Expression of the α6A integrin splice variant in developing mouse embryonic stem cell aggregates and correlation with cardiac muscle differentiation

Differentiation (1999) 64:173–184

© Springer-Verlag 1999

O R I G I NA L A RT I C L E

Sólveig Thorsteinsdóttir · Bernard A.J. Roelen Marie-José Goumans · Dorien Ward-van Oostwaard Ana C. Gaspar · Christine L. Mummery

Expression of the α6A integrin splice variant in developing mouse embryonic stem cell aggregates and correlation with cardiac muscle differentiation &misc:Accepted in revised form: 10 November 1998

&p.1:Abstract Mouse embryonic stem (ES) cells grown in aggregates give rise to several different cell types, including cardiac muscle. Given the lack of cardiac muscle cell lines, ES cells can be a useful tool in the study of cardiac muscle differentiation. The laminin-binding integrin α6β1 exists in two different splice variant forms of the α chain (α6A and α6B), the α6A form having been implicated as possibly playing a role in cardiac muscle development, based on its distribution pattern [4, 53]. In this study we characterise the ES cell model system in terms of the expression of the two different α6 splice variants. We correlate their expression with that of muscle markers and the transcription factor GATA-4, using the reverse transcription-polymerase chain reaction (RTPCR). We confirm that α6B is constitutively expressed by ES cells. In contrast, α6A expression appears later and overlaps in time with a period when the muscle marker myosin light chain-2V (MLC-2V) is expressed, but no MyoD is present, which indicates the presence of cardiac muscle cells in the aggregates. We further show that GATA-4 is present at the same time. Culturing the aggregates under conditions that stimulate (transforming growth factor β1 supplement) or inhibit (TGFβ1 plus 10− 9 M retinoic acid supplement) cardiac muscle differentiation does not lead to any qualitative differences in the timing of expression of these genes, but quantitative changes cannot be excluded. The TGFβ1 supplement does, however, lead to a relatively greater expression of α6A compared to α6B than the TGFβ1 plus 10−9 M RA supplement after 6 days in culture, suggesting that α6A expression is favoured under conditions that stimulate S. Thorsteinsdóttir (✉) · A.C. Gaspar Department of Zoology and Centre of Environmental Biology, Faculty of Sciences, University of Lisbon, P-1749-016 Lisbon, Portugal e-mail: [email protected], Fax: +351-1-7500028 B.A.J. Roelen · M.-J. Goumans · D. Ward-van Oostwaard C.L. Mummery Hubrecht Laboratory, Netherlands Institute for Developmental Biology, Upsalalaan 8, 3584CT Utrecht, The Netherlands&/fn-block:

cardiac muscle differentiation. The switch towards α6A expression in ES cell aggregates is paralleled by expression of the binding receptor for TGFβ (TβRII). Stable expression of a mutated (dominant negative) TβRII in ES cells, however, still resulted in (TGFβ-independent) upregulation of α6A, demonstrating that these events were not causally related and that parallel or alternative regulatory pathways exist. The initial characterisation of differentiating ES cell aggregates in terms of α6A integrin subunit expression suggests that this model system could be a valuable tool in the study of the role of the α6Aβ1 integrin in cardiac muscle differentiation.&bdy:

Introduction Integrin-mediated cell-matrix interactions play numerous roles during embryonic development [8, 22]. They not only mediate the connection between cells and the surrounding extracellular matrix, but also transmit specific signals to the interior of the cell, a process involving the integrin cytoplasmic domains [44, 59]. The interaction between integrins and laminins can lead to many developmentally relevant processes such as cell polarisation, cell differentiation and cell migration [1, 10, 29]. Laminins are a family of proteins that are assembled into trimeric complexes from genetically distinct subunit chains. Several genetic variants of the subunit chains have been identified and give rise to a number of laminin isoforms (reviewed in [6, 55]). A great number (α1β1, α2β1, α3β1, α6β1, α6β4, α7β1, α9β1, αIIβ3 and αvβ3) of integrin receptors interact with laminins [6]. Several of these integrins have subunits that undergo alternative splicing in their cytoplasmic domains. Given that the cytoplasmic domains specify the signal transmitted to the cell upon ligand binding [44, 59], these integrin subunit splice variants have caught the interest of several investigators. The α6 integrin subunit exists in two cytoplasmic splice variant forms, α6A and α6B [5, 17, 18, 52]. We and others have found that α6B is expressed throughout preimplantation

174

development, but that α6A first appears in inner cell mass (ICM) outgrowths [17, 51], presumably in parietal endoderm, since ES cells and F9 embryonic carcinoma cells induced to differentiate towards parietal endoderm also express α6A [17, 23, 34] and immunoreactivity for α6A is present on parietal endoderm cells in vivo [4]. In the embryo proper, the first site of α6A expression is in the heart [4, 53]. α6A expression is associated with the myocardium of the primitive heart tube of embryos at 8.5–9.5 days post coitum (p.c.), thus suggesting a role in the differentiation of cardiac muscle. β4 is not present in the embryo at these early stages [17, 53] and is not detectable in cell lysates from undifferentiated or differentiated F9 cells [23]. Thus, since the α6 subunit is only known to associate with β1 and β4 [49], the α6A-containing integrin present in both parietal endoderm and cardiac muscle is presumably α6Aβ1. Embryonic stem (ES) cells grown in aggregates can give rise to a variety of different cell types, including extraembryonic endoderm, mesenchymal cells, endothelial cells, neurons, cardiac and skeletal muscle, the extent of each cell type depending on the culture conditions used [9]. Cardiac muscle forms when aggregates are cultured in media supplemented with 20% fetal calf serum, and these aggregates express muscle genes in a stage-specific pattern reflecting the sequence in the embryo [32, 33, 39]. The rate and extent of cardiac muscle differentiation is stimulated by TGFβ1 and TGFβ2 [47], while a combination of TGFβ and low concentrations of retinoic acid (RA) [47], or high concentrations of RA [47, 60], were found to inhibit cardiac muscle formation. The ES cell system has some limitations as a model system for cardiac muscle differentiation, such as the simultaneous presence of several differentiated cell types other than cardiac muscle, but does provide the advantage that the extent of cardiac muscle formation can be manipulated with differing culture conditions. Overall, therefore, ES cells are a promising tool in the study of the factors involved in cardiac muscle differentiation. The objectives of this study were to characterise the ES cell culture system in terms of the expression of the α6 integrin splice variants, to determine whether a correlation exists between the expression of the α6A variant and cardiac muscle differentiation in the aggregates, and to establish whether TGFβ1 regulates the expression of this variant. Therefore, ES cells were cultured in aggregates for up to 12 days and sampled at several time points for reverse transcription-polymerase chain reaction (RT-PCR) analysis. The results obtained show a temporal overlap between the expression of the α6A integrin subunit and genes expressed in cardiac muscle. Furthermore, conditions that stimulate cardiac muscle formation in ES cell aggregates (TGFβ1 treatment) lead to an increase in α6A expression relative to α6B expression compared to conditions that inhibit this process (TGFβ1 plus 10−9 M RA treatment). We show, however, that upregulation of α6A is not due to the direct action of TGFβ1, and is thus most likely due to the activation of the cardiac muscle differentiation program per se. Our

results demonstrate that the ES cell system can be useful in the study of the mechanisms of cardiac muscle differentiation and of the potential role of the α6Aβ1 integrin in this process.

Methods ES cell culture and differentiation. Undifferentiated mouse ES cells, ES 5 [35, 36] and E14 [19], were grown on gelatinised flasks (Nunc, Roskilde, Denmark) in Minimal Essential Medium (MEM; Gibco, Paisley, UK) conditioned by buffalo rat liver (BRL) cells [48], supplemented with 20% heat-inactivated fetal calf serum (FCS; Gibco) and 10−4 M β-mercaptoethanol (Sigma, St. Louis, USA). Aggregation was carried out in hanging drops [37] essentially as described previously [47]. Briefly, ES cells were harvested by trypsin/EDTA and resuspended in MEM supplemented with 20% dextran-coated charcoal-treated FCS (DCC-FCS), thus devoid of retinoids [20], 5 mg/ml insulin (Sigma) and 10−4 M β-mercaptoethanol, at a density of 104 cells/ml. Twenty-microliter drops (containing approximately 800 cells) were placed on the inside of lids of Petri dishes, the lids placed on top of the dishes, which were filled with phosphate buffered saline (PBS), and then incubated at 37º C with 7.5% CO2. This was designated day 0. For RT-PCR analysis, aggregates were plated in gelatine-coated 24-well plates (Nunc) on day 3 and cultured in 0.5 ml medium. Cultures were either terminated on day 6 or the medium was changed on days 6 and 9 and the cultures terminated on day 12. For immunofluorescence analysis of differentiation marker expression, aggregates were plated on day 3 or 4 and the cultures fixed on day 4, 6, 9, 12, 14 or 15. The medium was changed on days 6, 9 and 12. A morphological assessment of the aggregates was performed on days 4, 5, 6, and (if applicable) days 9 and 12. In experiments designed to assess the effect of exogenously added TGFβ1 and/or 10−9 M RA on the expression of the α6 integrin and muscle markers, cells were aggregated under four different treatment regimes [47]. The control group was cultured in aggregation medium only (see above), the TGFβ1 group was cultured in the same medium supplemented with 2 ng/ml TGFβ1 (purified from platelets [57]), the TGFβ1 plus 10−9 M RA group was grown in the presence of 2 ng/ml TGFβ1 and 10−9 M all-trans RA (Sigma), and, finally, the last group was cultured in the presence of 10−9 M RA. ES cell lines expressing TβRII and dominant negative TβRII(∆TβRII). These cell lines were derived from E14 ES cells by electroporation transfection, of phosphoglycerate kinase (PGK) promoter-driven constructs (TβRII being full length and ∆TβRII lacking the entire serine/threonine kinase domain) containing a hygromycin selection cassette (see [12]). Their response to TGFβ1 was tested by transient transfection with 3 µg reporter construct, p3TP-lux, which contains the TGFβ1-sensitive promoter for plasminogen activator inhibitor (PAI-1). After transfection, cells were cultured for 16 h in the absence or presence of TGFβ1 (10 ng/ml) and luciferase activity measured as described previously [7]. The full-length construct was tested in undifferentiated ES cells, which do not express TβRII endogenously, and the ∆TβRII in cells differentiated by removal of leukemia inhibitory factor (LIF) for 8 days; under these conditions the endogenous receptor is expressed [12]. To answer the question as to whether TGFβ1 upregulates α6A expression, cells lines stably expressing either TβRII or ∆TβRII ectopically were aggregated for 3 days in control medium, medium supplemented with TGFβ1, TGFβ1 plus 10−9 M RA or 10−9 M RA as above, and then plated in the same medium. Samples for RT-PCR were collected on days 3 and 7.

175 Immunofluorescence Aggregates were either fixed in methanol for 10 min at −20º C or in 2% paraformaldehyde for 15 min, and then washed in PBS. When fixed with paraformaldehyde, aggregates were permeabilised with 0.1% Triton X-100 in PBS for 4 min. The following primary antibodies were used: the rat monoclonal antibody TROMA1, recognising the cytoskeletal protein endo A, and a commonly used marker for mouse endoderm [26] (a kind gift from Dr. Rolf Kemler), rabbit polyclonal anti-laminin (Sigma), monoclonal rat anti-PE-CAM1 (390) [3] (a kind gift from Dr. Clayton Buck), and mouse monoclonal anti-desmin (D3; developed by Dr. DA Fischman and obtained from the Developmental Studies Hybridoma Bank, maintained by the University of Iowa, USA under contract NO1-HD-7–3263 from the NICHD). Aggregates were blocked in 4% normal goat serum in PBS and incubations in primary antibody were done for 1 h at room temperature or overnight at 4º C. After washing in PBS, the coverslips were incubated with the appropriate affinity-purified FITC- or TRITC-conjugated secondary antibody (all from Sigma). Finally, the coverslips were washed in PBS, mounted in 2.5% 1,4-diazabicyclo 2,2,2 octane (DABCO; Sigma) in a 9:1 mixture of glycerol:PBS with 0.1% sodium azide or in Mowiol (Calbiochem, La Jolla, USA). RNA isolation and RT-PCR. Aggregates of various differentiation stages were lysed in Ultraspec TM (Biotecx, Veenendaal, The Netherlands), 15 mg polyinosonic acid (Boehringer, Almere, The Netherlands) was added as carrier, and total RNA was isolated according to the instructions of the manufacturer. cDNA was synthesised from the RNA with 200 U SuperScript reverse transcriptase (Gibco BRL, Breda, The Netherlands), 0.5 mg oligo (dT) primer, 0.5 mM dNTP, 10 mM DTT and 1 U RNasin (Promega, Leiden, The Netherlands) in reaction buffer (Gibco BRL). The reaction mixture was incubated at 41º C for 90 min, heated to 95º C for 5 min and then chilled on ice. PCR was performed in a DNA thermal cycler (Perkin Elmer GeneAmp PCR system 2400) essentially as described previously [53]. Samples were first denatured for 5 min at 94º C, after which they were amplified for 40 cycles, each consisting of 15 s denaturation at 94º C, 30 s annealing at a temperature adequate for each primer set (Table 1), and 45 s primer extension at 72º C. After the 40 cycles, the samples were kept at 72º C for 5 min. Amplification with β-actin primers verified the presence of cDNA in each sample, and samples where no β-actin was found were discarded. Table 1 Oligonucleotide primer sequences for: α6, α6 integrin splice variants; MLC-2V, myosin light chain-2V; MyoD; GATA-4; TβRII, transforming growth factor β receptor II; β-actin&/tbl.c:&

Amplified transcript

&/tbl.:

α6 Primer 1 Primer 2 MLC-2V Primer 1 Primer 2 MyoD Primer 1 Primer 2 GATA-4 Primer 1 Primer 2 TβRII Primer 1 Primer 2 β-actin Primer 1 Primer 2

Negative and positive controls were always included in each PCR amplification. Negative controls were samples where reverse transcription was excluded (RNA control) and samples where cDNA was excluded (water control). The 12.5-day-p.c. embryo heart cDNA and C2C12 myoblast cell line cDNA were used as positive controls. PCR products were visualised by electrophoresis on 1.2% agarose gels containing 0.4 mg/ml ethidium bromide, and the gels photographed under UV with a camera connected to the Imager system (Appligene). Calculation of the ratio of α6A and α6B from electrophoresed PCR products. Although the absolute amount of PCR product is not comparable between samples, the ratio of the α6A and α6B PCR products within a sample is comparable to the ratio obtained within other samples. We can, therefore, compare the relative amount of each splice variant between different time points during the development of ES cell aggregates. PCR products were run on ethidium bromide-containing polyacrylamide gels, and the image of the gel under UV light was documented using the Imager system. The intensity of the ethidium bromide staining could be quantified using the ImageQuant software (Molecular Dynamics). Corrections were made for the size of the products and for the heteroduplex formation [18], after which the ratio of the two products could be calculated. This ratio is thus the value that is comparable between the samples.

Results Morphological differentiation of ES cell aggregates after plating ES cells were grown as aggregates in hanging drops for 3 days and then plated. A morphological assessment of the cell types present was performed on days 4, 5, 6, 9, and 12 using the morphological features described previously [47]. In some cases, the nature of the cell types present was confirmed by immunostaining with differentiation markers (see Methods). Although there was some variability between experiments, the general pattern of

Oligonucleotide sequence

Size of PCR product

Annealing Reference temperature

5′-AAATACCAGACTCTCAACTGCA-3′ 5′-TGAAACTGTAGGTCCATACTGG-3′

A: 517 bp B: 387 bp

55ºC

5′-GCCAAGAAGCGGATAGAAGG-3′ 5′-CTGTGGTTCAGGGCTCAGTC-3′

499 bp

55º C

33

5′-CACTACAGTGGCGACTCAGA-3′ 5′-TGCTGCTGCAGTCGATCTCT-3′

425 bp

55º C

33

5′-TAACTCCAGCAATGCCACTAGC-3′ 5′-CTGATTACGCGGTGATTATGTC-3′

286 bp

55º C

5′-GGAAGTCTGCGTGGCCGTGTGG-3′ 5′-CTATGGCAATCCCAGCGGAGG-3′

299 bp

52º C

56

5′-TGAACCCTAAGGCCAACCGTG-3′ 5′-GCTCATAGCTCTTCTCCAGGG-3′

396 bp

55º C

54

176

177 Table 2 Major cell types in plated mouse embryonic stem (ES) cell aggregates and their approximate time of appearance&/tbl.c:&

ES cells Endoderm Mesenchyme-like cells Fusiform cells Endothelial vessels Contractile muscle

D4

D5

D6

D9

D12

+ + ± − − −

+ + + − − −

+ + + ± − −

+ + + + + ±

+ + + + + +

&/tbl.: Table 3 Expression of α6, MLC-2V, MyoD, GATA-4 and TβRII in developing ES cell aggregates cultured in control medium. Compilation of data obtained from 7 different experiments (presence of RT-PCR product/number of experiments where a particular gene was assayed).The age of the aggregates is shown in days (D3 to D12). D0 are ES cells taken directly from monolayer culture; na, not assayed&/tbl.c:&

α6A α6B MLC-2V MyoD GATA-4 TβRII β-actin

D0

D3

D4

D5

D6

D9

D12

0/2 2/2 0/1 0/1 0/2 0/2 2/2

3/7 7/7 2/3 0/1 2/2 1/2 7/7

1/2 2/2 na na 2/2 1/1 2/2

2/2 2/2 na na 2/2 1/1 2/2

7/7 7/7 3/3 0/1 2/2 2/2 7/7

4/4 4/4 2/2a 0/1 1/1 1/1 4/4

5/5 5/5 2/2 0/1 1/1 1/1 5/5

a

In one of these two samples, a PCR product for MLC-2V was only just detectable following 40 rounds of amplification, but was, however, clearly present after 42 rounds. This indicates that MLC2V expression was near the detection limit of PCR in this particular sample. In the other sample, an unambiguous PCR product was obtained after 40 rounds of amplification (see Fig. 2B) &/tbl.:

differentiation was as shown in Table 2. Endodermal cells were the first differentiated cells to become morphologically distinguishable (Fig. 1A insert) and their endodermal nature was confirmed with TROMA-1 antibody staining (Fig. 1A, B). Cells with a mesenchymelike morphology and that did not stain with the TROMA1 antibody (Fig. 1B) were evident by day 4 or 5. A TROMA-1-negative (data not shown) fusiform cell type (Fig. 1C) was present as early as day 6. These cells did not stain with antibodies to desmin or myosin (data not shown), and were thus not myoblasts. They did, however, stain strongly for laminin (Fig. 1C). Primary cultures

Fig. 1A–E Morphology and differentiation marker expression of some cell types present in differentiating mouse embryonic stem (ES) cell aggregates. Phase contrast image on the left, immunofluorescence image on the right. A Well-spread endodermal cells in a day-6 aggregate and the respective TROMA-1 antibody staining; insert typical morphology of endoderm (edge of a day-12 aggregate). B TROMA-1 staining on the edge of a central mass of (undifferentiated) cells in a day-14 aggregate. Mesenchyme-like, TROMA-1-negative cells are evident away from the central aggregate (arrow). C Fusiform cells in a day-15 aggregate show strong intracellular immunostaining for laminin (arrows). D Endothelial vessels in a day-14 aggregate are evident by immunolabeling for PE-CAM; insert morphology of endothelial vessels in a day-9 aggregate. E Spindle-shaped, desmin-positive cells (arrows) on top of a mass of cells in a day-15 aggregate; scale bar 50 µm&ig.c:/f

of muscle-derived fibroblasts have been shown to have strong intracellular staining for laminin [28], so these fusiform cells could have been fibroblasts. Endothelial vessels could be distinguished morphologically by day 9 (Fig. 1D insert), and immunlabeling with an antibody recognising the endothelial marker PE-CAM [3] confirmed the presence of endothelial cells in the cultures (Fig. 1D). Contracting (cardiac) muscle was sometimes present in aggregates scored on day 9, while by day 12 the great majority of culture experiments had several contractile foci. Although there was variation between culture experiments, in our hands 20%–30% of aggregates grown in control medium exhibited contractile foci by day 12. Finally, we did occasionally find spindle-shaped desminpositive staining in late-stage cultures (Fig. 1E), strongly suggesting the presence of skeletal myoblasts, which is in accordance with previous studies [43, 47]. RT-PCR analysis of the expression of α6, muscle markers, GATA-4 and TβRII during ES cell differentiation Our next step was to determine when the α6 integrin splice variants appeared in these developing aggregates and to compare their patterns of expression to those of different differentiation markers. RT-PCR was performed on ES cell aggregates collected at different time points in a total of seven different culture experiments. The results are summarised in Table 3. α6B was expressed in undifferentiated ES cells and at all stages of aggregate development, which is consistent with our earlier observation that α6B is present at all stages of mouse embryogenesis studied so far [17, 53]. α6A was not expressed in undifferentiated ES cells, but was present on day 3 in 3/7 (43%) of the experiments performed, and was then consistently present in aggregates from day 5 onwards. We then assayed for the expression of two muscle markers, myosin light chain-2V (MLC-2V), known to be expressed both in cardiac and skeletal muscle [38] and MyoD, a skeletal-muscle-specific marker [31]. MLC-2V expression was already detectable on day 3 in 2/3 of the experiments performed, and expression persisted in the other stages studied. In contrast, MyoD expression was not found at any stage studied when the aggregates were grown in control medium, but was, however, detected on day 12 when 10−9 M RA was present in the medium (see Fig. 2C). The early presence of MLC-2V is in contrast with a previous report [33] where MLC-2V transcripts were only found on day 8 of aggregation. The reason for this is presently not known, but could be due to different ES cell lines and culture methods being used. Although MLC-2V is expressed in slow skeletal muscle fibres [33], these appear only after MyoD is expressed [31]. We can therefore conclude that the MLC-2V expressed before day 12 of aggregate differentiation was due to the differentiation of cardiac muscle. The late appearance of MyoD was in accordance with previous studies in ES

178

cells [33, 43], which have shown that skeletal muscle appears considerably later than cardiac muscle, reflecting the situation in the embryo. We then determined the expression pattern of GATA4, which is essential for the early stages of visceral endoderm differentiation in vitro [50], and has also been implied as being functional in cardiac morphogenesis [2, 13, 14]. As shown in Table 3, GATA-4 is already expressed in day 3 aggregates and remains expressed at all the remaining time points studied. In a previous study [13], a small amount of GATA-4 expression was detected by Northern blotting in ES cell aggregates as early as day 5. Given the fact that RT-PCR is more sensitive than Northern blotting, our results appear to be in agreement with these findings. Finally, to know when TGFβ1 could start having an effect on the aggregates (see experiments below), we determined when its binding receptor, known as the TGFβ type II receptor (TβRII), was first expressed. The presence of this receptor, together with a type I receptor (TβRI or alk5) is a prerequisite for TGFβ-mediated signal transduction (reviewed in [30]). TβRII is not expressed by undifferentiated ES cells (Table 3; also see [41]), but we found expression on day 3 in one out of two experiments, and on day 4 onwards in the remaining experiments assayed. This suggests that the aggregates could respond to TGFβ at least from day 4 onwards, since ES cells express the necessary co-receptor (TβRI) endogenously, independent of their differentiation state [12]. Effect of added TGFβ1 and/or 10−9 M RA on the differentiating ES cell aggregates It has previously been reported that culturing ES cell aggregates in the presence of TGFβ1 or TGFβ2 increases the rate and amount of cardiac muscle formed in cultured aggregates, while in the additional presence of 10−9 M RA, differentiation of this cell type is retarded [47]. To determine whether these treatments have any effect on the expression of the genes under study, we aggregated ES cells in control medium containing DCC-FCS or in the same medium containing TGFβ1, TGFβ1 plus 10−9 M RA, or 10−9 M RA only. Four independent experiments with all four treatments were performed: in Experiments 1 and 2, aggregates were collected at days 3, 6, 9 and 12 and in Experiments 3 and 4, they were collected at days 3, 4, 5 and 6. The appearance of contractile muscle was monitored on days 6, 9 and 12 in Experiments 1 and 2. In Experiment 1, contractile muscle was observed on day 6 in aggregates growing in the presence of TGFβ1, and on day 9 in aggregates supplemented with 10−9 M RA. By day 12, all culture treatments had contractile muscle. In Experiment 2, no contractile muscle was evident on days 6 or 9, but at day 12, aggregates growing under all culture treatments had foci of contractile muscle. Thus, in this second experiment, there was no difference in the timing with which contractile muscle appeared. Howev-

er, in the presence of TGFβ1, 4 of a total of 13 aggregates (31%) developed foci of contractile muscle versus 3/11 (27%) in control medium. In the presence of TGFβ1 plus 10−9 M RA, only 2/12 aggregates (17%) had contractile muscle, while with 10−9 M RA, 4/14 aggregates (29%) developed beating foci. These results are in agreement with a previous study [47] where the same culture treatments were used. Although there is some heterogeneity between experiments evident in both this and the previous study [47], the stimulatory effect of TGFβ on the rate and/or extent of cardiac muscle development was reproducible, as was the suppressive effect of the additional presence of 10−9 M RA. RT-PCR analysis of the expression of the α6 splice variants (Fig. 2A, D) showed no qualitative difference between treatments. However, the relative amounts of α6A and α6B did vary between treatments (Fig. 2A, D; also see below). The timing of MLC-2V (Fig. 2B) and GATA-4 (Fig. 2E) expression was identical between treatments. However, unlike control and TGFβ1-treated aggregates, MyoD expression was evident by day 12 when 10−9 M RA was present in the medium (Fig. 2C). This is consistent with an earlier study [60] documenting a stimulatory effect of RA at certain concentrations and durations of exposure on skeletal myogenesis in ES cells differentiating as aggregates. Since there appeared to be a difference in the relative amounts of α6A and α6B between treatments in all experiments, we wanted to quantify this difference and determine whether a pattern could be detected for each treatment. We thus measured the relative intensities of the bands representing each splice variant for each treatment and time point, and calculated the ratio of α6A/α6B (see Methods). The results were consistent between experiments, and the data from one experiment (Experiment 4) are plotted in Fig. 3. Aggregates grown in control medium expressed the relatively highest ratio of α6A/α6B by the end of the culture period. Aggregates grown in the presence of TGFβ1 plus 10−9 M RA had consistently low ratios of α6A/α6B by the end of the culture period, and aggregates grown in the presence of TGFβ1 alone had a higher ratio of α6A/α6B by the end of the culture period than the ones grown in the presence of TGFβ1 plus 10−9 M RA. Aggregates grown in the presence of 10−9 M RA had low α6A/α6B ratios. Does TGFβ regulate differential expression of the α6A splice variant? The upregulation of TβRII and α6A occurred virtually simultaneously in differentiating ES cell aggregates (day 3–4 vs. day 3–5; see Table 3). Since TGFβ has been shown to upregulate the mRNA and protein of various β and α integrin subunits (reviewed in [27]), we addressed the question of whether there was a causal relationship between increased TβRII expression and the selective increase in α6A during the culture period, or whether these events were coincidental. To answer this question, we

179

Fig. 2A–E Reverse transcriptase-polymerase chain reaction (RTPCR) analysis of the expression of the α6 integrin splice variants, muscle markers and GATA-4 in differentiating ES cell aggregates grown in aggregation medium (DCC), medium supplemented with transforming growth factor-β1 (TGFβ1), TGFβ1 plus 10−9 M retinoic acid (RA) or only with 10−9 M RA. A–C Experiment 1 where aggregates were collected on days 3, 6, 9 and 12 and analysed for the expression of the α6 splice variants (A; α6A, 517 bp; α6B, 387 bp), MLC-2V (B) and MyoD (C). D–E Experiment 3 where aggregates were collected on days 3, 4, 5 and 6 and analysed for the expression of the α6 splice variants (D) and GATA-4 (E). No qualitative differences except for MyoD expression are observed between culture treatments&ig.c:/f

used ES cell lines that expressed either full-length or truncated (dominant negative) TβRII constitutively [12]. We firstly demonstrated that a TGFβ reporter construct (3TP-lux) transfected transiently into the ES cell lines responded as expected to TGFβ1 (Fig. 4A), i.e. the reporter was induced in an undifferentiated ES cell line expressing the full-length TβRII, whereas induction of the reporter by TGFβ1 in differentiated cells expressing the endogenous receptor was blocked by the dominant negative TβRII construct. α6A was not upregulated by TGFβ1 treatment of ES cells expressing the full-length TβRII when grown as a monolayer (data not shown), but since these ES cells grown in monolayer have been shown to be not responsive to TGFβ (presumably due to the lack of downstream molecules [12]), we also grew these ES cells in aggregates, where the TGFβ signalling pathway is intact [12]. When these cells were grown in aggregates, α6A expression was detected on day 3 (Fig. 4B), comparable to that in control ES cells (see Table 3). In ES cells containing the dominant negative TβRII, α6A expression was not blocked in aggregates formed from ES cells (Fig. 4B), showing that the absence of TGFβ signalling does not affect the onset of α6A expres-

180 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 D3

D4

D5

D6

Fig. 3 Ratio α6A/α6B for the different culture treatments and time points. Culture treatments were aggregation medium (DCC, ❚), medium supplemented with TGFβ1 (▲), TGFβ1 plus 10−9 M RA (•) and 10−9 M RA (✦). Results shown are from Experiment 4, where aggregates were collected on days 3, 4, 5 and 6. The ratio is highest when aggregates were cultured in DCC medium. Aggregates cultured with TGFβ1 had a higher ratio than the aggregates cultured with TGFβ1 plus 10−9 M RA or 10−9 M RA alone&ig.c:/f

sion. We concluded therefore that the ability to signal in response to TGFβ (i.e. TβRII expression) and α6A upregulation are not causally related and that the expression of these genes is independently regulated. We then calculated the ratio of α6A to α6B in the two different cell lines growing in control medium or in medium supplemented with TGFβ1. The results obtained are shown in Fig. 4C. It is clear from this figure that in both transfected cell lines there was no apparent difference between the α6A/α6B ratios in control versus TGFβ1-supplemented medium on day 7. This is in contrast to the data from wild-type ES cells (see Fig. 3) where the α6A/α6B ratio was about 2-fold higher in control medium than it was in TGFβ1-supplemented medium at the end of the culture period. Furthermore, it is interesting to note that the α6A/α6B ratio was higher in the cell line expressing the dominant negative construct as opposed to the one expressing the full-length TβRII construct. The possible reasons for this will be discussed below.

Discussion In the present study we have documented the expression of the α6 splice variants during the development of mouse ES cells in aggregates and correlated the expression of the α6A splice variant with the expression of various differentiation markers. Our results show a temporal correlation between the presence of the α6A integrin subunit, MLC-2V and GATA-4. We and others have previously shown that α6A is strongly expressed in the myocardium of the developing heart [4, 53] and that α6A is expressed in cells of early embryo outgrowths [17, 51], presumably in parietal endoderm [4, 17, 23, 34]. In vivo, MLC-2V is

expressed in the ventricular region of the primitive heart tube from day 8 p.c. and remains restricted to the ventricular compartment as development proceeds [38]. It is also expressed in slow skeletal muscle fibres later in development [33]. GATA-4 has been implied as having a role in cardiac muscle development [2, 13, 14]. It is also expressed by both F9 EC-derived parietal and visceral endoderm-like cells [2] and has been found to be essential for normal visceral endoderm development [50]. We identified the presence of endoderm in the ES cell aggregates based on its characteristic morphology and confirmed its presence with TROMA-1 antibody staining in plated aggregates as early as day 4. Contractile cardiac muscle was present from day 9 or 12, but the fact that MLC-2V expression was already detected at day 3 (and no MyoD was present), suggests that cardiogenic cells are present in the aggregates by that time. Thus, the α6A RT-PCR product obtained from ES cell aggregates after day 3 is likely to represent α6A expressed by parietal endoderm as well as differentiating cardiac muscle cells. In order to investigate further the potential correlation between α6A expression and cardiac muscle differentiation, we cultured aggregates under conditions that stimulate (TGFβ1 supplement) or inhibit (TGFβ1 plus 10−9 M RA supplement) cardiac muscle differentiation in mouse ES cell aggregates [47]. Although no qualitative differences were found between culture treatments, we found that there was a consistent difference between the ratio of α6A/α6B under the different conditions. The culture treatment that most favours α6A expression relative to α6B expression is the control condition (about 1.8-fold higher than with TGFβ1 supplementation and about 3.2fold higher than with TGFβ1 plus 10−9 M RA, see Fig. 3). This is not surprising, since TGFβ1 has been shown to inhibit the differentiation and outgrowth of parietal endoderm cells from the isolated inner cell mass (ICM) by about 8-fold [42] as well as to induce a 1.8-fold inhibition of the growth of already differentiated embryonal carcinoma-derived parietal endoderm-like cells [25]. Thus, in the absence of TGFβ, we would expect considerably more parietal endoderm to form than in the presence of TGFβ and, since these cells express α6A [4, 17, 23, 34], a high α6A/α6B ratio should be obtained. In the presence of TGFβ1, differentiation of parietal endoderm is inhibited but the differentiation of cardiac muscle is favoured (by 3–3.5 fold at culture day 14, [47]). However, in the additional presence of 10−9 M RA, cardiac muscle differentiation is inhibited [47] and, since TGFβ1 is present in the medium, parietal endoderm differentiation is presumably still inhibited. From our data it is clear that TGFβ1-treated aggregates have a higher α6A/α6B ratio than those exposed to TGFβ1 plus 10−9 M RA, thus suggesting that conditions favouring cardiac muscle differentiation lead to relatively higher levels of α6A expression than conditions that inhibit this process. In addition, since the α6A/α6B ratio in control aggregates is only about 1.8-fold higher than in TGFβ1-treated aggregates, while the reduction of parietal endoderm in the presence of TGFβ1 is potentially up to 8-fold [42], it is

181

A

B

Fig. 4A–C Expression of α6A in ES cell lines stably transfected with either a full-length or truncated TβRII construct. A Effect of TGFβ1 on a reporter construct (3TP-lux) sensitive to TGFβ1, in control ES cells (E14), in a cell line expressing the full-length TβRII construct (E14-TβRII), and a cell line expressing the truncated TβRII construct (E14-∆TβRII), before (white bars) and after (black bars) differentiation in monolayer. Control E14 cells respond to TGFβ1 only after differentiation, E14-TβRII cells show a low but significant (about 3.6-fold) induction in undifferentiated cells, and E14-∆TβRII cells show negligible induction both in undifferentiated and differentiated cells. B RT-PCR analysis of α6A and α6B expression in E14-TβRII cells (upper panel) and E14∆TβRII cells (lower panel) differentiated as aggregates in aggregation medium (DCC), medium supplemented with TGFβ1, TGFβ1 plus 10−9 M RA and 10−9 M RA and collected on days 3 and 7. Both cell lines expressed α6A (and α6B) on the 2 days sampled. C Ratio of α6A/α6B for the two cell lines on day 7 in aggregation medium (white bars) and in medium supplemented with TGFβ1 (black bars). No difference is observed between the two culture regimes. The E14-∆TβRII has a higher α6A/α6B ratio than the E14-TβRII line&ig.c:/f

tempting to speculate that a considerable amount of the α6A mRNA present in the aggregates treated with TGFβ1 is derived from cells differentiating into cardiac muscle. This will, however, have to be confirmed in

C

more detailed studies on the effect of TGFβ on ES cell aggregates. We then asked whether the upregulation of α6A in differentiating cardiac muscle [4, 53] could be a direct consequence of TGFβ signalling. A great number of studies have shown that TGFβs regulate integrin subunit expression in a variety of cell types [11, 15, 16, 21, 24, 40, 46, 58, 61]. Not many studies have addressed the effect of TGFβ on the expression of the α6β1 integrin (see [27]), but TGFβ1 has been shown to upregulate the expression of both laminin and the α6 integrin subunit in Moser colon cancer cells [21]. In addition, overexpression of the α6 integrin subunit in quail skeletal myoblasts inhibits proliferation but not differentiation, an effect that is similar to the effect of TGFβ on these cells, suggesting an interplay between the α6 integrin and TGFβ signalling pathways [45]. To answer our question, we used ES cell lines constitutively expressing a fulllength or dominant negative TβRII construct [12]. The cell line expressing the full-length TβRII was shown to be responsive to TGFβ after being induced to differentiate (by removal of LIF and addition of 10−6 M RA) in a monolayer and, under those conditions, gives rise to both mesodermal and endodermal derivatives [12]. In contrast, the dominant negative TβRII cell line showed no TGFβ response and gave rise to primarily endodermal derivatives in monolayer [12]. When these cell lines were grown in aggregates and then assayed for α6A expression in the presence of TGFβ1, we found that neither was α6A induced more strongly in the cells expressing the full-length receptor, nor was α6A expression inhibited in cells expressing the dominant negative TβRII construct. Thus TGFβ signalling does not directly affect the α6A/α6B ratio, but rather the upregulation of α6A appears to be related to the differentiated cell types formed. It is interesting to note that in both transfected cell lines there was apparently no difference between the α6A/α6B ratios with or without TGFβ1. This is not surprising in the case of the cell line expressing the dominant negative TβRII, since TGFβ signalling is blocked in that cell line. In the case of the cell line expressing the

182

full-length receptor constitutively, it is possible that receptor levels are high enough to produce a response due to endogenous TGFβ, present in the differentiating aggregates [12, 47]. Another interesting observation is that the α6A/α6B ratio in the cell line expressing the dominant negative TβRII was considerably higher after differentiation than the same ratio in the cell line expressing the full-length construct. Since α6A is expressed by parietal endoderm [4, 17, 23, 34], and since TGFβ inhibits parietal endoderm formation [42] and proliferation [25], it is not surprising that the α6A/α6B ratio is higher in cells where TGFβ signalling is blocked. Furthermore, when cells expressing the dominant negative receptor are differentiated in monolayer, predominantly endodermal cell derivatives are formed [12] and when these cells are grown in aggregates, more endoderm is formed than in aggregates from control cells or cells expressing the full-length receptor (Goumans and Mummery, unpublished observations). It thus appears that blocking TGFβ signalling with a dominant negative receptor construct leads to more endoderm formation and consequently a higher α6A/α6B ratio. In summary, this study shows a temporal correlation between the expression of the α6A integrin subunit, the muscle marker MLC-2V and the transcription factor GATA-4 during the differentiation of mouse ES cells in aggregates, MLC-2V being a cardiac-muscle-specific marker in the absence of MyoD expression. We show that conditions stimulating cardiac muscle differentiation lead to an increase in the ratio of α6A/α6B as compared to conditions that inhibit cardiac muscle formation, suggesting that the splicing of the A variant is favoured when differentiating cardiac muscle is present in the aggregates and parietal endoderm levels are constant. Furthermore we show that TGFβ signalling does not directly regulate the expression of the α6A variant and conclude that the stimulatory effect that TGFβ has on cardiac muscle development is not directly related to the upregulation of α6A in this cell type. Further studies, including quantitative measures of cardiac muscle and parietal endoderm differentiation, and their correlation with α6A integrin expression levels, should be performed to further elucidate the role of α6A as well as TGFβs during the differentiation of cardiac muscle in the mouse ES cell model system. &p.2:Acknowledgements We thank Eric Freund for help with ES cell culture and RT-PCR and Drs. Rolf Kemler and Clayton Buck for kindly giving antibodies. This work was supported by the European Science Foundation Programme on Developmental Biology (S.T.), the European Community (FEDER) STRIDE programme grant STRDA/C/CEN/527/92 (S.T., C.L.M), SLW (B.R.), the Hubrecht Fund (S.T.), and PRAXIS XXI/BTL/4621/95 (A.C.G).

References 1. Adams JC, Watt FM (1993) Regulation of development and differentiation by the extracellular matrix. Development 117:1183–1198 2. Arceci RJ, King AAJ, Simon MC, Orkin SH, Wilson DB (1993) Mouse GATA-4: A retinoic acid-inducible GATA-binding transcription factor expressed in endodermally derived tissues and heart. Mol Cell Biol 13:2235–2246 3. Baldwin HS, Shen HM, Yan HC, DeLisser HM, Chung A, Mickanin C, Trask T, Kirschbaum NE, Newman PJ, Albelda SM, Buck CA (1994) Platelet endothelial cell adhesion molecule-1 (PECAM-1/CD31): alternatively spliced, functionally distinct isoforms expressed during mammalian cardiovascular development. Development 120:2539–2553 4. Collo G, Domanico SZ, Klier G, Quaranta V (1995) Gradient of integrin α6A distribution in the myocardium during early heart development. Cell Adhesion Commun 3:101– 113 5. Cooper HM, Tamura RN, Quaranta V (1991) The major laminin receptor of mouse embryonic stem cells is a novel isoform of the α6β1 integrin. J Cell Biol 115:843–850 6. Delwel GO, Sonnenberg A (1995) Laminin isoforms and their integrin receptors. In: Horton MA (ed) Adhesion receptors: From basic science to clinical therapy. CRC Press, Boca Raton, pp 9–36 7. de Winter JP, Roelen BAJ, ten Dijke P, van der Burg B, van den Eijnden-van Raaij AJM (1997) DPC4 (SMAD4) mediates transforming growth factor-β1 (TGFβ1) induced growth inhibition and transcriptional response in breast tumor cells. Oncogene 14:1891–1899 8. DeSimone DW (1994) Adhesion and matrix in vertebrate development. Curr Opinion Cell Biol 6:747–751 9. Doetschman TC, Eistetter H, Katz M, Schmidt W, Kemler R (1985) The in vitro development of blastocyst derived embryonic stem cell lines: formation of visceral yolk sac, blood islands and myocardium. J Embryol Exp Morphol 87:27– 45 10. Ekblom P (1993) Basement membranes in development. In: Rohrbach DH, Timpl R (eds) Molecular and Cellular Aspects of Basement Membranes. Academic Press, San Diego, pp 359–383 11. Enenstein J, Waleh NS, Kramer RH (1992) Basic FGF and TGF-β differentially modulate integrin expression of human microvascular endothelial cells. Exp Cell Res 203:499– 503 12. Goumans M-J, Ward-van Oostwaard D, Wianny F, Savatier P, Zwijsen A, Mummery CL (1998) Embryonic stem cells with aberrant TGFβ signalling exhibit impaired differentiation in vitro and in vivo. Differentiation 63:101–113 13. Grépin C, Dagnino L, Robitaille L, Haberstroh L, Antakly T, Nemer M (1994) A hormone-encoding gene identifies a pathway for cardiac but not skeletal muscle gene transcription. Mol Cell Biol 14:3115–3129 14. Grépin C, Nemer G, Nemer M (1997) Enhanced cardiogenesis in embryonic stem cells overexpressing the GATA-4 transcription factor. Development 124:2387–2395 15. Heino J, Massagué J (1989) Transforming growth factor-β switches the pattern of integrins expressed in MG-63 human osteosarcoma cells and causes a selective loss of cell adhesion to laminin. J Biol Chem 264:21806–21811 16. Heino J, Ignotz RA, Hemler ME, Crouse C, Massagué J (1989) Regulation of cell adhesion receptors by transforming growth factor-β. Concomitant regulation of integrins that share a common β1 subunit. J Biol Chem 264:380–388 17. Hierck BP, Thorsteinsdóttir S, Niessen CM, Freund E, Iperen LV, Feyen A, Hogervorst F, Poelmann RE, Mummery CL, Sonnenberg A (1993) Variants of the α6β1 laminin receptor in early murine development: Distribution, molecular cloning and chromosomal localization of the mouse integrin α6 subunit. Cell Adhesion Commun 1:33–53

183 18. Hogervorst F, Kuikman I, van Kessel AG, Sonnenberg A (1991) Molecular cloning of the α6 integrin subunit. Alternative splicing of α6 mRNA and chromosomal localization of the α6 and β4 genes. Eur J Biochem 199:425–433 19. Hooper M, Hardy K, Handyside A, Hunter S, Monk M (1987) HPRT-deficient (Lesch-Nyhan) mouse embryos derived from germline colonization by cultured cells. Nature 326:292– 295 20. Horwitz KB, McGuire WL (1978) Estrogen control of progesterone receptor in human breast cancer cells. J Biol Chem 253:2223–2228 21. Huang S, Chakrabarty S (1994) Regulation of fibronectin and laminin receptor expression, fibronectin and laminin secretion in human colon cancer cells by transforming growth factor-β1. Int J Cancer 57:742–746 22. Hynes RO (1994) Genetic analyses of cell-matrix interactions in development. Curr Opinion Genet Dev 4:569–574 23. Jiang R, Grabel LB (1995) Function and differential regulation of the α6 integrin isoforms during parietal endoderm differentiation. Exp Cell Res 217:195–204 24. Kagami S, Kuhara T, Yasutomo K, Löster K, Reutter W, Kuroda Y (1996) Transforming growth factor β (TGFβ) stimulates the expression of β1 integrins and adhesion by rat mesangial cells. Exp Cell Res 229:1–6 25. Kelly D, Rizzino A (1989) Inhibitory effects of transforming growth factor-β on laminin production and growth exhibited by endoderm-like cells derived from embryonal carcinoma cells. Differentiation 41:34–41 26. Kemler R, Brûlet P, Schnebelen MY, Gaillard J, Jacob F (1981) Reactivity of monoclonal antibodies against intermediate filament proteins during embryonic development. J Embryol Exp Morphol 64:45–60 27. Kim LT, Yamada KM (1997) The regulation of expression of integrin receptors. Proc Soc Exp Biol Med 214:123–131 28. Kühl U, Timpl R, von der Mark K (1982) Synthesis of type IV collagen and laminin in cultures of skeletal muscle cells and their assembly on the surface of myotubes. Dev Biol 93:344–354 29. Lin CQ, Bissell MJ (1993) Multi-faceted regulation of cell differentiation by extracellular matrix. FASEB J 7:737–743 30. Lin HY, Lodish HF (1993) Receptors for the TGF-β superfamily: multiple polypeptides and serine/threonine kinases. Trends Cell Biol 3:14–19 31. Lyons GE, Buckingham ME (1992) Developmental regulation of myogenesis in the mouse. Sem Dev Biol 3:243–253 32. Maltsev VA, Rohwedel J, Hescheler J, Wobus AM (1993) Embryonic stem cells differentiate in vitro into cardiomyocytes representing sinusnodal, atrial and ventricular cell types. Mech Dev 44:41–50 33. Miller-Hance WC, LaCorbiere M, Fuller SJ, Evans S, Lyons G, Schmidt C, Robbins J, Chien KR (1993) In vitro chamber specification during embryonic stem cell cardiogenesis. Expression of the ventricular myosin light chain-2 gene is independent of heart tube formation. J Biol Chem 268:25244– 25252 34. Morini M, Piccini D, Barbieri O, Astigiano S (1997) Modulation of α6Bβ1 integrin expression during differentiation of F9 murine embryonal carcinoma cells to parietal endoderm. Exp Cell Res 232:304–312 35. Mummery CL, van den Eijnden-van Raaij J, Feijen A, Tsung H-C, Kruijer W (1989) Regulation of growth factors and their receptors in early murine embryogenesis. In: de Laat SW, Bluemink JG, Mummery CL (eds) Cell to Cell Signals in Mammalian Development. NATO ASI series H26, pp 231– 245 36. Mummery CL, Feyen A, Freund E, Shen S (1990) Characteristics of embryonic stem cell differentiation: a comparison with two embryonal carcinoma cell lines. Cell Differ Dev 30:195–206 37. Mummery CL, van Achterberg TAE, van den Eijnden-van Raaij AJM, van Haaster L, Willemse A, de Laat SW, Piersma AH (1991)Visceral-endoderm-like cell lines induce differenti-

38.

39.

40.

41.

42. 43.

44.

45.

46.

47.

48.

49.

50.

51. 52. 53.

54.

ation of murine embryonal carcinoma cells. Differentiation 46: 51–60 O’Brien TX, Lee KJ, Chien KR (1993) Positional specification of ventricular myosin light chain 2 expression in the primitive murine heart tube. Proc Natl Acad Sci USA 90:5157–5161 Robbins J, Gulick J, Sanchez A, Howles P, Doetschman T (1990) Mouse embryonic stem cells express the cardiac myosin heavy chain genes during development in vitro. J Biol Chem 265:11905–11909 Roberts CJ, Birkenmeier TM, McQuillan JJ, Akiyama SK, Yamada SS, Chen W-T, Yamada KM, McDonald JA (1988) Transforming growth factor β stimulates the expression of fibronectin and of both subunits of the human fibronectin receptor by cultured human lung fibroblasts. J Biol Chem 263: 4586–4592 Roelen BAJ, Lin HY, Knezevic V, Freund E, Mummery CL (1994) Expression of TGF-βs and their receptors during implantation and organogenesis of the mouse embryo. Dev Biol 166:716–728 Roelen BAJ, Goumans M-J, Zwijsen A, Mummery CL (1998) Identification of two distinct functions for TGF-β in early mouse development. Differentiation 64:19–31 Rohwedel J, Maltsev V, Bober E, Arnold H-H, Hescheler J, Wobus AM (1994) Muscle cell differentiation of embryonic stem cells reflects myogenesis in vivo: Developmentally regulated expression of myogenic determination genes and functional expression of ionic currents. Dev Biol 164:87–101 Sastry SK, Horwitz AF (1993) Integrin cytoplasmic domains: mediators of cytoskeletal linkages and extra- and intracellular initiated transmembrane signalling. Curr Opinion Cell Biol 5:819–831 Sastry SK, Lakonishok M, Thomas DA, Muschler J, Horwitz AF (1996) Integrin α subunit ratios, cytoplasmic domains, and growth factor synergy regulate muscle proliferation and differentiation. J Cell Biol 133:169–184 Sheppard D, Cohen DS, Wang A, Busk M (1992) Transforming growth factor β differentially regulates expression of integrin subunits in guinea pig airway epithelial cells. J Biol Chem 267:17409–17414 Slager HG, van Inzen W, Freund E, van den Eijnden-van Raaij AJM, Mummery CL (1993) Transforming growth factor-β in the early mouse embryo: Implications for the regulation of muscle formation and implantation. Dev Genet 14:212– 224 Smith AG, Hooper ML (1987) Buffalo rat liver cells produce a diffusible activity which inhibits the differentiation of murine embryonal carcinoma and embryonic stem cells. Dev Biol 121:1–9 Sonnenberg A, Linders CJT, Modderman PW, Damsky CH, Aumailley M, Timpl R (1990) Integrin recognition of different cell-binding fragments of laminin (P1, E3, E8) and evidence that α6β1 but not α6β4 functions as a major receptor for fragment E8. J Cell Biol 110:2145–2155 Soudais C, Bielinska M, Heikinheimo M, MacArthur CA, Narita N, Saffitz JE, Simon MC, Leiden JM, Wilson DB (1995) Targeted mutagenesis of the transcription factor GATA-4 gene in mouse embryonic stem cells disrupts visceral endoderm differentiation in vitro. Development 121:3877–3888 Sutherland AE, Calarco PG, Damsky CH (1993) Developmental regulation of integrin expression at the time of implantation in the mouse embryo. Development 119:1175–1186 Tamura RN, Cooper HM, Collo G, Quaranta V (1991) Cell type-specific integrin variants with alternative α chain cytoplasmic domains. Proc Natl Acad Sci USA 88:10183–10187 Thorsteinsdóttir S, Roelen BAJ, Freund E, Gaspar AC, Sonnenberg A, Mummery CL (1995) Expression patterns of laminin receptor splice variants α6Aβ1 and α6Bβ1 suggests different roles in mouse development. Dev Dyn 204:240–258 Tokunaga K, Taniguchi H, Yoda K, Shimizu M, Sakiyama S (1986) Nucleotide sequence of a full-length cDNA for mouse cytoskeletal β-actin mRNA. Nucleic Acids Res 14:2829

184 55. Tryggvason K (1993) The laminin family. Curr Opinion Cell Biol 5:877–882 56. Tsuchida T, Lewis KA, Mathews LS, Vale WW (1993) Molecular characterization of the rat transforming growth factor-β type II receptor. Biochem Biophys Res Commun 191:790–795 57. van den Eijnden-van Raaij AJM, Koornneef I, van Zoelen EJJ (1988) A new method for high yield purification of type β transforming growth factor from human platelets. Biochem Biophys Res Commun 157:16–23 58. Wang DH, Zhou G, Birkenmeier TM, Gong J, Sun LZ, Brattain MG (1995) Autocrine transforming growth factor β1 modulates the expression of integrin α5β1 in human colon carcinoma FET cells. J Biol Chem 270:14154–14159

59. Williams MJ, Hughes PE, O’Toole TE, Ginsberg MH (1994) The inner world of cell adhesion: integrin cytoplasmic domains. Trends Cell Biol 4:109–112 60. Wobus AM, Rohwedel J, Maltsev V, Hescheler J (1994) In vitro differentiation of embryonic stem cells into cardiomyocytes or skeletal muscle cells is specifically modulated by retinoic acid. Roux’s Arch Dev Biol 204:36–45 61. Zambruno G, Marchisio PC, Marconi A, Vaschieri C, Melchiori A, Giannetti A, De Luca M (1995) Transforming growth factor-β1 modulates β1 and β5 integrin receptors and induces the de novo expression of the αvβ6 heterodimer in normal human keratinocytes: implications for wound healing. J Cell Biol 129:853–865