Autoregulatory Loop in the Regulation of the Mammalian ftz-f1 Gene*

The mammalian ftz-f1 ( mftz-f1 ) gene encoding Ad4BP/ SF-1 has been demonstrated to be essential for the development of adrenal and gonadal glands. In a previous study, we identified an E box as the transcriptional element in the 5 (cid:42) -upstream region of the rat mftz-f1 gene. In the present study, we found a steroidogenic cell-specific transcriptional element in the first intron of the gene. Gel mobility shift and DNase I footprint analyses clearly revealed that Ad4BP itself binds to the element (Ad4 site). This finding was further supported by the positive effect of an Ad4BP expression vector on the transcription and by the significant decrease in the transcription caused by nucleotide substitutions within the Ad4 site. Similar loss was also caused by substitutions in the E box, indicating that the two elements are essential for the full transcriptional activity of the gene. DNase I hypersensitivity assays revealed that the chromatin structure around the Ad4 site and the E box was “open up” in the adrenal glands and Y-1 cells, whereas “closed down” in the liver. These observations indicated that the mftz-f1 gene is controlled by the autoregulatory loop in the steroidogenic tissues. The autoregulatory mecha- nism seems to be necessary to keep the mftz-f1 gene activated and thus maintain the tissues differentiated. Ad4BP (or SF-1) was identified as a tissue-specific transcription factor that regulates all the steroidogenic P-450 genes in the adrenal cortex and gonads (1, was described previously (22). DNase I Hypersensitivity Assays— DNase I hypersensitivity assay was performed on the nuclei isolated from Y-1 cells, rat adrenal glands, and a rat liver as described (23). The nuclei (35 (cid:109) g of DNA) were digested in a 100- (cid:109) l reaction mixture with DNase I at concentrations of 0, 10, 25, 64, and 160 units/ml at 25 °C for 6 min. Thirty-five (cid:109) g of the genomic DNA purified from each of the nuclear preparations was di- gested to completion with Hin dIII and then electrophoresed on a 1.5% agarose gel followed by Southern blotting. The 1.0-kb Hin dIII- Pma CI fragments from the rat and mouse mftz-f1 genes were 32 P-labeled and used as the probes for the rat tissue specimens and Y-1 cells, respectively.

The mammalian ftz-f1 (mftz-f1) gene encoding Ad4BP/ SF-1 has been demonstrated to be essential for the development of adrenal and gonadal glands. In a previous study, we identified an E box as the transcriptional element in the 5-upstream region of the rat mftz-f1 gene. In the present study, we found a steroidogenic cell-specific transcriptional element in the first intron of the gene. Gel mobility shift and DNase I footprint analyses clearly revealed that Ad4BP itself binds to the element (Ad4 site). This finding was further supported by the positive effect of an Ad4BP expression vector on the transcription and by the significant decrease in the transcription caused by nucleotide substitutions within the Ad4 site. Similar loss was also caused by substitutions in the E box, indicating that the two elements are essential for the full transcriptional activity of the gene. DNase I hypersensitivity assays revealed that the chromatin structure around the Ad4 site and the E box was "open up" in the adrenal glands and Y-1 cells, whereas "closed down" in the liver. These observations indicated that the mftz-f1 gene is controlled by the autoregulatory loop in the steroidogenic tissues. The autoregulatory mechanism seems to be necessary to keep the mftz-f1 gene activated and thus maintain the tissues differentiated.
Ad4BP (or SF-1) was identified as a tissue-specific transcription factor that regulates all the steroidogenic P-450 genes in the adrenal cortex and gonads (1,2). Analyses of the cDNA clone encoding Ad4BP revealed that the factor is a member of the steroid/thyroid hormone receptor superfamily (3), in which all the members have a zinc finger motif as the DNA binding domain (4). A functional study using an expression vector for Ad4BP confirmed that the transcription factor governs the tissue-specific expression of the steroidogenic P-450 genes (5). Recently, wider functions of Ad4BP in addition to the regulation of the steroidogenic P-450 genes has been shown by the following investigations.
The expression of Ad4BP was examined immunohistochemically using the specific antiserum (6) and in situ hybridization (7). Consistent with its role in the regulation of the steroidogenic P-450 genes, Ad4BP was confirmed to be expressed in the steroidogenic cells of the adrenal gland and gonads. In addition, Ad4BP was found to be expressed in the pituitary gonadotroph, which secretes follicule-stimulating hormone and luteinizing hormone, as well as in the ventromedial hypothalamic nucleus (8 -10), which controls female sexual behavior (11). Considering the physiological functions of those tissues, it is clear that Ad4BP is one of the essential factors which control the reproductive function of animals, although the genes controlled by Ad4BP in the ventromedial hypothalamic nucleus still remain to be clarified.
Further investigation of the developing tissues revealed that Ad4BP is expressed even in the primordial cells of the adrenal gland and gonads of the fetuses (7,12). In the fetal gonads, a significant amount of Ad4BP was expressed in the testes, whereas only a trace amount was expressed in the ovaries. Sexual differentiation of the gonads is known to be initiated by the transient expression of SRY in the somatic cells of the urogenital ridge (13). The sexually dimorphic expression of Ad4BP observed in the differentiating gonads suggested that the mammalian ftz-f1 (mftz-f1) gene encoding Ad4BP might be one of the genes located downstream to SRY. It is likely that the gonads express the steroidogenic P-450s and Mü llerian inhibitory substance in a sex-dependent manner as a consequence of the dimorphic expression of Ad4BP (12,14). In the adrenal glands, however, a significant amount of Ad4BP was continuously expressed throughout the life regardless of the sexes (12).
These postulated functions of Ad4BP have been supported by the targeted disruption of the mftz-f1 gene (10,(15)(16)(17). Ad4BP null mice showed a complete absence of the adrenal gland and gonads, indicating that Ad4BP governs the genes essential for the adrenal and gonadal differentiation. Therefore, information about the transcriptional mechanism of the mftz-f1 gene is essential for the elucidation of the molecular basis of adrenal and gonadal differentiation. In our previous report, an E box (18) responsible for the specific expression of the gene in the steroidogenic cells was identified in the 5Ј-upstream region (19). In the present study, we focused our attention on the contribution of the first intron to the gene expression and identified an Ad4 site as the transcriptional element. The function of the Ad4 site in the expression of the mftz-f1 gene clearly indicated that the gene is autoregulated by its own product, Ad4BP.

EXPERIMENTAL PROCEDURES
Plasmid Constructions for CAT Assay-As described previously (19), Ad4CAT2.0K and Ad4CAT0.8K contain the 2.0 and 0.8 kb 1 upstream regions from the EcoRI site in the first exon of the rat mftz-f1 gene, respectively, and CAT reporter gene (20). Ad4ECAT2.0K contains the 5.6-kb DNA fragment between the EcoRI site at 2.0-kb upstream region from the transcriptional initiation site and a SmaI site generated by site-directed mutagenesis in the second exon at 5 bp upstream from the initiation methionine. Ad4ECAT2.0K was employed to generate deletion constructs by either digestion or partial digestion at the restriction enzyme sites as shown in Fig. 1. For the construction of Ad4ECAT0.8K, the BamHI site at 0.8 kb upstream from the transcription initiation site was used. Ad4ECAT⌬3.0K, Ad4ECAT⌬2.2K, Ad4ECAT⌬1.8K, Ad4ECAT⌬0.8K, and Ad4ECAT⌬0.4K were constructed by using the XbaI site at 3.0 kb, BamHI site at 2.2 kb, HindIII site at 1.8 kb, SacI site at 0.8 kb, and XbaI site at 0.4 kb upstream from the splice acceptor site, respectively. Ad4ECAT⌬HX, Ad4ECAT⌬BX, Ad4ECAT⌬XX, and Ad4ECAT⌬PX were obtained by internal deletions of 1.4-kb HindIII-XbaI, 1.8-kb BamHI-XbaI, 2.6-kb XbaI-XbaI, and 3.2-kb PmaCI-XbaI fragments, respectively. Ad4ECAT⌬XX-A75, -A24, and -A0, and -B82, -B37, -B17, and -B1 were constructed from Ad4ECAT⌬XX by Bal31 nuclease digestion from the XbaI site to the upstream and downstream regions, respectively. The deletion end points were determined by DNA sequencing. The numbers preceded by A and B indicate the distances from the splice donor and splice acceptor sites, respectively. Ad4CAT0.8K-EF and -ER were generated by insertion of a 91-bp genomic fragment (ϩ146/ϩ236) at the Eco81I site (Ϫ265) of Ad4CAT0.8K in the forward and reverse directions, respectively. pCAT-EF and -ER contained the same 91-bp fragment at the XbaI site of the pCAT plasmid (Promega, Madison, WI) in the forward and reverse directions, respectively. Expression vectors for Ad4BP (RSV/ Ad4BP) and luciferase (RSV/luc) which are under the control of the Rous sarcoma virus enhancer/promoter were described previously (5). An expression vector for a truncated Ad4BP, ⌬Ad4BP207, was described previously (3).
Site-directed Mutagenesis-Ad4ECAT0.8KM carrying a four-nucleotide substitution within the E box was constructed from Ad4CATM3 (19). Site-directed mutagenesis within the first intron of the mftz-f1 gene was performed as described (21). In brief, the 2.8-kb long HindIII-HindIII fragment was subcloned into the HindIII site of the pUC119. Single strand DNA recovered from the above plasmid was used as the template for mutagenesis. An oligonucleotide, 5Ј-GAGGCCTGGGC-CCCGATATTCACTTA-3Ј, containing the three nucleotide substitution (underlined) was used as the primer. The nucleotide substitution was confirmed by DNA sequencing. The mutated fragment was subcloned back to Ad4ECAT0.8K and Ad4ECAT0.8KM to generate Ad4ECAT0.8KMA and Ad4ECAT0.8KMA, respectively.
Cell Culture and Transient Transfection-Y-1 and CV-1 cells were maintained as described previously (19). Three g of the CAT plasmids and 0.2 g of RSV/luc were transfected into the cultured cells by lipofection. The efficiencies of the transfections were normalized by the luciferase activities derived from the RSV/luc as described (5). For the cotransfection assays, 3 g of RSV/Ad4BP or the mock-expression plasmid, 3 g of Ad4ECAT0.8K or Ad4ECAT0.8KA, and 0.2 g of RSV/luc were transfected into CV-1 cells. The cells were harvested 42 h after the lipofection, and CAT assays were performed using 1-deoxy-[dichloroacetyl-1-14 C]chloramphenicol (56 mCi/mmol, Amersham, United Kingdom). All transfection experiments were performed at least 3 times.
DNase I Footprint Analysis-A DNA fragment spanning from ϩ57 bp to ϩ316 bp was obtained by polymerase chain reaction, inserted into the SmaI site of pUC19, and digested at either the EcoRI or HindIII site of the polylinker. After end labeling with polynucleotide kinase in the presence of [␥-32 P]ATP (222 TBq/mmol, Amersham), DNase I footprinting was performed with a nuclear extract prepared from Y-1 cells as described previously (22).
DNase I Hypersensitivity Assays-DNase I hypersensitivity assay was performed on the nuclei isolated from Y-1 cells, rat adrenal glands, and a rat liver as described (23). The nuclei (35 g of DNA) were digested in a 100-l reaction mixture with DNase I at concentrations of 0, 10, 25, 64, and 160 units/ml at 25°C for 6 min. Thirty-five g of the genomic DNA purified from each of the nuclear preparations was digested to completion with HindIII and then electrophoresed on a 1.5% agarose gel followed by Southern blotting. The 1.0-kb HindIII-PmaCI fragments from the rat and mouse mftz-f1 genes were 32 P-labeled and used as the probes for the rat tissue specimens and Y-1 cells, respectively.

Identification of the Transcriptional Element in the First
Intron of the Rat mftz-f1 Gene-We previously identified the E box as the transcriptional element in the 5Ј-upstream region of the rat mftz-f1 gene (19). In the present study, a further investigation of the gene transcription was performed with the CAT reporter gene constructs indicated in Fig. 1. Interestingly, Ad4ECAT2.0K which contains the 3.4-kb long first intron in addition to the 2.0 kb upstream from the EcoRI in the first exon showed an approximately 200-fold stronger CAT activity than Ad4CAT2.0K which lacks the first intron (19). The CAT activity of Ad4ECAT2.0K was examined in other steroidogenic I-10 testicular Leydig cells as well as non-steroidogenic CV-1 and NIH3T3 cells. The enhancement of the transcriptional activity by the first intron was observed only in the steroidogenic Y-1 and I-10 cells. Moreover, in Y-1 cells, the activity shown by Ad4ECAT2.0K was stronger than that by pSV2CAT containing an SV40 enhancer/promoter (data not shown). These results suggested the presence of a regulatory region responsible for the strong and the steroidogenic cell-specific transcription of the mftz-f1 gene in the first intron. Since Ad4ECAT2.0K contains the promoter regions and the first exons of both Ad4BP and ELP (19), we investigated which mRNA transcribed from the CAT plasmid in Y-1 cells by reverse transcriptase-polymerase chain reaction analyses. The amount of the ELP mRNA was significantly lower than that of Ad4BP, showing a good correlation with our previous observation (6) (data not shown). Therefore, the CAT activity of Ad4ECAT mostly reflects the Ad4BP promoter activity.
To identify the transcriptional element in the first intron, various deletion plasmids were constructed based on Ad4ECAT2.0K as shown in Fig. 1. A drastic decrease of the CAT activity was observed when the 5Ј deletion reached to the XbaI site (5Ј side) (Ad4ECAT⌬3.0K). Further truncations to the BamHI, HindIII, SacI, and XbaI (3Ј side) sites (Ad4ECAT⌬2.2K, ⌬1.8K, ⌬0.8K, and ⌬0.4K, respectively) abolished the remaining weak CAT activity. No remarkable change in the CAT activity was observed when Ad4ECAT⌬HX, ⌬BX, and ⌬XX having internal deletions were transfected. A significant decrease was, however, observed when the internal deletion reached to Ϫ79 bp in the 5Ј-upstream region (Ad4ECAT⌬PX). The decrease seems to be due to the lack of both the first exon and the E box element, the latter of which had been identified as a transcriptional element located from Ϫ82 to Ϫ77 bp (19). Judging from these observations, the transcriptional element in the first intron is likely to be located from the 5Ј splice junction to the XbaI site (5Ј side) and/or from the XbaI site (3Ј side) to the 3Ј splice junction. To locate the transcriptional element more precisely, the transcriptional activities of Ad4ECAT⌬XX-A and -B series of the CAT plasmids were investigated. In the case of the Ad4ECAT⌬XX-A series, a drastic decrease of the CAT activity was observed between Ad4ECAT⌬XX-A24 and -A0. In the Ad4ECAT⌬XX-B series, on the other hand, a 3-fold increase of the CAT activity was observed when the deletion reached to 82 bp from the 3Ј splice junction (Ad4ECAT⌬XX-B82), while a decrease of the CAT activity was observed with the plasmids from Ad4ECAT⌬XX-B37 to -B17. Further deletion to 1 bp (Ad4ECAT⌬XX-B1) abolished the CAT activity.
It is widely accepted that the splice junction sequences for correct splicing require not only a GT-AG rule but a pyrimidine nucleotide cluster just upstream from the 3Ј splice junction and the consensus several nucleotides at the 5Ј splice junction (24 -26). Indeed, the mftz-f1 gene has a pyrimidine-rich sequence at the 3Ј splice junction, (T/C) 10 ACAG/G, and the consensus sequence, CA/GTAAGT, at the 5Ј splice junction (Fig. 2). Taking the rule into consideration, the decrease of the CAT activity between Ad4ECAT⌬XX-B17 and -B1 seems to be caused by the impairment of the consensus 3Ј sequence. How- ever, since it was not clear whether the decrease between Ad4ECAT⌬XX-B37 and -B17 resulted from the deletion of a putative transcriptional element or a decrease in the splicing efficiency, we investigated the structure of the mRNA transcribed from Ad4ECAT⌬XX-B17. A significant portion of the mRNA was found to be unspliced (data not shown). Accordingly, it is likely that the decrease of the CAT activity mainly resulted from the decrease in the splicing efficiency. On the other hand, there was also a possibility that a putative transcriptional element is located in the 24-bp nucleotides from the 5Ј splice junction (ϩ154/ϩ177). To examine the transcriptional activity of the region, Ad4CAT0.8K-EF and -ER were constructed by insertion of a 91-bp fragment (ϩ146/ϩ236) including the 24-bp nucleotides at the Eco81I site (Ϫ265) of Ad4CAT0.8K. The CAT activities in Y-1 cells increased by about 7-fold by the insertion of the fragment regardless of their orientations as shown in Fig. 3. The fragment also increased the CAT activities of the SV40 core promoter by about 5-fold (pCAT-EF and -ER) (Fig. 3).
Binding of Ad4BP to the Transcriptional Element of Its Own Gene-The binding factor(s) to the transcriptional element described above was investigated with the nuclear extract prepared from Y-1 cells. DNase I footprint analyses were performed using a DNA fragment from ϩ57 to ϩ316 containing the element. When the nuclear extract prepared from Y-1 cells was used, a single region was protected from DNase I digestion as shown in Fig. 4A. Interestingly, the protected region, 5Ј-TGAAGGCCGGGGCCCA-3Ј, contained a possible sequence (underlined) recognized by Ad4BP (Ad4 site) (1). To determine whether Ad4BP or other proteins were responsible for the observed binding, gel mobility shift assays were performed with labeled dENC containing the protected sequence from the DNase I digestion. For characterization of the binding sequence, competition experiments were also performed with the oligonucleotides carrying nucleotide substitutions (dEM1 to dEM6) (Fig. 4B). As shown in Fig. 4C (left panel), a single complex formation with dENC was observed and was completely inhibited by an excess amount of nonradiolabeled dENC. The oligonucleotides, dEM2 and M3, carrying disrupted Ad4 sites did not function as competitors, whereas the other oligonucleotides, dENM1, M4, M5, and M6, carrying the intact Ad4 site were able to function as the competitors. As shown in Fig. 4C (right panel), the complexes with dENC showed the same mobility on a polyacrylamide gel as that with dAd4, an authentic Ad4 site in the bovine CYP11B (22). The signals with the two distinct probes completely disappeared by the nonradiolabeled oligonucleotides each other. Finally, both complex formations were inhibited by the addition of an antiserum to Ad4BP. The same results were also obtained with a nuclear extract prepared from rat adrenal glands.
Autoregulatory Loop in the mftz-f1 Gene Transcription-To investigate the function of the Ad4 site directly, the GAAG-GCCG sequence in Ad4ECAT0.8K was mutagenized to GAATATCG to generate Ad4ECAT0.8KA. The mutated sequence was confirmed to be incapable of binding to Ad4BP as described above. As shown in Fig. 5, the mutation within the Ad4 site caused more than a 80% decrease of the CAT activity. To examine the functional relationship between the Ad4 site and the E box in the transcriptional activity of the mftz-f1 gene, FIG. 4. Binding of Ad4BP to the regulatory region. A, DNase I footprint analysis of the regulatory region. The DNA fragment from ϩ57 to ϩ316 bp carrying the region was end-labeled at ϩ57 bp and used for footprint analysis. Increasing amounts of the nuclear extract prepared from Y-1 cells (from 0 to 25 g) were used. To determine the protected region, G ϩ A and T ϩ C ladders were prepared by chemical cleavage of the probe (41). The shaded oval indicates the portion protected from the DNase I digestion. The corresponding nucleotide sequence is shown in bold letters. B, the nucleotide sequences of the probes used in the gel mobility shift assays. dENC contains the protected region from the DNase I digestion. Each of the underlined three nucleotides is substituted to make dEM1 to dEM6. The closed box indicates the Ad4 site. C, gel mobility shift assays with the nuclear extract prepared from Y-1 cells. The 32 P-end labeled oligonucleotides, dENC or dAd4, were incubated with 5 g of the nuclear extract. For the competition assays, a 50-fold molar excess of each nonradiolabeled oligonucleotide was added prior to the addition of the probes. The antiserum to Ad4BP (␣Ad4BP) was added after the addition of the probes. The incubation mixtures were then examined on a 4.5% polyacrylamide gel. The complexes with dENC and dAd4 showing the same mobility are indicated by an arrowhead. the CAT plasmid carrying the intact Ad4 but a mutated E box (Ad4ECAT0.8KM) was transfected. The mutation within the E box caused more than a 75% decrease of the CAT activity. A similar extent of decrease was observed when both the Ad4 site and the E box were mutated (Ad4ECAT0.8KMA).
To obtain direct evidence that Ad4BP activates the transcription of the mftz-f1 gene, an expression vector for Ad4BP was cotransfected with Ad4ECAT2.0K or Ad4ECAT2.0KA into CV-1 cells which lacks endogenous Ad4BP. The Ad4BP expression vector activated the transcription of Ad4ECAT2.0K by about 3-fold as shown in Fig. 6. Because the disruption of the Ad4 site (Ad4ECAT2.0KA) completely abolished the activation, Ad4BP activated the gene as a consequence of the binding to the Ad4 site. A truncated form of Ad4BP (⌬Ad4BP207) which lacks a putative activator domain at the carboxyl-terminal half but has the ability to bind to the Ad4 site (3) failed to activate the transcription. An expression vector for the catalytic subunit of protein kinase A (5) was also cotransfected with that for Ad4BP. The protein kinase A expression vector had, however, no effect on the transactivation function of Ad4BP (data not shown).
DNase I Hypersensitive Region-One of the questions to be addressed is whether the E box element and the Ad4 site of the mftz-f1 gene are active or not in the adrenal cortex. To answer this question, we compared the chromosomal structure of the region containing both the elements by detecting hypersensitive sites to DNase I. The nuclei prepared from rat adrenal glands, Y-1 cells, and a rat liver were subjected to DNase I digestion. As shown in Fig. 7, a 3.1-kb long HindIII-HindIII fragment containing both the E box and the Ad4 site was detected without DNase I digestion. If sensitive sites are present in this locus, DNA fragments of small sizes should appear depending on the concentration of DNase I. As expected, a signal with a length estimated to be from 1.1 to 1.3 kb was observed when the nuclear preparations from the adrenal glands and Y-1 cells were used, whereas no signal was detected with the liver nuclei. Judging from the length of the digested fragment, the region sensitive to DNase I seems to cover both the E box and the Ad4 site. DISCUSSION In the present study, we investigated the transcriptional regulatory mechanism of the rat mftz-f1 gene and obtained strong evidence for the autoregulatory loop involved in the transcriptional regulation. The significant transcriptional enhancement by the first intron gave us a clue to find the autoregulatory loop. The regulatory region was located near the splice donor site in the first intron and the presence of an Ad4 site in the region was confirmed by using the specific antiserum to Ad4BP in the gel mobility shift analyses. In a previous paper (1), we described the purified Ad4BP bound to PuPuAGGTCA as well as PyCAAGGPyPyPu. Although the Ad4 site (GAAG-GCCG) identified in the present study satisfied neither of the two consensus sequences completely, it is likely to be an active derivative of the former sequence, where TCA at the 3Ј side are changed to CCG. Based on our previous observation that the former consensus sequence showed a weaker binding affinity than the latter (1), the Ad4 site identified in this study has relatively weak affinity in spite of its strong transcriptional activity. The inconsistency between the transcriptional activity and the binding affinity is probably explained by the following observation. In the mftz-f1 gene, the E box is the essential cis-element functioning cooperatively with the Ad4 site since the disruption of the E box caused a significant decrease of the transcriptional activity even in the presence of the intact Ad4 site. A similar decrease was also observed when only the Ad4 site was inactivated. These observations support the notion that the transcriptional activity of the Ad4 site is modulated by the function of the E box and thereby the Ad4 site shows the strong transcriptional activity in spite of the weak binding affinity.
The regulatory region including the Ad4 site was shown to activate both the original mftz-f1 gene promoter carrying the E box and the heterologous SV40 basal promoter. In both cases, however, the transcriptional activations by the region were insufficient to explain the significant transcriptional enhancement by the first intron. A possible reason for the discrepancy is as follow. It was reported in several genes that the splicing reaction itself seems to be critical for efficient production of the mRNA species (27,28). It is possible that the pre-mRNA from Ad4ECAT0.8K was efficiently spliced to produce the mature The function of the Ad4 site in the first intron of the mftz-f1 gene was further supported by the cotransfection assays with the expression vector for Ad4BP using CV-1 cells. The transcription was activated only when the CAT vector carrying the intact Ad4 site and the expression vector coding intact Ad4BP were used, although the activation by the coexpression of Ad4BP was smaller than expected. Since the transcription of the CYP11A and CYP11B genes was activated by Ad4BP only in the presence of an expression vector for the catalytic subunit of protein kinase A (5), the effect of the protein kinase A was also examined. However, a further activation was not achieved with the mftz-f1 gene. Considering the cooperative function of the E box and the Ad4 site, the low activation might be due to the absence of the E box binding factor in CV-1 cells. In fact, when the nuclear extracts prepared from Y-1 cells and CV-1 cells were examined by the gel mobility shift analyses, the E box binding factor was detectable with the Y-1 but not with the CV-1 nuclear extract (data not shown).
Differentiated tissues originate from the primordial cells through successive events. Various sets of genes express their functions at the destined time points along the differentiation steps, and finally the tissues acquire their specific functions by expressing the final set of the tissue-specific genes. In the case of the adernal cortex, the differentiated tissue is able to synthesize steroid hormones owing to the functions of the steroidogenic tissue-specific P-450s. When the differentiation process is considered, it seems reasonable to suppose that a gene regulatory cascade functions along the process, in which the genes encoding specific transcription factors are involved as the components. In the cascade required for adrenocortical differentiation, Ad4BP should be located upstream of the adrenocortical specific genes including the steroidogenic P-450 genes, while it is located downstream of other genes regulating the mftz-f1 gene transcription. It was clarified in the present study that the mftz-f1 gene is activated by the autoregulatory loop. Therefore, it is quite interesting to suppose that once the autoregulatory loop starts to function in the particular cell types, the upstream genes essential for the initial activation of the mftz-f1 gene transcription are no longer necessary thereafter to maintain the Ad4BP expression. Such the autoregulation was reported to function in the HOX4C (29), Hox 4.2 (30), MyoD1 (31), and pit-1 (32-34) genes in mammals, and the fushi tarazu (35), deformed (36), even-skipped (37), Ultrabithorax genes (38), and sex-lethal (39) genes in Drosophila. Interestingly, all the genes listed above have been reported to play significant roles in the differentiation of particular tissues or cell types. These observations, including the present one, seem to indicate that the autoregulatory mechanism is widely adopted as the transcriptional regulation of the key transcription factors during differentiation.
The chromatin structure of the mftz-f1 gene was also investigated by detecting the hypersensitive site to DNase I. The region containing the Ad4 site and the E box was observed to be open in the steroidogenic tissue but not in the non-steroidogenic tissue. The steroidogenic tissue-specific chromatin structure is probably essential to the mftz-f1 gene transcription by making the transcription factors such as Ad4BP and E box binding factor accessible to their binding sites. Accordingly, it is supposed that the tissue-specific transcription of the gene is also guaranteed by the tissue-specific chromatin structure in addition to the tissue-specific transcription factor.
The present study demonstrated that the rat mftz-f1 gene is autoregulated through the Ad4 site in the first intron and that the E box is essential for the function of the Ad4 site. These observations, however, were made with Y-1 cells which have differentiated features as the adrenocortical cells. Because of the crucial role of Ad4BP in adrenal and gonadal differentiation, it is of importance to study the transcriptional regulation of the gene during the differentiation processes of the animal tissues.
Acknowledgments-We thank Dr. Tsuneo Omura (Vanderbilt University) for critical reading of the manuscript and Dr. Hiroyuki Sasaki (Kyushu University) for technical advice on DNase I hypersensitivity assay.