Involvement of Livertine, a hepatocyte growth factor family member, in neural morphogenesis

(1)

Article

Reference

Involvement of Livertine, a hepatocyte growth factor family member, in neural morphogenesis

RUIZ ALTABA, Ariel, THÉRY, C

Abstract

The formulation of the nervous system in vertebrate embryos involves extensive morphogenetic movements that include the folding of the neural tube and the migration of neural crest cells. Changes in cell shape and cell movements underlie neural morphogenesis but the molecular mechanisms involved in these processes in vivo are not well understood.

Here, we show that a new member of the hepatocyte growth factor family, which we name Livertine, is expressed in frog embryos in neural cells including neural crest and midline neural plate cells which are undergoing pronounced morphogenetic movements. The ectopic expression of Livertine perturbs gastrulation and leads to positional changes in injected cells without apparently changing cell type. These results suggest that one of the normal functions of Livertine is the control of neural morphogenesis in the vertebrate embryo.

RUIZ ALTABA, Ariel, THÉRY, C. Involvement of Livertine, a hepatocyte growth factor family member, in neural morphogenesis. Mechanisms of Development , 1996, vol. 60, no. 2, p.

207-220

DOI : 10.1016/s0925-4773(96)00618-1 PMID : 9025073

Available at:

http://archive-ouverte.unige.ch/unige:154183

Disclaimer: layout of this document may differ from the published version.

1 / 1

(2)

E L S E V I E R Mechanisms of Development 60 (1996) 207-220

Involvement of Livertine, a hepatocyte growth factor family member, in neural morphogenesis

A. R u i z i

Altaba.,*,

C. T h 6 r y b

aDevelopmental Genetics Program, Skirball Institute of Biomolecular Medicine, and Department of Cell Biology, New York University Medical Center, 540 First Avenue, New York, NY IO016, USA

bDepartment of Genetics, Columbia University, 701 West 168th Street, New York, NY 10032, USA Received 30 August 1996; revision received 27 September 1996; accepted 7 October 1996

Abstract

The formation of the nervous system in vertebrate embryos involves extensive morphogenetic movements that include the folding of the neural tube and the migration of neural crest cells. Changes in cell shape and cell movements underlie neural morphogenesis but the molecular mechanisms involved in these processes in vivo are not well understood. Here, we show that a new member of the hepatocyte growth factor family, which we name Livertine, is expressed in frog embryos in neural cells including neural crest and midline neural plate cells which are undergoing pronounced morphogenetic movements. The ectopic expression of Livertine perturbs gastrulation and leads to positional changes in injected cells without apparently changing cell type. These results suggest that one of the normal functions of Livertine is the control of neural morphogenesis in the vertebrate embryo.

Keywords: Embryo; Hepatocyte growth factor; Livertine; Morphogenesis; Neural; Xenopus

1. Introduction

Pattern formation in vertebrate embryos is characterized by consecutive inductive and morphogenetic events that lead to the formation of the basic body plan, including a neural plan common to all vertebrates. Changes in cell shape and position are often signs of a change in cell type of the induced cells since specific morphogenetic programs are a result of the acquisition of distinct fates by groups of cells (e.g. Jacobson and Gordon, 1976; Symes and Smith, 1987; Schoenwolf and Smith, 1990). Whereas some of the genes involved in the normal control of cell fate specification are known, little is understood about the molecular control of morphogenesis per se in vivo. In vitro assays, however, have identified soluble factors able to affect cell behavior, such as motility and the ability to form organized three-dimensional structures (e.g. Stoker et al., 1987; Weidner et al., 1990; Montesano et al., 1991). In addition, membrane and extracellular matrix molecules have been shown to affect cell adhesion (e.g.

* Corresponding author. Fax: +1 212 2637760;

e-mail: ria@ saturn.med.nyu.edu

Friedlander et al., 1989; Takeichi, 1991), suggesting their participation in the control of morphogenesis in vivo.

Our interest in midline cells of gastrulating embryos as regulators of axis formation (Ruiz i Altaba and Jessell, 1993) prompted us to search for factors specifically expressed in the early gastrula organizer region (the dorsal lip in frog embryos) and midline neural cells: the notoplate and floor plate. Previously, we identified a winged- helix gene, Pintallavis (Ruiz i Altaba and Jessell, 1992;

XKFH1, Dirksen and Jamrich, 1992; XFD1/I", Kntchel et al., 1992) specifically expressed in midline cells of early frog embryos and showed that it was highly homologous to hepatocyte nuclear factor-3fl (HNF-3~), encoding a liver transcription factor (Lai et al., 1991). Subsequently, the HNF-3fl and HNF-3~z genes were found to be expressed in midline cells of early embryos in different species (Ruiz i Altaba et al., 1993a, 1995a; Monaghan et al., 1993; Sasaki and Hogan, 1993; Ang et al., 1993;

Striihle et al., 1993). The striking expression of transcription factors both in midline cells of gastrulating embryos and in the liver suggested the possible existence of additional regulatory or signaling molecules expressed both in midline and endodermal cell groups like the organizer

(3)

208 A. Ruiz i Altaba, C. Thdry / Mechanisms of Development 60 (1996) 207-220 a n d the liver. F o r this reason, we searched for the expres-

sion in m i d l i n e cells o f g e n e s related to hepatocyte growth factor ( H G F ) , w h i c h e n c o d e s a secreted m o l e c u l e c o n t a i n i n g an a c h a i n with f o u r k r i n g l e d o m a i n s a n d a fl c h a i n s i m i l a r to that o f b l o o d serine proteases b u t without protease activity ( N a k a m u r a et al., 1989; Gherardi a n d Stoker, 1991; W e i d n e r et al, 1991; G o l d b e r g a n d Rosen, 1993).

Here we report o n the identification a n d functional analysis o f Livertine, a n e w m e m b e r o f the H G F family specifically expressed in a subset o f neural cells, includ- i n g m i d l i n e cells, a n d in the liver in frog e m b r y o s . Differ- e n t types o f e m b r y o n i c n e u r a l cells u n d e r g o i n g changes in cell shape or active m o v e m e n t s display p r o n o u n c e d e x p r e s s i o n o f Livertine. E m b r y o s o v e r e x p r e s s i n g Liver-

tine display gastrulation defects and ectopic e x p r e s s i o n of Livertine induces positional c h a n g e s o f e x p r e s s i n g cells.

W e propose that L i v e r t i n e is n o r m a l l y i n v o l v e d in the control o f neural m o r p h o g e n e s i s in the early vertebrate e m b r y o .

Recently, Xhl, a frog g e n e highly related to Livertine s h o w i n g a similar b u t n o t identical pattern o f expression has b e e n reported ( A b e r g e r et al., 1996).

2. Results

2.1. Identification and cloning o f Livertine f r o m f r o g embryos

To search for genes related to H G F expressed in

A u

L i v e r t i n e M14RRRT S P S N M K F S L F L L C F H A A T L F V T A H R SALNDYQR S KGL E LVHMNNGGVKQE I Q S E I QVCAKQC S DL L D C R S b"VYNWK S QTCR L L pWTQN SANVL L i 00

L i v e r t i m e Q R N V Q Y D L Y Q K K D Y I R ~ A G N G N T Y R G T V S K T K S G R T C Q R W R L K F P H D H K F S P I H W P E L E E N Y C R N P D S D P E G P W C Y T T D K N I R H Q Y C G I K K + D A ~ 200

L i v e r t i n e ITCNGEDYRGSg"DRTE S G K E C Q R W D L Q T P H A H P Y K P E K Y P D K S L D D N Y C R N P D S S E R P W C Y T T D P N V E K E F C R I T K ~ K Q R L SN I E I T S t F K E R G E G Y R ~ 300

L i v e r t i m e ~ T T S G I P C Q R W D S Q T P Q S H R F L P E K Y P C K G L D E N Y C R N - P D G S E A P W C - - F T T L - P - - - G M R M A Y C F Q I K R % D D V L - E P ~ Y H G N G E L Y S G R V S K ~ 392 T - L i v e r t i n e ~ . . . - ... - -. ... - - -. ... S. V F P . ~ R C W I - V . . ~ * |

R~-~ca / cwr . . . sran'm~SQS. R u ~ s . ~L . . . ~ r a ~ w . . -I- . . . -I

L i v e r t i n e

~.KGIKCRRWEEKRNDLELSLDQ~YLvPLEENYCRNPDRD~HGPWcYTMDPNTPFDYcAIKPtGEKvLTLEEAESIVFD~GKRNDRIFQKS~[VGG~G

^{4 9 2}

R T - P C R . . . F . T .

L i v e r t i n e

NSPWTV~LRNRQGEHFCGGSLVKENWV~TR~FSS~DADLSGYEAVMGTLFKNPS~DDPDKQS~PINKI~CG~D~-SLVMLKLERPITLNSR~ALI~LP

⁵⁹²

C H A I N

L i v e r t i n e P E R Y I V P E E T K C E I A G W G D T R G T G H D N V L K I A T F Y I I S N D N C N K N Y R S Q R N K V S D N E M C T K P M P I D A G A C E G D ~ G G P L A C L T H D C L V L E G V I V P A R G C G K 692

L i v e r f i n e K N Q PAl F T R V S V Y V D W I N K V M K M V * 716

L 4 v ~ r t 4 n ~

~ - Liv~%~c 4 m e

R T - I ~ R

L i v e r t i n e

T - L i v e r t 4 n e ~ T A ~ C ~ C ~ G G A A G A A A C C G A C T A C T G C A C G ~ A C C A G ~ T C ~ T C ~ C A G T C A C C ~ T G T T T A T A G T A T A T A A A A T A A A C C A C A T T T T T A A A A A T

L i ~ r t i n o T - L i ~ t l n e R T - ~

• ~ r A ~ G C C ~ A ~ W A ~ T ~ ~ ~ ~ T ~ @ ~ . A ~ C C CTGQT@TTJ~.~A~AATGG~CC C C J & ~ C A ~ C C C ~ I ~ A ~ d ~ T J ~ A ~ A A T T T T C C T T T T T C T C T G T A A T A A T A A A A C A ~ T A G T T T G T A C T T A T ~ C A ~ A C A A A G A T A T A A T T G A T C A T T T A T T T G ~ G A A A T C A A A A C C A ~ C C T A C T G G

. . . ~ . . . C . C . . . T . .

Fig. 1. Sequence of Livertine. (A) Deduced amino acid sequence of the Livertine sequence and of the variant forms, T-Livertine and the original RT- PCR product (see text). The variant sequences are shown only in the regions of divergence. The signal sequence (ss) is underlined and the kringle domains with the conserved cysteines are boxed and numbered (K1-K4). The cysteines of the loop in between the signal sequence and the kringle domains are denoted by asterisks. An inverted triangle shows the position of the predicted cleavage site between the a and the fl chains. The primer sequences in the RT-PCR product are shown underlined. Note the end of the T-Livertine sequence occurs before the start of kringle 4. The final cysteine of k_ringle 3 in the RT-PCR sequence is substituted. The position of termination codons is shown by asterisks following the sequences of T- Livertine and Livertine. The sequence of Livertine has been deposited in Genbank under accession number U57455. (B) Nucleotide sequence of the region of Livertine where the divergence occurs in T-Livertine and the RT-PCR product. The positions of the sequences coding for the end of kringle 3 and the beginning of kringle 4 are indicated. Primer sequences in the RT-PCR product are underlined by arrows. The points of divergence occur after AG duplets which are boxed. The non-coding 3' untranslated sequence of T-Livertine is shown in plain font. The TGA termination triplet of T- Livertine is underlined. Dots represent sequences identical to those of Livertine.

(4)

A. Ruiz i Altaba, C. Th~ry / Mechanisms of Development 60 (1996) 207-220 209 midline cells, we designed primers to conserved regions

in the second and fourth kringle domains of the rat and human H G F genes (Miyazawa et al., 1989; Nakarnura et al., 1989; Tashiro et al., 1990; Weidner et al., 1991).

These primers (see Section 4) were used to perform degenerate PCR at low stringency with frog early gastrula (stage 10) dorsal lip eDNA. Sequence of a PCR product of the predicted size (Fig. 1A) revealed the presence of an open reading frame encoding partial kringle domains related to, but distinct from, frog H G F (Nakamura et al., 1995).

This PCR product was used as a probe to screen a stage 11 gastrula eDNA library and two distinct clones were isolated of 3 kb and 2.2 kb in length. The DNA sequence showed a partial open reading frame encoding a member of the H G F family. The long and short clones are two independent cDNAs of the same gene. The two clones, however, appear to derive from two different copies of the same gene in the Xenopus laevis tetraploid genome since they show a few single base substitutions (Fig. 1B).

The 716 amino acid sequence of the full-length open reading frame resulting from the combination of the long and short clones showed the presence of an N-terminal signal sequence, a four-cysteine loop structure, four kringle domains and a C-terminal region homologous to the r-chain of blood serine-proteases showing substitutions of the three residues critical for protease activity (Fig. IA). All these regions were highly homologous to those of HGF and those of the related factor HGF-like.

The long clone started at amino acid 93, located between the cysteine loop and the kringle domains, whereas the shorter clone contained the entire 5' region (Fig. 1A).

Overall, the new protein showed 55%, 54%, and 61%

identity to human, mouse and chick HGF-like proteins, respectively (Degen et al., 1991; Hart et al., 1991; Th6ry et al., 1995), and 42-44% identity to human, mouse, chick and frog HGF proteins (Nakamura et al., 1989, 1995; Th6ry et al., 1995). The relatively low identity between HGF-like molecules (58% identity between chick and mouse, and 76% identity between mouse and human proteins) as compared to that between HGF molecules (74% identity between chick and mouse, 90% identity between human and mouse and 70% between frog and human proteins) suggests that there may be a subfamily of HGF-like related genes. For this reason, and because the gene described in this paper shows a pattern of expression distinct from those published previously for H G F and HGF-like in different species (De Frances et al., 1992; Sonnenberg et al,, 1993; Nakamura et al. 1995;

Th6ry et al. 1995), we consider it a novel member of the H G F family and named it Livertine.

Sequence divergence was detected in the short clone in a region corresponding to the end of kringle 3 displaying a stop codon before kringle 4 (Fig. 1A,B). We named this truncated clone T-Livertine. Within kringle 3, sequence

divergence was also present in the open reading frame of the original PCR product (Fig. 1A). These differences could result from differential splicing as the points of divergence were marked by putative splice sites (Fig. 1B).

2.2. Embryonic Livertine expression occurs in neural tissue

The pattern of expression of Livertine in the frog embryo was determined by in situ hybridization. Using a full-length probe from the long clone, we detected expression of Livertine transcripts beginning at the early gastrula stage (stage 10-10 +) in a small area of ectoderm adjacent to the dorsal lip (not shown). This expression is consistent with the isolation of the original PCR product from dorsal lip RNA. Livertine was detected in neural plate cells during gastrulation (stages 101/2-11 to 121/2;

Fig. 2A,B), and it was highest towards the midline, becoming expressed exclusively in midline cells by the late gastrula-early neurula stage (stages 13-14; Fig. 2C).

Within the neural plate, the expression of Livertine was confined to the outer layer (Fig. 2D,E).

These results showed that Livertine is first expressed in cells of the outer layer of the dorsal non-involuting marginal zone (dorsal NIMZ) which along with the deep layer of that region is characterized by convergent extension movements (Keller and Danilchik, 1988). Descen- dants of these cells include those in the notoplate, midline neural plate cells overlying the notochord, where Liver- tine expression is detected (Fig. 2C,D). Livertine expression in these cells could reflect expression in notoplate cells per se or in midline cells becoming floor plate by induction from the underlying notochord (Jessell and Dodd, 1992; Ruiz i Altaba, 1992; Ruiz i Altaba and Jes- sell, 1992). To distinguish between these two possibili- ties, we assayed for the expression of Livertine in the ectoderm of complete exogastrulae, in which involuted axial mesoderm and dorsal ectoderm are not apposed vertically as in normal embryos. Livertine was expressed in a nar- row band of cells along the A-P axis of the neurectoderm of exogastrulae (stage 15-16; Fig. 2F) which approxi- mates the position of the notoplate as determined previously by lineage tracing analysis (Ruiz i Altaba, 1992).

This suggests that expression of Livertine unlike that of the floor plate markers Pintallavis, HNF-3~ and F- spondin (Ruiz i Altaba and Jessell, 1992; Ruiz i Altaba et al., 1993a,b) in midline neural plate cells is not dependent on persistent signaling by the notochord, although we cannot rule out a possible contribution from notochord- derived signals to Livertine expression during normal development.

After neural tube closure, Livertine was detected in the dorsal neural tube and in migrating head neural crest ceils (stage -20-22; Fig. 3A). Expression was detected in the three major crest populations invading the branchial arches, in a pattern similar to that of Slug in frog embryos

(5)

210

2

A. Ruiz i Altaba, C. Thdry / Mechanisms of Development 60 (1996) 207-220

r

17. 3~

E

e

n c

dnt

B

f..

ba

fp

G

(6)

A. Ruiz i Altaba, C. Tl~ry I Mechanisms of Developraent 60 (1996) 207-220 211 (Mayor et al., 1995) and in the frontonasal process (Fig.

3A,B).

In the spinal cord of tadpole stage embryos (stage 30- 36), Livertine mRNA was detected in dorsal regions including neural crest cells and cells within the neural tube (Fig. 3D,E), as well as in neural crest-derived dorsal (Fig.

3F) and ventral (Fig. 3E) fin mesenchymal cells. In addition, Livertine was detected in the floor plate only in the posterior, and therefore the youngest, regions (Fig. 3D, E). This expression was transient as it was not detected in more anterior regions or in the same region of older embryos (Fig. 3F). In the youngest neural tube area, close to the tailbud, expression of Livertine was detected throughout the D-V extent of the neural tube (Fig. 3E). This expression was also transient and mimicked that found at early neural plate stages where it was expressed first throughout the neural plate becoming later confined to the midline (Fig. 2A-C).

In the tadpole brain, Livertine mRNA was found in the telencephalon, ventral diencephalon ventral midbrain and retina (Fig. 3A,B). Within the retina, the expression was widespread in tailbud and early tadpole stage embryos (Fig. 3A,B), whereas in older tadpoles (stage 36) it was highest in the ciliary marginal zone (CMZ) region (Fig.

3C), a center of mitotic activity. At larval stages (stage 40--45), expression of Livertine was maintained in the forebraln and new expression was observed in rhombomere centers (Fig. 3G). At these stages, Livertine was also expressed in the liver (not shown).

The distribution of T-Livertine remains unclear as 3' specific probes showed no labeling. However, RT-PCR analysis demonstrated its presence from gastrula to tadpole stages (not shown). It is possible that T-Livertine

expression occurs in a very small subset of cells, that the mRNA is very unstable or that it is expressed at very low levels in a widespread manner.

2.3. Overexpression of Livertine affects early morphogenesis

To test for the function of Livertine during early embryonic development, we ectopically expressed this putative signaling molecule by injection of plasmid vectors driving its expression into 1- or 2-cell stage frog embryos.

The cDNAs of Livertine and T-Livertine used in these experiments are functional as assessed by their in vitro transcription and translation which yielded protein prod- ucts of the predicted sizes (not shown). Expression from injected plasmids resulted in the abundant expression of Livertine mRNA in the progeny of the injected cells from late blastula stages onwards. Embryos injected with Livertine into the animal area showed anterior defects involving a deformed head often smaller than in control embryos. Similar results were obtained by the injection of synthetic Livertine RNA (not shown).

Because anterior defects are a common phenotype that results from perturbations of dorsal cells in the early gastrula and the fate map of the early embryo is known (Dale and Slack, 1987), we attempted to localize the cells affected by overexpression of Livertine. We performed lo- calized injections of plasmid DNA into either dorsal or ventral blastomeres at the 4-cell stage, which give rise to dorsal or ventral gastrula regions, respectively. Injection of Livertine into dorsal regions led to the development of anterior defects in injected tadpoles (23%, n = 242; Fig. 4, Table 1). These defects were not observed when T-

Fig. 2. Expression of Livertine in the neural ectoderm of gastrula and neurula stage frog embryos. (A-C) Whole-mount in situ hybridization pictures of the expression of Livertine in early gastrula (stage -10.5-11; A), late-gastrola (stage -12.5; B) and late gaswala/early neurula (stage -13-14; C) frog embryos. In all cases dorsal side is up and arrows point to the sites of expression. Note the decreasing size of the blastopore (bp) during gastrulation, a, anterior; p, posterior; md, midline neural plate (np) cells. (D,E) Expression of L/vert/ne in the superficial layer of the neural plate seen in a side view of a whole-mount (D) at stage 11-11.5 and in cross section through the anterior region of a stage -14 embryo (E). Arrows point to sites of expression and the brackets show the position of the different tissues, bp, blastopore; d, deep layer of the neural plate; en, endoderm of the archenteron (ar) roof; me, axial mesoderm; n, notochord; ne, neural ectoderm; s, superficial layer of the neural plate; so, somite. (F) Dorsolateral view of the expression of Liver- tine in the ectoderm (ec) of an early exogastrula (stage -15) in cells occupying the position characteristic for the notoplate, a, anterior in the ectoderm;

m, mesoderm; p, posterior in the ectoderm. Scale bar: (A,B,C,F) 240tim; (D) 60/~m; (E) 100#m.

Fig. 3. Expression of Livertine in late neurula, tailbud and tadpole stage embryos. (A) Livertine mRNA is detected in the migrating neural crest (nc) of the head, in the eye (e) and in the dorsal neural tube (dnt). Expression in the crest is detected in the mandibular, hyoid and branchial groups (from anterior to posterior, left to right, respectively). These three groups can be clearly seen in the bottom embryo (stage -20). Dorsal side is up for the embryo on top (stage -22) and the embryo on the bottom is a more lateral view. (B) Expression of Livertine in the bead of a stage -30 tadpole showing its mRNA in cells of the branchial arches (ba), frontonasal process (fn), the retina of the eye (e), the telencephalon (t), ventral diencephalon (vdi) and ventral midbrain (vmb). (C) Unilateral view of a cross section of the head of a stage -36 tadpole showing the expression of Livertine in the ventral diencephalon (vdi) and in the ciliary marginal zone (CMZ) of the retina. (D) Cross section through the posterior spinal cord of a stage -36 tadpole showing the presence of Livertine in the cells of the floor plate (fp) and in those of the dorsal neural tube (dnt). The notochord (n) is unlabeled. (E) Side view of the tail of a stage -30--32 embryo showing the expression of Livertine in the newly formed floor plate (fp), dorsal neural tube (dnt) and in the ventral fin (vf). Expression is also detected throughout the neural tube (nt) in the most posterior region, n: notocbord. (F) Side view of the trunk of a stage ~36 embryo showing the expression of Livertine in the mesenehymal cells of the dorsal fin (df). These cells are scattered throughout the fin.

Note also the absence of expression in the floor plate (fp). Brackets denote the position of tissues behind the segmented somites, dnt, dorsal neural tube; n, notochord; nt, neural tube. (G) Side view of the hindbrain (hb) at larval stages (stage ,-40) showing expression of Livertine at rhombomere centers. Numbers denote rhombomeres. The dark spots anterior to rhombomere 1 and adjacent to rhombomere 6 are pigment cells. Scale bar: (A) 450#m; (B) 170/~m; (C,D) 50#m; (E,F) 70#m; (G) 540#m.

(7)

212 A. Ruiz i AItaba, C. Thdry / Mechanisms of Development 60 (1996) 207-220 Table 1

Qnantitation of phenotypes detected in injected tadpoles

pDNA Normal op bl A def P def n

Dorsal

Livertine 43 29 23 5 242

T-Livertine 67 20 8 5 185

Ventral

Live~ine 64 17 3 16 211

T-Livertine 53 40 0 5 126

Uninjected 93 4 2 1 795

Plasmid DNA (pDNA) driving the expression of the Livertine or T- Livertine cDNAs was injected into either the dorsal or ventral marginal zones of 4-cell embryos and the resulting phenotypes scored at the tadpole stage (stage -32-36). Tadpoles were scored as having a normal morphology (normal) or displaying an open blastopore (op bl), anterior defects (A def) affecting the head or posterior defects (P def) affecting the tail or trunk and tail. All numbers represent percentages except the' total number of embryos scored (n).

Livertine was similarly injected (8%, n = 185; Table 1).

Anterior defects included an abnormal morphology of the head. The forebrain was reduced or absent and facial features such as the presence of a cement gland, eye and brancial arches were not distinguishable. Such embryos often displayed abnormal notochord positioning or development which resulted in the absence of floor plate development anteriorly (Fig. 4B). Injection of Livertine into ventral regions led to the development of posterior defects (16%, n = 211; Table 1) which were not observed in embryos similarly injected with T-Livertine (5%, n = 129;

Table 1). Posterior defects included an abnormal morphology of the tail or trunk and tail regions often appear- ing shorter and displaying abnormal somites.

A n u m b e r of embryos injected with either Livertine or T-Livertine displayed overt gastrulation defects as determined by the presence of an open blastopore at the neurula and tailbud stages (between 17% and 40%; Table 1;

Fig. 4C). At later stages, this led to the development of abnormal, often duplicated trunks and tails. These defects are not due to toxicity of the injected DNA, as injection of

identical amounts of plasmids driving the expression of sonic hedgehog or HNF-3/~ (not shown) did not produce such defects.

Livertine is more similar to m a m m a l i a n HGF-like than to HGF. For this reason we tested the effects of expression of the human HGF-like receptor, Ron (Ronsin et al., 1993; Gaudino et al., 1994; W a n g et al., 1994), in frog embryos. Both the injection of a full-length Ron receptor and a dominant-negative form, with a truncation of the tyrosine-kinase domain like in the dominant-negative forms of F G F receptors (Amaya et al., 1991), resulted in pronounced gastrulation defects similar to those obtained in Livertine injected embryos (C. Th6ry, C.D. Stern and A. Ruiz i Altaba, unpublished results). However, the expression of the dominant-negative receptor was unable to balance the effects of overexpression of Livertine in the whole embryo or in injected animal pole cells, making these results difficult to interpret.

Because cells normally expressing Livertine are neural and include the early floor plate, we tested the possibility that the phenotypes observed after ectopic expression of Livertine could be related to changes in cell type. Em- bryos ectopically expressing Livertine in the animal region, or animal caps derived from such embryos, did not display ectopic neural, floor plate or mesodermal development (not shown) as assessed by the expression of a variety of molecular and histological markers including the pan-neural marker Xen-1 and the floor plate marker HNF-3fl (Ruiz i Altaba, 1992; Ruiz i Altaba et al., 1995b). Thus, the data do not support a role of Livertine in the induction of neural or floor plate cells. In contrast, the results point to the possibility that ectopic expression of Livertine affects their morphogenetic movements during gastrulation. This, together with the expression of Livertine in a variety of cell types undergoing cell movements or changes in shape suggests that Livertine could be normally involved in regulating cell behavior.

2.4. Livertine expression leads to changes in cell position To test directly the possibility that Livertine may induce cell movements or changes in cell shape, we first

Fig. 4. Morphological defects observed in embryos injected with Livertine. (A) Side view of a stage -36 tadpole injected with plasmids driving the expression of Livertine into dorsal blastomeres at the 4-cell stage. The arrow points to the head region which is deformed and small, a, anterior; p, posterior. (B) Cross section through the hindbrain region of an embryo similar to that shown in (A) displaying a ventrally displaced notochord (n). The arrow points to the presence of axons at the ventral midline of the neural tube. The ventral displacement of the notochord lead to the absence of floor plate differentiation anteriorly. This was confirmed by labeling with anti-HNF-3fl antibodies (not shown), ov, otic vesicle. (C) Embryos injected with Livertine (bottom) showing the presence of an open blastopore (arrow). The embryo on top is a normal uninjected control. The middle row shows dorsal views whereas the bottom row shows side views of affected embryos injected with Livertine. Head and tail structures were sometimes detected.

Scale bar: (A) 260/zm; (B) 60/~m; (C) 440/~m.

Fig. 5. Effects of Livertine of cell position. (A,B) Pattern of RLDx fluorescent labeling in embryos (stage -14-15) injected with RLDx alone into blastomere B1 at the 32-ceil stage (A) and in those similarly injected with RLDx and Livertine mRNA (B). In both cases dorsal views are shown. Note the greater scattering of cells (arrows) detected in embryos injected with Livertine. (B) Pattern offl-gal expression in embryos (early stage 9) injected with plasmids driving the expression of LacZ only (C) or LacZ plus Livertine (D) into a small area of the animal pole at the 2-4-cell stage. In both cases views of the animal pole are shown. Some embryos show damage in their ventral sides that occurred during preparation for photography. Note the greater dispersion offl-gal + cells in Livertine injected embryos. Scale bar: (A,B) 330/.tm; (C,D) 720/~m.

(8)

A. Ruiz i Altaba, C. Thdry I Mechanisms of Development 60 (1996) 207-220 213

4A

C

-}

o

I._acZ t . a ~ +

(9)

214 A. Ruiz i Altaba, C. Th~ry / Mechanisms of Development 60 (1996) 207-220

assessed its ability to change the position of expressing cells derived from specific blastomeres as predicted from the fate map (Dale and Slack, 1987). We chose to inject mRNA into the blastomere B 1 of 32-cell albino embryos since its descendants mostly populate the notochord and floor plate making a variance of this pattern easy to score.

Descendants of blastomere B 1 include midline cells that normally express Livertine. To follow the position of the progeny of the injected blastomere, a lineage tracer, rhodamine-lysine-dextran (RLDx), was co-injected with the synthetic mRNA.

Injection of RLDx into blastomere B1 resulted in the predicted labeling of midline cells as well as cells in more anterior or posterior regions depending on the fidelity of the individual injected embryos to the canonical fate map (Fig. 5A). A fraction of these embryos (20%, n = 25) showed a small number (-10-50) of cells located lateral to the contiguous group of labeled midline cells. Injection of RLDx plus Livertine mRNA into blastomere B1 resulted in the labeling of midline cells and a more robust scattering of cells located further away from the midline

Table 2

Quantitation of the effects of Livertine on the position of injected animal pole cells

6 Q Q O

pDNA ~ 100% >50% 50% <50% n

Stage ⁹

Lac Z 40 27 33 0 15

Livertine + Lac Z 0 12 28 60 33

Stage 10

Lac Z 30 40 22 8 83

Stage 11

Lac Z 15 50 33 2 46

T-Livertine + Lac Z 10 80 10 0 63

Stage 12

Lac Z 38 36 19 7 47

T-Livertine + Lac Z 20 70 10 0 10

The drawings depict examples of the different distributions of injected cells. Each drawing is located above the corresponding column showing the variability from a homogeneous clone of injected cells (100%) to a highly dispersed clone containing less than 50% of labeled cells within the maximal area of the clone. Plasmid DNAs (pDNA) driving the expression of LacZ, LacZ plus Livertine or LacZ plus T-Livertine were injected into the animal pole at the l-2-cell stage and scored at blastula (stage 9) and gastrula (stages 10-12) stages. All numbers represent percentages except the total number of embryos scored (n).

than in control embryos (51%, n = 45; Fig. 5B) suggesting that expression of Livertine, but not T-Livertine (n = 20; not shown), was causing the cells to undergo pronounced positional changes. However, the variance of individual embryos from the canonical fate map made the quantitation of this effect difficult.

To quantify the effects of Livertine on cell position we developed an assay using animal pole cells as test cells in vivo. Plasmids driving the expression of LacZ were fo- cally injected into the animal most region of 2- or 4-cell albino embryos alone as control or together with plasmids driving the expression of Livertine or T-Livertine. fl- Galactosidase (fl-gal) expressed from the LacZ plasmids was used to monitor the position of the progeny of injected cells in the developing embryo via the X-Gal reaction. Because not all cells formed at the injected sites express the two plasmids and not all cells inherit plasmid DNA there is a mosaic distribution of the injected mate- rial. Moreover, there is no certainty that all cells expressing fl-gal are also expressing Livertine. Thus, since the analysis of the position of fl-gal ÷ cells may include cells that do not express Livertine and cells expressing Liver- tine alone will not be counted, the results will represent at worst an underestimation of the effects of Livertine and thus can be used to reliably test and quantify the effects of this secreted factor on cell behavior. The position of superficial cells descendant of injected cells was assayed before and during gastrulation (Table 2). In each injected embryo, the fl-gal ÷ cells were counted and considered a 'clone' and the degree of dispersion of the cells of each clone was measured by estimating the ratio between labeled and unlabeled cells inside the maximal area encom- passed by the clone.

At stage 9, injection of LacZ alone resulted in 40% of clones showing a homogeneous distribution (-100%) of labeled cells inside the area of the clone and there were no clones of fl-gal ÷ cells with less than 50% of labeled cells (Fig. 5B; Table 2). This is consistent with previous findings on the normal dispersion of animal pole cells (Wetts and Fraser, 1989). In contrast, no clones offl-gal ÷ cells in LacZ and Livertine injected embryos showed a homogeneous distribution and 60% of clones displayed a highly mixed distribution of labeled and unlabeled cells with less than 50% of these being labeled (Fig. 5C; Table 2).

During gastrulation, as epiboly progresses, there is a normal increase in the mixing of animal cap cells. This was reflected in the presence of unlabeled cells in fl-gal ÷ clones of embryos injected with LacZ only, assayed at later stages (stages 10-12, Table 2). However, there was in all cases a clear difference in the ratio of labeled to unlabeled cells within the maximal area of any fl-gal ÷ clone in LacZ only versus LacZ and Livertine injected embryos. For example, at stage 12, 38% of lacZ only versus 5% of LacZ plus Livertine clones showed a homogeneous distribution of labeled cells whereas 7% of LacZ

(10)

A. Ruiz i Altaba, C. Th[,ry I Mechanisms of Development 60 (1996) 207-220 215 only versus 35% of LacZ plus Livertine clones showed a

ratio of labeled to unlabeled cells inside each clone of less than 50% (Table 2). The effect on the distribution of the progeny of cells injected with Livertine was spe- cific as it was not observed in embryos injected with LacZ and T-Livertine (Table 2). The number of cells per clone did not appear to greatly differ between clones of LacZ alone versus LacZ plus Livertine (Fig. 5C,D and not shown).

Together, these results show that expression of Liver- tine leads cells to change neighbors more extensively (thus its name from a liver growth factor-like gene causing libertine behavior) and suggest that it may normally be involved in the control of cell movements or changes in shape that lead to changes in cell position.

3. Discussion

3.1. Livertine: a member o f the H G F family of signaling molecules

Livertine represents a new member of the HGF family.

HGF or scatter factor, and HGF-like or macrophage stimulating protein, have been described in different species (see Thtry et al. 1995; Goldberg and Rosen, 1993) but the identity among HGF-like proteins is lower than that among HGF proteins, raising the possibility that there are additional related genes such as Livertine. This possibility is also suggested by the existence of c-sea, an or- phan receptor (Huff et al. 1993) highly homologous to the transmembrane tyrosine kinases c-met and Ron, the receptors for HGF and HGF-like, respectively (Park et al., 1987; Bottaro et al., 1991; Naldini et al., 1991; Ronsin et al., 1993; Gaudino et al., 1994; Iwama et al., 1994; Wang et al., 1994). Moreover, Livertine and Xhl, a Xenopus HGF-like gene recently reported (Aberger et al., 1996) may be distinct. Even though these two genes share 84%

identity at the nucleotide level and 91% identity at the amino acid level the T-Livertine eDNA appears to derive from the other copy of the Livertine gene in the Xenopus tetraploid genome as it shows fewer base changes from the Livertine eDNA than that of Xhl (Fig. 1B, not shown) making Livertine and Xhl distinct loci. Alternatively, differences between the T-Livertine and Livertine cDNAs could represent polymorphisms.

The pattern of expression of Livertine is distinct from those previously reported for HGF-like genes in other species. Livertine, like chick HGF, is expressed in the cranial neural crest and branchial arches. It is also expressed, like chick HGF-like, throughout the early neural tube becoming restricted to the floor plate at later stages.

However, chick HGF-like, is expressed in the notochord and myotomes, two cell groups that do not express Liver- tine, and Xhl, unlike Livertine, is not expressed in the notoplate (Figs. 2,3) (Thtry et al., 1995; Alberger et al., 1996).

3.2. Involvement o f Livertine in the regulation of neural cell positioning

Livertine is expressed in neural cells undergoing morphogenetic movements or changes in cell shape such as midline cells of the neural plate and the neural crest. In this paper, we provide evidence to support a role of Livertine in neural cell positioning. In our first assay, a number of descendant cells from blastomere B 1 showed a variant distribution from that predicted by the fate map when they expressed Livertine (Fig. 5A,B). We have not determined the exact location of or eventual fate imposed on scattered cells. For example, presumptive notochord cells that end up lateral to the mature notochord due to the effects of exogenous Livertine may become part of the somitic mesoclerm. In our second assay, animal pole cells expressing Livertine show a higher degree of dispersion than non-expressing cells (Fig. 5C,D; Table 2). The mechanism of positional change is not clear as it could involve active movements and/or changes in shape of expressing cells in relation to their neighbors. The dispersion induced by Livertine on injected animal pole cells could also be due, in principle, to a mitogenic effect. This is unlikely since a large degree of dispersion of injected cells is observed by stage 9, a time when all animal pole cells are rapidly dividing and have been doing so syn- chronously up to 2 h earlier (stage 8). Moreover, since the injected plasmids are not transcribed before the midblas- tula transition (stage 8) there is little time for Livertine to affect cell position by having only a mitogenic effect.

The existence of T-Livertine is reminiscent of that of a truncated form of HGF which can act as a competitive antagonist (Chan et al., 1991; Miyazawa et al., 1991;

Lokker and Godowski, 1993). However, T-Livertine ends at the end of kringle 3 whereas the truncated form of HGF ends after kringle 2, making a direct functional compari- son difficult. The location of the variant sequences and the inability of T-Livertine to induce changes in position of animal pole cells suggests that kringles 3 and 4 are important functional parts of the Livertine molecule. This contrast with the ability of a truncated form of HGF lacking kringles 3 and 4 and the p chain to induce moto- genie activity (Hartmann et al., 1992; Lokker et al, 1992).

In normal development, the onset of expression of Livertine is detected in the dorsal NIMZ cells of the early gastrula embryo. Dorsal NIMZ cells are known to undergo convergent-extension movements that lead to the elongation of the midline of the neural plate, the notoplate (Jacobson and Gordon, 1976; Keller et al., 1992; Ruiz i Altaba, 1992), where Livertine expression is detected.

Since Livertine is detected in the neural ectoderm of stage 15 exogastrulae, its expression in midline neural plate cells could be related to changes in their position and not to their induction by the underlying notochord to become floor plate. However, it is not yet clear whether expressing cells lateral to the midline maintain Livertine expres-

(11)

216 A. Ruiz i Altaba, C. Th~ry / Mechanisms of Development 60 (1996) 207-220 sion and move to the midline or remain in their position

and turn off Livertine expression. It is also possible that Livertine expressed in midline superficial cells of the neural plate is involved in inducing the rearrangements of deeper non-expressing cells which undergo pronounced convergent-extension movements. Another possibility is that Livertine is also involved in the redial intercalation of superficial cells into the deep layer of the neural plate during neurulation. Expression in the newly induced floor plate of tailbud and tadpole stage embryos could be similarly related to the rearrangement of midline cells to form a one cell-thick floor plate from a cavitating neural tube primordium (the caudal eminence) or to the apical con- striction of floor plate cells required to act as a hinge during neural tube formation.

The expression of Livertine in moving cells is best ex- emplified by its presence in migrating neural crest cells in the head and in neural crest-derived dorsal fin cells (Fig.

3). Dorsal fin mesenchymal cells scatter as single cells whereas head neural crest cells migrate in groups. The control of the context of cell movements (as single cells or as groups) and the direction of moving cells appears to be independent of Livertine. The expression of Livertine in subsets of cells that are themselves changing position as single cells raises the possibility that Livertine can act in an autocrine manner, similar to the effect of HGF on some cell types (Adams et al., 1991; Bellusci et al, 1994;

Itakura et al, 1994) and the folded gastrulation gene product in Drosophila (Costa et al., 1994). In other con- texts it could have a paracrine function.

The expression of Livertine in all rhombomere centers in the frog larval hindbrain (Fig. 3G) contrasts with the expression of H G F and HGF-like at rhombomere boundaries in the chick embryo hindbrain (Th6ry et al., 1995). Assessment of cell movements in the hindbrain has shown that cells can freely move within the rhombomeres but generally cannot cross their boundaries (Fraser et al., 1990). Thus, it is possible that expression of Livertine would be related to cell movements within a rhombomere to change their position. In this case, Livertine would appear to have different effects to those of HGF and HGF-like. Indeed, Livertine lacks the ability to induce motile activity in transfected MDCK cells in vitro (B. Han and A. Ruiz i Altaba, unpublished results), a property of HGF (Stoker et al., 1987) but not of HGF-like (C. Th6ry and C.D. Stern, unpublished results).

Expression of Livertine in the tadpole brain, including the CMZ of the retina (Fig. 3) is unlikely to be related to cell movements as is its expression in the liver. In regions such as the CMZ, the presence of Livertine could be related to a putative mitogenic activity similar to that of HGF family members (Russell et al., 1984; Nakamura et al., 1989; Matsumoto et al., 1991; Bussolino et al., 1992;

Naidu et al., 1994) or to the competence of newly induced cells to acquire distinct fates.

3.3. Molecular control of neural morphogenesis

As we suggest for Livertine, other HGF family members have been proposed to regulate cell behavior and morphogenesis (reviewed in Th6ry and Stern, 1996). In vitro, HGF, or scatter factor, can induce motile behavior in epithelial (MDCK) cells and other cell types (e.g.

Stoker and Perryman, 1985; Stoker et al., 1987; Matsu- moto et al., 1991; Bussolino et al., 1992). This factor can also induce MDCK cells to form branched epithelial tu- bules in collagen gels that mimic kidney morphogenesis (Montesano et al., 1991; Santos and Nigam, 1993; Santos et al., 1994). HGF-like, or macrophage stimulating protein, can induce the motile behavior in macrophages and their response to chemoattractants (Leonard and Skeel, 1976; Skeel et al., 1991; Yoshimura et al., 1993; Skeel and Leonard, 1994).

In vivo grafting experiments have implicated HGF in neural induction in the chick embryo (Stern et al., 1990;

Streit et al., 1995). The analysis of a loss of function mu- tation in mice shows its requirement only in liver and placental development (Schmidt et al., 1995; Uehara et al., 1995) and the HGF receptor, the transmembrane tyrosine kinase c-met, is required for the migration of myo- genic precursors from the somites into the growing limbs (Bladt et al., 1995). Loss of function of neither HGF nor of c-met affect early morphogenesis. In contrast, our re- suits show that ectopic Livertine can affect early morphogenesis and induce cell positional changes suggesting a normal function in these processes although its requirement in the early embryo has not yet been assessed.

In frog embryos, both the misexpression and the in- hibition of cadherin function has been shown to result in abnormal development including the disruption of gastrulation movements (Detrick et al., 1990; Fujimori et al., 1990; Kintner, 1992; Dufour et al., 1994; Levine et al., 1994; Heasman et al., 1994; Lee and Gumbiner, 1995).

Gastrulation defects were similarly observed in embryos overexpressing Xwnt5A (Moon et al., 1993) and in those in which the function of the tyrosine phosphatase SH- PTP2 function was impaired (Tang et al., 1995). It is possible, therefore, that Livertine cooperates with cadherin, wnt or PTP factors to regulate neural cell behavior.

Finally, the involvement of genes of the HNF-3 and HGF families in early neural development suggests that the ontogeny of neural and endodermal tissues may have common principles.

4. Materials and m e t h o d s

4.1. Emb~os

Xenopus laevis female frogs were induced to lay eggs by injection of human gonadotropin. Fertilization in vitro was done with testis homogenates. Embryo rearing was as described (Ruiz i Altaba, 1993a) and staging was accord-

(12)

A. Ruiz i Altaba, C. TIv~ry I Mechanisms of Development 60 (1996) 207-220 217 ing to Nieuwkoop and Faber (1967). Exogastrulae were

obtained as described previously (Ruiz i Altaba, 1992).

4.2. Nucleic acid manipulations

Degenerate oligonucleotides corresponding to the sequences CMTCK/NG (5' AAGAATrCTGT/CATGACI- TGT/CAAIGG 3') and PWCFT (5' AAGGATCCGTIG- TA/GAAA/G CACCAIGG 3') found at the beginning of kringle 2 and towards the end of kringle 4, respectively, of the human and rat H G F genes (Nakamura et al., 1989;

Tashiro et al., 1990; Weidner et al,. 1991) were used in RT-PCR. These primers contain EcoRI and BamHI sites for cloning. First strand eDNA was prepared from dorsal lip (stage 10) organizer mRNA obtained from dissected pieces. RT-PCR yielded a -0.4 kb product. This was subcloned into pBluescript (Stratagene) and sequenced.

The PCR-derived insert was then used to screen both a stage 11 lithium-treated (kindly provided by C. Hume and J. Dodd, Columbia University) and a stage 17 (Kintner and Melton, 1987) cDNA libraries. The frequency of positive clones was 1/100 000 for the stage 11 library and 1/17 000 for the stage 17 library. Two positive clones were obtained in pSport (BRL) from the stage 11 library.

A full-length cDNA clone was made by joining the 3' SphI fragment of the long clone into the 5' SalI-SphI fragment of the short clone. The combined clone corresponds to Livertine and the short clone to T-Livertine (see text).

Whole-mount in situ hybridization of albino embryos was performed as described (Harland, 1991) but without a prehybridization step and without hydrolysis of the digoxygenin-labeled probes. To synthesize RNA probes for hybridization, the combined clone was cut with SalI and transcribed with SP6 RNA polymerase in the presence of digoxygenin-labeled UTP. A 3'-specific RNA probe for the long clone was made by cutting the plasmid with SspI and transcribing it with SP6 RNA polymerase. A 3'- specific RNA probe for the short (T-Livertine) clone was made by cutting a Bal-31 deletion subclone in pBluescript (Stratagene), lacking the 3' polyA tail of the cDNA, with NcoI and transcribing it with T3 RNA polymerase. La- beled embryos were inspected after clearing in benzyl alcohol/benzyl benzoate with a Zeiss Axiophot micro- scope. Photographs were taken with Kodak 160 ASA Tungsten-balanced slide film.

Histological sections of labeled embryos were obtained after embedding the embryos in wax and cutting in a mi- crotome at 10-15/am as described (Ruiz i Altaba, 1992).

4.3. Microinjection and lineage analysis

In vitro transcription of sense RNAs was performed by cutting the clones with NotI and transcribing with T7 RNA polymerase. Two nanograms of synthetic RNA were injected into each embryo at the 1-cell stage. For plasmid injections, the Livertine (combined) and T-

Livertine cDNAs were cloned into pcDNA1-Amp (Invitrogen), a plasmid in which ubiquitous expression of cloned eDNAs is driven by cytomegalovirus regulatory sequences. The Livertine cDNA was digested with SalI and SspI, filled in and cloned into filled-in EcoRV digested pcDNA1-Amp. The T-Livertine eDNA was digested with SalI, filled in, digested with NotI and ligated to filled-in-EcoRV plus NotI digested pcDNA1-Amp.

Between 20 and 200 pg of purified pDNA (Qiagen) were injected into each embryo at different locations. Injections into the animal most area were as described (Ruiz i AI- taba et al., 1995b). The injection procedure was as previously described (Ruiz i Altaba, 1993a). Histological analysis of injected pigmented embryos was as described above with the exception that in this case the sections were stained by the Feulgen/OrangeG/Light Green method (see Smith, 1987).

Lineage tracing in RNA-injected embryos was performed by co-injecting rhodamine-lysine-dextran at (15- 10 nl at 25 mg/ml in water; Molecular Probes). The distribution of RLDx was detected by fluorescent micros- copy after fixing in MEMFA and clearing the injected albino embryos in benzyl alcohol/benzyl benzoate (Har- land, 1991). Lineage tracing in plasmid-injected embryos was performed by co-injecting equal amounts of the desired plasmid and a pcDNA plasmid carrying the bacterial LacZ gene (kindly provided by Dr. C. Dulac, Columbia University). At the desired stage the injected albino embryos were fixed with glutaraldehyde and processed for the X-Gal reaction to yield an insoluble blue precipitate.

These embryos were inspected and photographed without clearing.

Acknowledgements

The work presented here was initiated in the laboratory of Thomas M. Jessell at the Howard Hughes Medical In- stitute, Center for Neurobiology and Behavior, Columbia University were ARA was a postdoctoral fellow. His support is greatly appreciated. C.T. was a postdoctoral fellow in the laboratory of Claudio D. Stern at Columbia Uni- versity and was funded by a grant of the Muscular Dys- trophy Association to C.D.S. Funding also derived from a start-up grant from the Skirball Institute to A.R.A. We thank Cliff Hume and Jane Dodd for supplying the stage 11 lithium-treated frog cDNA library, C. Dulac for supplying the pcDNA-lacZ plasmid and J. Smith for the Feulgen/Orange G/Light Green staining method. We are grateful to Claudio Stern, Gord Fishell, Will Talbot, Col- leen Ring, Alain Prochiantz, Jessica Treisman, Alex Joy- ner, Alex Schier and Tom Jessell for comments on the manuscript.

References

Aberger, F., Schmidt, G. and Richter, K. (1996) The Xenopus homo- logue of hepatocyte growth factor- like protein is specifically ex-