Epithelial–mesenchymal interactions are very important in the development of the vertebrate gut. In the avian embryonic stomach (proventriculus), expression of embryonic chick pepsinogen (ECPg) gene, which is specific to developing glandular cells in stomach epithelium, is regulated by mesenchymal influence. Molecular mechanisms of tissue-specific transcriptional regulation of the ECPg gene and the molecular nature of the mesenchymal signals were analyzed using a combination of the classic organ culture system and gene transfer strategies. In the present review, three methods for the introduction of DNA into tissues are described: lipofection, electroporation and retroviral infection, and characteristics of each system are discussed.
The embryonic gut of vertebrates consists of the endodermal epithelium and surrounding mesenchyme. Although the gut is initially a simple tube, it becomes divided anteroposteriorly into various organs, such as the esophagus, stomach (proventriculus (PV) and gizzard (GZ) in the chicken), duodenum, small intestine and large intestine. We have been studying the mechanism of differentiation of PV epithelium, which is characterized by the formation of compound glands, using both experimental and molecular approaches.
Glandular epithelial cells of PV produce and secrete embryonic chicken pepsinogen (ECPg) during embryonic stages of development under the influence of the PV mesenchyme, as demonstrated by the epithelial–mesenchymal recombination experiments ( Takiguchi et al. 1986 , 1988; Urase & Yasugi 1993). For example, when the epithelium of embryonic GZ was recombined and cultured with the PV mesenchyme, the epithelium formed PV-like glands and expressed ECPg. The GZ mesenchyme completely inhibits PV-specific differentiation even in the PV epithelium. Embryonic chicken pepsinogen cDNA was isolated ( Hayashi et al. 1988a ) and the induction of ECPg gene expression by the mesenchyme was shown to be regulated at the transcriptional level ( Hayashi et al. 1988c ).
To analyze the molecular mechanism of organogenesis in the proventriculus, recently we combined the organ culture method and gene transfer into the epithelium or mesenchyme by lipofection, electroporation and virus infection. Using these efficient and specific gene transfer systems, we tried to elucidate the mechanism of tissue-specific expression of the ECPg gene in PV.
Regulatory element of the pepsinogen gene
We cloned the genomic DNA of the ECPg gene ( Hayashi et al. 1988b ) and analyzed the 5′ regulatory region. We first examined the expression of reporter genes such as β-galactosidase and luciferase connected to the 5′-flanking region of the ECPg gene after introducing them into PV or GZ epithelial cells by lipofection. For efficient introduction of foreign genes into epithelial cells, cells were dissociated into single cells before transfection and transfected cells were mixed with dissociated PV or GZ mesenchymal cells, and the resulting cell aggregates were cultured organotypically. The results demonstrated that, for the induction of transcription of the ECPg gene by PV mesenchymal cells, a 1.1 kb 5′-flanking region of the ECPg gene is needed. This region is also required for the inhibition of ECPg transcription by the GZ mesenchyme. The present study was the first demonstration that mesenchymal signals affect transcription of a gene through its 5′- regulatory region ( Fukuda et al. 1994 ). In the study, with extensive improvement in the conditions of gene transfer, efficiency was augmented up to 5%, and still we must use many (up to 240) embryos in one series of experiments to detect weak activity of luciferase. It was, therefore, rather difficult to analyze further the role of each segment of the promoter region of the ECPg gene for its transcription.
In the 1.1 kb 5′-flanking region of the ECPg gene, there is one Sox binding motif and four GATA binding motifs. Both Sox and GATA are transcription factors and, in PV, Sox2 is expressed in luminal (non-glandular) epithelium but not in gland epithelium. Expression of GATA5 is stronger in glandular epithelium than in luminal epithelium. Therefore, it is possible that Sox2 and GATA5 regulate the transcription of the ECPg gene. To improve the efficiency of transfection and to simplify the manipulation, we adopted electroporation as a method for transfection. We noticed that, when pulses were generated, the solution near the cathode was electrolyzed and solution harmful to cells was produced. To avoid contact of this solution with cells, we made a small gel vessel that was put into an electrode chamber ( Fig. 1A). Use of this vessel also saved DNA solution. The efficiency of transfection by this method estimated by expression of the GFP gene in epithelial cells was about 30% ( Fig. 1B,C). 5′-Flanking regions of the ECPg gene of various lengths were connected with the reporter gene and introduced into PV epithelial cells. After 4 days of organ culture, the activities of the reporter gene were assayed. The results confirmed that a 1.1 kb 5′-flanking region is sufficient for cell type- specific expression of ECPg, and also demonstrated that four GATA motifs are indispensable for the expression of the reporter gene in glandular cells of PV epithelium ( N. Sakamoto et al. 2000 ).
We also analyzed the effects of GATA5 and Sox2 more directly by cotransfection of reporter constructs with GATA5- or Sox2-expression vectors. Overexpression of GATA5 activated expression of the reporter gene 1.7-fold higher than the control. This activation requires four GATA binding sites. Overexpression of Sox2 inhibits reporter gene activity, regardless of the existence of a Sox-binding motif ( N. Sakamoto et al. 2000 ).
These results strongly suggest that GATA5, which is expressed more strongly in glandular epithelium than in luminal epithelium, upregulates transcription of the ECPg gene by binding directly to GATA motifs. Sox2, which shows suppressed expression in glandular epithelium, downregulates the transcription of the ECPg gene indirectly. Embryonic chicken pepsinogen gene may be expressed solely in glandular epithelium as a consequence of interactions of these gene products.
Molecular mechanism of epithelial– mesenchymal interaction in PV
We are also addressing the problems regarding the molecular nature of mesenchymal factor(s) involved in the induction of gland formation in the epithelium. The inductive molecule(s) can pass through the porous membrane (the pores are small enough to inhibit cell–cell contacts) and can be trapped in a gel ( Koike & Yasugi 1999), suggesting that it may be a soluble factor. As candidates for molecules responsible for these inductive events, we paid attention to bone morphogenetic proteins (BMP), which have been reported as mediators of epithelial–mesenchymal interaction in various organs ( Dassule & McMahon 1998; Kim et al. 1998 ; Merino et al. 1998 ). In the embryonic chicken gut, BMP2 messenger RNA (mRNA) was detected only in PV mesenchyme from day 5 to 6.5 of embryonic development, when gland formation in the epithelium begins. To elucidate the possible function of BMP2 in gland formation of PV epithelium, the effect of overexpression of BMP2 was analyzed.
Electroporation is an efficient method for introducing plasmids as mentioned earlier, but the transcription of foreign DNA ceases in a few days. In addition, by electroporation, only one or a few cell layers in the mesenchymal tissue can be transfected with DNA. Therefore, the overexpression of BMP2 was carried out using a retrovirus. For specific transfection to the epithelium or mesenchyme, tissues from embryos, two white leghorn strains were used ( Fig. 2A). For example, the GZ epithelium from the embryo of C/AB strain, which is resistant to retrovirus, was recombined with virus-infected PV mesenchyme from embryos of the C/O strain, which is sensitive to retrovirus. A few glands developed in the epithelium cultured with PV mesenchyme infected with control virus ( Fig. 2B,C), while in the explants with mesenchyme transfected with BMP2 virus, there were many glands expressing ECPg ( Fig. 2D,E). Moreover, when the noggin gene encoding an antagonist of BMPs was expressed in the PV mesenchyme by retrovirus, the epithelium did not form glands and express ECPg ( Narita et al. 2000 ).
These data suggest that BMP2 is a mesenchymal factor that acts directly on the epithelial cells and induces gland formation and ECPg expression in the epithelium.
Characteristics of each method
We used three methods for the introduction of foreign DNA into embryonic chicken tissues to analyze the function of some genes that are expressed in these tissues. Here we describe the characteristics of each method.
Lipofection is a very mild and easy way to introduce DNA as this method is based on the phenomenon that the liposome-DNA complex fuses with the cell membrane. The plasmids introduced by this method were transcribed within a few hours and the transcription continued for a few days. As this method was established originally for the introduction of DNA into cultured cells, the transfection efficiency into tissues was low. Recently, however, various reagents for lipofection have become available and the transfection efficiencies are variable among them. Thus, it is possible, and necessary, to choose an appropriate reagent for each experiment.
Electroporation is an easy and efficient way to introduce DNA into tissues. The efficiency reaches up to 50% of cells and the handling time is very short. The transfected plasmid DNA is transcribed as early as in the case of lipofection. It is also easy to introduce two or more kinds of DNA into one cell simultaneously by this method. Electroporation is very useful for epithelial tissue, but the transfection efficiency into mesenchymal tissue is much lower. Furthermore, DNA is transfected into only a few cell layers. Therefore, if one wants to transfect DNA into cells of thick mesenchymal tissue, methods other than electroporation must be tried.
Retrovirus-mediated transfection is a most efficient method. Replication-competent retrovirus can efficiently infect both the epithelium and mesenchyme and expression of a transgene continues for a long time in a stable manner. We can infect viruses into the epithelium or mesenchyme selectively using tissues from embryos of two strains, one resistant to retroviral infection and the other sensitive to it. We noted the following points as disadvantages of retrovirus-mediated transfection. Preparation of retroviral constructs and high titer retrovirus for efficient infection require many steps. As some types of retrovirus can infect human cells ( Iba 2000), we should naturally handle them with special care. In addition, expression of the transgene was not detected until 1 day after infection (Narita et al., unpubl. data, 2000). Finally in the viral system, when two or three genes are to be transfected into one cell using different subgroups of the viral envelope, we must be careful regarding the order of the infection to avoid the problem of interference ( Iba 2000).
In conclusion, we are elucidating the molecular mechanisms that regulate the organogenesis of PV using a combination of organ culture and DNA transfection. We chose lipofection and electroporation for analysis of the regulatory element of the ECPg gene, and retrovirus for overexpression of genes in the mesenchyme, considering the advantages of each method. In the chicken, there have been many reports concerning tissue interactions of various organs using organ or tissue culture methods. There are many candidate genes expressed in these organs that may be involved in tissue interactions. Our methods may provide a link between classical embryological studies and molecular analyses in organogenesis.