SEARCH

SEARCH BY CITATION

Keywords:

  • activin-A;
  • alkaline phosphatase;
  • bone morphogenetic protein;
  • cardiac specification;
  • Dlx-5;
  • fibroblast growth factor;
  • GATA-4;
  • Nkx-2.5;
  • non-precardiac mesoderm, serum response factor

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We previously reported that combined treatment with bone morphogenetic protein-2 (BMP-2) and fibroblast growth factor-4 (FGF-4) induces cardiogenic events culminating in full cardiac differentiation of non-precardiac mesoderm explanted from stage 6 avian embryos (Lough et al. [1996] Dev. Biol. 178:198–202.). To elucidate the respective functions of BMP and FGF in initiating and maintaining the cardiogenic process, we have used these ectopic cells as a cardiac specification model to ascertain requirements for growth factor specificity and extent of application, as well as induction of cardiac transcription factors. The inability of some BMP isoforms to replace the inductive activity of BMPs-2/4 indicated a specific requirement for this signaling pathway; moreover, neither activin-A nor insulin, which support terminal differentiation of precardiac mesoderm, nor leukocyte inhibitory factor (LIF), which promotes hypertrophy in cardiac myocytes, could replace BMP's cardiogenic activity. A similarly specific requirement for FGF-2/4 signaling was revealed since neither FGF-7, activin-A nor insulin could replace this activity. The effect of both factors was concentration-dependent; maximal incidence of explant differentiation for each occurred at 50 ng/ml. Surprisingly, the majority of explants treated with high BMP levels (250 ng/ml) exhibited a non-cardiac phenotype that was characterized by intense expression of alkaline phosphatase, suggesting differentiation toward an alternative mesodermal phenotype. Experiments to assess the duration of exposure to each factor that was required revealed that while exposure to BMP and FGF during only the initial 30 min of a 48-hr culture period was sufficient to induce cardiogenesis in a significant percentage of explants, 100% incidence of explant differentiation was obtained only when FGF treatment was restricted to the first 30 min and BMP was continuously present during the 48-hr culture period. Treatment with both growth factors was required to induce the cardiac transcription factors cNkx-2.5 and SRF; neither mRNA was induced by BMP or FGF alone. These findings indicate that: (1) specific members of the BMP and FGF families are required to induce cardiogenesis in non-precardiac mesoderm; (2) BMPs-2/4 may function as a morphogen; (3) brief application of both factors can induce cardiogenesis in a modest number of explants whereas (4) 100% incidence of explant differentiation can only be attained by brief FGF treatment combined with continuous BMP treatment and (5) both factors are necessary to induce downstream cardiac transcription factors. These findings are interpreted in terms of these factors' possible roles during cardiac specification and differentiation. Dev Dyn;218:383–393. © 2000 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Previous work has shown that embryonic anterior lateral (AL) plate endoderm cells, which are necessary to support terminal cardiogenesis in stage 6 precardiac mesoderm (Sugi and Lough, 1994), can also induce cardiogenesis in cells that are not in the cardiogenic pathway (Schultheiss et al., 1995; Barron and Lough, unpublished observations). To ascertain the molecular basis of these effects, this laboratory is characterizing secretory products of AL endoderm. To date, the vitamin A transport proteins (Barron et al., 1998), as well as growth factors in the fibroblast growth factor (FGF) and transforming growth factor-β (TGF-β) families, have been identified. Members of the FGF family include FGFs 1, 2, alt-2 and 4 (Parlow et al., 1991; Zhu et al., 1996). Members of the TGFβ family include activin-A (Kokan-Moore et al., 1991) and bone morphogenetic proteins (BMPs) 2 and 4 (Lough et al., 1996; Schultheiss et al., 1997; Andree et al., 1998). All of the endoderm-secreted growth factors except the BMPs can mimic the ability of anterolateral (AL) endoderm to support terminal cardiac differentiation in precardiac mesoderm (Sugi and Lough, 1995; Lough et al., 1996).

Pursuant to demonstrations that AL endoderm can (re-)specify stage 4 posterior primitive streak (Schultheiss et al. 1995) and stage 6 posterior lateral plate (non-precardiac) mesoderm cells (Barron and Lough, unpublished observations) to the cardiogenic pathway, it was investigated whether endoderm products could mimic this effect. Although when present alone neither FGF-4 nor BMP-2 could induce cardiogenesis, treatment with both factors induced non-precardiac mesoderm to enter the cardiogenic pathway and complete the full differentiative program as indicated by the development of a rhythmically contractile multicellular vesicle within 48 hr (Lough et al., 1996).

This finding suggested that FGF functions as a survival factor while BMP, in accord with evidence from other laboratories, is a cardiac specification factor. However, demonstrating these respective roles in the context of the embryo is difficult for several reasons, including the unavailability of promoters that confer specific transgene expression to signaling and responding cells, as well as uncertainties regarding the precise location and identity of these cells at the time of cardiac specification, which occurs very early during embryonic development. While fate-mapping has shown that cells destined to occupy the heart are aligned in the primitive streak at stage 3 (Garcia-Martinez and Schoenwolf, 1993), near BMP- (Schultheiss et al., 1997) and FGF- (Shamim and Mason, 1999) expressing cells, the unavailability of markers to identify cardiac cells at the time of specification has prevented precise determination of the time and place of this event, precluding biochemical studies on isolated progenitor cells. As an alternative, experiments using ectopic tissue have shown that BMP-2 induces expression of the cardiac lineage marker Nkx-2.5, but not terminal differentiation, in stage 6 anteromedial mesoderm in vivo (Schultheiss et al., 1997; Andree et al., 1998) and full differentiation when explants are cultured in vitro (Schultheiss et al., 1997). These findings, along with the ability of combined BMP and FGF to induce cardiogenesis (Lough et al., 1996), suggested mechanisms regarding the respective roles of these factors during cardiogenic induction that have been experimentally tested in this report using ectopic stage 6 posterior lateral non-precardiac mesoderm as a model for specification. First, the specificity of BMPs-2/4 and FGFs-2/4 for cardiogenic signaling has been examined, in comparison with other members of their respective families as well as with other endoderm-secreted growth factors. Second, to test the hypothesis that brief BMP treatment confers cardiac specification while continuous FGF treatment confers survival, growth factor application has been regulated to determine their respective concentration- and time-dependent effects. Finally, because BMP and FGF respectively induce expression of cNkx-2.5 (tinman: Frasch, 1995; Schultheiss et al., 1997) and serum response factor (SRF; Parker et al., 1992; Moss et al., 1994), heterodimers of which strongly promote cardiac gene expression (Chen and Schwartz, 1996), it was assessed whether these cardiac transcription factors are respectively upregulated in non-precardiac mesoderm. We report that (1) specific members of the BMP and FGF families, especially isoforms 2 and 4 of each, are required to support the cardiogenic process; (2) high BMP levels may generate a non-cardiac phenotype; (3) brief application of either factor, in the presence of the other, is sufficient to induce cardiogenesis; (4) 100% incidence of explant differentiation is only attained by brief FGF treatment combined with continuous BMP treatment and (5) both factors are necessary to induce SRF and cNkx-2.5.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

BMP and FGF Are Specific in Their Ability to Induce Cardiogenesis

Determinations were performed to identify the most cardiogenic homologues of BMP and FGF. Regarding the latter, FGF-2, which was observed to have cardiogenic efficacy that was indistinguishable from FGF-4 (data not shown), was used in all these experiments. Based on availability and biological activity in other systems, BMPs 2, 4, 6, 7, 12,and 13 were selected for evaluation. As shown in Figure 1, it was observed that BMPs 2, 4,and 7 had similar activity, supporting cardiogenesis in 50–60% of explants. These factors were followed in potency by BMPs 6 and 12, which respectively generated cardiogenic vesicles in 30% and 10% of explants. BMP-13 was not cardiogenic.

thumbnail image

Figure 1. Relative ability of BMP isoforms to induce non-precardiac mesoderm. Non-precardiac mesoderm was explanted from the posterior lateral plate of stage 6 embryos as described in Experimental Procedures. Explants were exposed to a combination of 50 ng/ml FGF-2 and 50 ng/ml of each BMP isoform for 48 hr, after which cardiogenic differentiation was assessed from observations of rhythmic contractility. The number above each bar indicates the total number of explants that were evaluated. The cardiogenic efficacy of BMPs 12 and 13 was significantly less than that of BMPs 2, 4, and 7, as determined by the modified z-test (P ≤ 0.05).

Download figure to PowerPoint

Activin-A, a related member of the TGFβ family, can mimic the ability of hypoblast to induce cardiac myogenesis in stage 1 epiblast (Yatskievych et al., 1997) and can mimic the ability of stage 6 endoderm to support the completion of cardiogenesis in precardiac mesoderm (Sugi and Lough, 1995); insulin similarly mimics endoderm's effects (Sugi and Lough, 1995). Leukocyte inhibitory factor (LIF) is transduced via the cardiotrophin receptor to induce hypertrophy in differentiated cardiac myocytes (Sheng et al., 1996). It was therefore of interest to ascertain whether activin-A, insulin, or LIF could replace the cardiogenic effect of BMP-2 on posterior non-precardiac mesoderm. As shown in Figure 2A, explants in which BMP-2 was replaced with 10–100 ng/ml activin-A, 50 ng/ml insulin or LIF did not differentiate. (Evidence that the activin-A used in these determinations was bioactive was shown by its ability to support terminal differentiation in simultaneously prepared precardiac mesoderm explants.)

thumbnail image

Figure 2. Neither BMP nor FGF can be replaced by activin-A or insulin. A: Posterior non-precardiac mesoderm from stage 6 embryos was cultured in Medium 199 plus FGF-2 (50 ng/ml) plus either activin-A, insulin or BMP-2 at the indicated concentrations. As shown in the solid bars, in combination with FGF-2, neither activin-A nor insulin could replace BMP-2's inductive effect on cardiogenesis. (Data indicated by the open bar at far left was a positive control to ensure the efficacy of activin-A, which when present alone (100 ng/ml) induced terminal cardiogenesis in precardiac mesoderm.) B: The experiments shown in B were similar to those in A, except that non-precardiac mesoderm explants were treated with 50 ng/ml BMP-2, plus either activin-A, insulin, FGF-7, or FGF-2 at the indicated concentrations. In both panels, the number above each bar indicates the total number of explants that were evaluated. The cardiogenic effect of either BMP-2 or FGF-2 could not be replaced by activin-A, insulin or FGF-7, as indicated by statistical significance at P ≤ 0.01.

Download figure to PowerPoint

Because activin-A and insulin, like FGFs 1, 2, or 4, can mimic the ability of stage 6 endoderm to support terminal differentiation in precardiac mesoderm (Sugi and Lough, 1995; Zhu et al., 1996), it was also of interest to ascertain whether activin-A or insulin could replace the cardiogenic effect of FGF-2 on posterior non-precardiac mesoderm. As shown in Figure 2B, with the exception of one explant that differentiated in the presence of activin-A, none of these factors could replace FGF-2's cardiogenic effect. Because FGF-2 is highly homologous to FGF-4, which also has cardiogenic potency (Lough et al., 1996), it was of interest to determine whether a more distantly related FGF protein such as FGF-7 could induce cardiogenesis. As shown in Figure 2B, FGF-7 could not induce cardiogenesis.

Optimal Concentrations of BMP-2 and FGF-2

Having determined that BMPs 2 and 4 were the most potent cardiogenic isoforms, it was decided to utilize BMP-2 for the remaining experiments. To assess the optimum concentration of BMP-2, explants were exposed to a range of 0–500 ng/ml, while maintaining FGF-2 at 50 ng/ml. Explants were evaluated for rhythmic contractility and sarcomeric α-actin immunostaining. As shown in Figure 3, concentrations lower than 5 ng/ml did not support cardiogenesis, while treatment with 5–10 ng/ml was minimally effective. Treatment with 25 ng/ml caused approximately 50% of the explants to differentiate, whereas maximal cardiac differentiation was obtained at a concentration of 50 ng/ml. When levels were increased to 100–500 ng/ml, the percentage of cardiogenic explants declined. These results indicate that 50 ng/ml, which incidentally was the concentration used to evaluate efficacy of the BMP isoforms in Figure 1, was optimal for cardiogenesis. Similar determinations to assess the concentration dependent effects of FGF-2, in the presence of a constant level of 50 ng/ml BMP-2, revealed essentially identical results (data not shown).

thumbnail image

Figure 3. Dose-dependent induction of cardiogenesis by BMP-2. Non-precardiac mesoderm was explanted from the posterior lateral plate of stage 6 embryos as described in Experimental Procedures. Explants were continuously treated with 50 ng/ml FGF-2 in combination with the indicated concentrations of BMP-2 for 48 hr. Evidence of cardiogenesis was assessed by rhythmic contractility. The number above each bar indicates the total number of explants that were evaluated. Comparison of paired values using the modified Z-test revealed that 25 and 50 ng/ml BMP-2 was significantly more cardiogenic than the respectively lower dosage (i.e., 10 and 25 ng/ml), and that 50 ng/ml was more cardiogenic than either 250 or 500 ng/ml (P ≤ 0.05).

Download figure to PowerPoint

High BMP-2 Induces a Non-Cardiac Phenotype

It was consistently observed that explants treated with 250 ng/ml BMP that did not undergo cardiogenesis formed a multilayer of cells that was solid, in distinction with the hollow vesicle that is indicative of cardiogenesis. Because evaluation by electron microscopy revealed an expansive extracellular matrix reminiscent of osteogenic tissue (not shown), explants were histochemically reacted to detect alkaline phosphatase activity, high levels of which are associated with cells undergoing osteogenesis. It was consistently observed that explants treated with 250 ng/ml BMP-2 which did not undergo cardiogenesis as revealed by absence of beating and sarcomeric α-actin immunostaining (Fig. 4E) exhibited high levels of alkaline phosphatase activity (Fig. 4C). By contrast, cardiogenic vesicles that formed during treatment with 250 ng/ml BMP-2 were indistinguishable from cardiogenic explants treated with 50 ng/ml BMP-2: alkaline phosphatase was never detected (Fig. 4B) while sarcomeric α-actin was always detected (Fig. 4D). Alkaline phosphatase activity was never detected in non-cardiogenic explants that had been treated with lower BMP-2 levels (Fig. 4A). Because such intense alkaline phosphatase expression is a characteristic of osteogenesis, expression of an early osteogenic marker, the BMP-inducible homeobox transcription factor Dlx-5, was evaluated (Miyama et al., 1999). As shown in the RT/PCR determination depicted in Figure 4F, Dlx-5 was up-regulated in BMP concentration-dependent fashion within 30 min of BMP+FGF application.

thumbnail image

Figure 4. High levels of BMP-2 induce alkaline phosphatase (AP) in non-cardiogenic explants. Non-precardiac mesoderm was explanted from the posterior lateral plate of stage 6 embryos as described in Experimental Procedures and cultivated in the presence of FGF (50 ng/ml) at the indicated dosage of BMP-2. Explants were reacted to detect AP activity (histochemistry: A, B, C), presence of sarcomeric α-actin (immunohistochemistry: D, E), or expression of the Dlx-5 (RT/PCR: F) as described in Experimental Procedures. A shows that non-cardiogenic (undifferentiated) explants treated with low levels of BMP-2 did not express AP. By contrast, non-cardiogenic explants treated with high levels of BMP-2 always expressed AP, indicated by the blue reaction product in C. Curiously, the yellow hue produced by the reaction mixture was observed only in multilayered explants (B, C; in the latter it is obscured by the AP reaction product; by contrast cellular monolayer in C was colorless. Expression of AP and α-sarcomeric actin was always mutually exclusive (compare B, D and C, E). F is an EtBr-stained agarose gel showing that Dlx-5 was induced by BMP in concentration-dependent fashion within 30 min treatment. Explants in A–E are shown at the same magnification.

Download figure to PowerPoint

Transient Exposure to BMP and FGF Is Sufficient to Induce Cardiogenesis

It was of interest to consider whether a transient signaling event is sufficient to initiate cardiogenesis, consistent with the notion that migrating mesoderm cells may be only transiently positioned to receive a cardiogenic signal during gastrulation. Therefore, it was determined whether transient exposure of non-precardiac mesoderm to BMP was sufficient to induce the cardiogenic pathway, whereas constant exposure to FGF, in accord with its perceived role as a “survival” molecule, was necessary to maintain specified cells in the cardiogenic pathway.

Figure 5A shows results from experiments to determine the duration of BMP exposure required to specify cardiogenesis. Cultures were initiated with 50 ng/ml BMP-2 and FGF-2. At the indicated intervals, BMP-containing medium was exchanged for medium containing only FGF-2 and cultures were continued to the 48-hr endpoint. As expected, cardiogenesis did not occur in the absence of BMP-2 (Fig. 5A, hr exposure). However, exposure for only the first 15 min of the 48-hr culture period was sufficient to cause cardiac differentiation in one of six explants, and, only 30 min' treatment induced cardiogenesis in a significant percentage of 13 explants tested. The percentage of cardiogenic explants increased with duration of BMP-2 treatment, with the exception of a consistent decline in explants that were treated for only 2 hr. These experiments demonstrate that brief exposure to BMP-2 at the beginning of the culture period is sufficient to initiate cardiogenesis and that increasing the duration of exposure increases the incidence of contractile explants. In related experiments, it was determined that treatment with BMP-2 at the beginning of the culture period was crucial; explants from which BMP-2 was withheld for the first 24 hours of the culture period did not undergo cardiogenesis (not shown).

thumbnail image

Figure 5. Incidence of cardiogenesis in explants treated with BMP or FGF for defined intervals of the culture period. Cultures of non-precardiac mesoderm were initiated with 50 ng/ml FGF-2 and 50 ng/ml BMP-2 in Medium 199. In A, BMP-2 was removed at the indicated intervals by discarding the medium and adding fresh medium containing only FGF-2. After 48 hr, explants were evaluated for cardiac differentiation as indicated by formation of a rhythmically contractile vesicle. The decline in incidence of contractility caused by increasing exposure from 30 min to 2 hr was significant at P ≤ 0.07; the increase caused by extending treatment from 2 to 24 hours was significant at P ≤ 0.05. In B, cultures were initiated with 50 ng/ml BMP-2 and 50 ng/ml FGF-2. At the indicated intervals, FGF-2 was removed by discarding the medium and adding fresh medium containing only BMP-2. After 48 hr, explants were evaluated for cardiac differentiation. Pair-wise comparison revealed that the decline in explant contractility caused by increasing exposure time from 30 min to 2 hr was significant at P ≤ 0.01; the increase caused by extending treatment from 2 to 24 hr was significant at P ≤ 0.05. Note that 100% incidence of cardiac explants was obtained by restricting FGF treatment to 30 min.

Download figure to PowerPoint

Reciprocal experiments were performed in which non-precardiac mesoderm was continuously treated with BMP-2 for 48 hr while FGF-2 was applied for variable periods. As shown in Fig. 5B, although cardiogenic differentiation was dependent on treatment with FGF-2, only 15 min' exposure was sufficient to support differentiation in the majority of explants. Surprisingly, only 30 min' exposure supported differentiation in 100% of the explants, a finding that has been observed in 40 consecutive repetitions. As in the case of BMP, extending FGF treatment to 2 hr decreased the incidence of contractile explants, followed by increases after longer exposures which however did not approach the 100% incidence of differentiation observed when FGF was limited to the first 30 min of the culture period.

Based on the findings in Figure 5, it was of interest to ascertain the incidence of cardiogenic explants generated by exposure to FGF and BMP for only the first 30 min of the culture period. It was observed that such treatment resulted in a cardiogenic incidence of 40%, suggesting, in accord with the data in Figure 5A, that prolonged treatment with BMP is required to attain 100% cardiogenic explants (data not shown).

FGF and BMP Cooperate to Induce SRF and cNkx-2.5

Serum response factor (SRF), which is involved in the transcription of several cardiac genes including the cardiac and skeletal α-actins and sm22α, is induced by FGF (Parker et al., 1992; Moss et al., 1994). Chicken Nkx-2.5, the homologue of Drosophila tinman, is a transcription factor that is expressed in mesoderm and endoderm cells of the cardiac domain in the embryo; in both Drosophila and avians, tinman/cNkx-2.5 has been shown to be induced by BMP (Schultheiss et al., 1997). SRF and Nkx-2.5 heterodimers strongly up-regulate the transcription of several cardiac genes (Chen and Schwartz, 1996). Since BMP and FGF cooperatively induce cardiogenesis, it was of interest to determine whether BMP-2 and FGF-2 respectively up-regulate cNkx-2.5 and SRF in non-precardiac mesoderm. Determinations were performed in which explants were treated for 10 or 24 hr with either FGF-2, BMP-2 or both, followed by conventional RT/PCR analysis using primer pairs that amplify SRF, cNkx-2.5 and GAPDH. As shown in Table 1, freshly explanted non-cultured posterior mesoderm (0 hr) revealed the presence of GAPDH in all instances, whereas Nkx-2.5 was detected in only 1 of 7 explants and SRF was barely detectable in 3 of 7 explants. Treatment with BMP alone for 10 or 24 hr induced no ethidium bromide-detectable Nkx-2.5 (or SRF) cDNA. Similarly, treatment with FGF alone induced neither transcription factor after 10 hr, although SRF (and Nkx-2.5) was detected in 1 of 4 explants after 24 hr. By contrast, explants treated with both BMP and FGF for 10 or 24 hr exhibited both Nkx-2.5 and SRF in nearly every instance.

Table 1. Results of Conventional RT/PCR Determinationsa
Growth factor addedDuration (hr)“n”GAPDHNkx-2.5SRF
+++
  • a

    Thirty-five cycles of PCR amplification, followed by EtBr staining, were performed to test the hypothesis that BMP and FGF respectively up-regulate the cardiac transcription factors Nkx-2.5 and serum response factor (SRF). (+) indicates detection of EtBr-stained PCR product; (−) indicates that PCR products were not seen.

None07611634
BMP-2103300303
FGF-2103300303
FGF-2 + BMP-2103303030
BMP-2243210303
FGF-2244311313
FGF-2 + BMP-2246606051

To more sensitively perform this assessment, as well as to determine whether the cardiac transcription factor GATA-4 was induced, the semi-quantitative PCR determination shown in Figure 6 was performed. Individual explants were cultured for 24 hr in the presence of the indicated growth factors prior to preparation for RT/PCR analysis. Each template for PCR consisted of precisely one-fourth of the RT product from a single explant, and only 28 cycles of PCR amplification were employed, after which accumulation of all cDNAs was linear and quantifiable. Thus, normalization of each transcription factor cDNA to the amount of amplified GAPDH cDNA indicates the extent of induction by each growth factor. A shows that explants treated with BMP only did not induce Nkx-2.5 (or SRF or GATA-4). Similarly, explants treated with FGF only (B) did not induce SRF (or Nkx-2.5 or GATA-4). However, explants treated with FGF and BMP (C) induced approximately 7- and 15-fold increases in SRF and Nkx-2.5, respectively, as assessed by ImageQuant analysis. Although GATA-4 was not appreciably amplified after 28 cycles, conventional PCR using 40 cycles revealed the presence of GATA-4 after treatment with BMP and FGF for 24 and 48 hr (not shown). Also noteworthy in Figure 6 was the increasing amplification of GAPDH in explants treated with BMP, FGF, and BMP+FGF, reflecting the respective increases in cell proliferation that we previously noted in identically treated explants (Lough et al., 1996).

thumbnail image

Figure 6. Semi-quantitative RT/PCR assessment of transcription factors induced by BMP and FGF. Posterolateral non-precardiac mesoderm was explanted from stage 6 embryos and cultured in the presence of 50 ng/ml of the indicated growth factors. RNA from each explant was reverse-transcribed (RT) and one-fourth of each RT product was subjected to PCR amplification in the presence of 32P-α-dCTP for 28 cycles, during which the accumulation of PCR products was linear. The radioactive PCR products were separated on a 4.5% polyacrylamide gel followed by phosphorimaging and ImageQuant analysis.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Recent reports have indicated that BMP may be a cardiac specification molecule. In in vivo stage 6 embryos, BMP-2 can induce cNkx-2.5, a marker for the cardiac situs, in anteromedial mesoderm, which is an ectopic site (Schultheiss et al., 1997; Andree et al., 1998). Although this induction was not accompanied by cardiogenic differentiation (Schultheiss et al., 1997; Andree et al., 1998), explantation of anteromedial mesoderm into cell culture medium fortified with BMP-2, serum and embryo extract resulted in terminal cardiac differentiation that included rhythmic contractions (Schultheiss et al., 1997). Considering that the medium supplements in the latter experiment in all likelihood contained an FGF-like activity, these findings are consistent with our previous assessments using non-precardiac mesoderm which demonstrated that combined treatment with BMP-2 and FGF-2 in chemically defined medium can support full cardiac differentiation in these explants (Lough et al., 1996). Because these findings indicated that BMP and FGF have complementary roles in the cardiogenic process, it was hypothesized, based on above evidence suggesting that BMP has specification activity, that BMP (re)-specifies non-precardiac mesoderm cells to the cardiac lineage while FGF, in accord with previous observations (Sugi and Lough, 1995), supports their terminal differentiation, perhaps by providing survival activity. The experiments described in this paper were designed to extend our earlier findings while beginning to examine these possibilities.

Relative Cardiogenic Potency of BMP Homologues

BMP-2-null mice exhibit early embryonic lethality that is partly caused by abnormal development of the heart in the exocoelomic cavity (Zhang and Bradley, 1996). The observation that cardiac specification nonetheless occurs probably reflects genetic redundancy mediated by BMP homologues. In this regard, BMP-4 is also expressed by anterior lateral plate endoderm, as well as by ectoderm in the cardiogenic region at stage 6 (Schultheiss et al., 1997). To ascertain which BMP homologues may function as cardiogenic inducers, the determinations shown in Figure 1 were performed. This revealed the following hierarchy of inductive potency, from most to least potent: BMP-2, BMP-4, BMP-7, BMP-6, BMP-12, BMP-13. Not surprisingly, this corresponds with these molecules' structural relatedness at the amino acid level. For example, BMPs 2 and 4 share 86% structural identity, while these factors share only ∼57% homology with BMPs 6 and 7. The finding that BMP 7 is significantly more cardiogenic than BMP-6 may reflect their low structural identity (71%). The low cardiogenic efficacy of the recently discovered BMPs 12 and 13 was not surprising since, despite their homology to BMPs 2 and 4 (∼57%), these factors are apparently transduced by unknown, novel receptors (Inada et al., 1996).

Activin, Insulin, and LIF Cannot Replace BMP

We previously reported that activin-A, which like FGF is a product of anterior lateral plate endoderm (Kokan-Moore et al., 1991), can mimic the ability of cardiogenic endoderm to effect terminal differentiation in precardiac mesoderm (Sugi and Lough, 1995). By contrast, BMP cannot mimic endoderm's effect on terminal differentiation of precardiac mesoderm (Lough et al., 1996). Because activin-A and BMP-2 are related molecules within the TGFβ superfamily, sharing 42% amino acid identity in their respective mature signaling domains, it was of interest to determine whether activin-A could replace BMP's cardiogenic effect on non-precardiac mesoderm. Up to 100 ng/ml activin-A could not induce cardiogenesis (Fig. 2A). Moreover neither insulin, which can support terminal differentiation of precardiac mesoderm (Yamazaki and Hirakow, 1991), nor LIF, which induces cardiac cell hypertrophy via the cardiotrophin receptor (Sheng et al., 1996), was effective. These findings, plus the inability of some BMP homologues to induce specification, provide strong evidence for BMP family specificity in the induction of cardiac specification. Thus, BMP and activin appear to have different roles during cardiogenesis. While endoderm-derived BMP-2 may have cardiac specification activity, activin may function at an earlier developmental stage, perhaps at the time of endoderm/mesoderm diversification (Yatskievych et al., 1997), and/or later, during terminal cardiac differentiation (Sugi and Lough, 1995; Antin et al., 1996) and cardiac cushion formation (Moore et al., 1998).

Optimal BMP Concentration for Cardiogenesis: Dose-Dependent Induction of an Alkaline Phosphatase Phenotype

The optimum dosage for cardiogenic induction shown in Figure 3, 50 ng/ml, is somewhat high in comparison with growth factor activities that are usually functional in the 1–20 ng/ml range. This could reflect circumstances including the relative inefficacy of recombinant, as opposed to native, BMP (Wang et al., 1990; Yamaguchi et al., 1991), the use of human rather than avian isoforms, or the possibility that exogenous BMP is transduced by a non-cognate receptor; regarding the latter, however, our unpublished observations indicate that non-precardiac mesoderm contains BMP receptor IA and IB transcripts. The most remarkable concentration-dependent effect of BMP was the observation that explants treated with 250 ng/ml BMP which did not undergo cardiogenesis formed a solid, rather than hollow, cellular multilayer. These explants, which neither contracted nor exhibited α-sarcomeric actin immunostaining, displayed intense histochemical staining for alkaline phosphatase that is suggestive of cells undergoing osteogenesis (Fig. 4C). In addition, RT/PCR determinations revealed that the osteogenic homeodomain transcription factor Dlx-5 was induced in BMP concentration-dependent fashion by only 30 min' treatment with BMP+FGF. However, expression of neither Dlx-5 nor that of other osteogenic markers including the avian homologues of osteocalcin and cbfa (as well as [chondrogenic] sox-9) has been detected after 48 hr' culture with FGF and high levels of BMP. While this seemingly contraindicates osteogenic induction, it is noted that the appearance of the latter factors in osteogenic assays requires weeks, rather than days, an in vitro requirement that is not feasible in non-precardiac mesoderm explants which begin to deteriorate after 2 days' cultivation. It therefore remains possible that the observations in Figure 4 indicate that high levels of BMP initiate an alternative mesodermal pathway. Interestingly, a requirement to attenuate BMP signaling in precardiac mesoderm, which is surrounded by BMP expressing tissues (Schultheiss et al., 1997), is consistent with two recent findings. First, Yamada et al. (1999) have demonstrated that Smad6, an inhibitory Smad which antagonizes BMP signaling, is induced by BMP in precardiac mesoderm. More recently, Galvin et al. (2000) have shown that targeted disruption of the Smad6 gene, expression of which is largely confined to the cardiovascular system, results in multiple cardiovascular defects including endocardial cushion malformation and aortic ossification. The possibility that BMP functions as a morphogen in regulating the differentiation pathways of mesoderm during early embryogenesis is the focus of continued investigation.

Brief BMP Treatment Can Induce Cardiogenesis, although Continuous Treatment Is Necessary to Attain Maximal Incidence of Explant Differentiation

The data in Figure 5A demonstrate that, in the continuous presence of FGF, treatment with BMP for only the first 30 min of the culture period was sufficient to induce cardiogenesis in a modest number of explants. This is consistent with findings indicating that freshly explanted non-precardiac mesoderm is competent to transduce the BMP signal; as determined by RT/PCR, these explants contain mRNAs for BMP receptors IA and IB as well as for intracellular BMP signal transducers Smads 1 and 5 at the time of explantation (unpublished observations). Interpreting this finding in the context of the embryo, it is consistent with the notion that although gastrulating mesoderm cells may be only transiently positioned to receive a BMP signal, a brief signaling event may be sufficient to initiate cardiac myogenesis. Although these observations could be considered to reflect BMP retention by the fibronectin substrate resulting in a prolonged exposure, a similar experimental paradigm has demonstrated that skeletal myogenesis in the developing somite is regulated by a similar mechanism wherein myoblasts are specified by transient exposure to two growth factors: an “unknown” TGFβ homologue and FGF-2 (Stern et al., 1997).

Although the effect of brief BMP treatment satisfied the hypothesis that cardiogenesis would occur, it was nonetheless surprising that continuous BMP exposure was necessary to increase the incidence of cardiogenic explants. Three possibilities that are not mutually exclusive may explain this finding. First, this may reflect increases in BMP receptor numbers during time in culture, resulting in attainment of a threshold level of occupied receptors in each explant that evokes cardiogenesis. Second, this may indicate that BMP signaling is required to maintain cardiogenic specification, a possibility consistent with the proximity of precardiac mesoderm to BMP-expressing tissues (Schultheiss et al., 1997) and with recent findings indicating that although initial expression of Nkx-2.5 is not inhibited in Xenopus embryos expressing dominant-negative BMP receptors, subsequent Nkx-2.5 expression and heart development are significantly inhibited (Katsev et al., 1999). Finally, this may reflect the likelihood that cardiac specification is not an isolated event, but a continuously occurring phenomenon resulting in the recruitment of cells to the lineage with increasing time in culture and perhaps in vivo. Despite the time-dependent increases in incidence of explant differentiation, a consistent decrease was noted in explants that were exposed to BMP for only 2 hr; it is speculated that this reflects occupation of an excessive number of BMP receptors at this time, causing a switch in gene expression toward an alternate lineage according to the “ratchet” mechanism proposed by Dyson and Gurdon (1998) to explain activin's morphogenic activity.

Brief FGF Treatment Is Sufficient to Induce Cardiogenesis. Maximum Incidence of Cardiogenesis Is Attained by Limiting FGF Exposure to 30 Minutes

Experiments in which FGF treatment was limited to intervals of the culture period also yielded unexpected results (Fig. 5B). Contrary to the hypothesis that continuous presence of FGF is necessary to maintain the survival of cells in the cardiogenic pathway, treatment with FGF for only 30 min was not only sufficient to initiate cardiogenesis but also necessary to attain 100% incidence of differentiation. One explanation for this result is that, similar to a putative mechanism for regulating growth of definitive myocardium (Sugi et al., 1993), an autocrine FGF signaling pathway may be activated by brief FGF exposure, effectively providing an optimal level of endogenous FGF. In addition, this might indicate that FGF, rather than functioning as a survival molecule, is a cardiac specification factor that is only briefly required to signal migrating progenitor cells during gastrulation. Regarding the observation that treatment for 120 min resulted in minimal explant differentiation, a phenomenon identical to BMP's effect (Fig. 5A), it is speculated that this decline is also caused by the selection of an alternative differentiative pathway predicted by the ratchet mechanism (Dyson and Gurdon, 1998).

Both FGF and BMP Are Required to Induce SRF and cNkx-2.5 Expression

Previous work has indicated that SRF expression is downstream of FGF of signaling (Parker et al., 1992; Moss et al., 1994). And, evidence from Drosophila (Frasch, 1995) and avians has indicated that cNkx-2.5, a transcription factor expressed in mesoderm and endoderm of the cardiac domain in the embryo, is downstream of BMP (Schultheiss et al., 1997; Andree et al., 1998). The possibility that the cardiogenic effects of BMP+FGF are transduced by up-regulating these transcription factors is suggested by findings in 10T1/2 cells that transcription complexes containing SRF/cNkx-2.5 heterodimers cause a ∼100-fold increase in α-actin gene transcription, as compared to SRF homodimers (Chen and Schwartz, 1996). However, the hypothesis that BMP and FGF respectively induce cNkx-2.5 and SRF was not satisfied inasmuch as both growth factors are necessary to induce cNkx-2.5 and SRF (Table 1 and Fig. 6). Although explants treated with BMP/FGF for 24 hr did not exhibit appreciable amounts of GATA-4 cDNA after 28 amplification cycles (Fig. 6), GATA-4 was easily detected after 40 PCR cycles (not shown), suggesting a sequence of BMP/FGF-induced transcription factor expression in which Nkx-2.5 and SRF induction is followed by GATA-4.

Is FGF a Cardiac Specification Molecule?

The findings in Figure 6 demonstrate that FGF, in addition to BMP, is required to induce cNkx-2.5, the earliest known marker for cardiac specification. Moreover, only brief exposure to FGFs-2/4 is required to activate signaling that culminates in cardiac differentiation (Fig. 5B). We propose that these findings argue against FGF's role as a mere survival factor in the cardiogenic process. Indeed, the spatial/temporal expression of FGFs, which proteins are present throughout the gastrulating embryo (Mitrani et al., 1990; Riese et al., 1995) as well as in anterolateral endoderm at stage 6 (Parlow et al., 1991; Zhu et al., 1996), is consistent with FGF's role as a cardiac specification factor. During mouse development, FGF-8 mRNA is present at E6.25 in the visceral embryonic endoderm (VEE), a tissue which has been shown by Arai et al. (1997) to be the functional cardiogenic homologue of chick AL endoderm; later, FGF-8 is expressed in the definitive endoderm (Crossley and Martin, 1995). Recently, expression of FGF-4 mRNA has been demonstrated in the anterior primitive streak of the avian embryo at stage 3 (Shamim and Mason, 1999), a site occupied by cardiac progenitor cells (Garcia-Martinez and Schoenwolf, 1993) that may be undergoing specification. While it is perhaps most significant that mechanistic experiments have shown that FGF signaling is required to induce formation of heart tissue in avian blastula stage explants (Gordon-Thomson and Fabian, 1994; Ladd et al., 1998), this could be interpreted as an indirect effect in which FGF's immediate function is to induce a cascade that begins with the formation of mesoderm and endoderm from epiblast and culminates in endoderm-induced cardiac specification. Similarly, experiments utilizing non-precardiac mesoderm explants, such as those described in this paper, may represent indirect effects of FGF, as well as of BMP. Whether signaling by FGF, as well as BMP, is indispensable for cardiac specification awaits investigations, preferably utilizing transgenic embryos, which address the effect of specifically disrupting FGF and BMP signaling pathways that occur between cardiogenic signaling cells—perhaps newly formed endoderm—and responding cardiac progenitor cells during the specification process. Results from these and related determinations to ascertain the cellular context in which BMP and FGF-induced transcription factors cooperatively induce cardiogenesis, perhaps by integration with transcriptional co-activator proteins as recently described for BMP+LIF-induced astrocyte differentiation (Nakashima et al., 1999), should be compelling.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Explantation and Culture of Non-Precardiac Mesoderm

Chicken embryos judged according to the criteria of Hamburger and Hamilton (1951) to be at stage 6 were exclusively used in this study. Anterior lateral plate precardiac mesoderm, and non-precardiac mesoderm from the posterior lateral plate, were explanted and cultured as previously described (Lough et al., 1996). Growth factors were added to the indicated final concentrations after explants attached to the fibronectin substrate. Activin-A and human recombinant FGFs 2, 4, and 7 were purchased from R&D Systems (Minneapolis, MN). Insulin was purchased from Sigma Chemical Company (St. Louis, MO). Leukocyte inhibitory factor (LIF: murine, cat. no. 13275-029) was purchased from Gibco BRL, Gaithersburg, MD. The Genetics Institute (Cambridge, MA) generously contributed human recombinant BMPs 2, 4, 6, 7, 12, and 13. Medium, including growth factors, was changed daily except as otherwise noted.

Formation of cardiac muscle in explants was verified by the morphogenesis of multilayered vesicles containing cells whose contractions were always rhythmic and synchronous. At the biochemical level contractile explants expressed sarcomeric α-actin as detected by immunostaining (Fig. 4), ventricular myosin heavy chain (VMHC; Stewart et al., 1991) as detected by RT/PCR (not shown) and, in particular, Nkx-2.5 (Fig. 6), a transcription factor that is not expressed in skeletal muscle tissue. Explants that did not become multilayered neither contracted nor exhibited biochemical differentiation; these were scored as non-contractile. Multilayered explants that exhibited contractility, which was always rhythmic (indicative of cardiac myogenesis), were scored as contractile. The appropriateness of these assessments for cardiogenesis and the absence of skeletal muscle differentiation in these explants has previously been discussed (Sugi and Lough, 1994; Yatskievych et al., 1997). Statistical analysis of explant differentiation was performed by pairwise comparisons using a modified z-test with a pooled estimate of standard error.

Immunohistochemistry

Biochemical differentiation was monitored by immunohistochemistry as described previously (Sugi and Lough, 1994), using a monoclonal antibody that recognizes sarcomeric α-actin (Sigma, St. Louis, MO; Cat. No. A-2172); the secondary antibody was fluorescent isothiocyanate (FITC)-labeled goat anti-mouse IgM (Organon Teknika (Cappel), Durham, NC).

Reverse Transcription/Polymerase Chain Reaction (RT/PCR)

Determinations of gene expression in these explants, each of which contains only approximately 10,000 cells, required sensitivity provided by the reverse transcription/polymerase chain reaction (RT/PCR). RNA from individual non-precardiac mesoderm explants was purified, using 5 μg linear polyacrylamide as carrier, with RNAstat (Tel-Test, Friendswood, TX). Purified RNA was treated with DNase I (Boehringer Mannheim, Indianapolis, IN) to remove any contaminating genomic DNA. Reverse transcription (RT) was performed using oligo-dT as the primer and with M-MLV reverse transcriptase (Promega, Madison, WI). To ensure the absence of contaminating genomic DNA, samples containing RNA that was not reverse-transcribed were simultaneously processed. One-tenth of the resultant RT product was used as template for PCR reactions performed in a 25 μl reaction mixture containing 1.5 mM MgCl2 that was catalyzed with Thermus aquaticus (Taq) DNA polymerase (Promega). Standard PCR amplifications were performed using 35 cycles of denaturation (94°C, 30 sec), annealing (60°C, 60 sec), and extension (72°C, 120 sec). Two-fifths of each PCR product were separated on a 1.5% agarose gel and stained with ethidium bromide. PCR products were sized by comparing migration to that of standard base pair markers (100 bp ladder; Gibco BRL). Semi-quantitative PCR was performed by including 1.0 μCi α-32P-dCTP in the reaction mixture and using only 28 cycles, during which accumulation of PCR products was linear and which amplified quantifiable amounts of Nkx-2.5 and SRF cDNAs without generating saturating amounts of GAPDH PCR product. Two-fifths of each PCR product were separated on a 4.5% acrylamide gel and bands were visualized on a Storm 860 Optical Scanner (Molecular Dynamics, Sunnyvale, CA). Amounts of PCR product relative to GAPDH were estimated by ImageQuant analysis. Size markers were provided by the 100 bp ladder (Pharmacia, Gaithersburg, MD) which was end-labeled with α-32P-dCTP using the Klenow reaction.

All oligodeoxynucleotide primers were purchased from Operon (Alameda, CA). The primer pair for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was 5′-ACGCCATCACTATCTTCCAG-3′ (forward) and 5′-CAGCCTTCACTACCCTCTTG-3′ (reverse), designed to amplify a 579 bp PCR product corresponding base pairs 265-843 of chicken GAPDH (Panabieres et al., 1984). The primer pair for Nkx-2.5 was 5′-CTACGAACTGGAGAGAAGGT-3′ (forward) and 5′-GTAGGCGTTGTAGCTATAGG–3′ (reverse), designed to amplify a 295 bp PCR product corresponding to base pairs 471–765 of chicken Nkx-2.5 (Schultheiss et al., 1995). The primer pair for serum response factor (SRF) was 5′-CAGCAACTCTCTCGTACGGA-3′ (forward) and 5′-TCTGCAGGACAGCTCCAGGT-3′ (reverse), designed to amplify a 343 bp PCR product corresponding to base pairs 1183–1525 of chicken SRF (Croissant et al., 1996). The primer pair for GATA-4 was 5′-CTCCTACTCCAGCCCTTACC-3′ (forward) and 5′-GCCCTGTGCCATCTCTCCTC-3′ (reverse), which amplifies a 224-bp segment of chicken GATA-4 (bp 300–523; Laverriere et al., 1994). The primer pair used to amplify Dlx-5 was 5′-GCTCCGCCGGCACCTACCC-3′ (forward) and 5′-GGAGCGCGACGAGCCCTGAG-3′ (reverse), which amplifies a 452-bp segment of chicken Dlx-5 (bp 296–747; Ferrari et al., 1995). The primer pair used to amplify VMHC was 5′-GGCGACTCTTGATGAGAACA-3′ (forward) and 5′-GCTTCCAGCTCCTCTTCCAG-3′ (reverse), generating a 425-bp PCR product corresponding to bp 255–680 in the sequence reported by Stewart et al. (1991).

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Matthew Barron was supported by a Predoctoral Fellowship from the American Heart Association of Wisconsin. We thank Parker Antin for discussions and reading of the manuscript, Andrea Ladd for help with statistical analysis, and Donna McAllister for expert technical assistance.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES