Purification, partial characterization and biological

591

Development 103, 591-600 (1988) Printed in Great Britain © The Company of Biologists Limited 1988

Purification, partial characterization and biological effects of the XTC mesoderm-inducing factor

J. C. SMITH, M. YAQOOB and K. SYMES Laboratory of Embryogenesis, National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 1AA, UK

Summary The mesoderm of Xenopus laevis is formed through an inductive interaction in which a signal from the vegetal hemisphere of the blastula acts on overlying animal pole cells. We have recently reported that the Xenopus XTC cell line secretes a mesoderm-inducing factor (MIF) which may resemble the natural signal. In this paper, we describe the purification and biological effects of XTC-MIF. XTC-MEF is a hydrophobic protein with an isoelectric point of 7-8 and an apparent relative molecular mass (A/r) of 23500. On reduction, XTC-MIF loses its biological activity and the protein dissociates into two inactive subunits with apparent MT of about 15 000. These properties closely resemble those of transforming growth factor type /3 (TGF-/3), and it is interesting that TGF-/32 has recently been shown to have mesoderm-inducing activity.

The biological response to XTC-MIF is graded. After exposure to 0 2-1 O n g m l 1 XTC-MIF, stage-8 animal pole explants form mesenchyme and mesothelium. At higher concentrations, up to about Sngml" 1 , muscle is formed, occasionally with neural tissue. In response to concentrations of XTC-MIF greater than 5-10 ng ml"1, notochord and neural tissue are usually formed. The formation of notochord and neural tissue in response to XTC-MIF represents a qualitative difference between this inducing factor and the other known group of MIFs, the heparinbinding growth factors.

Introduction

Kirschner (1987), who also show that mRNA for an FGF-like molecule is present in early Xenopus embryos. In addition, Kimelman & Kirschner (1987) find that transforming growth factor-type /Jl (TGFfi\), although not an inducing factor itself, acts synergistically with FGF, such that'the two molecules together do induce large amounts of muscle. By contrast, TGF-/32, which is closely related to TGF-/J1, is capable of inducing muscle from animal pole ectoderm in the absence of other factors (Rosa et al. 1988). Interestingly, an mRNA localized to the vegetal hemisphere of the Xenopus embryo, Vgl (Rebagliati et al. 1985; Melton, 1987), codes for a factor related to TGF-/3 (Weeks & Melton, 1987). It is not yet known whether the Vgl protein is an inducing factor. In this paper, we describe the purification and partial characterization of the XTC mesoderm-inducing factor. Our results indicate that the factor is active at picomolar concentrations, and biochemical data indicate that it is related to TGF-/3. Unlike FGF,

The earliest inductive interaction in amphibian development is mesoderm induction, in which an equatorial mesodermal rudiment is induced from animal hemisphere tissue under the influence of one or more signals from the vegetal hemisphere (Nieuwkoop, 1969, 1973; Dale etal. 1985; Gurdon etal. 1985; Jones & Woodland, 1987; reviewed by Smith, 1988). Recently, significant progress has been made towards identifying the mesoderm-inducing factors (MIFs) involved in this process. The Xenopus XTC cell line has been found to secrete a factor that induces isolated Xenopus animal pole ectoderm to form mesodermal cell types, including notochord and muscle (Smith, 1987). Slack etal. (1987) report that bovine fibroblast growth factor (FGF) induces mesoderm from isolated ectoderm, but that this mesoderm is ventral in character, including mesenchyme and mesothelium but little muscle and no notochord. This observation has been confirmed by Kimelman &

Key words: Xenopus, mesoderm induction, morphogen, fibroblast growth factor (FGF), transforming growth factor type fi (TGF-/3), amphibian embryo.

592

/. C. Smith, M. Yaqoob and K. Symes

it can induce notochord from Xenopus animal pole tissue. Materials and methods XTC conditioned medium Serum-free conditioned medium from the XTC cell line was obtained as described by Smith (1987) except that the cells were grown on glass roller bottles. The medium was heated to 100°C (Smith, 1987), concentrated 10- to 40-fold using a Minitan Ultrafiltration System equipped with 10000 Mr cutoff filter packets and exchanged into 20 mM-Tris-HCl pH80. Batches of concentrated conditioned medium were stored at —70°C. The large-scale purification described in this paper used 585 ml of concentrated medium which was derived from an initial volume of 181. Embryos Embryos of Xenopus laevis were obtained by artificial fertilization as described by Smith & Slack (1983). They were chemically dejellied using 2 % cysteine hydrochloride (pH7-8-8-l), washed and transferred to Petri dishes coated with 1 % Noble Agar and containing 10 % normal amphibian medium (NAM: Slack, 1984). The embryos were staged according to Nieuwkoop & Faber (1967). Mesoderm induction assay Aliquots of test fractions were diluted into three-quarters strength NAM containing 0-1% bovine serum albumin (BSA: Sigma). Inducing activity was assayed as described by Smith (1987) and Symes & Smith (1987). Briefly, animal pole regions were dissected from stage-8 Xenopus embryos and transferred, blastocoel-facing surface up, to the test solution. Fractions containing inducing activity could be identified within 5-6 h; instead of remaining as spheres, induced explants begin to elongate (Fig. 1). This change in shape is an early and reliable marker of mesoderm induction and represents an attempt by the induced cells to undergo gastrulation movements (Symes & Smith, 1987). To minimize the interval between successive purification steps the elongation of test explants was used to select the fractions to be applied to the next column. However, the explants were allowed to develop until control embryos

Fig. 1. Gastrulation movements provide a marker of mesoderm induction. (A) A Xenopus animal pole explant cultured in the absence of mesoderm-inducing activity remains as a sphere. (B) An induced explant at the equivalent of the late gastrula stage. It has undergone 'convergent extension' movements and formed an elongated structure. Bars, 200[tm.

reached stage 40 when induced tissue can be recognized visually by its swollen appearance (see Slack et al. 1987; Godsave et al. 1988), and this always confirmed the earlier assessment. We define one unit of mesoderm-inducing activity as the minimum quantity that must be present in 1 ml medium for induction to occur. Thus if the highest active dilution is 1/1000, the solution contains 1000 units ml"1 (see Cooke et al. 1987; Godsave et al. 1988). Chromatographic procedures All steps were carried out at room temperature. Ion exchange chromatography Concentrated heated XTC conditioned medium, exchanged into 20 mM-Tris-HCl pH8-0, was pumped at a flow rate of 60mlh~' on to a column (2-6 cm diameter x 20-5 cm) of DEAE-Sepharose (Pharmacia, UK) equilibrated in the same buffer. The column was washed with 20 mM-Tris-HCl pH8-0 until the absorbance of the effluent (monitored at 280 nm) fell to zero. Mesoderminducing activity was then eluted with a gradient of 0-1 MNaCl in 20mM-Tris-HCl pH80. The flow rate was 90mlmin~' and the gradient volume was 11. Fractions of 13-5 ml were collected. Hydrophobic interaction chromatography Active fractions from ion exchange chromatogTaphy were pooled and the NaCl concentration of the pool was calculated from its conductivity. Solid NaCl was added to bring the concentration to 1 M and this material was pumped at 20mlh~' onto a column (l-6cm diameter x 11-5cm) of phenyl-Sepharose (Pharmacia, UK) equilibrated in 1 MNaCl, 20mM-Tris-HCl pH80. The column was washed with lM-NaCl, 15mM-Tris-CH3COOH pH80 until the absorbance of the effluent fell to zero. The column was then washed again with 15mM-Tris-CH3COOH pH80 and mesoderm-inducing activity was eluted with a gradient of 0-80 % ethylene glycol in the same buffer. The gradient volume was 300 ml, the flow rate was 20 ml h~' and fractions of 3-3 ml were collected. Fractions of 10 ml were collected while the column was being loaded and washed. Chromatofocusing Active fractions from hydrophobic interaction chromatography were pooled and concentrated to 7 ml using an Amicon pressure cell with a YM5 membrane. Tween 20 was added to 002 % and Tris base was added to bring the final concentration to 75 mM and the pH to 9-2. Chromatofocusing 'start buffer' (see below) was added to bring the volume to 10 ml. Chromatofocusing was carried out with a Mono PHR5/20 column (Pharmacia, UK) connected to an FPLC System (Pharmacia, UK) using a pH interval of 9-6. The start buffer was 75mM-Tris-CH3COOH, pH9-3, 10% ethylene glycol, 002% Tween. The eluent was 10% Polybuffer 96-CH3COOH, pH60, 10% etheylene glycol, 002 % Tween. A flow rate of 0-5 ml min"1 was used. After loading the sample, the column was washed with 4 ml start buffer and inducing activity was then eluted with 35 ml eluent. Fractions of 3 ml were collected while loading the

Mesoderm-inducing factor in Xenopus

593

sample, 2 ml during washing and 0 5 ml during elution. The pH gradient was determined by measuring the pH of each fraction (except those containing inducing activity, to avoid contamination of the samples). It was not possible to monitor the absorbance of the effluent.

stained polyacrylamide gels with those of known amounts of carbonic anhydrase, trypsinogen and soybean trypsin inhibitor. These three proteins stain with silver to slightly different extents and the concentration of XTC-MIF was taken as the average of the three deduced concentrations.

HPLC size-exclusion chromatography Active fractions from the chromatofocusing step were concentrated to 120 jA using a Centricon-10 microconcentrator (Amicon). HPLC size exclusion chromatography was performed with a 7-5 mm x 60cm TSK-G2000 SW column (Anachem) and .a 7-5mmx7-5cm TSK guard column (Anachem), connected to an LKB Ultrochron GTi HPLC system. The solvent was 0-2M-NaCl, 50mM-sodium phosphate pH7-0, 0 0 2 % Tween, with a flow rate of 0-5mlmin~'. Fractions of 0-5ml were collected, and active fractions were pooled and concentrated using a Centricon10 microconcentrator.

Histological analysis Explants treated with pure XTC-MIF were cultured at 20 °C for three days until control embryos reached stage 40. They were then fixed for 48 h in a solution of 10 % formalin, 2 % glacial acetic acid, 50 % ethanol and 38 % NAM, followed by 48 h in 10 % formol saline. The specimens were embedded in paraffin wax and sectioned at 8^m before being stained by the Feulgen/Light Green/Orange G technique ofCooke (1979).

SDS-polyacrylamide gel electrophoresis Samples to be analysed by SDS-PAGE were mixed with an equal volume of 2 x gel sample buffer, omitting a reducing agent. The samples were usually applied to mini vertical slab gels (Bio-Rad model 360) containing 12-5 % or 13-5 % acrylamide and using the buffer system of Laemmli (1970). Occasionally conventional 17x15 cm gels were used. Gels were silver stained using a modification of the technique of Wray et al. (1981), which detects less than 1 ng of some protein species. Protein could be recovered from gel slices by elution into 0-1% BSA in NAM, using the technique described by Hughes & Raff (1987). For semipreparative purposes the gel was first stained with CuCl2 to locate the band of interest (Lee et al. 1987) and then destained with 0-25 MEDTA/0-25M-Tris-HCl, pH90. Preparative gels were eluted into PBS without BSA.

Preliminary characterization of the XTC mesoderminducing activity has established that the active principle is a heat-stable protein with an apparent relative molecular mass (Afr) of 16000 (Smith, 1987). Further experiments have demonstrated that over 80 % of the inducing activity is lost on exposure to 1 M-acetic acid (data not shown); the initial stages of the purification procedure described below therefore employed neutral pH conditions. XTC mesoderm-inducing activity does not bind to heparin-Sepharose CL-6B (Pharmacia) (data not shown). This indicates that the XTC factor is not a member of the heparin-binding growth factor family, which includes acidic and basic FGF and embryonal carcinoma-derived growth factor (ECDGF) (see Slack et al. 1987).

Protein quantification Protein was determined by the modified Lowry et al. (1951) technique of Sargent (1987). When interfering substances were present they were removed by selective precipitation of the proteins with acetone, as described by Sargent (1987) or by repeated dilution and concentration using Centricon10 microconcentrators. Quantification of small amounts of XTC-MIF was achieved by comparing their staining intensities on silver-

Purification of XTC-MIF A summary of the purification of XTC-MIF is given in Table 1. The initial step employed ion exchange chromatography on DEAE-Sepharose. All the detectable inducing activity of the XTC conditioned medium bound to the anion exchange column, and when the column was developed with a gradient of 0-lM-NaCl, inducing activity eluted at a concentration of approximately 0-4M (Fig. 2). The peak

Results

Table 1. Purification of XTC-MIF

Fraction from 1. Heated XTC conditioned medium 2. DEAE Sepharose 3. Phenyl-Sepharose 4. Mono P 5. TSK G2000*

Total volume (ml)

Total protein (mg)

Total activity (units)

585

292-5

175500

600

18-9 1-54 012 3xlO~ 3

126000 35000 15000 15000

6667 22727 125000 5-OxlO6

63 35 1 006

Specific activity (units mg~!)

Yield (%)

Purification (-fold)

100 72 20 8-6 8-6

* The active fractions from step 5 were concentrated to 60y\ from 1 ml. Total protein was estimated1 from silver-stained polyacrylamide gels (see Materials and methods). All other protein determinations were according to Sargent (1987).

1 111 37-9 208-3 8333.3

594


ft !\ \\ \ \ \

1-41-31-21110-

0-9|0-8: o-70-60-50-40-3- \ 0-2- \ 01-

\ \ v 1 Vfl

/ /

/ /Q\ / /Y // / / L» r/ v\

^

/

/

rO-8

/

•0-7

/

•0-6 •0-5

-0-4 •0-3

•0-2

V

-01

4 8 1216 202428 32 36 40 44 48 52 56 60 64 68

Fraction number

Fig. 2. DEAE-Sepharose chromatography of XTC mesoderm-inducing activity. 585 ml of concentrated heated XTC conditioned medium was applied to a column of DEAE-Sepharose which was then washed until the absorbance of the eluate, monitored at 280 nm, fell almost to zero. The figure shows the elution profile when the column was developed with a gradient of 0-1 M NaCl. The solid horizontal bar represents the fractions containing inducing activity, which were pooled for the next purification step.

coincided with the elution of the phenol red which was present in the original culture medium. This provides a useful marker of the position of inducing activity but the absorbance of these fractions more reflects the presence of the indicator than the protein concentration (see Fig. 2). The active fractions from DEAE—Sepharose chromatography were adjusted to 1 M-NaCl and applied to a column of phenyl-Sepharose. All the inducing activity bound to the column, which was washed and then developed with a gradient of 0-80 % ethylene glycol in 15 miu-Tris-acetic acid pH8-0.

Inducing activity eluted at approximately 30 % ethylene glycol (Fig. 3). Phenol red, concentrated at the ion exchange chromatography step, did not bind to the column. Thus the absorbance of the unbound material shown in Fig. 3 is due in part to the absorbance of phenol red. Recovery from the phenyl-Sepharose step was poor; 72 % of the activity applied to the column was lost. Hydrophobic interaction chromatography is a mild technique which usually gives high recoveries (see Scopes, 1982). It is possible, therefore, that a factor acting synergistically with XTC-MIF was lost at this stage. The active fractions from hydrophobic interaction chromatography were concentrated, adjusted to 75mM-Tris-acetic acid pH9-0, and Tween 20 was added to 0-02%. This material was injected onto a Mono P chromatofocusing column which was developed with an eluent buffer adjusted to pH6-0. Mesoderm-inducing activity eluted at a pH7-85 (Fig. 4), which represents the isoelectric point of the protein. Samples from the chromatofocusing step were concentrated and subjected to high-pressure sizeexclusion chromatography (Fig. 5A,B). Mesoderminducing activity eluted at an apparent Mr of approximately 16000, slightly in advance of a major peak of u.v. absorbance at 10000. Fractions 42 and 43, which contained most of the inducing activity, were pooled and concentrated to about 60 \A. Fractions on either side of the inducing activity peak were also concentrated and 3jil samples were analysed by minipolyacrylamide gel electrophoresis under nonreducing conditions. After staining the gel with Coomassie Brilliant Blue, the highest level of inducing activity was seen to be correlated with a band of Mr23 500. No other bands were visible, except for a small amount of Coomassie-stainable material ahead of the buffer front. Much larger amounts of this material

IM NaCl 0-5-, •70

-60

0-4-

50 3 0-3-40 0-2-

-30

-20 01-10

-0

12 34 56 7891015 2025 30 35 40 45 50 55 60 65 70 7580 85 90 Fraction number

|

Fig. 3. Phenyl-Sepharose chromatography of XTC mesoderm-inducing activity. Fractions from the DEAE-Sepharose step were adjusted to 1 M-NaCl and loaded onto a column of phenyl-Sepharose. The column was then washed with 1 M-NaCl until the absorbance of the eluate fell almost to zero, and washed again with a buffer containing no NaCl before mesoderm-inducing activity was eluted with a gradient of 0-80% ethylene glycol. The solid horizontal bar represents the fractions containing inducing activity, which were pooled for the next purification step.


595

were present in later fractions. After silver staining, the 23 500 band was also visible in fractions adjacent 15000to numbers 42 and 43, and their intensities correlated with the levels of inducing activity (Fig. 5B, inset). Nonspecific bands with apparent relative molecular masses of 55 000 and 14000 were visible in all tracks !_ 10000-8 but no other silver-staining material was visible in X fractions 42 and 43; the Coomassie-staining material a. at the buffer front did not stain with silver. u. To confirm that the 23500Mr band represents the -7 5000XTC mesoderm-inducing factor, two experiments were performed. In the first, a sample of XTC-MIF was adjusted to pH2-0 and applied to an RP-300 (C8) column, which was developed with a gradient of 0-60% acetonitrile in 0-1% trifluoroacetic acid. 0 1 2 3 4 5 10 15 20 25 30 35 40 45 50 55 60 65 70 7580 Inducing activity eluted at approximately 39 % acetoFraction number nitrile and mini-polyacrylamide gel electrophoresis revealed a single band at 23 500 in the active fractions. Fig. 4. Fractionation of XTC mesoderm-inducing activity Recovery of inducing activity by this procedure was by chromatofocusing. After loading the column and only about 20%, presumably due to the low pH washing with 4 ml 'start buffer' a pH gradient of 9-6 was conditions. In the second experiment, duplicate 2^x1 developed. Fractions of 0-5 ml were collected and assayed samples were subjected to mini-polyacrylamide gel for inducing activity. Fractions 29 and 30 were pooled for electrophoresis under nonreducing conditions. Two further purification. •3

15000-1 2912-4 6-5

38 40 42 44 46 48 39 41 43 45 47 49

10000E

2 3

5000-

a o-oi-

s

0- 1

5

10 15 20 25 30 35 40 45 50 55 60 Fraction number

\

32 3 4 3 6 3 8 4 0 42 4 4 4 6 4 8 5 0 52 5456 Fraction number Fig. 5. Fractionation of XTC mesoderm-inducing activity by HPLC size-exclusion chromatography. (A) Arrows indicate relative mobility of standards: bovine serum albumin (66), carbonic anhydrase (29), cytochrome C (12-4) and aprotinin (6). The solid bar indicates the fractions of highest inducing activity. (B) Inducing activity of each fraction. Inset, analysis of the indicated fractions by polyacrylamide gel electrophoresis and silver staining. Solid circles to the right of the gel indicate relative mobility of standards: bovine serum albumin (66000), ovalbumin (45000), glyceraldehyde-3-phosphate dehydrogenase (36000), carbonic anhydrase (29000), trypsinogen (24000), soybean trypsin inhibitor (20100) and a--lactalbumin (14200).

596

J. C. Smith, M. Yaqoob and K. Symes 1

r

2

3 rO

-1

66 45 36

ft

• 2 °° CL

29

-3

24

-4

o o

201

-5

14 2, 1500

I I 1000 500 MIF (units m r 1

additional lanes contained samples that had been boiled for 3min in the presence of 50mM-dithiothreitol. One pair of reduced and nonreduced samples was silver stained. The other tracks were sliced and protein eluted from each slice was assayed for mesoderm-inducing activity. Activity was only present in the nonreduced sample and coincided with the 23500Mr protein (Fig. 6). No activity was present in the reduced sample, and the 23 500 band had disappeared to be replaced by a doublet consisting of bands of approximately 15 500 and 14500Afr. To confirm the impression that XTC-MIF is a heterodimer, reduced and nonreduced samples were analysed in four additional experiments using 13-5% acrylamide gels. All four gels suggested that the reduced form of XTC-MIF consists of two distinct species, but the intensity of the higher molecular weight band varied in different experiments. In two cases, the higher and lower molecular weight species were of similar intensities, but in the other two there was much less of the higher molecular weight form; it was barely visible in one gel. The reason for this variation is unclear. The procedure described above achieves a 8333fold purification of XTC-MIF (Table 1). No contaminating proteins are detectable on silver-stained polyacrylamide gels, under conditions where a contaminant of less than 10% would be noted. On Coomassie-stained gels a minor contaminant is visible ahead of the buffer front, but the nature of this material, which lacks biological activity itself and is not required for the activity of XTC-MIF (Fig. 6) is unclear. It may represent Polybuffer from the earlier chromatofocussing step. A total of 3/ig of XTC-MIF was recovered from 181 of XTC-conditioned medium, representing a recovery of 8-6%.

Fig. 6. Mesoderm-inducing activity can be eluted from a polyacrylamide gel; activity coincides with a 23 500Mr protein and is lost on reduction with 50mM-dithiothreitol. (A) Silver-stained gel showing, lane 1: relative mobility of standards (as in Fig. 5B); lane 2: 2\x\ of mesoderm-inducing activity from HPLC size-exclusion chromatography, run under nonreducing conditions; lane 3: identical sample to lane 2, but treated with 50mM-dithiothreitol. (B) Inducing activity of protein eluted from parallel tracks run under nonreducing conditions ( ) or after reduction ( ). Biological effects of XTC-MIF To study the biological effects of XTC-MIF, groups of twelve stage-8 animal pole explants were exposed to serial dilutions of the purified material from sizeexclusion HPLC. The earliest response that could be observed was a dramatic elongation of the induced explants, as described by Symes & Smith (1987). The extent of elongation depended on the concentration of XTC-MIF; none could be observed at concentrations less than 0-2ngmr'. It is not surprising that XTC-MIF should cause elongation of animal pole explants, of course, since elongation was used as the test for active fractions (see Fig. 1). However, it is noteworthy that a single substance can both cause elongation and, as we see below, induced mesodermal cell types. In the absence of inducing factors, isolated Xenopus animal pole regions differentiate as 'atypical epidermis' (Fig. 7A). The lowest concentration of XTC-MIF which causes a diversion from this mode of differentiation is 0-2ngml~', which induces explants to form 'bubbles' of thin epidermis containing mesothelium and loose mesenchyme (Fig. 7B; see also Cooke et al. 1987; Godsave et al. 1988). After induction with l-2ngml~' XTC-MIF and above, explants usually contain substantial amounts of muscle, often with neural tissue (Fig. 7C) and 25 ng ml~' usually causes formation of notochord and neural tissue (Fig. 7D). However, concentrations as low as 0-8ngml~' have caused notochord formation in some cases. Formation of notochord in response to XTC-MIF represents a qualitative difference between this factor and fibroblast growth factor; even at concentrations of l^gml" 1 FGF does not induce notochord or neural tissue (Godsave et al. 1988). To ensure that XTC-MIF alone was responsible for inducing notochord and that protein species refractory to silver


597

me

mus Fig. 7. Biological effects of XTC-MIF. Stage-8 Xenopus animal pole explants were cultured in the absence (A) or presence (B-D) of XTC-MIF. They were fixed and sectioned after 3 days at 20°C, when controls had reached stage 40. (A) An explant cultured in the absence of inducing activity forms 'atypical epidermis'. (B) 0-4ngml~' XTC-MIF causes the formation of a partial 'bubble' of thin epidermis, containing mesenchyme and mesothelium. (C) 3ngml~' XTC-MIF induces muscle and neural development. (D) 25ngml~' XTC-MIF induces notochord, cement gland and neural development. Scale bar in A is 100/OT), and applies to all four frames. Abbreviations: epi, epidermis; me, mesenchyme; mst, mesothelium; mus, muscle; not, notochord; eg, cement gland; nt, neural tissue.

staining were not acting synergistically with it, a sample of HPLC-purified XTC-MIF was applied to a 17x15cm 13-5 % acrylamide gel, in the absence of a reducing agent. After electrophoresis the XTC-MIF band was located by CuCl2 staining (Lee et al. 1987) and eluted into PBS. Recovery of inducing activity by this procedure was poor, presumably due to nonspecific adherence of XTC-MIF to the walls of the tube. However, the gel-purified XTC-MIF induced notochord at concentrations between 3ngml~' and 50ngml~'. On the batches of embryos used for this experiment, induction of ventral mesodermal tissue required XTC-MIF concentrations of about lngml" 1 . Discussion XTC-MIF and TGF-fi In this paper, we describe the purification, partial characterization and biological effects of the XTC mesoderm-inducing factor. The work represents one of the few cases in which an inducing factor has been identified through its inducing activity and purified using this as an assay (see also Schaller & Bodenmuller, 1981; Kay et al. 1983; Morris et al. 1987). XTC-MIF is a hydrophobic protein with an isoelectric

point of 7-8. Under nonreducing conditions, its relative molecular mass, according to polyacrylamide gel electrophoresis, is 23 500. On reduction, XTC-MIF loses its activity and polyacrylamide gel electrophoresis reveals one or two bands of approximately 15 000Mr. The reason for this variability in the electrophoretic behaviour of the reduced form of XTC-MIF is unclear. It may reflect unstable posttranslational modification of the protein. The properties of XTC-MIF described above most closely resemble those of transforming growth factortype p (TGF-/3) (see reviews by Moses et al. 1985; Roberts & Sporn, 1987; Massague, 1987), and indeed XTC-MIF comigrates with TGF-/3 on polyacrylamide gels (data not shown). Another point of similarity is that the inducing activity of XTC conditioned medium is activated by heat (Smith, 1987) and low pH (Rosa et al. 1988); the latent form of TGF-/3 is also activated by these treatments (Lawrence et al. 1985; O'Connor-McCourt & Wakefield, 1987). It is of interest, therefore, that Kimelman & Kirschner (1987) find that TGF-/31 acts synergistically with FGF in inducing muscle from Xenopus animal pole explants, and it is of even greater interest that Rosa et al. (1988) have demonstrated that TGF-yS2 can induce muscle in the absence of any other factor. Furthermore, Rosa et al. (1988) find that XTC con-

598


ditioned medium contains TGF-yS-like activity, and that some of the inducing activity of XTC conditioned medium can be abolished by antibodies to TGF-/G but not antibodies to TGF-/21. The simplest interpretation of the results of Rosa et al. (1988) and of those described here is that the XTC mesoderm-inducing factor is Xenopus TGF-/J2. However, XTC-MIF induces muscle (see Fig. 7C) at a concentration of about 2 ng ml"', whereas 12 ng ml"' of porcine TGF-/J2 is required to activate musclespecific actin gene expression (Rosa et al. 1988). Another difference is that, unlike TGF-£1 and TGFp2, XTC-MIF loses some activity after prolonged acidification. These discrepancies could be due to the species difference, but the TGF-/3s show a remarkable degree of sequence conservation between species (see Massague, 1987). It is possible, therefore, that XTC-MIF represents a novel third form of TGF-/3. We intend to investigate this by obtaining Nterminal sequence data from XTC-MIF and screening an XTC cDNA library. This will also enable us to discover whether XTC-MIF mRNA is present in early Xenopus embryos and if so, whether, like Vgl (Melton, 1987), it is localized to the vegetal hemisphere. Biological effects of XTC-MIF The earliest biological effect of XTC-MIF on isolated Xenopus animal pole regions is to induce a change in shape, involving elongation and constriction. We believe that these movements mimic the gastrulation movements of the normal dorsal mesoderm (Symes & Smith, 1987). Subsequently, induced explants differentiate into a variety of cell types, including neural tissue, notochord, muscle, mesenchyme, mesothelium and epidermis. Like Godsave et al. (1988), who used several partially purified MIFs (including XTC-MIF, a MIF from the WEHI cell line, and one from chick embryos) and Grunz (1983), who used Tiedemann's 'vegetalizing factor' (Born et al. 1972), we find that the range of cell types formed depends upon the concentration of MIF. At low concentrations (