Report

Mixer, a Homeobox Gene Required for Endoderm Development

See allHide authors and affiliations

Science  03 Jul 1998:
Vol. 281, Issue 5373, pp. 91-96
DOI: 10.1126/science.281.5373.91

Abstract

An expression cloning strategy in Xenopus laevis was used to isolate a homeobox-containing gene, Mixer, that can cause embryonic cells to form endoderm. Mixer transcripts are found specifically in the prospective endoderm of gastrula, which coincides with the time and place that endodermal cells become histologically distinct and irreversibly determined. Loss-of-function studies with a dominant inhibitory mutant demonstrate thatMixer activity is required for endoderm development. In particular, the expression of Sox17α andSox17β, two previously identified endodermal determinants, require Mixer function. Together, these data suggest that Mixer is an embryonic transcription factor involved in specifying the endodermal germ layer.

Specification of vertebrate germ layers has been extensively studied in recent years and considerable progress has been made in understanding genes and signaling pathways involved in formation of the mesoderm, ectoderm, and the embryonic nervous system (1). For example, the Brachyury transcription factor is a key regulator of embryonic mesodermal cell specification and, for the most part, the signals inducing neural and epidermal tissues have been identified (2). In contrast, relatively little is known about genes that are required to make endoderm in vertebrates. Understanding the embryological specification of the endoderm is a first step toward explaining the organogenesis of all endodermal derivatives, including the thymus, liver, pancreas, intestine, and respiratory tract.

Xenopus endoderm derives from the yolk-rich vegetal region of the blastula. Before gastrulation, these vegetal cells are not irreversibly determined or committed to form endoderm and will differentiate as ectodermal or mesodermal tissues when transplanted into the blastocoel of a host (3). Moreover, prospective endodermal cells can be converted to mesodermal and ectodermal cell types by expression of dominant-negative constructs that block transforming growth factor-β (TGF-β) signals (4). As gastrulation proceeds, the labile nature of prospective endoderm is lost and the fate of the endoderm becomes restricted to cells that will line the gut tube (3).

Differentiation of the numerous endodermal derivatives along the respiratory and gastrointestinal tracts normally occurs in close association with adjacent mesoderm (5), but prospective endoderm can undergo some differentiation in the absence of mesoderm, as evidenced by the fact that, when cultured in isolation, vegetal explants from Xenopus blastula express endoderm-specific molecular markers (6). In this case, expression of these markers reveals a prepattern in the blastula prospective endoderm: anterior vegetal cells autonomously express a pancreas-specific transcription factor, whereas posterior vegetal cells express a small intestine–specific transcript (4).

The ability of blastula stage vegetal cells to differentiate in isolation suggests that key regulatory genes are expressed in prospective endoderm. Following this assumption, we constructed a cDNA library from prospective endoderm (blastula to early gastrula vegetal poles) and used a functional screen for endodermal determinants. The objective of the screen was to find genes capable of inducing endoderm in the absence of other complicating germ layers such as the mesoderm. This was accomplished by an expression cloning strategy similar to that used to isolate neuralizing factors and mesoderm inducers (7). In this case, pools of prospective endoderm-derived mRNAs were injected into the animal poles of one-cell embryos (7) and animal cap explants were removed at the blastula stages and scored 2 days later for expression of endodermal markers in the absence of markers for other germ layers such as mesoderm or neurectoderm (Fig. 1A). Secreted factors such as cerberus, chordin, and Vg1can all induce endoderm in this animal cap assay, but these factors fail the screening test as they also induce either neural or mesodermal cell types.

Figure 1

Expression cloning of Mixer, a transcription factor that causes embryonic cells to form endoderm. (A) A zygotic endodermal cDNA expression library was screened by injecting pools of mRNAs into the animal hemisphere of one-cell embryos. Pools that specifically induced endodermal, but not mesodermal or neural marker, expression were sib-selected and further reduced to a single clone (29). (B) Mixer induces endodermal marker expression in the absence of mesodermal and neural differentiation. EDD (9), LFABP (10),XlHbox8 (homologue of Pdx1) (11), IFABP (12) are all endoderm-specific markers. A twofold dilution series ending with 500 pg of mRNA per egg is shown. UnlikeMixer, the secreted TGF-β–like factor B-Vg1 induces both mesodermal and endodermal markers (4). Mix.1 does not induce markers of any germ layer in this assay, as demonstrated by a twofold dilution series ending with 500 pg of mRNA (30). Higher doses of either Mix.1 or Mixer cause cell death in injected animal caps. The control lane represents uninjected animal caps. All markers were assayed at stage 35 exceptXbra, which was scored from caps of the same experiment at initiation of gastrulation (stage 10.25). The ubiquitous metabolic factor ornithine decarboxylase (ODC) serves as a loading control and the WE-RT lane is a control of whole embryos assayed by PCR without RT (31). The amount of B-Vg1 mRNA (30 pg) used in this experiment was not sufficient to drive a secondary NCAM induction.

Woodland and colleagues (8) recently identified two transcription factors, Sox17α and Sox17β, in a subtractive polymerase chain reaction (PCR) screen designed to find genes whose expression is enriched in embryonic endoderm. BothSox genes induce endoderm in animal caps, and endodermal development is blocked in embryos expressing a dominant inhibitory mutant of Sox17β. Significantly, the Sox17genes are expressed exclusively in the endoderm, which supports the hypothesis that key regulatory factors should be expressed in this germ layer. Here we describe Mixer (Mix-like endodermal regulator), a third endoderm-specific factor that, like theSox genes, specifically converts blastula cells to an endodermal fate (GenBank accession number AF068263). Moreover, through loss-of-function experiments, we show that expression of bothSox17 genes requires Mixer activity.

Using the animal cap endoderm induction assay, we screened 50,000 independent cDNA clones by separating them into 100 pools of 500 clones each. The activity of one clone isolated by this strategy is shown (Fig. 1B). Mixer potently induces endoderm as judged by its ability to induce the expression of four markers: endodermin (Edd), a pan-endodermal marker (9); the liver fatty acid binding protein (LFABP), an anterior small intestine and liver marker (10); Xlhbox-8, a pancreatic homeobox gene (homologous to Pdx1) (11); and the intestinal fatty acid binding protein (IFABP), a marker of small intestine (12). With the exception of Edd, all these endodermal markers initiate expression approximately a day after gastrulation ends, just before conversion of the endoderm to epithelium (stages 25 to 30). Edd expression begins just before gastrulation, but it is not endoderm specific until the other three markers are expressed (9). Animal caps injected with Mixer turn on all endodermal markers that are autonomously expressed in mesoderm-free endodermal explants (4, 13). However, a dose response of endodermal marker expression is not observed, which suggests thatMixer is a general endodermal determinant that by itself is incapable of generating an anteroposterior pattern (Fig. 1B).

Mixer induces endoderm in the absence of both mesodermal and neural cell types. Brachyury (Xbra), a pan-mesodermal marker muscle actin that marks somitic muscle;Xtwist, an early ventral-lateral mesodermal marker; globin, a ventral mesodermal marker; and NCAM, a pan-neural marker are all absent in explants expressing Mixer (14). LikeMixer, both Sox17 genes convert blastula cells specifically to endoderm (8). Animal caps induced by the T-box transcription factor Xbra express only mesodermal genes (15). Thus, Xbra, Mixer, and the recently described Sox genes are unusual in that they are early determinants that specifically cause blastula cells to adopt the fate of a single germ layer.

The 1.4-kb Mixer cDNA isolated in the screen is the same size as a single transcript observed in Northern blots (13). Analysis of the Mixer open reading frame shows that it has high homology to two previously isolated homeobox genes,Mix.1 and Mix.2 (16). The three genes are 84% identical within the homeobox and 28% identical in the putative transcriptional effector domains. Mixer is the least homologous of the three genes: Mix.1 andMix.2 are 70% identical along their entire lengths, whereasMix.1 is only 37% identical to Mixer(13). The Mix homeoboxes are most similar to thePaired subclass, which have been shown to bind DNA as dimers (17).

Because of the sequence similarity between previously identifiedMix genes and Mixer, we tested Mix.1in an animal cap assay for endoderm induction. As shown (Fig. 1B),Mix.1 does not induce endodermal marker expression, consistent with its previously described role in patterning the mesoderm toward a ventral fate (18). There are extremely low levels of globin and endodermin expression in animal caps expressingMix.1, but none of the other endodermal markers is appreciably expressed (Fig. 1B).

A time course of expression (Fig. 2A) shows that Mixer transcripts are detected for only 6 hours during the gastrula stage. Transcripts for Mix.1,Sox17α, and Sox17β all appear at the midblastula transition (MBT), approximately 3 hours before those of Mixer. Both Mix.1 and Mixertranscripts abruptly disappear as gastrulation ends at stage 13.Sox17α and -β gene expression continues throughout the neurula and tadpole stages (8) (Fig. 2A). Expression ofMixer during the gastrula stages coincides with the first point in amphibian development when a trilaminar germ layer arrangement can be histologically discerned. Additionally, it is not until the early gastrula stages that endodermal cells are irreversibly determined toward their eventual fate of lining the gut tube (3).

Figure 2

Mixer expression is turned on and off specifically during gastrulation. (A)Mixer transcripts are detected by RT-PCR at the indicated stages. Expression of Mix.1 and both Sox17 genes is observed approximately 3 hours before Mixer is detected, but both Mix.1 and Mixer transcripts abruptly disappear with the end of gastrulation at stage 13. The equivalent of one-third of a whole embryo's RNA was used to synthesize cDNA followed by PCR on 1/20th of the synthesized cDNA. ODC serves as a loading control and the 35-RT lane is a control of RT-PCR on stage-35 whole embryo RNA in the absence of RT (31). (B) Paraffin section of an early gastrula (stage 10.25) embryo hybridized to an antisense Mixer probe. Expression is highest at the endodermal edge of the mesendodermal boundary. (C) A section from a midgastrula embryo (stage 12) hybridized as in (B). (D) A section adjacent to that in (C) hybridized to a probe for the pan-mesodermal marker Xbra. Yellow arrows in (C) and (D) highlight the fact that Xbra expression is observed where Mixer expression is absent. (E) Expression of Mixer is distinct from that of the closely related geneMix.1. An antisense Mix.1 probe was hybridized to the section in (E), which is 30 μm away from the sections in (C) and (D). (F) A midgastrula section hybridized to aSox17β antisense probe. Expression of Sox17β is nearly identical to expression of Mixer at this stage with uniform staining across the endoderm. (G) A horizontal section of a midgastrula embryo treated as in (C) displays no strong dorsoventral differences in Mixer expression. (H) Control of a midgastrula section hybridized to a sense Mixer probe. Albino embryos were not used in these in situ experiments and, as a result, a dark rim of pigment is observed across the animal pole of sections shown in (B) to (F) and (H). Arrowheads mark the mesendoderm boundary and are placed in the same position in each panel (32). (I) Schematic of a midgastrula embryo with endoderm (e) in red and mesoderm (m) in green.

In situ hybridizations on sections of gastrula embryos show thatMixer transcripts are found specifically in the prospective endoderm (Fig. 2, B, C, and G). Comparing adjacent serial sections that were hybridized with either Mixer or Xbraantisense probes (Fig. 2, C and D) best shows the endodermal specificity of Mixer. Mixer expression is strongest at the mesendodermal boundary, which, as gastrulation proceeds, shifts toward the vegetal pole of the embryo (19). Consistent with this observation, the edges of the Mixerexpression domain descend vegetally during gastrulation (Fig. 2, compare B and C). Hybridization to a Mix.1 probe of a section similar to that in Figure 2D shows expression in both the presumptive endoderm and the marginal zone mesoderm as described (Fig. 2E) (20). The gastrula expression of Sox17β is shown (Fig. 2F) and it is identical to that observed forMixer except that Sox17β expression appears to be more uniform across the interior of the embryo. The Sox17genes have identical expression patterns during gastrulation (8).

Mixer-induced animal caps are dumbbell shaped with one end more white or yolky in appearance (Fig. 3, A and B). Histological analysis of animal caps coinjected with mRNA encoding the lineage tracer β-galactosidase (βgal) shows that endodermin expression coincides with cells that received Mixer mRNA, which are also those that appear yolky or lighter in color in whole explants (Fig. 3, C and D). This suggests that Mixer induction of endoderm is cell autonomous, as expected for a determinant that is a transcription factor.

Figure 3

Induction of endoderm by Mixer is cell autonomous and Mixer is turned on by Vg1, but not by BMP-4. (A) Mixer-injected animal caps. (B) Uninjected animal caps. Animal caps induced byMixer have a dumbbell shape by the late neurula–early tail bud stages. (C) Section from an animal cap coinjected withMixer and βgal mRNAs and hybridized with an antisense endodermin probe (dark blue). (D) A section adjacent to that in (C) hybridized with a sense control endodermin probe. In this case, the light blue βgal stain marks the injected cells that coincide with those in (C) expressing endodermin. (E) Control of an animal cap injected with βgal mRNA alone and hybridized with an endodermin antisense probe. Light blue stain is for βgal; explants lack the dumbbell shape of the Mixer-injected caps shown in (C) and (D). (F) Control of a vegetal pole endodermal explant hybridized to an endodermin antisense probe showing ubiquitous expression of this pan-endodermal marker (32). (G) A 30-fold dilution series of both B-Vg1 (ending at 80 pg of mRNA per egg) and BMP-4 (ending at 2000 pg of mRNA per egg) injected into the animal hemisphere of one-cell embryos, followed by animal cap removal at the blastula stages. Both B-Vg1 and BMP-4 induce Xbra, but only B-Vg1 turns on Mixer transcription. A similar experiment was done with Smad1 and Smad2 and using a twofold dilution series ending with 500 pg of mRNA per egg (30). Mix.1, unlikeMixer, is induced both by BMP-4 and Vg1 (18). ODC is a loading control and the WE10-RT control lane represents a RT-PCR on whole embryo (Stage 10) RNA lacking RT (31). (H) One animal blastomere of an embryo at the four-cell stage was coinjected with B-Vg1 and βgal mRNAs. Section of an animal cap dissected at the blastula stage, cultured until initiation of gastrulation, and hybridized with a Mixerantisense probe. Cells expressing Mixer RNA (dark blue) and containing βgal (light blue) overlap in the left two-thirds of the explant. (I) A section adjacent to that in (H) was hybridized with an Xbra antisense probe. Xbraexpression (dark blue) is in a crescent of cells adjacent to those expressing Mixer and containing the nuclear βgal signal. Red dashes denote the boundary of βgal (injected) cells to the left and uninjected cells to the right expressing Xbra(32). Explants in (H) to (I) were harvested at stage 10.25 with the remaining explants, (A) to (F), isolated at stage 35.

Mixer is turned on in animal caps by the secreted protein Vg1, but not by BMP-4 (Fig. 3G). Moreover, Smad2, but not Smad1, can ectopically induce Mixer expression, consistent with the finding that Smad2 but not Smad1 transduces Vg1/activin signals (Fig. 3G) (21). Activin has been shown to induce bothSox17 genes in animal caps (8) and we find that B-Vg1 also induces both genes (Fig. 4C). Induction of Mixer and Sox17 expression by Vg1 is consistent with previous work demonstrating a role for Vg1 signaling, but not BMP-2/4 signaling, in differentiation of the endoderm (4).

Figure 4

Structure and activity of Mixer and a dominant inhibitory mutant, Mixer-ENR, made with the engrailed repressor domain. Box diagram shows engrailed repressor domain with a triple bar and Mixer homeobox domain as a black box. The engrailed repressor was fused to the NH2-terminus of the Mixer homeobox (28). Various fusion positions were tested and all behaved similarly. (A) Mixer inducesSox17β in animal cap explants, but Sox17β does not induce Mixer. A twofold dilution series ofMixer and Sox17β, from 65 to 500 pg of mRNA injected per egg. Data from the animal caps injected withSox17β and Mixer are shown on the left and right, respectively. ODC, Mixer, and Sox17β data were obtained from stage 11 animal caps, whereas the EDD assays were performed on RNA isolated from stage 35 animal caps. Although both genes induce endoderm (EDD), Mixer induces Sox17α (data not shown) and Sox17β, but neitherSox17 gene induces Mixer (left). For coinjection experiments, the inducer (Mixer or Sox17β) is injected at 250 pg per egg and the engrailed repressor fusions (Mixer-ENR or Sox17-ENR) are injected at 500 pg per egg. Engrailed repressor fusions alone were at 500 pg of mRNA per egg. Sox17β-ENR is a COOH-terminal fusion to the HMG DNA binding domain (8). (B) Mixer induces cerberus, DKK-1, and Xnr-3, whereas the Sox17 genes induce cerberus only. A weak induction of Dkk-1 by the Sox17 genes is reproducibly observed. All RNAs were injected at 500 pg per egg; lower doses of all three genes have the same activity. Animal caps were isolated at stage 11. (C) Mixer-ENR blocks the endoderm activity of Vg1; 50 pg of B-Vg1 mRNA was coinjected with 3, 30, or 300 pg of Mixer-ENR mRNA. Xbra, cerberus, DKK-1 and Xnr-3 were assayed at stage 11, stage 19 caps were used to assay expression of both Sox17 genes, and stage 35 animal caps were used for ODC, EDD, and Xlhbox-8 assays. ODC for stage 11 and 19 data were similar to those for the stage 35 experiment. ODC serves as a loading control and the WE-RT lane is a control of RT-PCR on stage 11,19, and 35 whole embryo RNA in the absence of RT (31).

Histological analysis of animal caps coinjected with Vg1 and βgal shows that Vg1-expressing cells induce Mixerexpression, whereas distant (uninjected) cells express the mesodermal marker Xbra (Fig. 3, H and I). This suggests a mechanism by which a single TGF-β can induce both mesoderm and endoderm in a single animal cap. High levels of Vg1 signaling lead to endodermal specification and lower levels of signaling lead to mesodermal induction (22). The endodermal localization of Vg1 transcripts is consistent with such a model (23).

Because the expression pattern of the Sox genes is essentially identical to that of Mixer during the gastrula period and all three genes induce endoderm in animal cap explants, we tested whether one gene might act upstream of the others.Mixer induces the expression of both Sox17 genes in animal cap explants, but neither Sox17 gene inducesMixer (Fig. 4A) (24). Further analyses of animal caps in the gastrula stage reveal other differences among these endodermal determinants. Mixer induces the expression of cerberus (25), Xenopus nodal related-3 (Xnr-3) (26), and DKK-1(27), whereas Sox17α and Sox17β induce only cerberus (Fig. 4B). Both cerberus and DKK-1 are expressed in the anterior endoderm of embryos in the gastrula stage and have been implicated in formation and function of the head organizer.Xnr-3 is also expressed in the gastrula endoderm and has been shown to be a downstream target of the Wnt signaling pathway. The endodermal marker endodermin is induced by Mixer and theSox17 genes, although the latter do so at lower doses of injected mRNA (Fig. 4A).

A dominant inhibitory mutant of Mixer was constructed by fusing the NH2-terminal transcriptional repressor domain of the Drosophila engrailed protein to the homeobox ofMixer (Fig. 4A) (28). This construct, Mixer-ENR, blocks the ability of Mixer, but not Sox17β, to induce endoderm in animal cap assays. In a similar experiment, expression of both Mixer and Sox17β is blocked by a Sox17β engrailed repressor fusion (Fig. 4A). These experiments suggest that Mixer is upstream of Sox17α andSox17β. However, both Sox genes appear 3 hours before Mixer in normal development. A plausible explanation for these data is that Mixer positively maintains expression of the Sox17 genes in the endoderm during gastrulation, with all three genes initiating expression independently of one another.

To test whether Mixer is required for the maintenance of Sox17 expression, we analyzed animal caps dissected from embryos coinjected with Mixer-ENR and B-Vg1 RNAs. As shown (Fig. 4C), the dominant inhibitory Mixer mutant blocked the ability of B-Vg1 to induce late endodermal markers such as EDD andXlhbox-8. In animal caps from the same experiment, isolated at the gastrula stages, DKK-1, cerberus, and Xnr-3 are all blocked. At stage 9, both Sox genes are induced by B-Vg1 in the presence of the Mixer-ENR fusion (data not shown). However, during gastrulation and later at the neurula stages, Sox17α andSox17β transcripts are severely reduced (Fig. 4C). Thus,Sox17 gene expression is not maintained whenMixer activity is blocked. Additionally, mesodermal development in animal caps coinjected with B-Vg1 and Mixer-ENR is unaffected, consistent with Mixer being an endoderm-specific determinant.

A key test for Mixer function is the developmental phenotype of whole embryos expressing the Mixer-ENR fusion. As shown (Fig. 5A), Mixer-ENR–injected embryos are severely affected by the repressor fusion and exhibit anterior truncation and loss of head structures as well as defective gut development; 30% of embryos injected with 150 to 200 pg of mRNA encoding the engrailed repressor fusion develop as shown (Fig. 5A) and the remainder display a similar, but less severe, phenotype with shortening of the anteroposterior axis and defective head formation (n = 220). Late endodermal markers are variably expressed in these embryos. Injecting higher doses of Mixer-ENR leads to lethality during gastrulation, presumably because of the complete failure to form the leading anterior (pharyngeal) endoderm.

Figure 5

Loss of Mixer function blocks endodermal development. (A) Mid-tail bud, control injected embryo (stage 31) is shown above two embryos derived from fertilized eggs injected in the vegetal hemisphere with 150 pg of Mixer-ENR mRNA. Gross morphology of the embryos injected with Mixer-ENR is similar to that observed after injection of a Sox17β-ENR fusion (8). (B, D, and F) Transverse sections of control midgastrula (stage 11) embryos for comparison to transverse sections of mid-gastrula (stage 11) embryos injected at the one-cell stage with 500 pg of Mixer-ENR mRNA per egg (C, E, andG). (B and C) In situ hybridization using an Xbraantisense probe. Arrowheads in (B) highlight notochord expression ofXbra. (D and E) Cerberus in situ hybridization. Note stronger expression of cerberus on the dorsal side in (D). (F and G)Sox17β in situ hybridization. (H)Mixer function is required for Sox17 expression in endodermal explants. A twofold dilution series of Mixer-ENR mRNA, from 65 to 1000 pg of mRNA per egg, was injected at the one-cell stage and vegetal pole explants were dissected at late blastula and harvested at stage 20. Note that it is possible to use higher levels of the Mixer-ENR repressor fusion in explant experiments. (I) Experiment similar to that in (H) except that 30 and 300 pg of Mixer-ENR mRNA per egg was injected at the one-cell stage. The specificity test or rescue test was performed by coinjectingMixer (500 pg of mRNA per egg) and Mixer-ENR (300 pg of mRNA per egg). Although larger amounts of Mixer-ENR mRNA can be used in explants than in whole embryos, it is difficult to culture these vegetal pole explants past the mid-tail bud period (stage 30 or more).

The block to gastrulation and accompanying lethality led us to examine the Mixer loss-of-function phenotype at earlier stages of development. Gastrula stage embryos in which Mixer function was blocked were analyzed by in situ hybridization. The expression ofSox17β, cerberus, and Xnr-3 are all strongly reduced in embryos expressing Mixer-ENR mRNA (Fig. 5, D to G). The lack of cerberus expression is consistent with the severe head phenotype (anterior loss) observed in late stage embryos (Fig. 5A).Brachyury expression is not reduced by the Mixer-ENR fusion and, in fact, appears to be higher or less concentrated in injected embryos (Fig. 5, B and C). It is possible that this effect onBrachyury expression is a result of blocking the activity of other Mix transcription factors. Alternatively, this alteration in mesodermal marker expression could reflect defects in cell movements, and we note that late gastrula embryos do not form an embryonic gut lumen or archenteron (Fig. 5, B, and C). These injected embryos initially form a blastopore, but it regresses before stage 13. Similar gastrulation defects are observed in embryos lackingSox17 activity (8), and together these findings suggest that the endoderm is required for mesodermal cell movements and gastrulation.

We have further assessed the effect of the Mixer-ENR fusion by assaying gene activities in endodermal explants. Vegetal pole explants dissected from embryos injected with Mixer-ENR lack IFABP andXlhbox-8 expression (Fig. 5I). This block in gene expression is completely rescued by coinjection of wild-type Mixer RNA, which suggests that the block is specific. Mix.1 does not rescue in similar explant assays. Unlike IFABP andXlhbox-8, Edd expression is not affected by the Mixer-ENR fusion. There are several possible explanations for this finding, one of which is that Mixer is involved in development of a subset of endodermal cells. In younger endodermal explants, both Sox17 genes initiate expression at the MBT, but by the neurula stages they are strongly down-regulated in the absence of Mixer function (Fig. 5H). Thus, as suggested by the animal cap induction experiments, Mixer is required to maintain the expression of both Sox17 genes in vegetal cells.

In summary, this report provides several lines of evidence to support the conclusion that Mixer is a key endodermal determinant. Mixer expression drives cells into an endodermal lineage, diverting them from their specification toward ectoderm. In accord with this inducing activity, Mixertranscripts are found exclusively in gastrula stage prospective endoderm, a period recognized as the first point at which the endoderm is visible histologically and irreversibly determined (3).Mixer is required to maintain expression of bothSox17 genes, themselves previously shown to be required for endodermal development. Finally, loss of Mixer function blocks development of the endoderm. Future experiments should aim to determine whether a homologous Mixer gene functions to specify endodermal cell lineages in other vertebrates including zebrafish, chickens, and mice.

  • * To whom correspondence should be addressed. E-mail: dmelton{at}biohp.harvard.edu

REFERENCES AND NOTES

View Abstract

Navigate This Article