keratan sulphatein the notochord (an embryonic form of vertebralcolumn also formed by mesoderm induction) has

been described by Smith & Watt (1985), but thisappears to bind only extracellular material and maynot be diagnostic for single cells. The next majorinduction in amphibian development takes placeduring gastrulation when mesodermal cells at thedorsal blastopore lip and inside the embryo inducethe overlying ectoderm to form nerve (neural orprimary induction). Most molecular markers ofneural induction appear rather late in development(Takata, Yamamoto & Ozawa, 1981, 1984; Duprat,Kan, Gualandris, Foulquier & Marty, 1985; Godsave,Anderton & Wylie, 1986). However, two early mol-ecular markers have recently been described forneural induction in Xenopus. One is the extracellularglycoprotein N-CAM, a neural cell adhesion mol-ecule. This can be detected at the neurula stage by anantibody (Jacobson & Rutishauser, 1986), or at thelate gastrula stage by a cDNA probe as seen in Fig. 5(Kintner & Melton, 1987). The other is a Xenopushomeobox-containing gene XlHbox6, part of whichserves as a probe for spinal cord induction as earlyas the late gastrula (Sharpe, Fritz, DeRobertis &Gurdon, unpublished data). It is also worth notingthat sodium pump activation can be detected byelectrophysiological means in neural cells a few hoursafter they have been induced (Blackshaw & Warner,1976). Fortunately there are excellent cytokeratin orother markers for epidermal differentiation (Jonas,Sargent & Dawid, 1985); these are expressed in theabsence of induction but are suppressed by neuraldifferentiation (Slack, 1985; Jones & Woodland,1986; Akers, Phillips & Wessells, 1986). In situhybridization has been used successfully in relativelylate larval stages of Xenopus (stage 42) to detectmuscle, epidermal or neural differentiation (Dwor-kin-Rastl, Kelley & Dworkin, 1986; Fig. 3). The thirdmuch-studied induction in amphibian development islens induction, which results mainly from the opticlobes of the embryonic brain inducing overlyingectoderm to invaginate and form a lens vesicle(Jacobson, 1966). Nucleic acid probes and antibodiesare available for lens-specific crystallins (e.g.Grainger, Hazard-Leonards, Samaha, Hougan, Lesk& Thomsen, 1983; Kondoh & Okada, 1986).

In fact early amphibian inductions are more com-plex than just suggested. Each major inductive step,such as mesodermal induction or neural induction,includes more precise regional information. Thusvegetal cells on the dorsal side of an embryo inducenotochord and muscle, whereas vegetal cells on theventral side induce blood and mesenchyme, con-sidered to be lower grades of mesodermal induction(Boterenbrood & Nieuwkoop, 1973). Similarly, Man-gold (1933) and Horst (1948) concluded, in respect ofneural induction in newts, that anterior mesoderminduces brain, while posterior mesoderm induces

Embryonic induction - molecular prospects 287

EXTERNAL VIEWSFROM SIDE

Animal pole

TRANSVERSESECTIONS

Greycrescent/(dorsal)

Inducedmesoderm

Spermentry(ventral)

Vegetal pole

MID BLASTULA.

Responsiveanimal.cells

Blasto-coel

Inducingvegetalcells

MESODERM INDUCTION

Dorsalblastopore^l i p -

Induced

'Yolkplug

Responsiveectoderm

Inducingmesoderm

Arch-enteron

Residualblastocoel

B MID GASTRULA. NEURAL INDUCTION

Dorsal Responsiveectoderm

Section through future eye

NEURULA.

Inducingoptic lobeof brain

LENS INDUCTION

Fig. 1. Sequential inductions in amphibian development. Inducing tissues are indicated by arrows, cells competent torespond to induction by open circles, and cells which actually respond by filled in circles.

288 J. B. Gurdon

WholeUndi-gested ^probe

Cardiacmuscle kactin

An Conj Veg

Fig. 2. A molecular probe for an early inductiveresponse. A single stranded 380-nucleotide length ofDNA, complementary to part of the 3' region of Xenopuslaevis cardiac actin mRNA, is 32P-labelled by in vitrosynthesis from an M13 vector. After hybridization toRNA, SI nuclease digestion shortens the probe to a 250-nucleotide length protected by the complementarymRNA. Animal (An), vegetal (Veg), or animal-vegetalconjugates (Conj) were prepared from blastulae andcultured to the neural folds stage when RNA wasextracted and analysed. Animal (responding) or vegetal(inducing) tissues are negative for muscle actin genetranscription when cultured in isolation, but stronglypositive, as are normal whole embryos, when placed incontact. Similar results can be obtained using anti-senseRNA probes synthesized with SP6 or T7 RNApolymerase (from Gurdon et al. 1985).

spinal cord. The regional nature of early inductionshas been strongly emphasized and explored in detailby Slack & Forman (1980), Dale, Smith & Slack(1985), and Smith et al. (1985), who believe that anindependent dorsalizing induction spreads from thedorsal equatorial region during the early gastrulastage, causing some ventral equatorial cells to be-come muscle. There appear to be two possibleinterpretations of these events. One is that eachmajor induction in reality consists of two or moreseparate processes, such that vegetal and dorsalinducers differ from each other, as would anteriorand posterior neural inducers. Indeed, there is noreason to limit the presumed diversity of inducers;perhaps there are separate mesoderm inducers andreceptors for notochord, muscle, blood, mesen-chyme, etc. The other .interpretation is that dorsalvegetal cells of a blastula and early gastrula emit onlyone kind of inducer, which in high concentration(dorsal side) would induce notochord and muscle, butin lower concentration (ventral side) would induceblood and mesenchyme. The same argument could

apply to neural induction, one inducer having differ-ent effects in anterior and posterior regions. Thereseems to be no decisive experiment that distinguishesthese interpretations. However, for the immediateprospects of molecular analysis, the difference is notcrucial; what matters most is the availability of earlyexpressed quantitative markers, as discussed abovefor muscle. It should be mentioned that globin genetranscription, as a marker for blood, first occursrather late, at stage 30, in amphibian development(Banville & Williams, 1985a,b), and a molecularmarker has not been described for mesenchyme.

In avian development, cartilage formation in thefuture vertebral column is induced in somite cells byspinal cord; chondroitin sulphate serves as a quanti-tative measure of inductive effect (Kosher, Lash &Minor, 1973), though this is synthesized at a signifi-cant but much lower level in noninduced tissue.Similarly, collagen synthesis serves as a marker forthe enhanced differentiation of chick cornea by lens(Meier & Hay, 1975). Epidermal differentiationinto keratinized, mucous or ciliated epidermis isdependent on induction by the underlying dermis.Although molecular markers have not been used insuch experiments, it would not appear difficult todevelop markers for products characteristic of thesekinds of differentiation.

In mammals, pancreas differentiation (Fig. 6) isaccompanied by the synthesis of many well-definedgene products (Rutter, Wessells & Grobstein, 1964).Specialized cells of the pancreas are derived from theendodermal (gut) epithelium, after induction by theadjacent mesodermal mesenchyme. Exocrine cellssecrete such enzymes as trypsin, ribonuclease andcarboxypeptidase; the endocrine cells synthesize glu-cagon and insulin, the assays for insulin being exceed-ingly sensitive.

A huge amount of very careful work has been doneon many other inductive systems (excellent summaryby Wessells, 1977, where references can be found).These include kidney (formed from mesoderm in-duced by other mesoderm of the ureteric bud), mousemammary gland (formed from mammary epitheliuminduced by mammary or other sources of mesen-chyme), and avian feathers and scales (formed in theepidermis induced by mesodermal dermis). In allthese cases, the emphasis has been on the morpho-genetic patterns formed by populations of cells.Pattern formation is very hard to analyse by molecu-lar means, though recent work suggests that celladhesion molecules may be involved in the morpho-genetic response to induction (Gallin, Chuong, Fin-kel & Edelman, 1986).

(3) Instructive and permissive inductions

A key question in the analysis of any induction iswhether the kind of response is determined by theinducing or responding tissue. The distinction empha-sized by Saxen (1977) between a permissive and an


instructive induction is particularly helpful. A per-missive induction is one in which the respondingtissue is already so far committed to its final state ofdifferentiation that any, often unspecific, influencewill serve to complete the process; in such a case thenature of the response is almost entirely determined

Fig. 3. The use of in situ hybridization to recognize gene activation in induced tissues. Single-stranded DNA probes,50-250 nucleotides long, were 3H-labelled by nick-translation of plasmids containing cDNAs complementary to tadpolemRNAs. The probes were hybridized to sections of fixed embryos, which were subsequently autoradiographed;autoradiographic grains, representing hybridized probe, are seen as white areas in the figures shown. (A) This probehybridizes preferentially to the nervous system, the brain and neural retina being clearly labelled; it also labelsepidermis. (B) This probe labels muscle very strongly, but not epidermis or nerve. Both sections are of normal hatchedtadpoles. In situ hybridization nearly always generates more background labelling than antibodies: compare Fig. 3B with4A (from Dworkin-Rastl et al. 1986).

Fig. 4. The use of an antibody to recognize muscle induction, showing low background and single cell resolution. The12/101 monoclonal antibody of Kintner & Brockes (1984), against an as yet unidentified muscle antigen, has beenapplied to sections of Xenopus embryos, and visualized by a secondary fluorescent rhodamine-labelled antibody.(A,B) Whole embryo, hatching stage. (C,D) An animal-vegetal conjugate, cultured until controls reached the muscularresponse stage. (A and C) Stained with antibody; note the absence of nonmuscle background in A. (B and D) Similarsections stained with Hoechst to show the distribution of cells. In C and D, the junction between animal and vegetaltissues is indicated by white arrows.

290 /. B. Gurdon

Fig. 5. A probe for a neural cell adhesion moleculeserves as an early marker of neural differentiation. Thefigures show transverse sections through the neural foldsregion of a Xenopus embryo. (A) Phase contrast; NT,neural tube; N, notochord; M, muscle. (B) In situhybridization with an 35S-RNA probe synthesized from anN-CAM cDNA; the neural tube shows stronghybridization, the autoradiographic grains appearingwhite under dark-field illumination. (From Kintner &Melton, 1987.)

by the responding tissue. In contrast, an instructiveinduction is one in which the responding tissue is tosome extent uncommitted and requires a specificsignal from the inducing tissue telling it in which oftwo or more directions to differentiate.

Examples will make the distinction more clear.One is provided by the later stages of mouse pancreasdifferentiation. From the 15-somite stage, the pancre-atic mesenchyme can be replaced by mesenchymefrom many other organ primordia such as the salivarygland, or even by embryo extract, and yet pancreasdifferentiation will take place (Rutter et al. 1964). Atthe 15-somite stage the endoderm epithelium is there-fore largely committed to pancreas differentiation,and needs only a relatively unspecific inductive stimu-lus to complete the process (Wessells & Cohen,1967). Other examples of permissive inductions occurin various adult tissues when enzymes are used tophysically separate inducing and responding tissues.In such cases, the readdition of collagen, glycos-aminoglycans and other extracellular materials willenhance the amount of terminal cell differentiation aseffectively as inducer tissue, and may act by replacingmaterials that were removed during preparation andwhich are normally required to provide permissive

conditions for terminal differentiation (Meier & Hay,1975).

Instructive inductions can be recognized in differ-ent ways. A tissue is generally considered to induceinstructively if it can force a responding tissue tochange its path of differentiation and conform to thefate normally determined by that inducing tissue.This clearly happens in chick and mouse skin devel-opment. For example, a piece of ectoderm from afuture feathered area of a chick embryo combinedwith mesoderm from the lower leg area will formscales characteristic of the leg and not feathers whichwould otherwise have been formed in normal devel-opment (Rawles, 1963; Fig. 7). Even more remark-able is the combination of chick corneal epithelium(destined to form the translucent cornea in normaldevelopment) with mouse mesoderm (which wouldnormally induce mouse ectoderm to form hair). Theresult is that the chick epithelium forms feathersthough these are morphologically fairly abnormal(Sengel, 1976). In these cases, the mesoderm iswholly responsible for redirecting the way the ecto-derm differentiates. It is restricted, obviously, by thegenetic repertoire of the responding cells, such thatchick epithelium forms feathers and not hair.

A more stringent criterion for recognizing aninstructive induction requires that a tissue can beinduced to differentiate in two different ways apartfrom the way it would if receiving no induction at all.Thus Xenopus early gastrula ectoderm will formepidermis if cultured in isolation, but can be inducedto form nerve if induced by mesoderm, or muscle ifinduced by blastula vegetal cells (section 2 above).At least one of the two latter must give specificinformation during induction and must therefore beinstructive.

The answer to the question whether tissue induc-tions are permissive or instructive is that there areclear examples of each kind. Indeed both types seemto occur within the same system. In the mouse, forexample, endoderm isolated at the 9-somite stage willonly form pancreas if associated with its own mesen-chyme. At the 15-somite stage, the endoderm willform pancreas if induced by any kind of mesenchyme,a permissive effect described above. Similarly, liverinduction in the chick is at first instructive andsubsequently permissive (Le Douarin, 1964). It seemslikely that it may be a general principle that aninstructive induction is needed at first, with progress-ively more permissive inductions at later stages, asdevelopmental options are narrowed, to produce amorphologically normal tissue. However, it is import-ant to appreciate that, even when inductions are

Embryonic induction — molecular prospects 291

9-day embryo — longitudinal section

B Stomach

Mesoderm

Pancreas

Connectionto yolk sac

9-day gut — longitudinal section

IntestineEpithelium

Islets of Langerhans(Endocrine: secreteinsulin, glucagon, andsomatostatin)

Pancreatic acini (Exocrine:

secrete proteases, peptidases,

nucleases, amylase,

lipases, phospholipases)Mesoderm

Pancreas at 15 days

Fig. 6. Development of the pancreas in the mouse. (A,B) Longitudinal sections through the pancreas forming as anevagination of the gut. (C) Diagram of the pancreas at 15 days; acini and islets are derivatives of the epithelium inducedby the adjacent mesoderm. (From Wessells & Rutter, 1969.)

described as instructive, the ways in which the re-sponding tissue can differentiate are already limited;one of the most important aspects of an induction istherefore the predisposition of cells to respond in aparticular way.

Does the classification of an induction as permiss-ive or instructive give any useful information about itsmolecular basis? A permissive inducer could besomething as unspecific as amino acids, glucose orother nutrients, which would be unlikely to tell usanything useful about the inductive response. On theother hand, an instructive response resembles inseveral ways those signalling mechanisms in whichwell-characterized receptors have been identified,

hormones and growth factors being examples. It istherefore a reasonable assumption at present thatinstructive inductions involve specific receptor mol-ecules. The fact that LiCl has strong inductive effects(see p. 297) does not exclude the involvement of areceptor, since Li+ might affect the pathway ofinduction beyond the stage of receptor activation.

We will return later to the important questionwhether the instructive or permissive nature of induc-tion, so far discussed at the tissue level, can beextended to individual cells. But first it is useful toreview temporal aspects of induction, which empha-size the importance of the responding tissue.

292 J. B, Gurdon

High

A. Leg mesoderm in wing -> leg feathers

B. Feather-area mesoderm plusnonfeather-ectoderm -> feathers

C. Scale-area mesoderm plusfeather-area ectoderm —> scales

Fig. 7. The mesodermal induction of epidermaldifferentiation in chick development is instructive. (FromWessells, 1977.)

(4) The timing of the response

Much useful information has come from a carefuldescription of timing in inductions. The first import-ant generalization is that the time for which tissuescan respond to induction (their state of competence),as well as the time for which other tissues can induce,are both strictly limited in duration. This has emergedmost clearly from early embryonic inductions such asthose that form mesodermal and neural structures inAmphibia. The stage at which the inductive andresponsive capacities are acquired is unclear, but bothterminate abruptly soon after the time when theseinductions are complete in normal embryos (Fig. 8).Inductions that take place much later in life, and on amuch longer time scale, appear to have similar timingcharacteristics. For example, the dermis and epider-mis in chick development acquire and lose inductivecapacity and responsiveness over just the few dayswhen this induction normally takes place (Rawles,1963).

We can now ask an important question, whetherthe time of response to an induction is determined bythe inducing or responding tissues. This requires thecombination of inducing and responding tissues of

£ 800 Oc o'5 D.

II

Low

i* Hind-brain

Mesoderm(inducing) -*•

5 10

Blastula

Mesodermal induction

15

Gastrula

Neural induction

20 Hours

Neurula

Fig. 8. Inductive capacity and competence to respond arerapidly lost after the time in normal development whenthese activities normally take place. The two curves onthe left of the figure apply to mesodermal induction andthose on the right to neural induction. (From results ofGurdon et al. (1985) and Dale etal. (1985).)

different ages, an experimental situation that can beachieved once the duration of the inductive andresponsive phases is known. Using a sufficiently well-defined criterion by which to recognize when induc-tion has taken place, such as muscle actin transcrip-tion, it is clear that the timing of induction isdetermined almost entirely by the developmentalstate of the responding tissue, and not by the timesince it was exposed to inducing tissue (Fig. 9). Oneway in which this could operate is as follows. Anendogenous series of events, perhaps including theactivation of one gene by another, could permitthe synthesis of muscle-specific actin, but only at thenormal time in development and only if the cells havereceived an inductive signal at some previous but notprecisely defined time.

Another useful piece of temporal information is forhow long the inducing signal must be applied to beeffective. Using the amphibian muscle inductionsystem, inducer contact must be maintained for 1-2 hto give a response, but the amount of induction (actingene transcripts, in this case) increases with contacttimes of up to about 5h (Gurdon et al. 1985). Withheterogenous inducers, much shorter exposure timesof lh or less are sufficient to elicit a response (e.g.Smith, 1987), though it seems hard to eliminate thepossibility that some part of the inducing pellet orsolution used in such experiments has been leftbehind at the time of its removal. The later in

StageHrsat""23°C

Normal developmentof fertilized eggs"

An st 7 + veg st 8

An st 10 + veg st 8

An st 7 + veg st 8Veg removed after 2 hrs

6 8 11 13 14 171 3 5 7 9 11 13 15 17 19

• •

| | Animal tissue

/$^ Vegetal tissue

^M First detection of muscle gene activation

Fig. 9. The time of an inductive response is determinedby the responding, not inducing, tissue. Animal andvegetal tissues, removed from a blastula (stages 7 or 8) orfrom an early gastrula (stage 10), were placed in contactaccording to the time scale at the top of the figure. Theywere then frozen at different times ranging from 11 to17 h after fertilization, and analysed using a muscle actingene probe as indicated in Fig. 2. (From Gurdon et al.1985.)

development that induction takes place, the longer isthe minimum exposure time; in the case of themesenchymal induction of pancreas differentiation,24 h are required (Wessells, 1977).

The most significant temporal aspects of embryonicinductions can be summarized as follows. First, theminimum time from the start of induction to the finalresponse is long, being measured in hours or days as isalso the case for the action of most hormones andgrowth factors. Second, the abilities to respond andinduce are of strictly limited duration. Third, the timeof response is governed by the state of the respondingtissue and not by the time that has elapsed since theinducer is placed in contact. In these last two re-spects, inductions appear to differ fundamentallyfrom all other known cell signalling systems. In thecase of hormones, growth factors and neurotransmit-ters, the competent state (i.e. possession of a recep-tor) is a permanent property of responding cells, andthe response is timed by the arrival of signallingmolecules.

(5) The localization of the response

From the point of view of developmental significance,much the most important aspect of an induction is itslocalization, i.e. the selection of cells that are torespond or not to respond to the induction. We haveemphasized above how the responding tissue plays a


predominant part in the timing of the response. Onthe other hand the inducing tissue seems to be amajor contributor to the position of the response.

The first important principle is that the number ofcells that are capable of responding to an inductivestimulus greatly exceeds the number of cells thatactually do so. Good examples of this come fromearly amphibian inductions. At a gross level, there isclearly some predisposition of a responding tissuesince not all cells in an embryo are equally com-petent; for example only cells in the animal half of ablastula can respond to inducing vegetal cells byforming mesoderm. However, only cells in closeproximity to the inducing tissue actually respond(Fig. 4C). It is also clear, from experiments withvarious configurations of inducing and respondingtissues, that the proximity of inducing and respondingcells to each other is important, rather than theirdistance from the animal and vegetal poles of anembryo (Nieuwkoop et al. 1952; Gurdon, unpub-lished data). Therefore within a tissue, it is theproximity of inducing cells as well as geographicallylimited competence that determines the localizationof the response.

At the level of single cells, the situation is morecomplicated. All blastula cells in the animal half of anembryo are capable of responding to vegetal induc-tion to become mesodermal, but not more than 40 %in fact do so (Dale et al. 1985). Similarly, both innerand outer layers of Xenopus gastrula ectoderm canrespond to a mesodermal inducer by becoming brain,but only the inner cells actually do so (Asashima &Grunz, 1983). Is it just a question of proximity toinducing cells, or do other factors come in?

We have so far discussed induction from the pointof view of tissues, which consist of hundreds or eventhousands of cells. It would greatly simplify furtheranalysis if we could analyse the response at the levelof individual cells. The possible complications ofdescribing inductive effects in terms of whole tissuesare exemplified by referring back to our discussionon instructive versus permissive effects. We quotedinstructive inductions in which a particular tissue canbe made to differentiate in divergent ways accordingto the kind of inducing tissue; this is the case whenXenopus ectoderm is induced by mesoderm to formnerve or by endoderm to form muscle. The simplestinterpretation of this result is that individual cells arecaused to divert their uniriduced direction of differen-tiation from epidermis into nerve or muscle. How-ever, there are several other interpretations. Theresponding tissue might consist of a mixture of cellswhich are individually predisposed to respond todifferent inducers. The interpretation is further com-plicated by the fact that nearly all tissues that respondto induction undergo some proliferation and cell

294 /. B. Gurdon

rearrangement and in some cases there is selectivecell death. For all of these reasons it is very hard toknow how individual cells in an induced tissue werepositioned with respect to the inducing tissue at thetime of the induction. There are two further reasonswhy it is very desirable to be able to eliminate celldivision in induction experiments. One is that anybiochemical description of postinductive events ishard to interpret if these events could be related tocell proliferation and not to induction. The otheris that the elimination of cell division would testwhether 'quantal mitosis' is involved in this exampleof cell differentiation (Dienstman & Holtzer, 1975).

In the case of amphibian ectoderm, it has beenpossible to suppress all cell division and cell move- -ment with colchicine or cytochalasin, and to showthat muscle gene transcription is nevertheless induced(Gurdon & Fairman, 1986). This does not mean thatthe later morphogenetic aspects of induction proceedunder these conditions, but at least the mechanismsleading to gene, activation in roughly the normalnumber of cells is independent of cell division andrearrangement. The same conclusion does not appearto have been established for other inductive systems,probably because the much slower time scale in-volved would require a longer exposure to inhibitorsthat might be harmful. In no system has it so far beenpossible to suppress DNA synthesis and yet obtain aninductive effect.

To eliminate cell division and cell movement sim-plifies the analysis of inductive systems, but a majorbreakthrough would be achieved if the process couldbe made to take place in single responding cells. Thisappears never to have been achieved, and we mustask if this is significant. Is an inductive responsesimply the sum of individual cell responses, or is somecooperation between responding cells required? So-called 'mass effects' have been described, mostly as aresult of experiments in which the size of the respond-ing tissue is progressively reduced (Lopashov, 1935;Muchmore, 1957; Grunz, 1979). It is usually foundthat there is a size of tissue or number of cells belowwhich no response takes place. However, this couldmerely reflect the common observation that cellsdifferentiate best when surrounded by other cells,rather than by culture medium (e.g. Jones & Elsdale,1963).

A particularly valuable series of experiments hasbeen carried out by Heasman, Snape, Smith & Wylie(1986) in which it is found that single vegetal cells of ablastula, which will differentiate only poorly in iso-lation or in small groups, will nevertheless do sonormally if surrounded by other cells, but it isunimportant whether these surrounding cells are thesame as, or different from, the differentiating cell.This therefore fits the idea that the mass effect is

unspecific and may merely represent a permissiveenvironment. The most informative experimentwould be one in which a single responsive cell wouldbe exposed to normal inducing tissue, since thissingle cell should receive as much inductive influenceand have as favourable a cellular environment aswould many such cells. Some important experimentsleading in this direction have been described byHeasman et al. (1984, 1985), who injected singlemarked vegetal cells into blastulae. They found thatsingle early blastula cells differentiate in conformitywith their surrounding cells, whatever type these are(Fig. 10A). Elegant single-cell transfer experimentshave also been performed by Gimlich & Gerhart(1984) & Gimlich (1986); however, the one or twoblastomeres transferred from the 64-cell stage sub-sequently divide (Fig. 10B), and the transplantedcells are inducing rather than responding. Our ownwork in progress, using a muscle-specific antibody,suggests that single animal cells are much less easilyinduced than a group of these cells, when underlaidor surrounded by vegetal cells. These unpublishedresults and those of Heasman et al. (1984) suggest aprocess by which gene activation or differentiation ofa cell is facilitated by other cells of like kind. This maybe described as a 'community effect' to distinguish itfrom the unspecifically beneficial effect of any othersurrounding cells.

A community effect is not the same as homoio-genetic induction (Mangold & Spemann, 1927; Spe-mann, 1938), by which cells induced to differentiatein one way can in turn induce their neighbours todo the same. Homoiogenetic induction is usuallythought to include an amplification step, by which acell having received an inductive signal itself makesmore inducer, thereby influencing its neighbours inthe same way. The inductive effect can therefore bepassed along within the responding tissue. Resultsconsistent with this idea were obtained by Kurihara &Sasaki (1981) who found that an inducing signalcannot be transmitted through aged, and thereforenonresponsive, tissue. An amplification process isknown to take place during amphibian oocyte matu-ration, in which the maturation-promoting factor(MPF), when phosphorylated, induces oocyte matu-ration as well as the phosphorylation of other MPFmolecules (review by Masui & Clarke, 1979). Whilehomoiogenetic induction may well involve an ampli-fying or relay process, it can also be explained by amuch simpler process, in which inducer molecules arepassed on, without amplification, from one inducingcell to the next; the response would cease when theamount passed on is below a threshold concentration.It is important to appreciate that homoiogeneticinduction is not the same as a community effect. The


Fig. 10. Single-cell transfers in amphibian embryos. (A) A single rhodamine-labelled animal cell of a midblastula wasinjected into the blastocoel of another, unlabelled, host blastula. When the host embryo had reached the swimmingtadpole stage, progeny of the injected cell, which would normally have formed epidermis or nerve, were found insections of the gut (shown). The differentiation of the transferred cell conforms to that of the surrounding host tissue.(From Heasman et al. 1984.) (B) A single dorsal vegetal cell was labelled by injection of fluoresceinated-lysine-dextranat the 64-cell stage, transferred to the ventral vegetal region of another unlabelled 64-cell embryo. The labelled cellintegrated into the host and divided (as shown); dorsal vegetal cells transplanted in this way induce a secondary axis inthe host embryo. This shows that a single 64-cell blastomere (or its daughter cells) possesses axis-inducing activity.(From Gimlich & Gerhart, 1984.)

two would be distinguished by placing a single re-sponsive cell in normal inducing tissue; failure torespond could be explained by a community effect,but not by homoiogenetic induction.

When single-cell assays are used, another import-ant characteristic of induction responses becomesevident. This is the sharp demarcation betweengroups of cells which do or do not respond toinduction. Many induced tissues in a vertebratecontain coherent blocks of cells of the same type,such as muscle, notochord and the neural tube. Onemight expect such groups of cells to be formed in partby the effects of localized cell proliferation and cellmovement. However, when these processes are in-hibited as mentioned above, induced cells neverthe-less form a localized group with a sharp demarcationbetween cells that are positive or negative for themuscle marker antibody (Fig. 11). This seems mostsimply explained by a threshold effect such that anyindividual cell responds fully or not at all (Slack,1983, for a theoretical discussion). A demarcation ofresponse could also be enhanced by the communityeffect suggested above.

In conclusion, the localization of an inductiveresponse in those cases where it has been appropri-ately investigated may be most simply explained asfollows. First of all, responsiveness is restricted tocompetent cells. It is further limited by the rate ofdiffusion of inducing substances and by the proximity

of competent cells to its source. Therefore the onlycells to respond would be those near enough theinducing tissue to receive a concentration of inducerabove their threshold level of response during theircompetent life. The precise localization of the re-sponse in tissues such as muscle and nerve would beachieved by homoiogenetic induction and by a com-munity effect in which cells can respond to aninduction in a given way only if their neighbours areundergoing the same response.

(6) The nature of an inducer

Much of what we know about molecules involved ininduction concerns the nature of the inductive signal.We may first ask whether it is necessary for inducingand responding cells to make direct physical contactwith each other, or whether the inductive signal canbe conveyed by diffusible molecules. Since the orig-inal experiments of Grobstein (1956), numerous testshave been carried out with inducing and respondingtissues separated by a filter. Millipore filters havetortuous pores of somewhat variable size, and thismakes it hard to know whether the cytoplasmicprocesses often seen in the pores are continuous fromone side of the filter to the other. In contrast,nucleopore filters have straight pores of ratherconstant size (Wartiovaara, Lehtonen, Nordling &Saxen, 1972). In early work (e.g. Grobstein, 1956), it

296 J. B. Gurdon

Fig. 11. Induction in the absence of cell division. Animaland vegetal tissues from a Xenopus blastula were placedin contact, and the conjugate immediately transferred tosolid medium containing lOjUgml"1 cytochalasin B (CB).This drug inhibits cytoplasmic cell division, whilepermitting nuclear division. When control embryos hadreached the heart-beat stage, the conjugate was fixed,sectioned, and processed for binding of muscle-specificantibody (see Fig. 4). No cell division has taken placesince the start of induction, although control conjugates(without CB) would have increased their cell numberabout five times. Animal cells close to the inducingvegetal cells have undergone muscle gene activation andthere is a sharp demarcation between animal cells thathave responded to induction and those that have not.(A) Stained with 12/101 antibody; (B) stained withHoechst. White arrows indicate the junction betweenanimal and vegetal cells.

was assumed, perhaps correctly, that cytoplasmicprocesses did not reach right through a Milliporefilter, but nearly all recent work has used nucleoporeniters in which it can be seen when cytoplasmicpenetration has been achieved. It is found that thethicker the filter and the smaller the holes, the longerit takes for cytoplasmic penetration to occur. Inseveral very careful quantitative analyses, a clear

correlation has been established between the passageof cell processes right through the filter and aninductive response: the removal of the inducer beforefilter penetration eliminates induction and the induc-tive effect is stronger the greater the amount ofpenetration (e.g. Wartiovaara, Nordling, Lehtonen& Saxen, 1974; Meier & Hay, 1975; Saxen & Leh-tonen, 1978). This conclusion suggests that cellularcontact between inducing and responding tissues maybe required for successful induction. However, it isalso compatible with the view that the signal isconveyed by a network of extracellular materials(Grobstein, 1967) or even by labile diffusible mol-ecules. Most inductions where contact seems necess-ary are of the permissive type; these include mousekidney tubule induction by mesenchyme and chickcorneal induction by lens (Wartiovaara et al. 197'4;Meier & Hay, 1975; Saxen & Lehtonen, 1978), andmay involve extracellular materials (section 2). It issomewhat easier to interpret those inductions wherecell contact is not required. These include instructiveinductions such as chick lens by the optic vesicle(Karkinen, 1978) and Xenopus mesoderm by vegetalcells (Grunz & Tacke, 1986), as well as permissiveones such as the neuralization of newt ectoderm(Toivonen, Tarin & Saxen, 1976). In these cases, it isclear from the use of filters, as well as from theexistence of soluble inducing extracts, that an induc-tive signal can be conveyed by diffusible factors.

The fact that some inductions can be transmittedwithout physical contact does not mean that this ishow it happens in normal development. Duringmesoderm induction in normal Xenopus blastulae,gap junctions are known to link Xenopus blastulacells (Palmer & Slack, 1970; Regen & Steinhardt,1986). However, mesodermal induction causesmuscle gene activation to a quantitatively normalextent in embryonic cells which are in Ca2+-/Mg2+-free medium (i.e. in close proximity but not stablyattached to each other) (Gurdon, Brennan, Fairman& Mohun, 1984; Sargent, Jamrich & Dawid, 1986), aswell as in embryos in which gap junction communi-cation has been suppressed by anti-gap-junction anti-bodies (Warner & Gurdon, 1987). The conclusionthat gap junctions or other stable cell contacts are notrequired in induction is at present applicable to theonly case where this has been precisely tested. Pre-vious experiments with the same anti-gap-junctionantibodies gave larval defects that were consistentwith the idea that they resulted from impaired neuralinduction (Warner, Guthrie & Gilula, 1984); how-ever, as the authors point out, the injected antibodiesbegin to lose their effect by the neural stage and theimportance of gap junctions for neural induction is asuggested rather than proven conclusion.


Attempts to identify inducing substances have hada long history of slow progress. The difficulty ofworking with the more permissive inductions, such asthe neuralization of newt ectoderm, soon becameevident when it was found that almost any disturbingcondition will cause induction (see Introduction).Much the most useful information has come from theuse of heterogenous inducers. (Heterogenous meansof different origin, i.e. not derived from the naturalinducing tissue. Heterogeneous, sometimes used inerror, means of diverse composition.) The great meritof heterogenous inducers is that they are available infairly large quantities, that they have strong effectsand therefore that they are suitable for biochemicalpurification. Very largely through the work of Tiede-mann's laboratory over several decades, it seemsclear that some inducers are macromolecular andcontain proteins (Table 1). Recently Smith (1987) hasdiscovered a very convenient source of heterogenousmesodermal inducer, from the culture medium of aXenopus cell line, derived originally from a meta-morphosing tadpole. This seems to be a heat-stableprotein of about 16 x 103 MT (16K), and since it has thegreat merit of being soluble, there appear to be verygood prospects for the eventual purification of this

material. However, as yet no macromolecular in-ducer molecule has been fully purified or identified.There will always be concern that a heterogenousinducer may be substantially different from naturalinducers, and some effort has been made to purifynatural inducers from amphibian blastulae (Faul-haber, 1972). But the source of material is limiting,the effect of it weak during assay, and direct purifi-cation at present seems a forbidding task. It must beborne in mind that a few pure substances have stronginductive effects. For example, LiCl has long beenknown to have vegetalizing effects on sea urchin andamphibian embryos (Masui, 1961; Slack, 1983;Davidson, 1986). However, recent studies show thatLi+ can also have dorsalizing and neuralizing influ-ences (Kao, Masui & Elinson, 1986; Breckenridge,Warren & Warner, 1987). It seems that the effects ofLi+ on development may well include secondaryconsequences of a generally disruptive effect on cellinteractions, and may not give useful insight into anynormal inductive process.

While discussing inducers, we should comment onan apparent characteristic that different concen-trations or exposure times of the same inducer canhave dramatically different developmental effects.For example, a 10-fold increase in the concentration

Table 1. Many different substances act as embryonic inducers

Inducer Source Assay Composition Reference

Vegetalizing

Vegetalizing

Vegetalizing

Vegetalizing

Neuralizing

NeuralizingNeuralizing

Cartilage enhancing

Pure substances

Lithium ions (vegetalizingand other effects - seetext)

Concanavalin A(neuralizing)

Guinea pig bone marrow(ethanol extract)

9-13 day chick embryo

Carp swimbladder(ethanol pellet)

Xenopus cultured cellmedium (soluble)

Guinea pig liver (ethanolpellet)

HeLa cells (ethanol pellet)Xenopus eggs and embryos

10 day chick cartilage

Triturus gastrula implant

Triturus gastrula orectoderm sandwichimplants

Triturus gastrula ectodermsandwich implant

Xenopus blastula ectoderm

Triturus gastrula implant

Triturus gastrula implantTriturus gastrula implant

Chick somite chondroitinsulphate synthesis

Amphibian embryos (manyspecies)

Newt (Cynops) gastrulaectoderm

30 K protein

16 K heat stable protein

Toivonen (1953)

Born et al. (1972)Schwartz et al. (1981)

Kawakami et al. (1976, 1977)

Smith (1987)

Toivonen & Saxen (1955)

Saxen & Toivonen (1958)Protein (from microsomes Faulhaber (1972)

and yolk)Chondromucoprotein

LiCl

Purified ConA

Janeczek et al. (1984)Kosher et al. (1973)

Masui (1961)Kao et al. (1986)

Takata et al. (1981)

Inducers effective in early embryos are classified only as vegetalizing or neuralizing. Vegetalizing includes 'mesoderm-inducing', on the grounds thatthe formation of mesodermal structures such as muscle and notochord probably results from the induction of vegetal cells which themselves induceanimal cells into mesodermal cells. 'Neuralizing' factors include those described as inducing archencephalon (forebrain), deuterencephalon (hind-brain),and spinal cord, since these different morphological structures can be induced by different concentrations and exposure times of the same extract. It isassumed, as explained in the text, that neuralization of test embryos is caused by a direct neuralizing factor, and not by induced mesoderm cells whichin turn induce ectoderm into neural structures. The chick-cartilage-enhancing factor acts permissively, increasing a substantial background (uninduced)level of chondroitin synthesis.

298 J. B. Gurdon

of vegetalizing inducer or in the time of exposure toit, can cause Xenopus ectoderm to form substantialamounts of muscle and nerve; by comparison, bloodis formed at low doses or epidermis with no inducer atall (Grunz, 1983). This might suggest that differentconcentrations of the same factor can activate differ-ent genes in equivalent cells. However, this effectcould well be explained if higher concentrations ofinducer 'vegetalize' (make into vegetal cells) increas-ing numbers of animal or ectoderm cells, and if thesevegetalized cells cause secondary inductions, accord-ing to the scheme shown in Fig. 1A, as suggested byMinuth & Grunz (1980). Another simpler interpret-ation, and one to be preferred until disproved, is thatthe heterogenous inducer used in this experimentcontains several factors which independently inducedifferent cell types, a conclusion supported by someinducer mixing experiments of Asahi, Born, Tiede-mann & Tiedemann (1979).

This leads us to consider an extreme proposition,namely that any one cell at a particular stage indevelopment can respond to only one kind of in-ducer; the concept is that a cell has only two options,to be induced or not, and that if induced it canrespond in only one way. It would have to besupposed that amphibian early gastrula ectodermcells though superficially similar are in fact hetero-geneous, some being able to respond to vegetal cellsor vegetalizing inducers by becoming mesoderm, andothers to mesoderm cells or neuralizing inducers byforming nerve. It might also be supposed that a strongvegetal inducer would vegetalize a sufficient numberof blastula animal cells sufficiently quickly for them inturn to induce some of the remaining animal cells tobecome mesoderm, which could subsequently inducenerve. It might therefore be that neuralizing inducersdiffer from vegetalizing inducers only in their fasteror stronger action, enabling secondary inductions toproceed as far as neural differentiation. The strongestargument against such sequential effects of one induc-tion is that a primary inducer is unlikely to be able tohave its effect fast enough for secondary inductions totake place before competence, which is strictly lim-ited in duration, has been lost. Another is that lowdoses or short exposure times of a neural inducerwould be expected to give mesodermal inductions,but this is not observed (Saxen & Toivonen, 1962).However, these points do not argue against a heter-ogeneous population of responsive cells in the earlygastrula. A decisive test of this important point willprobably require not only early molecular markers(section 2), but also single cell inductions (section 5).

(7) Early steps in the inductive response

A molecular understanding of embryonic inductionrequires a knowledge of events that lead from theappearance of inducer at a responding cell's surfaceto the earliest differentiation event, such as geneactivation.

Do inducers need to enter cells? In a few cases itseems clear that they can act at the surface of therecipient cell and do not need to enter, as do steroidhormones, to have their effect. This is true for a crudechick embryo fraction as well as for concanavalin A,both of which induce neural differentiation in newtembryos, when covalently bound to Sepharose beadswhich do not enter cells (Tiedemann & Born, 1978;Takata et al. 1981). In contrast, a vegetalizing inducerof the same chick origin and assayed in the same wayloses activity if bound to Sepharose, though it can besubsequently recovered in active form (Born, Grunz,Tiedemann & Tiedemann, 1980).

Do initial responses include changes in free Ca2+

and cyclic AMP content, protein phosphorylation,oncogene activation, and other events often associ-ated with major changes in cell activity? Pictet &Rutter (1977) found that neither cAMP nor cGMPderivatives can substitute in pancreatic epitheliumfor the mesenchyme-inducing factor, an effect whichwould have been expected if an increase in thesesubstances was a direct effect of induction. Similarresults have been obtained by Grunz & Tiedemann(1977). The strong effect of LiCl on developmentmight suggest the involvement of inositol triphos-phates, since Li+ is a specific inhibitor of inositolmetabolism (Berridge, 1986); but if Li+ has a gener-ally disruptive effect on development (p. 297),inositol metabolism might well be disturbed forreasons unconnected with induction. It will require aninduction system in which cell proliferation eitherdoes not accompany induction, or one in which it canbe suppressed, before a useful investigation of earlypostinduction events can be undertaken (as discussedabove).

It is important to know whether protein synthesisis required during early inductive events. A shortexposure to cycloheximide is sufficient to largelysuppress protein synthesis in embryonic amphibiantissue, in a way that is nontoxic and reversible.Grunz (1970) established that cycloheximide treat-ment causes a prolongation of the period ofcompetence, suggesting that the termination ofcompetence may be actively induced by mRNAtranslation. Recently it has been found, by transcriptand two-dimensional gel protein analyses, that pro-tein synthesis, and therefore mRNA translation, isabsolutely required during the first two thirds (6 h) ofthe vegetal induction of Xenopus blastula tissue into


muscle (Cascio & Gurdon, 1987), but not during thelast 3 h. It is not known whether transcription is alsorequired during this induction period, or whetherthe mRNA required is preformed (and therefore inthis case maternal). However, these results clearlysuggest the value of screening cDNA libraries forinduction-related clones (p. 301).

Much recent work concerned with gene activationhas been successfully directed towards the identifi-cation of cis-acting sequences in front of genes,presumed to interact with gene-specific proteinsor RNAs. Success has recently been achieved inobtaining correct tissue-specific expression of muscleactin genes injected as DNA into Xenopus eggs,and the same transferred genes can be activated inanimal cells by vegetal induction (Wilson, Cross &Woodland, 1986; Mohun, Garrett & Gurdon, 1986;Fig. 12). Mohun etal. (1986) have also found that thesequences needed for gene activation by induction liewithin 400 bases upstream from the promoter of thisgene. This seems an encouraging start on an analysisthat aims to work backwards from an early inductiveresponse (actin gene activation). The hope will be to

Acetylated

forms of 14C-

chloramphenicol

14Gchloramphenicol*» # #

aa>oCD

Eo

>,_ca

AN VEG CONJFig. 12. A rapid gene transfer assay for DNA sequencesinvolved in a response to embryonic induction. AXenopus cardiac actin gene promoter (3 Kb) has beenfused to a bacterial chloramphenicol acetyl transferase(CAT) gene. DNA of this kind is injected into fertilizedeggs of Xenopus from which animal and vegetal regionsare isolated at the blastula stage and cultured separately(AN, VEG) or after conjugation (CONJ). The embryopieces are cultured overnight until controls have reachedthe neurula stage, when extracts are assayed for CATactivity. Acetylated [14C]chloramphenicol shows that theCAT gene has been expressed by activation of themuscle-specific cardiac actin gene promoter. The positivesignal in the conjugates results from the activation of theactin gene promoter in animal cells by induction fromvegetal cells. (From Mohun et al. 1986.)

find factors that bind to this upstream region of theinduced actin gene; these might be the products of agene whose translation occurs during the cyclohexi-mide-sensitive phase at the start of induction. Sincethe whole induction process is, in this case, completedin 7-9 h, this is about the timing expected if inductionwere to cause the transcription and translation of agene or genes, whose products directly activate theactin gene.

(8) Comparison with other cell signallingsystems

Of all the cell signalling systems that operate betweencells, embryonic induction is the least understood atthe molecular level. In most other cases the ligand,and often the receptor, have been well characterizedand even sequenced at the nucleic acid or proteinlevels. It is therefore useful to compare the biologicalproperties of embryonic induction with other cellinteractions. How similar is embryonic induction tothese other systems?

Table 2 summarizes most of the conclusions wehave reached in the preceding discussion. In severalrespects embryonic induction appears to differ funda-mentally from all other known types of cell interac-tion. First, induction appears to be the only case inwhich the responsive or competent state disappearssoon after the reaction is completed. In others, thepossession of a functional receptor is an indefiniteproperty of responding cells once they have acquiredit. One exception to this generalization, and thereforeresemblance to embryonic competence, exists in theaction of a lepidopteran eclosion hormone, a neuro-secretory peptide; the central nervous system, onwhich this hormone acts, is competent to respond foronly a few hours, which coincide with the appearanceand disappearance of two phosphorylatable proteins(Morton & Truman, 1986). A second differencebetween induction and other systems, is that thetiming of an inductive response is determined byproperties of the responding tissue (section 4), andnot as in other systems by the time when the ligand,such as hormone, growth factor, neurotransmitter,etc. reaches responsive cells. A third difference is thatseveral totally unrelated chemical substances, such asLi+ and a guinea-pig bone marrow protein, can havethe same specific effect (such as mesodermal differen-tiation) emphasizing the large extent to which induc-tions seem to promote a response to which cells arealready committed. In all these respects, inductiondiffers from other cell interactions in the overridingimportance of the responsive or competent state. Itseems as if responding cells undergo a series ofendogenous processes which require, at any timewithin quite wide limits, the contribution of inducer

300 /. B. Gurdon

molecules, which can cause a switch from the unin-duced to one of a small number of induced directionsof differentiation. When we consider the purposeof embryonic induction in development, differencesfrom other cell-signalling systems make sense. Em-bryonic induction brings about an irreversible differ-entiation, or more precisely determination, of cells inone part of an embryo. The exact position and time ofthe induction needs to be precisely coordinated withmany other events in development. In contrast, mosthormones, growth factors and neurotransmitters are

required to give reversible effects, often timed inrelation to an effect of the external environment onan organism.

In a few respects embryonic induction sharescharacteristics with certain kinds of cell interaction,but not others. Hormones are effective over very longdistances, being conveyed in the blood stream be-tween their origin in an endocrine organ and theresponding tissue; on the other hand, normal in-ducers are effective only over short distances of a fewcell diameters, and often affect only adjacent cells

Table 2. Embryonic induction compared with other cell interactions

Interacting system

Embryonic induction

e.g. Xenopus embryo

Hormones1'2

e.g. Oestradiol (steroid)

Erythropoietin (46 Kglycoprotein)

Lymphokines3

e.g. Interleukin-2 (15 Kprotein)

Growth factors4

e.g. Nerve GF(2xll8aa)

Epidermal GF (53 aa)

Prostaglandins5

Thromboxane,PGE2, PGI2 (smallfatty acid derivatives)

Neurotransmitters6

Acetylcholine, enkephalinsand endorphins (shortpeptides)

Morphogens7

Hydra head activator(11 aa)

Hydra head inhibitor(<500Mr)

Slime mould DIF andcAMP(<300Mr)

aa, amino acids; K, xlO3Mr

References1 Standard textbooks.

Main biologicaleffect

Mesoderm and nervedifferentiation

Increased gene activityin oviduct

Proliferation of earlyerythrocytes in bonemarrow

Cell division anddifferentiation ofT cells

Survival and growth ofneurones

Stimulates cell division

Blood plateletaggregation

Modulation of nerveconduction

Control of headregeneration

Control of headregeneration

Spore or stalkdifferentiation

Duration ofresponsive

state

Terminates rapidly

Long term

Long term (in earlyerythrocytes)

Long term

Long term

Long term

Long term

Long term

Long term

Long term

Long term

Responsetimed by

Responding cells

Hormone release

Hormone release

Lymphokine release

NGF release anddistance

EGF release anddistance

Prostaglandinrelease

Transmitter release

Head activator andinhibitor release

Head activator andinhibitor release

Morphogen release

Time frominduction to

main response

Hours or days

Hours

Days

Hours

Hours or days

Hours

Seconds or minutes

Microseconds tomilliseconds

Hours or days

Hours or days

Seconds or minutes

Distance frominducer source toresponding cells

Few cell diameters

Blood stream (fromovary)

Blood stream (fromkidney)

Few cell diameters(from T cells)

Few cell diameters

Few cell diameters

Few cells (unstable)

Nanometers tomicrons (rapiddegradation)

Activator (few cells)

Inhibitor (manycells)

Few cells (rapidlydegraded)

2Goldwasser (1975); Metcalf (1981).3Farrarefa/. (1982)."Levi-Montalcini & Calissano (1979); Yankner & Shooter (1982); Carpenter (1985).5 Samuelson et al. (1975).6Iversen (1984).7Schaller et al. (1986); Gross et al. (1981); Loomis (1982).


less than 1 fira away. Neurotransmitters are effectiveover the same short distances as inducers, but differin having extremely rapid effects, of the order ofmilliseconds, compared to inducers whose main ef-fects are seen hours or days later. The single mostimportant characteristic of embryonic inducers is thatthey cause major changes in the direction of celldifferentiation. In this respect, they differ fromgrowth factors, whose effects are usually to enhancethe proliferation of an already determined cell type.The nerve growth factor does not initiate cell div-ision, nor does it cause a change in the direction ofcell differentiation; it stimulates the outgrowth ofaxons from cells already committed to nerve differen-tiation (Levi-Montalcini & Calissano, 1979). Thereare several other 'local mediators' of cell interactions,such as prostaglandins, histamine, enkephalins, etc.which share many properties with inducers, but whichdo not cause substantial changes in cell differen-tiation.

(9) Prospects for further analysis at themolecular level

It has commonly turned out, in molecular biology,that progress on a difficult problem has been achievedby selecting a particularly favourable biological sys-tem for analysis. A good example is the identificationof the nerve growth factor. This substance is pro-duced in very small amounts by many tissues buthappens to be present in the submaxillary gland at aconcentration 103 times higher than elsewhere inthe mouse, thereby permitting its initial purification(Cohen, 1960) and the subsequent characterization ofits receptor. It may therefore be helpful to considerwhich kinds of inductive system are likely to be mostuseful, and what attributes such a system shouldhave, for future molecular analysis.

At the cellular level, a great simplification would beachieved if an experimental induction system couldbe made to work at the level of single cells cultured inisolation, as has been done for terminal differen-tiation of astrocyte and oligodendrocyte nerve cells(Temple & Raff, 1985). The fact that inductivesystems so far analysed appear to depend on aninteraction between multiple responding cells compli-cates analysis (p. 294). Single cell assays are, ofcourse, impossible for inductions recognized by themorphogenetic arrangement of multiple cells, butthese kinds of induction processes are likely to behardest to analyse in molecular terms. A secondsubstantial simplification could be achieved at thecellular level, if cell division, cell rearrangement andeventually DNA synthesis can be eliminated, asdiscussed in section 5.

A particularly successful route towards the molecu-lar analysis of any complex biological process is toobtain mutants that inactivate, one at a time, thevarious gene products involved. Once a mutation ofinterest has been secured, it is a matter of time, inDrosophila, before the gene sequence and codingcapacity can be determined, though this does notnecessarily reveal its function. The organisms mostfavourable for genetic analysis, namely Drosophilaand the nematode Caenorhabditis, are also thosewhose development is least obviously affected byinductions, and there is at present no experimentalinductive system with isolated tissues available inthese species, like those discussed for vertebrates.Nevertheless it seems certain that interesting mutantsaffecting cell interactions will be found and analysed.A good example is the nematode mutant lin 12(Greenwald, 1985), in which the DNA sequenceencodes multiple copies of a cysteine-rich 40-50amino acid peptide; this is present many times in thesequence of the precursor to the mammalian epider-mal growth factor, and once in the mature growthfactor itself. A Drosophila mutation that has turnedout to be of special interest is Notch, the developmen-tal effects of which are complex but suggest theinvolvement in cell interactions (Akam, 1986). TheDNA sequence of Notch also reveals many repeats ofa sequence capable of coding the same cysteine-richpeptide found in epidermal growth factor genes(Wharton, Johansen & Artavanis-Tsakonas, 1985).We should bear in mind the common observation,especially from cell hybrid experiments, that cellcomponents involved in gene activation seem to beconserved between distant species, but are quitedifferent from one tissue to another (Ringertz &Savage, 1976; Blau, Chiu & Webster, 1983). It iscommonly found that a sequence from one organism,such as the Drosophila homeobox, can be used to findsimilar sequences (though not necessarily function-ally equivalent) in other species, such as mice andfrogs. Putting these observations together, it may wellbe rewarding to screen cDNA libraries prepared fromdevelopmental stages where inductive processes takeplace, using low stringency hybridization with probesfrom other organisms for receptors of known growthfactors, hormones, etc. Using subtracted cDNAlibraries (Sargent & Dawid, 1983), or subtractivehybridization, it should be feasible to isolate cDNAsfor mRNAs present at a frequency of 10~4 in the totalpoly(A)+ RNA of an embryo. This would corre-spond, in a Xenopus blastula, to about 100 moleculesof one type of mRNA per cell, a minimum concen-tration expected for a message whose product needsto be synthesized in substantial amounts within only afew hours.

302 /. B. Gurdon

(10) Conclusion

The overall impression conveyed by this survey ofembryonic inductions is that these processes are verycomplex, probably consisting of several reciprocalinteractions between inducing and responding tis-sues, each step enhancing, little by little, progressiontowards the eventual differentiation. Most inductionsseem to take place as follows. Both competence andinducing ability are acquired by restricted popu-lations of cells for a limited time. The proximity of thefirst, apparently instructive, inducer further restrictsthe number of cells within the competent populationthat will respond. Especially with inductions that takeplace late in development, several further, usuallypermissive, inductions from other inducing tissuesenhance the initial response. Tissue structure may befinalized by interactions among responding cells witheach other.

In retrospect it is not surprising that so many stepsshould be required for the reliable formation ofsomething as complicated, as an embryo. Any manu-facturing process in which a complicated machine isassembled requires controls to ensure the coordinateoperation of different steps. If the formation of anembryo were to take place without frequent interac-tions between different regions, errors in the finalproduct would often arise.

I believe that the greatest obstacle to the molecularanalysis of induction over many decades may havebeen the imprecision and late appearance of theassays used, which often depend on morphologicalassessment many days after the inductive responsehas started. It may also have been necessary, faute demieux, to concentrate on identifying inducer mol-ecules; but compared to ligands for other cell interac-tions, embryonic inducers seem less specific, sincethey can be substituted for by other substances, andtherefore harder to purify. For embryonic inductionto be accessible to the powerful methods of molecularanalysis now available, it seems essential to use, as anassay, a single early response, such as the expressionof one gene. Nucleic acid technology has probablynow reached a sufficient level of precision and ef-ficiency of operation to be usefully applied to theanalysis of inductive responses, working from theresponse backwards, rather than from the inducerforwards.

I am indebted to J. C. Smith, A. E. Warner, C. C. Wylie,J. M. W. Slack and C. Sharpe for invaluable comments onparts of the manuscript, and to Barbara Rodbard for help inits preparation.

References

AKAM, M. E. (1986). Mediators of cell communication?Nature, Lond. 319, 447-448.

AKERS, R. M., PHILLIPS, C. R. & WESSELLS, N. K.

(1986). Expression of an epidermal antigen used tostudy tissue induction in the early Xenopus laevisembryo. Science 231, 613-616.

ASAHI, K., BORN, J., TIEDEMANN, H. & TIEDEMANN, H.(1979). Formation of mesodermal pattern by secondaryinducing interactions. Wilhelm Roux' Arch, devl Biol.187, 231-244.

ASASHIMA, M. & GRUNZ, H. (1983). Effects of inducerson inner and outer gastrula ectoderm layers of Xenopuslaevis. Differentiation 23, 206-212.

BANVILLE, D. & WILLIAMS, J. G. (1985a). The pattern ofexpression of the Xenopus laevis tadpole a-globin geneand the amino acid sequence of the three majortadpole a"-globin polypeptides. Nucleic Acids Res. 13,5407-5421.

BANVILLE, D. & WILLIAMS, J. G. (1985b). Developmentalchanges in the pattern of larval ^-globin geneexpression in Xenopus laevis. J. molec. Biol. 184,611-620.

BARTH, L. G. (1968). Neural differentiation withoutorganizer. J. exp. Zool. 87, 371-383.

BERRIDGE, M. J. (1986). Cell signalling throughphospholipid metabolism. /. Cell Sci., Supplement 4,137-153.

BLACKSHAW, S. E. & WARNER, A. E. (1976). Alterationsin resting membrane properties during neural platestages of development of the nervous system./. Physiol, Lond. 255, 231-247.

BLAU, H. M., CHIU, C.-P. & WEBSTER, C. (1983).Cytoplasmic activation of human nuclear genes instable heterokaryons. Cell 32, 1171-1180.

BORN, J., GEITHE, H.-P., TIEDEMANN, H., TIEDEMANN, H.& KOCHER-BECKER, U. (1972). Isolation of avegetalizing inducing factor. Hoppe-Seyler's Z. Physiol.Chem. 353, 1075-1084.

BORN, J., GRUNZ, H., TIEDEMANN, H. & TIEDEMANN, H.

(1980). Biological activity of the vegetalizing factor:decrease after coupling to polysaccharide matrix andenzymatic recovery of active factor. Wilhelm Roux'Arch, devl Biol. 189, 47-56.

BOTERENBROOD, E. C. & NIEUWKOOP, P. D. (1973). Theformation of the mesoderm in Urodelean Amphibians.V. Its regional induction by the endoderm. WilhelmRoux' Arch. EntwMech. Org. 173, 319-332.

BRECKENRIDGE, L. J., WARREN, R. L. & WARNER, A. E.

(1987). Lithium inhibits morphogenesis of the nervoussystem, but not neuronal differentiation in Xenopuslaevis. Development 99, 353-370.

CARPENTER, G. (1985). Epidermal growth factor: biologyand receptor metabolism. J. Cell Sci. Supplement 3,1-10.

CASCIO, S. & GURDON, J. B. (1987). The initiation of newgene transcription during Xenopus gastrulation requiresimmediately preceding protein synthesis. Development,submitted.


COHEN, S. (1960). Purification of a nerve-growthpromoting protein from the mouse salivary gland andits neuro-cytotoxic antiserum. Proc. natn. Acad. Sci.U.S.A. 46, 302-311.

DALE, L., SMITH, J. C. & SLACK, J. M. W. (1985).Mesoderm induction in Xenopus laevis: a quantitativestudy using a cell lineage label and tissue-specificantibodies. J. Embryol. exp. Morph. 89, 289-312.

DAVIDSON, E. H. (1986). Gene Activity in EarlyDevelopment. 3rd Edn. New York: Academic Press.

DIENSTMAN, S. R. & HOLTZER, H. (1975). Myogenesis: acell lineage interpretation. In Cell Cycle and CellDifferentiation (ed. J. Reinert & H. Holtzer), pp. 1-25.New York: Springer.

DUPRAT, A. M., KAN, P., GUALANDRIS, L., FOULQUIER, F.& MARTY, J. (1985). Neural induction: embryonicdetermination elicits full expression of specific neuronaltraits. J. Embryol. exp. Morph. 89 Supplement,167-183.

DWORKIN-RASTL, E., KELLEY, D. B. & DWORKIN, M. B.(1986). Localization of specific mRNA sequences inXenopus laevis embryos by in situ hybridization.J. Embryol. exp. Morph. 91, 153-168.

FARRAR, J.-J., BENJAMIN, W. R., HILFIKER, M. L.,HOWARD, M., FARRAR, W. L. & FULLER-FARRAR, J.(1982). The biochemistry, biology and role ofinterleukin 2 in the induction of cytotoxic T cell andantibody-forming B cell responses. Immunol. Rev. 63,129-166.

FAULHABER, I. (1972). Die Induktionsleistungsubzellularer Fraktionen aus der Gastrula von Xenopuslaevis. Wilhelm Roux' Arch. EntwMech. Org. 171,87-108.

GALLIN, W. J., CHUONG, C.-M., FINKEL, L. H. &EDELMAN, G. M. (1986). Antibodies to liver celladhesion molecule perturb inductive interactions andalter feather pattern and structure. Proc. natn. Acad.Sci. U.S.A. 83, 8235-8239.

GIMLICH, R. L. (1986). Acquisition of developmentalautonomy in the equatorial region of the Xenopusembryo. Devi Biol. 115, 340-352.

GIMLICH, R. L. & GERHART, J. C. (1984). Early cellularinteractions promote embryonic axis formation inXenopus laevis. Devi Biol. 104, 117-130.

GODSAVE, S. F., ANDERTON, B. H. & WYLIE, C. C.(1986). The appearance and distribution ofintermediate filament proteins during differentiation ofthe central nervous system, skin, and notochord ofXenopus laevis. J. Embryol. exp. Morph. 97, 201-223.

GOLDWASSER, E. (1975). Erythropoietin and thedifferentiation of red blood cells. Fedn Proc. 34,2285-2292.

GRAINGER, R. M., HAZARD-LEONARDS, R. M., SAMAHA,F., HOUGAN, L. M., LESK, M. R. & THOMSEN, G. H.(1983). Is hypomethylation linked to activation ofdelta-crystallin genes during lens development? Nature,Lond. 306, 88-91.

GREENWALD, I. (1985). Lin-12, a nematode homeoticgene, is homologous to a set of mammalian proteinsthat includes epidermal growth factor. Cell 43, 583-590.

GROBSTEIN, C. (1956). Transfilter induction of tubules inmouse metanephrogenic mesenchyme. Expl Cell Res.10, 424-440.

GROBSTEIN, C. (1967). Mechanisms of organogenetictissue interaction. Nat. Cancer Inst. Monogr. 26,279-299.

GROSS, J. D., TOWN, C. D., BROOKMAN, J.-J., JERMYN, K.A., PEACEY, M. J. & KAY, R. R. (1981). Cellpatterning in Dictyostelium. Phil. Trans. Roy. Soc. B295, 497-508.

GRUNZ, H. (1970). Abhangigkeit der Kompetenz desAmphibien-Ektoderms von der Proteinsynthese.Wilhelm Roux' Arch. EntwMech. Org. 165, 91-102.

GRUNZ, H. (1979). Change of the differentiation patternof amphibian ectoderm after the increase of the initialcell mass. Wilhelm Roux' Arch, devl Biol. 187, 49-57.

GRUNZ, H. (1983). Change in the differentiation patternof Xenopus laevis ectoderm by variation of theincubation time and concentration of vegetalizingfactor. Wilhelm Roux' Arch, devl Biol. 192, 130-137.

GRUNZ, H. & TIEDEMANN, H. (1977). Influence of cyclicnucleotides on amphibian ectoderm. Wilhelm Roux'Arch, devl Biol. 181, 261-265.

GRUNZ, H. & TACKE, L. (1986). The inducing capacity ofthe presumptive endoderm of Xenopus laevis studied bytransfilter experiments. Wilhelm Roux' Arch, devl Biol.195, 467-473.

GURDON, J. B., BRENNAN, S., FAIRMAN, S. & MOHUN, T.J. (1984). Transcription of muscle-specific actin genesin early Xenopus development: nuclear transplantationand cell dissociation. Ce//38, 691-700.

GURDON, J. B., FAIRMAN, S., MOHUN, T. J. & BRENNAN,S. (1985). Activation of muscle-specific actin genes inXenopus development by an induction between animaland vegetal cells of a blastula. Cell 41, 913-922.

GURDON, J. B. & FAIRMAN, S. (1986). Muscle geneactivation by induction and the nonrequirement for celldivision. /. Embryol. exp. Morph. 97 Supplement,75-84.

HEASMAN, J., WYLIE, C. C , HAUSEN, P. & SMITH, J. C.(1984). Fates and states of determination of singlevegetal pole blastomeres of X laevis. Cell 37, 185-194.

HEASMAN, J., SNAPE, A., SMITH, J. C. & WYLIE, C. C.(1985). Single cell analysis of commitment in earlyembryogenesis. J. Embryol. exp. Morph. 89Supplement, 297-316.

HEASMAN, J., SNAPE, A., SMITH, J. C. & WYLIE, C. C.(1986). The nature of developmental restrictions inXenopus laevis embryos. J. Embryol. exp. Morph. 97Supplement, 65-73.

HOLTFRETER, J. (1934). Der Einfluss thermischer,mechanischer und chemischer Eingriffe auf dieInduzierfahigkeit von Triton-keimteilen. Wilhelm Roux'Arch. EntwMech. Org. 132, 225-306.

HOLTFRETER, J. (1947). Neural induction in explants whichhave passed through a sublethal cytolysis. J. exp. Zool.106, 197-222.

HORST, J. T. (1948). Differenzierungs undInduktionsleistungen verschiedener Abschnitte derMedullarplatte und des Urdarmdaches von Triton im

304 /. B. Gurdon

Kombinat. Wilhelm Roux' Arch. EntwMech. Org. 143,275-303.

HORSTADIUS, S. (1973). Experimental Embryology ofEchinoderms. Oxford: Clarendon Press.

IVERSEN, L. L. (1984). Amino acids and peptides: fast andslow chemical signals in the nervous system? Proc. Roy.Soc. B 221, 245-260.

JACOBSON, A. (1966). Inductive processes in embryonicdevelopment. Science 152, 25-34.

JACOBSON, M. & RUTISHAUSER, U. (1986). Induction ofneural cell adhesion molecule (NCAM) in Xenopusembryos. Devi Biol. 116, 524-531.

JANECZEK, J., JOHN, M., BORN, J., TIEDEMANN, H. &TIEDEMANN, H. (1984). Inducing activity of subcellularfractions from amphibian embryos. Wilhelm Roux'Arch, devl Biol. 193, 1-12.

JONAS, E., SARGENT, T. D. & DAWID, I. B. (1985).Epidermal keratin gene expressed in embryos ofXenopus laevis. Proc. natn. Acad. Sci. U.S.A. 82,5413-5417.

JONES, E. A. & WOODLAND, H. R. (1986). Developmentof the ectoderm in Xenopus: tissue specification andthe role of cell association and division. Cell 44,345-355.

JONES, K. W. & ELSDALE, T. R. (1963). The culture ofsmall aggregates of amphibian embryonic cells in vitro.J. Embryol. exp. Morph. 11, 135-154.

KAO, K. R., MASUI, Y. & ELINSON, R. P. (1986).Lithium-induced respecification of pattern in Xenopuslaevis embryos. Nature, Lond. 322, 371-373.

KARKINEN-JAASKELAINEN, M. (1978). Transfilter lensinduction in avian embryo. Differentiation 12, 31-37.

KAWAKAMI, I. (1976). Fish swimbladder: an excellentmesodermal inductor in primary embryonic induction.J. Embryol. exp. Morph. 36, 315-320.

KAWAKAMI, I., NODA, S., KURIHARA, K. & OKUMA, K.(1977). Vegetalizing factor extracted from the fishswimbladder and tested on presumptive ectoderm ofTriturus embryos. Wilhelm Roux' Arch, devl Biol. 182,1-7.

KINTNER, C. R. & BROCKES, J. P. (1984). Monoclonalantibodies identify blastemal cells derived fromdedifferentiating muscle in newt limb regeneration.Nature, Lond. 308, 67-69.

KINTNER, C. R. & MELTON, D. A. (1987). Expression ofXenopus N-CAM RNA in ectoderm as an earlyresponse to neural induction. Development 99, 311-320.

KONDOH, H. & OKADA, T. S. (1986). Dual regulation ofexpression of exogenous delta-crystallin gene inmammalian cells: a search for molecular background ofinstability in differentiation. Current Top. devl Biol. 20,153-164.

KOSHER, R. A., LASH, J. W. & MINOR, R. R. (1973).

Environmental enhancement of in vitro chondrogenesis.IV. Stimulation of somite chondrogenesis by exogenouschondromucoprotein. Devl Biol. 35, 210-220.

KURIHARA, K. & SASAKI, N. (1981). Transmission ofhomoiogenetic induction in presumptive ectoderm ofnewt embryo. Devl. Growth Diffn 23, 361-369.

LE DOUARIN, N. (1964). Induction de l'endoderme pre-hepatique par le mesoderme de l'aire cardiaque chez

l'embryon de poulet. J. Embryol. exp. Morph. 12,651-664.

LEVI-MONTALCINI, R. & CALISSANO, P. (1979). The nerve-growth factor. Sci. Amer. 240(6), 44-53. .

LEWIS, W. H. (1904). Experimental studies on thedevelopment of the eye in Amphibia. I. On the originof the lens. Rana palustris. Amer. J. Anat. 3, 505-536.

LOOMIS, W. F. (1982). The Development of Dictyosteliumdiscoideum. New York: Academic Press.

LOPASHOV, G. (1935). Die Entwicklungsleistungen desGastrulamesoderms in Abhangigkeit vonVeranderungen seiner Masse. Biol. Zbl. 55, 606-615.

MANGOLD, O. VON (1933). Uber die Induktionsfahigkeitder Verschiedenen Bezirke der Neurula von Urodelen.Naturwissenschaften 21, 761-766.

MANGOLD, O. & SPEMANN, H. (1927). Uber Induktionvon Medullarplatte durch Medullarplatte im jiingerenKeim, ein Beispiel homoogenetischer oderassimilatorischer Induktion. Wilhelm Roux' Arch.EntwMech. Org. I l l , 341-422.

MASUI, Y. (1961). Mesodermal and endodermaldifferentiation of the presumptive ectoderm of Triturusgastrula through influence of lithium ion. Experientia17, 458-459.

MASUI, Y. & CLARKE, H. J. (1979). Oocyte maturation.Int. Rev. Cytol. 57, 185-282.

MEIER, S. & HAY, E. D. (1975). Stimulation of cornealdifferentiation by interaction between cell surface andextracellular matrix. I. Morphometric analysis oftransfilter "induction". /. Cell Biol. 66, 275-291.

METCALF, D. (1981). Haemopoietic colony stimulatingfactors. In Handbook of Experimental Pharmacology,vol. 57 (ed. R. Baserga), pp. 343-384. New York:Springer-Verlag.

MINUTH, M. & GRUNZ, H. (1980). The formation ofmesodermal derivatives after induction withvegetalizing factor depends on secondary cellinteractions. Cell Differentiation 9, 229-238.

MOHUN, T. J., BRENNAN, S., DATHAN, N., FAIRMAN, S. &GURDON, J. B. (1984). Cell-type specific activation ofactin genes in the early amphibian embryo. Nature,Lond. 311, 716-721.

MOHUN, T. J., GARRETT, N. & GURDON, J. B. (1986).Upstream sequences required for tissue-specificactivation of the cardiac actin gene in Xenopus laevisembryos. EMBO J. 5, 3185-3193.

MORTON, D. B. & TRUMAN, J. W. (1986). Substratephosphoprotein availability regulates eclosion hormonesensitivity in an insect CNS. Nature, Lond. 323,264-267.

MUCHMORE, W. B. (1957). Differentiation of the trunkmesoderm in Ambystoma maculatum. J. exp. Zool. 134,293-313.

NIEUWKOOP, P. D. et al. (1952). Activation andorganization of the central nervous system inAmphibians. /. exp. Zool. 120, 1-31.

PALMER, J. F. & SLACK, C. (1970). Some bio-electricparameters of early Xenopus embryos. /. Embryol. exp.Morph. 24, 535-553.

PICTET, R. L. & RUTTER, W. J. (1977). The molecularbasis of the mesenchymal-epithelial interactions in


pancreatic development. In Cell Interactions inDifferentiation (ed. M. Karkinen), pp. 339-350. Londonand New York: Academic Press.

RAWLES, M. E. (1963). Tissue interactions in scale andfeather development as studied in dermal-epidermalrecombinations. J. Embryol. exp. Morph. 11, 765-789.

REGEN, J. & STEINHARDT, R. (1986). Global properties ofXenopus blastulae are mediated by a high resistanceepithelial seal. Devi Biol. 113, 147-154.

RINGERTZ, N. R. & SAVAGE, R. E. (1976). Cell Hybrids.New York: Academic Press.

RUTTER, W. J., WESSELLS, N. K. & GROBSTEIN, C. (1964).Control of specific synthesis in the developingpancreas. J. natn. Cancer Inst. Monogr. 13, 51-65.

SAMUELSSON, B., GRANSTROM, E., GREEN, K., HAMBERG,M. & HAMMARSTROM, S. (1975). Prostaglandins. A.Rev. Biochem. 44, 669-695.

SARGENT, T. D. & DAWTD, I. B. (1983). Differential geneexpression in the gastmla of Xenopus laevis. Science222, 135-139.

SARGENT, T. D., JAMRICH, M. & DAWID, I. B. (1986). Cellinteractions and the control of gene activity duringearly development of Xenopus laevis. Devi Biol. 114,238-246.

SAXEN, L. (1977). Directive versus permissive induction:a working hypothesis. In Cell and Tissue Interactions(ed. J. Lash & M. Burger), pp. 1-9. New York: RavenPress.

SAXEN, L. & TOIVONEN, S. (1958). The dependence of theembryonic inductive action of HeLa cells on theirgrowth media. /. Embryol. exp. Morph. 6, 616-633.

SAXEN, L. & TOIVONEN, S. (1962). Primary EmbryonicInduction. London: Logos Press and Academic Press.

SAXEN, L. & LEHTONEN, E. (1978). Transfilter inductionof kidney tubules as a function of the extent andduration of intercellular contacts. J. Embryol. exp.Morph. 47, 97-109.

SCHALLER, H . C , ROBERGE, M . , ZACHMANN, B . ,HOFFMEISTER, S., SCHILLING, E . & BODENMULLER, H .(1986). The head activator is released fromregenerating Hydra bound to a carrier molecule.EMBOJ. 5, 1821-1824.

SCHWARTZ, W., TIEDEMANN, H. & TIEDEMANN, H. (1981).High performance gel permeation of proteins. Mol.Biol. Rep. 8, 7-10.

SENGEL, P. (1976). Morphogenesis of Skin. CambridgeUniv. Press.

SLACK, J. M. W. (1983). From Egg to Embryo. CambridgeUniv. Press.

SLACK, J. M. W. (1985). Peanut lectin receptors in theearly amphibian embryo: regional markers for thestudy of embryonic induction. Cell 41, 237-247.

SLACK, J. M. W. & FORMAN, D. (1980). An interactionbetween dorsal and ventral regions of the marginalzone in early amphibian embryos. J. Embryol. exp.Morph. 56, 283-299.

SMITH, J. C. & WATT, F. M. (1985). Biochemicalspecificity of Xenopus notochord. Differentiation 29,109-115.

SMITH, J. C , DALE, L. & SLACK, J. M. W. (1985). Celllineage labels and region-specific markers in the

analysis of inductive interactions. J. Embryol. exp.Morph. 89 Supplement, 317-331.

SMITH, J. C. (1987). A mesoderm-inducing factor isproduced by a Xenopus cell line. Development 99, 3-14.

SPEMANN, H. (1901). Uber Korrelationen in derEntwicklung des Auges. Verh. Anat. Ges., 15 Vers.Bonn (Anat. Anz. 15), 61-79.

SPEMANN, H. & MANGOLD, H. (1924). Uber Induktionvon Embryonalanlagen durch Implantation artfremderOrganisatoren. Arch, mikrosk. Anat. EntwMech. 100,599-638.

SPEMANN, H. (1938). Embryonic Development andInduction. New Haven: Yale Univ. Press.

STERNBERG, P. W. & HORVITZ, H. R. (1986). Patternformation during vulval development in C. elegans. Cell44, 761-772.

TAKATA, K., YAMAMOTO, K. Y. & OZAWA, R. (1981). Useof lectins as probes for analysing embryonic induction.Wilhelm Roux' Arch, devl Biol. 190, 92-96.

TAKATA, K., YAMAMOTO, K. Y., ISHII, I. & TAKAHASHI,N. (1984). Glycoproteins responsive to the neural-inducing effect of concanavalin A in Cynopspresumptive ectoderm. Cell Differentiation 14, 25-31.

TEMPLE, S. & RAFF, M. C. (1985). Differentiation of abipotential glial progenitor cell in a single cellmicroculture. Nature, Lond. 313, 223-225.

TIEDEMANN, H. & BORN, J. (1978). Biological activity ofvegetalizing and neuralizing inducing factors afterbinding to BAC-cellulose and CNBr-sepharose.Wilhelm Roux' Arch, devl Biol. 184, 285-299.

TOIVONEN, S. (1953). Bone-marrow of the guinea-pig as amesodermal inductor in implantation experiments withembryos of Triturus. J. Embryol. exp. Morph. 1, 97-104.

TOIVONEN, S. & SAXEN, L. (1955). The simultaneousinducing action of liver and bone marrow of theguinea-pig in implantation and explantationexperiments with embryos of Triturus. Expl Cell Res.Suppl. 3, 346-357.

TOIVONEN, S., TARIN, D. & SAXEN, L. (1976). Thetransmission of morphogenetic signals from amphibianmesoderm to ectoderm in primary induction.Differentiation 5, 49-55.

WADDINGTON, C. H., NEEDHAM, J. & BRACHET, J. (1936).Studies on the nature of the amphibian organizationcentre. III. The activation of the evocator. Proc. Roy.Soc. B 120, 173-198.

WARNER, A. E., GUTHRIE, S. C. & GILULA, N. B. (1984).Antibodies to gap junctional protein selectively disruptjunctional communication in the early amphibianembryo. Nature, Lond. 311, 127-131.

WARNER, A. E. & GURDON, J. B. (1987). Functional gapjunctions are not required for muscle gene activationby induction in Xenopus embryos. J. Cell Biol. 104, (inpress).

WARTIOVAARA, J., LEHTONEN, E., NORDLING, S. & SAXEN,L. (1972). Do membrane filters prevent cell contacts?Nature, Lond. 238, 407-408.

WARTIOVAARA, J., NORDLING, S., LEHTONEN, E. & SAXEN,L. (1974). Transfilter induction of kidney tubules:correlation with cytoplasmic penetration into

306 J. B. Gurdon

Nucleopore filters. /. Embryol. exp. Morph. 31,667-682.

WESSELLS, N. K. (1977). Tissue Interactions andDevelopment. Calif. USA: W. A. Benjamin.

WESSELLS, N. K. & COHEN, J. H. (1967). Early pancreasorganogenesis: Morphogenesis, tissue interactions, andmass effects. Devi Biol. 15, 237-270.

WESSELLS, N. K. & RUTTER, W. J. (1969). Phases in celldifferentiation. Sci. Amer. 220(3), 36-44.

WHARTON, K. A., JOHANSEN, K. M., TIAN, X. &ARTAVONIS-TSAKONAS, S. (1985). Nucleotide sequence

from the neurogenic locus Notch implies a geneproduct that shares homology with proteins containingEGF-like repeats. Cell 43, 567-581.

WILSON, C , CROSS, G. S. & WOODLAND, H. R. (1986).Tissue specific expression of actin genes injected intoXenopus embryos. Cell 47, 589-599.

WITKOWSKI, J. (1985). The hunting of the organizer: anepisode in biochemical embryology. TIBS 10, 379-381.

YANKNER, B. A. & SHOOTER, E. M. (1982). The biologyand mechanism of action of nerve growth factor. A.Rev. Biochem. 51, 845-868.

Development 1987 GURDON 285 306

Documents

Transcript of Development 1987 GURDON 285 306