The role of the Rx homeobox gene in retinal progenitor proliferation and cell fate specification.

The Retinal homeobox gene (Rx; also Rax) plays a crucial role in the early development of the vertebrate eye. Germline deletion of Rx in mice results in the failure of optic vesicle formation, leading to anophthalmia. Recent research using conditional mouse knockout models provides some clues to the role of Rx in eye development following optic vesicle formation. However, the functions of Rx in embryonic retinogenesis are still not fully understood. We investigated the function of Rx in the mouse neural retina using a conditional knockout where the Pax6α-Cre driver deletes Rx activity in early retinal progenitors. The deletion of Rx activity causes a loss of retinal lamination, a depletion of retinal progenitors, and a change in retinal cell fate in our conditional knockout model. The deletion of Rx leads to an absence of late-born retinal neurons (rods and bipolar cells) and Müller glia at postnatal ages, as well as a loss of the early-born cone photoreceptors. Decreased BrdU labeling in the Rx-deleted portion of the retina suggests a loss of retinal progenitors via early cell cycle exit, which likely prevents the formation of late-born cells. As early-born cells, cone photoreceptors should not be as affected by early cell cycle exit of retinal progenitors. However, embryonic cone photoreceptor labeling is also markedly reduced in Rx-deleted retinas. Together these data demonstrate the importance of Rx for retinal progenitor proliferation and a specific requirement of Rx for cone formation in mice.

Analysis of mouse development with conditional mutagenesis.

Explorations into the molecular embryology of the mouse have played a vital role in our understanding of the basic mechanisms of gene regulation that govern development and disease. In the last 15 years, these mechanisms have been analyzed with vastly greater precision and clarity with the advent of systems that allow the conditional control of gene expression. Typically, this control is achieved by silencing or activating the gene of interest with site-specific DNA recombination or transcriptional transactivation. In this review, I discuss the application of these technologies to mouse development, focusing on recent innovations and experimental designs that specifically aid the study of the mouse embryo.

Conditional control of gene expression in the mouse.

One of the most powerful tools that the molecular biology revolution has given us is the ability to turn genes on and off at our discretion. In the mouse, this has been accomplished by using binary systems in which gene expression is dependent on the interaction of two components, resulting in either transcriptional transactivation or DNA recombination. During recent years, these systems have been used to analyse complex and multi-staged biological processes, such as embryogenesis and cancer, with unprecedented precision. Here, I review these systems and discuss certain studies that exemplify the advantages and limitations of each system.

Fgf8 signalling from the AER is essential for normal limb development.

Vertebrate limb development depends on signals from the apical ectodermal ridge (AER), which rims the distal tip of the limb bud. Removal of the AER in chick results in limbs lacking distal skeletal elements. Fibroblast growth factor (FGF) proteins can substitute for the AER (refs 4-7), suggesting that FGF signalling mediates AER activity. Of the four mouse Fgf genes (Fgf4 , Fgf8, Fgf9, Fgf17) known to display AER-specific expression domains within the limb bud (AER-Fgfs), only Fgf8 is expressed throughout the AER. Moreover, Fgf8 expression precedes that of other AER-Fgfs (refs 8-13), suggesting that Fgf8 may perform unique functions early in limb development. In mice, loss of function of Fgf4 (refs 13,14), Fgf9 (D. Ornitz, pers. comm.) or Fgf17 (ref. 15) has no effect on limb formation. We report here that inactivating Fgf8 in early limb ectoderm causes a substantial reduction in limb-bud size, a delay in Shh expression, misregulation of Fgf4 expression, and hypoplasia or aplasia of specific skeletal elements. Our data identify Fgf8 as the only known AER-Fgf individually necessary for normal limb development, and provide insight into the function of Fgf signalling from the AER in the normal outgrowth and patterning of the limb.

Cre-mediated recombination in the pituitary gland.

Organ-specific expression of a cre recombinase transgene allows for the analysis of gene function in a particular tissue or cell type. Using a 4.6 kb promoter from the mouse glycoprotein hormone alpha-subunit (alphaGSU or Cga) gene, we have generated and characterized a line of transgenic mice that express cre recombinase in the anterior and intermediate lobes of the pituitary gland. Utilizing a cre-responsive reporter transgene, alphaGSU-cre transgene expression was detected in the pituitary primordium and in all five cell types of the adult anterior pituitary. alphaGSU-cre transgene activity was also detected in the cardiac and skeletal muscle. Little or no activity was evident in the gonads, adrenal glands, brain, ventromedial hypothalamus, or kidneys. The alphaGSU-cre transgenic mice characterized here will be a valuable tool for examining gene function in the pituitary gland.

Conditional inactivation of Fgf4 reveals complexity of signalling during limb bud development.

Development of the vertebrate limb bud depends on reciprocal interactions between the zone of polarizing activity (ZPA) and the apical ectodermal ridge (AER). Sonic hedgehog (SHH) and fibroblast growth factors (FGFs) are key signalling molecules produced in the ZPA and AER, respectively. Experiments in chicks suggested that SHH expression in the ZPA is maintained by FGF4 expression in the AER, and vice versa, providing a molecular mechanism for coordinating the activities of these two signalling centres. This SHH/FGF4 feedback loop model is supported by genetic evidence showing that Fgf4 expression is not maintained in Shh-/- mouse limbs. We report here that Shh expression is maintained and limb formation is normal when Fgf4 is inactivated in mouse limbs, thus contradicting the model. We also found that maintenance of Fgf9 and Fgf17 expression is dependent on Shh, whereas Fgf8 expression is not. We discuss a model in which no individual Fgf expressed in the AER (AER-Fgf) is solely necessary to maintain Shh expression, but, instead, the combined activities of two or more AER-Fgfs function in a positive feedback loop with Shh to control limb development.

Targeted disruption of Fgf8 causes failure of cell migration in the gastrulating mouse embryo.

Fgf8 and Fgf4 encode FGF family members that are coexpressed in the primitive streak of the gastrulating mouse embryo. We have analyzed the phenotype of Fgf8(-/-) embryos and discovered that they fail to express Fgf4 in the streak. In the absence of both FGF8 and FGF4, epiblast cells move into the streak and undergo an epithelial-to-mesenchymal transition, but most cells then fail to move away from the streak. As a consequence, no embryonic mesoderm- or endoderm-derived tissues develop, although extraembryonic tissues form. Patterning of the prospective neuroectoderm is greatly perturbed in the mutant embryos. Anterior neuroectoderm markers are widely expressed, at least in part because the anterior visceral endoderm, which provides signals that regulate their expression, is not displaced proximally in the absence of definitive endoderm. Posterior neuroectoderm markers are not expressed, presumably because there is neither mesendoderm underlying the prospective neuroectoderm nor a morphologically normal node to provide the inductive signals necessary for their expression. This study identifies Fgf8 as a gene essential for gastrulation and shows that signaling via FGF8 and/or FGF4 is required for cell migration away from the primitive streak.

A Huntington's disease CAG expansion at the murine Hdh locus is unstable and associated with behavioural abnormalities in mice.

Huntington's disease (HD) is a dominant disorder characterized by premature and progressive neurodegeneration. In order to generate an accurate model of the disease, we introduced an HD-like mutation (an extended stretch of 72-80 CAG repeats) into the endogenous mouse Hdh gene. Analysis of the mutation in vivo reveals significant levels of germline instability, with expansions, contractions and sex-of-origin effects in evidence. Mice expressing full-length mutant protein display abnormal social behaviour in the absence of acute neurodegeneration. Given that psychiatric changes, including irritability and aggression, are common findings in HD patients, our data are consistent with the hypothesis that some clinical features of HD may be caused by pathological processes that precede gross neuronal cell death. This implies that effective treatment of HD may require an understanding and amelioration of these dysfunctional processes, rather than simply preventing the premature death of neurons in the brain. These mice should facilitate the investigation of the molecular mechanisms that underpin the pathway from genotype to phenotype in HD.

Muscle-specific cell ablation conditional upon Cre-mediated DNA recombination in transgenic mice leads to massive spinal and cranial motoneuron loss.

We describe here a binary transgenic system based on Cre-mediated DNA recombination for genetic cell ablation in mice that enabled us to obtain skeletal muscle-deficient embryos by mating two phenotypically normal transgenic lines. In those embryos, skeletal muscles are eliminated as a consequence of the expression of the gene encoding the diphtheria toxin A fragment. Cell ablation occurs gradually beginning approximately on embryonic day (E) 12.5, and by E18-5 almost all skeletal muscle is absent. Analysis of the consequences of muscle cell ablation revealed that almost all spinal motoneurons are lost by E18.5, providing strong evidence that survival of spinal motoneurons during embryogenesis is dependent on signals from their target tissue, skeletal muscle, and that trophic signals produced by nonmuscle sources are sufficient to support survival of no more than 10% of embryonic spinal motoneurons in the absence of muscle-derived signals. There was also substantial loss of cranial (hypoglossal and facial) motoneurons in the muscle-deficient embryos, thus indicating that cranial motoneuron survival is also dependent on trophic signals produced by their target tissue. Although spinal motoneurons are a major target of spinal interneurons, the loss of motoneurons did not affect interneuron survival. Muscle-deficient embryos had a cleft palate and abnormalities of the lower jaw, raising the possibility that they might serve as a mouse model for the human disorder, Robin sequence. The data reported here demonstrate the utility of a binary transgenic system for obtaining mouse embryos in which a specific cell population has been ablated, so that its role in embryonic development can be studied.

Mitochondrial transcription factor A is necessary for mtDNA maintenance and embryogenesis in mice.

The regulation of mitochondrial DNA (mtDNA) expression is crucial for mitochondrial biogenesis during development and differentiation. We have disrupted the mouse gene for mitochondrial transcription factor A (Tfam; formerly known as m-mtTFA) by gene targetting of loxP-sites followed by cre-mediated excision in vivo. Heterozygous knockout mice exhibit reduced mtDNA copy number and respiratory chain deficiency in heart. Homozygous knockout embryos exhibit a severe mtDNA depletion with abolished oxidative phosphorylation. Mutant embryos proceed through implantation and gastrulation, but die prior to embryonic day (E)10.5. Thus, Tfam is the first mammalian protein demonstrated to regulate mtDNA copy number in vivo and is essential for mitochondrial biogenesis and embryonic development.

An Fgf8 mutant allelic series generated by Cre- and Flp-mediated recombination.

We describe a strategy for generating an allelic series of mutations at a given locus that requires the production of only one targetted mouse line. The 'allelogenic' mouse line we produced carries a hypomorphic allele of Fgf8, which can be converted to a null allele by mating to cre transgenic animals. The hypomorphic allele can also be reverted to wild-type by mating the allelogenic mice to flp transgenic animals, thereby generating a mouse line suitable for Cre-induced tissue-specific knockout experiments. Analysis of embryos carrying different combinations of these alleles revealed requirements for Fgf8 gene function during gastrulation, as well as cardiac, craniofacial, forebrain, midbrain and cerebellar development.

Cre-mediated chromosome loss in mice.

Chromosome loss in early human embryos is thought to cause a large proportion of spontaneous abortions; when it occurs in specific cell lineages in older embryos or adults, it can result in neoplasia. Although early embryonic chromosome loss can be modelled by breeding mice carrying robertsonian translocation chromosomes, there is currently no method for producing mice with tissue-specific monosomies. Here we demonstrate that DNA recombination mediated by the site-specific recombinase Cre causes loss of a chromosome carrying loxP sites (Cre recognition sites) in an inverted orientation. Thus, when male mice carrying a Y-linked transgene containing inverted loxP sites are mated with females carrying a cre gene that is obiquitously expressed in the early embryo, almost all their XY progeny lose the Y chromosome early in embryogenesis and develop as XO females. Because inverted loxP sites can be targetted to any mouse chromosome and mice can be produced that express cre in specific cell lineages, these data suggest a method for engineering tissue-specific loss of particular chromosomes to provide mouse models for human diseases caused by or associated with specific monosomies.

Specification of the anterior hindbrain and establishment of a normal mid/hindbrain organizer is dependent on Gbx2 gene function.

Analysis of mouse embryos homozygous for a loss-of-function allele of Gbx2 demonstrates that this homeobox gene is required for normal development of the mid/hindbrain region. Gbx2 function appears to be necessary at the neural plate stage for the correct specification and normal proliferation or survival of anterior hindbrain precursors. It is also required to maintain normal patterns of expression at the mid/hindbrain boundary of Fgf8 and Wnt1, genes that encode signaling molecules thought to be key components of the mid/hindbrain (isthmic) organizer. In the absence of Gbx2 function, isthmic nuclei, the cerebellum, motor nerve V, and other derivatives of rhombomeres 1-3 fail to form. Additionally, the posterior midbrain in the mutant embryos appears to be extended caudally and displays abnormalities in anterior/posterior patterning. The failure of anterior hindbrain development is presumably due to the loss of Gbx2 function in the precursors of the anterior hindbrain. However, since Gbx2 expression is not detected in the midbrain it seems likely that the defects in midbrain anterior/posterior patterning result from an abnormal isthmic signaling center. These data provide genetic evidence for a link between patterning of the anterior hindbrain and the establishment of the mid/hindbrain organizer, and identify Gbx2 as a gene required for these processes to occur normally.

Zp3-cre, a transgenic mouse line for the activation or inactivation of loxP-flanked target genes specifically in the female germ line.

The site-specific DNA recombinase Cre is being used to develop a new generation of tools for controlling gene expression in mice [1]. Cre mediates the recombination of two directly repeated target (loxP) sites to a single loxP site, with concomitant excision of the DNA segment flanked by the loxP sites (the 'floxed' DNA). Such recombination can function to activate a gene by excising a floxed DNA segment that blocks expression because it either separates the regulatory and coding sequences of the gene [2] or interrupts the gene's open reading frame. Conversely, DNA excision can inactivate a gene if an essential fragment of the gene is floxed [3]. Gene activation or inactivation in vivo can be achieved by mating two different animals, one carrying a 'target gene' with appropriately placed loxP sites and one carrying a cre transgene. In most cases, the specificity of the system is dependent upon stringent regulation of cre expression. We describe here a mouse line in which cre expression is controlled by regulatory sequences from the mouse zona pellucida 3 (Zp3) gene, which is normally expressed exclusively in the growing oocyte prior to the completion of the first meiotic division [4]. We show that in target-bearing Zp3-cre mice, Cre-mediated recombination of the target gene apparently occurs in 100 % of oocytes. Moreover, Cre activity is not detected in the somatic tissues of most target-bearing Zp3-cre mice. Potential uses for this mouse line are discussed.

Temporal regulation of the Bacillus subtilis early sporulation gene spo0F.

The initiation of sporulation in Bacillus subtilis depends on seven genes of the spo0 class. One of these, spo0F, codes for a protein of 14,000 daltons. We studied the regulation of spo0F by using spo0F-lacZ translational fusions and also measured Spo0F protein levels by immunoassays. spo0F-lacZ and Spo0F levels increased as the cells entered the stationary phase, and this effect was repressed by glucose and glutamine. Decoyinine, which lowers GTP levels and allows sporulation in the presence of normally repressing levels of glucose, induced spo0F-lacZ expression and raised Spo0F levels. The expression of spo0F-lacZ was dependent on spo0A, -0B, -0E, -0F, and -0H genes, a spo0H deletion causing the strongest effect. In most respects, the spo0F gene was regulated in a manner similar to that of spoVG. However, the presence of an abrB mutation did not relieve the dependence of spo0F gene expression on spo0A, as it does with spoVG (P. Zuber and R. Losick, J. Bacteriol. 169:2223-2230, 1987).

Use of a versatile lacZ vector to analyze the upstream region of the Bacillus subtilis spoOF gene.

We have constructed a versatile vector, pIS112, in which lacZ translational fusions can be made in Escherichia coli and then analyzed in Bacillus subtilis in three contexts, without recloning: in multicopy during propagation of the plasmid, in single copy integrated via a Campbell-type mechanism into the wild-type locus of the cloned fragment, or in single copy integrated into a heterologous locus. Upstream regions are reconstituted in the integration into the wild-type locus, but not into the heterologous locus, allowing the identification of upstream regulatory sequences. We have used this vector to analyze the expression of the early sporulation gene, spoOF, which, during early stationary phase, is induced 10-fold from a basal vegetative level. When a region, -50 to -150 bp relative to the transcriptional start site, is removed in the spoOF-lacZ gene, stationary phase induction of beta-galactosidase is lost. The same deletion in the upstream region of the functional spoOF gene results in cells which sporulate very poorly, although they are not blocked at the onset of sporulation, as in an spoOF null mutant. This suggests induction of spoOF expression during the beginning of stationary phase is necessary for wild-type sporulation.

Transcriptional regulation of the spo0F gene of Bacillus subtilis.

We have cloned the early sporulation gene spo0F, which encodes an open reading frame of 124 codons. The putative Spo0F protein derived from this open reading frame, which has been shown to share homology with the Spo0A protein as well as several other regulatory proteins from Escherichia coli, Salmonella typhimurium, and Klebsiella pneumoniae, also shares homology with the E. coli EcoRI methyltransferase. We have shown by S1 nuclease mapping of in vivo transcripts that spo0F is regulated from dual promoters: RNA II was transcribed from an upstream promoter, and RNA I was initated 30 base pairs downstream from RNA II. The promoter sequences for RNA II, but not those for RNA I, conformed to the -10 region consensus sequence for sigma 43 promoters. RNA II was found in low amounts in exponentially growing cells but was not observed in stationary-phase cells, and the presence of RNA II was glucose insensitive. RNA I was found in low amounts in exponentially growing cells, increased three- to fivefold at the end of exponential growth, and remained at this higher level for at least 3 h into stationary phase. RNA I was repressed by glucose during exponential growth but not during stationary phase.