Tissue-specific Expression of an 1 I@-Hydroxysteroid Dehydrogenase with a Truncated N-terminal Domain A POTENTIAL MECHANISM FOR DIFFERENTIAL INTRACELLULAR LOCALIZATION WITHIN MINERALOCORTICOID TARGET CELLS*

The enzyme 1 la-hydroxysteroid dehydrogenase (1 1- HSD) is thought to confer specificity on the nonselective Type I adrenocorticoid receptor by converting glucocorticoids to receptor-inactive metabolites in mineralocorticoid target tissues. S1 nuclease analyses using a rat liver 11-HSD probe demonstrated tissue-specific expression of the 6’ region of the 11-HSD gene in the liver, lung, and kidney not evident in previous studies. Renal tissue contained a unique protected species which mapped to a position within the coding region, consistent with a divergence in liver and kidney protein sequences. Screening of a rat kidney cDNA library resulted in the isolation of several clones (11- HSDlB) noncolinear in their 6’ regions with the liver sequence (1 1-HSD1 A). Nucleic acid sequence analysis showed that the divergent clones code for a protein lacking a 26-amino acid NHz-terminal putative mem-brane-spanning signal peptide. The deletion of the leader sequence from the microsomal 11-HSDlA protein may result in a nuclear localization of the 11- HSDlB isoform. The renal 11-HSDlA and 11-HSDlB species increased coordinately during ontogeny and in parallel with the developmental surge in glucocorticoids. At least three alternate sites of polyadenylation were found to be utilized by the

The enzyme 1 la-hydroxysteroid dehydrogenase (1 1-HSD) is thought to confer specificity on the nonselective Type I adrenocorticoid receptor by converting glucocorticoids to receptor-inactive metabolites in mineralocorticoid target tissues. S1 nuclease analyses using a rat liver 11-HSD probe demonstrated tissuespecific expression of the 6' region of the 11-HSD gene in the liver, lung, and kidney not evident in previous studies. Renal tissue contained a unique protected species which mapped to a position within the coding region, consistent with a divergence in liver and kidney protein sequences. Screening of a rat kidney cDNA library resulted in the isolation of several clones (11-HSDlB) noncolinear in their 6' regions with the liver sequence (1 1-HSD1 A). Nucleic acid sequence analysis showed that the divergent clones code for a protein lacking a 26-amino acid NHz-terminal putative membrane-spanning signal peptide. The deletion of the leader sequence from the microsomal 11-HSDlA protein may result in a nuclear localization of the 11-HSDlB isoform. The renal 11-HSDlA and 11-HSDlB species increased coordinately during ontogeny and in parallel with the developmental surge in glucocorticoids. At least three alternate sites of polyadenylation were found to be utilized by the 11-HSD gene. Southern blot analysis showed the presence of a single gene in the rat. This study shows the expression of a kidneyspecific 11-HSD isoform which may protect the Type I adrenocorticoid receptor from occupation by glucocorticoids in the nucleus of a mineralocorticoid target cell.
The mineralocorticoid (or Type I adrenocorticoid) receptor exhibits equal affinity for mineralocorticoid and glucocorticoid hormones (Krozowski and Funder, 1983;Beaumont and Fanestil, 1983). It has been proposed that the enzyme 110hydroxysteroid dehydrogenase (11-HSD)' confers mineralocorticoid specificity on the Type I receptor by converting the much higher levels of circulating glucocorticoids to receptor * This work was supported by the National Health and Medical Research Council of Australia. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in thispaper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) M77835.
inactive metabolites. The action of 11-HSD on corticosterone produces an ll-keto metabolite which has about 0.3% of the affinity of the parent compound for Type I receptors and a correspondingly lower affinity for the Type I1 or classical glucocorticoid receptor . The mineralocorticoid aldosterone is not metabolized by 11-HSD due to the presence of a highly reactive aldehyde group at C18, which in solution cyclizes to the 11,18 hemiketal or the 11,18,20 hemiacetal.
In vivo studies have shown that inhibition of 11-HSD activity by administration of carbenoxolone or glycyrrhetinic acid abolishes mineralocorticoid specificity of the Type I receptor (Edwards et al., 1988, Brem et al., 1989; in man a congenital deficiency in 11-HSD activity results in the syndrome of apparent mineralocorticoid excess which is characterized by salt retention and elevated blood pressure (Ulick et al., 1979). Although these effects may be due to the inappropriate occupation of the Type I receptor by glucocorticoids, there is now increasing evidence that glucocorticoids acting through the Type I1 receptor can also effect mineralocorticoid responses in whole cell' and in in uiuo systems (Funder et al., 1990). The Type I1 adrenocorticoid receptor seemingly activates the mineralocorticoid response element, exacerbating the nonselectivity of glucocorticoid hormones. 11-HSD, and probably of other enzymes, thus endow mineralocorticoid specificity both to the receptor and to the mineralocorticoid response element.
Recently, an NADP+-dependent 11-HSD has been purified from rat liver microsomes and cloned from a rat liver expression library (Lakshmi and Agarwal et al., 1989). Northern blot analysis has revealed the presence of a 1700nucleotide (nt) mRNA in all tissues examined with the exception of the kidney where we have recently identified multiple unique species of the message (Krozowski et al., 1990). The pattern of 11-HSD gene expression in the renal papilla also differs from that observed in the medulla and cortex where mineralocorticoids are known to modulate sodium retention.
Although it has been proposed that in mineralocorticoid target cells the Type I receptor is protected by abundant 11-HSD activity, antibodies raised against the hepatic 11-HSD enzyme have failed to colocalize 11-HSD and Type I receptor immunoreactivity (Rundle et al., 1989). Northern blot analyses have readily detected 11-HSD mRNA in those regions of the kidney with high concentrations of Type I receptors (Krozowski et al., 1990). Together these results suggest that the kidney may contain an 11-HSD enzyme(s) which differs from that found in hepatocytes. Indeed, Western blot analysis has revealed that the kidney contains immunoreactive species Naray-Fejes-Toth, A., Watlington, C., and Fejes-Toth, G . (1991) Endocrinology 129, 17-21. which are absent from tissue homogenates of the liver (Monder and Lakshmi, 1990).
Southern blot analyses indicate that there may be more than one 11-HSD gene in the rat and human genomes (Agarwal et al., 1989). There is also increasing evidence that there may be several unrelated 11-HSD enzymes present in the rat kidney. Studies on immunopurified cortical-collecting cells have demonstrated an 11-HSD with a K,,, 2 orders of magnitude lower than that of the hepatic species,' whereas histochemical studies in the rat kidney have identified an NAD+dependant enzyme with a clearly different distribution to that of the NADP+-requiring species (Mercer and Krozowski, 1991). Further studies are needed to characterize 11-HSD gene expression in the kidney in order to determine the functional significance of the unique mRNA species observed in this tissue.
In the present study we have investigated the heterogeneity of 11-HSD mRNA expression in the rat kidney and have identified 11-HSD cDNAs containing an alternate 5' U T region and an open reading frame which codes for a n 11-HSD enzyme with a deleted signal peptide. Recent studies (Mercer and Krozowski, 1991) from our laboratory have also identified a n NAD+-dependent renal 11-HSD which we have named 11-HSD2. In the present study we refer to the NADP+-dependent enzyme as 11-HSD1 (Lakshmi and Agarwal et al., 1990).

MATERIALS AND METHODS
Isolation of RNA-Total RNA was isolated from the liver, lung, and kidney of male Sprague-Dawley rats by the guanidinium isothiocyanate method as described previously (Krozowski et al., 1990). SI Nuclease Mapping-Antisense RNA probes were used to perform solution hybridization analyses of tissue RNA. Plasmid constructs used for the generation of riboprobes consisted of fragments of the pllDH-1 insert (Agarwal et al., 1989) ligated into either the pGEM-3Z or Bluescript KS vector. Plasmid pRPA consisted of nucleotides 192-673 of the pllDH-1 insert ligated into the BamHI/ EcoRI site of the pGEM-3Z plasmid. The 481-bp fragment was generated by the polymerase chain reaction using primers flanked by BamHI or EcoRI. Probe A was transcribed from the T7 promoter after linearizing pRPA with HindIII. Plasmid pRPB was constructed by insertion of the 466-bp EcoRI/EcoRV fragment into Bluescript KS. Probe B was then transcribed from the T3 promoter after linearizing the pRPB plasmid with XbaI. The plasmid pRPC was constructed by inserting the 439-bpAccI/EcoRI fragment into pGEM-32. pRPC was linearized with HindIII and transcribed from the T7 promoter to yield probe C. Plasmid pRPD consisted of the complete pllDH-1 insert ligated into Bluescript KS. Probe D was synthesized from pRPD by transcription from the T3 promoter after linearizing at the internal NcoI sites. All riboprobes were synthesized with the aid of the Gemini Riboprobe Synthesis Kit (Promega, Madison, WI).
Solution hybridization and S1 nuclease analyses were performed by a modification of a method described previously (Albiston et al., 1990). Briefly 5 X lo4 cpm of riboprobe was added to 30 pg of total tissue RNA or transfer RNA (tRNA), denatured for 5 min at 85 'C, and hybridized at 60 "C overnight. Each reaction mixture was then digested with 500 units of S1 nuclease for 50 min at 37 "C. Reaction products were analyzed on a 4% polyacrylamide, 8 M urea sequencing gel. Protected fragment sizes were calculated from a sequencing ladder run in parallel.
Screening of a Rat Kidney cDNA Library-A commercial rat kidney cDNA library constructed in X g t l O was used (Clontech, CAI. Plaques (3 X 10') were transferred to 14-cm nylon filters and hybridized overnight in a buffer containing 5 X Denhardt's, 5 X SSPE, 0.1% SDS, 50% formamide, and 100 pg/ml herring sperm DNA at 42 "C with the 32P randomly primed insert from pllDH-1 (Agarwal et al., 1989). Filters were washed in 2 X SSC, 0.1% SDS at room temperature for 20 min and then twice in 0.2 X SSC, 0.1% SDS at 50 "C for 20 min. Positive plaques were identified and purified by standard procedures (Maniatis et al., 1982).
Primer Extension Analysis-A synthetic oligonucleotide, oligo 1301, was 5' end-labeled using T4 polynucleotide kinase (Maniatis et al., 1982). An aliquot of lo7 cpm was added to 50 pg of kidney RNA, the mixture denatured at 85 "C for 5 min, and the probe annealed for 6 h at 50 "C. Primer extension was performed using 200 units of Moloney murine leukemia virus reverse transcriptase under conditions recommended by the supplier (Bethesda Research Laboratories). The reaction mixture was extracted with phenol/chloroform and precipitated with ethanol and the extension products run on a 6% polyacrylamide, 8 M urea sequencing gel. The sizes of the extended fragments were determined by comparison with a dideoxy sequencing reaction performed on pGEM-3Z containing clone c13 (see Fig. 2a) and using oligo 1301 as primer.
Northern Blot Anulysis-Oligo 1301 was end-labeled with 32P and hybridized to Northern blots in a buffer containing 5 X Denhardt's, 5 X SSPE, 0.1% SDS, 50% formamide, and 100 pg/ml herring sperm DNA at 42 "C overnight and washed in 5 X SSC, 0.1% for 20 min at room temperature, followed by a second wash in 1 X SSC, 0.1% SDS at room temperature (Maniatis et al., 1982). Oligo 1301 is complementary to the sequence mapping between positions 23-58 in 11-HSDlB (Fig. 26). Filters were stripped by briefly boiling in deionized water before reprobing with the insert from pllDH-1 as described previously (Krozowski et al., 1990).
Southern Blot Analysis-Genomic DNA was isolated from the livers of Sprague-Dawley rats (Maniatis et al., 1982). Aliquots (10 pg) were digested with EcoRI, BglII, XbaI, HindIII, and HincII, fractionated on a 0.8% agarose gel, and transferred to a nylon membrane. The blots were probed with an oligo-labeled (10' cpmlpg) kidney cDNA probe (clone c8) and washed under the same conditions as used to screen the cDNA library.
DNA Sequence Anulysis-DNA sequencing was performed on cDNA fragments cloned into pGEM3Z. Sequence reactions incorporated the dideoxy chain terminating method using the Sequenase sequencing kit (United States Biochemical Corp.).

SI Nuclease
Mapping of Liver, Lung, and Kidney RNA-Total RNA was prepared from liver, lung, and kidney tissue and subjected to S1 nuclease analysis using antisense RNA probes. Probes A, C, and D (Fig. l a ) showed no discernable differences in probe protection between the three tissues examined (results not shown), but probe B gave a distinctly different pattern between liver, lung, and kidney RNA (Fig.   lb). In all three tissues a band was observed at 466 nt corresponding to the fully protectedprobe B minus polylinker sequences. In the liver protected RNA species were present a t 420 nt and 369 nt, whereas in the lung and kidney bands were seen at 410 nt and 369 nt. Furthermore, the kidney alone contained a protected species a t 307 bp. The presence of identical truncated species in all tissues may be due to a divergence in sequence between probe and message or it may be due to RNA secondary structure. However, protected bands not present in all three preparations of RNA are more likely to be the result of noncolinearity, and an unequivocal resolution of this issue can only come from the isolation of cDNAs containing sequences which show dishomology with the probe.

11-HSDIB (kidney)
, . . . . . . -, , . . . . . . . . . . . . E f R """"""""""""""""""""- FIG. 1. a, schematic representation of antisense RNA probes used for S1 nuclease analysis. The cDNA clone pllDH-1 (Agarwal et al., 1989) is shown above with the coding region represented by an open box. Fragments of pllDH-1 were subcloned into pGEM-3Z or Bluescript KS and antisense RNA probes transcribed from the T7 or T3 promoters as described under "Materials and Methods." The region covered by each probe is represented by an arrow. b, S1 nuclease analysis of rat liver, lung, and kidney mRNA using probe B. Total RNA from liver, lung, and kidney was annealed with 5 X lo4 cpm of probe B prepared as described under "Materials and Methods." A control hybridization was also performed in the presence of transfer RNA (tRNA) instead of tissue RNA. The RNA-RNA hybrids were digested with S1 nuclease and subjected to electrophoresis on a 4% polyacrylamide sequencing gel. An aliquot of undigested probe B (10' cpm) was also loaded and is shown at the extreme right of the autoradiogram. Autoradiography was performed overnight with an intensifying screen. c12) and 11-HSDlB (clone c20) are compared with the liver 11-HSDlA sequence (pllDH-1). The 5' region of 11-HSDlB diverges from 11-HSDlA at a position corresponding to 159 nt in the liver 11-HSDlA sequence (Fig. 2b), consistent with the band obtained at 307 nt during S1 nuclease studies.
Sequence analysis of clone c13 also showed that the TAC codon coding for TyrR4 in pllDH-1 was replaced by a TAT, resulting in a silent mutation. This single base change was not unique to the 11-HSDlB cDNAs but was also found in all 11-HSDlA clones, suggesting the mutation is due to allelic polymorphism.
The noncontiguous region does not contain translation initiation codons (Fig. 2b). Instead Met' in the 11-HSDlB protein is translated from an ATG equivalent to a codon coding for Met2' in the 11-HSDlA enzyme. Thereafter the two peptide sequences are identical.
Sequence analysis of the 3' region of clones isolated with poly(A) tails showed that three alternate sites of polyadenylation were used (Fig. 2c). Poly(A) extension commenced either 16,24, or 30 bp downstream of two putative overlapping poly(A) signal sequences. 11-HSDlA and 11-HSDlB clones did not show an obvious preference for any of the polyadenylation sites as all three sites of poly(A) addition were found in cDNAs coding for both isoforms of the enzyme.
Primer Extension and Northern Blot Analysis of 11-HSDlB-When primer extension analysis was performed on kidney RNA using oligo 1301 (Fig. 3a), the length of extended products observed corresponded to positions 11 and 15 nucle- The pllDH-1 insert was isolated and labeled with "P by random priming. A rat kidney cDNA library in X g t l O was screened (450,000 plaques) and clones c3 to c19 isolated. Partial restriction mapping and sequencing of 200-300 bp of the flanking regions of all cDNAs identified clones c13 and c18 as noncolinear with pllDH-1. The noncolinear regions of these clones are represented by filled boxes. A oligonucleotide (oligo 1301) complementary to the noncolinear region of c13 (see b) was then synthesized, end-labelled with ' ?P and used to screen another 300,000 plaques. Screening with the oligo 1301 probe resulted in the isolation of clone c20. Sequencing of c20 showed that it contained the longest 5' region noncolinear with pllDH-I. All 5' UT regions noncolinear with the pllDH-1 clone were found to be colinear with each other. b, sequence alignment of the 5' regions of 11-HSDlA and 11-HSDlB cDNA clones. The 5' region of the liver cDNA clone pllDH-1 is represented by 11-HSDlA (liver) and is aligned with the 5' regions of the kidney 11-HSDlA and 11-HSDlB cDNAs (clones c12 and c20, respectively, in a ) . Derived amino acid sequences are given below their respective cDNAs. Putative initiation codons are boxed. The oligonucleotide oligo 1301 was made complementary to the nucleotide sequence double underlined in 11-HSDlB. c, nucleotide sequences of the 3' termini of clones c6, c8, and c18. All cDNA clones containingpoly(A) tails were found to be polyadenylated in one of three positions, as represented here by the nucleotide sequences of clones c6, c8, and c18. Two overlapping polyadenylation signals (AATAAA) are underlined. otides 5' of the 11-HSDlB sequence shown in Fig. 2b, indicating that the total length of the 5' UT region in 11-HSDlB is about 75 bp. When oligo 1301 was used as a probe Northern blot analysis of kidney RNA showed that the 11-HSDlB message migrated as a 1.5-kilobase species (Fig. 3b).
Hydrophilicity Analysis of the 11-HSDlA Protein-An analysis of the hydrophilicity profile of the 11-HSDlA protein showed the presence of a strongly hydrophobic NH, terminus characteristic of a signal peptide (Fig. 4). Alignment of the 11-HSDlA and 11-HSDlB proteins revealed that the latter molecule is a truncated protein with a deleted signal sequence. However, the truncation stops short of the NADP+ cofactor binding domain. It is likely the shortened protein has retained enzyme activity given the high homology of the NH, terminus of 11-HSDlB with the NH2 termini of other dehydrogenases (The et al., 1989, Debelle andSharma, 1986). Tissue-specific 11 -HSD Gene Expression FIG. 3. a, primer extension analysis of 11-HSDlB mRNA. An oligonucleotide primer (oligo 1301, see Fig. 26) was end-labeled and hybridized to total kidney RNA ( l a n e I ) or tRNA (lane 2) and extended using reverse transcriptase in the presence of all four unlabeled deoxyribonucleotides as described under "Materials and Methods." To obtain size markers a dideoxy sequencing reaction was performed on clone c13 using oligo 1301 as a primer and the reactions run in parallel on a 6% polyacrylamide sequencing gel. The two primer extension products evident in lane 1 correspond to positions 11 and 15 nt upstream of clone c20. b, Northern blot analysis of liver and kidney using oligo 1301. Twenty pg of total liver ( l a n e 1 ) or kidney ( l a n e 2) RNA were denatured with glyoxyl and electrophoresed on a 1.2% agarose gel. After transfer to Hybond membrane the RNA was hybridized with ""P-end-labeled oligo 1301, washed, and subjected to autoradiography for two days with an intensifying screen as described under "Materials and Methods." The results are depicted in a. The filter was then stripped and reprobed with the "'P-labeled pllHD-1 cDNA probe as described previously (Krozowski et al., 1990). The results obtained with the cDNA probe are depicted in b. Size markers shown on the left are in kilobases. The amino acid sequence of the 11-HSDlA protein was subjected to hydrophilicity analysis by the Hopp and Woods algorithm using a sliding window of 6 residues (Hopp and Woods, 1981). Positive hydrophilicity values are indicated above the horizontal axis. The 11-HSDlA and 11-HSDlB proteins also represented by horizontal bars and are aligned with the hydrophilicity plot. The filled region in the 11-HSDlA diagram represents the putative signal peptide. The positions and approximate extents of the putative NADP+ and steroid binding domains are also shown. These domains were deduced from the regions of conserved amino acids across several dehydrogenases (Agarwal et al., 1989).
Ontogeny of 11-HSDlA and 11-HSDIB mRNA Expression in the Kidney-Using Northern blot analysis we have previously shown that the 11-HSD1 gene is differentially expressed during development (Krozowski et dl., 1990). Modest levels of message are present in the liver and lung of the neonate, whereas renal 11-HSD1 mRNA remains undetectable until three weeks of age. In the present study we used the more sensitive technique of S1 nuclease analysis to determine the developmental expression of renal 11-HSD1 mRNA. An autoradiogram showing the protected 11-HSDlA and 11-HSDlB mRNA species is shown in Fig. 5. Equal amounts of both species of mRNA were found to be present in kidney tissue from 1-week old rats. Between 1 and 2 weeks of age the 11-HSDlA and 11-HSDlB messages were found to be coordinately expressed and increase in abundance in an exponential fashion. There was a further coordinate increase in messages up to 4 weeks of age but after this time there were no further increases up to the age of 16 weeks.
Southern Blot Analysis-Expression of the two 11-HSD1 messages may arise as a result of transcription from a single gene or they may be the product of two separate but closely related genes. It has previously been suggested that there are multiple 11-HSD1 genes in the rat (Agarwal et al., 1989), and we have also obtained multiple bands on Southern blots when using the full-length pllDH-1 clone as a probe (results not shown). However, these data are also consistent with the presence of a single large gene containing internal restriction sites recognised by enzymes used in the Southern blot analysis. In order to resolve this issue we probed a rat genomic blot with clone c8 (see Fig. Za), a cDNA which extends 3' of 772 bp in pllDH-1. If there are multiple genes coding for 11-HSD1, one would still expect to see more than one band, in the majority of restriction enzyme digests, when using the truncated probe. However, EcoRI, BglII, HindIII, and HincII digests gave a single band while only XbaI appeared to give two bands (Fig. 6). These results are consistent with the existence of a single 11-HSD1 gene in the rat genome.

DISCUSSION
In the present study we have demonstrated that the liver, lung, and kidney express the 11-HSD1 gene in a tissue-specific manner and that the kidney expresses an mRNA which codes for a truncated form of the enzyme. It has been proposed that 466 -4 1 0 -369 -307 -1 2 4 8 1 6 Weeks of Age FIG. 5. Ontogeny of 11-HSDlA and 11-HSDlB gene expression in the rat kidney. Total RNA was extracted from whole rat kidneys and subjected to S1 nuclease analysis using Probe B as described in Fig. lb. Autoradiography was performed for 3 days with an intensifying screen. The equivalence of RNA added to each sample was checked by subjecting an aliquot to Northern blot analysis and probing for 18 S ribosomal RNA as described previously (Krozowski et al., 1990). Total rat genomic DNA (10 ug) was digested with EcoRI, BglII, XbaI, HindIII, and HincII and subjected to Southern blot analysis using probe c8 as described under "Materials and Methods." the 11-HSD1 enzyme is involved in mediating aldosterone specificity in mineralocorticoid target tissues where high levels of circulating glucocorticoids would otherwise occupy the nonselective Type I adrenocorticoid receptor (Edwards et al., 1988;Funder et al., 1988). The widespread distribution of the enzyme suggests that it may also modulate occupancy of the Type I1 adrenocorticoid or classical glucocorticoid receptor, although it is not known whether the same form of 11-HSD1 is present in all tissues. Indeed, the unique species of 11-HSDl RNA expressed in the kidney suggest that some renal cells may produce tissue specific isoforms. S1 nuclease analysis revealed a complexity of tissue specific 11-HSD1 gene expression not evident in previous studies. The heterogeneity of renal RNAs observed on Northern blot analysis (Krozowski et al., 1990) is mainly due to the existence of alternate 5' sequences. The presence of multiple sites of poly(A) addition also make a small contribution to the differences in message size. We originally observed five species of 11-HSD1 RNA in the kidney by Northern blot analysis (Krozowski et al., 1990), whereas in the present studies only four RNAs were found with differing 5' regions. This apparent inconsistency can be reconciled by the presence of RNA species which extend 5' of the probe, consistent with the identification of 1900 n t RNAs in renal medulla and cortex (Krozowski et al., 1990).
Sequence analysis of the 3' end of renal 11-HSD1 clones showed the presence of three alternate sites of poly(A) addition. Alternate polyadenylation may be due to the presence of the two overlapping poly(A) addition signals present in the sequence AATAAAATAAA. The insertion of an AU-rich sequence into the 3' noncoding region of genes has been shown to destabilize transcripts (Shaw and Kamen, 1986); the presence of the ATAAATT sequence in the 3"extended 11-HSDl cDNAs indicates that the corresponding messages may also be less stable. Since no correlation was apparent between the site of poly(A) addition and 11-HSDlA or 11-HSDlB clones, it is likely that similar mechanisms are used to modulate the stability of both messages. These observations suggest that 11-HSD1 messages can be regulated at the level of mRNA turnover by the selection of different 3' UT sequences.
The heterogeneity of mRNA species observed in the kidney suggests a complex scenario of enzyme expression. Several eucaryotic genes are known to display 5' heterogeneity in their mRNAs by the use of alternative promoters with or without alternate splicing (Chobert et al., 1990, Mukai et al., 1986, Frunzio et al., 1986. The noncolinearity near the start of the 11-HSDlA and 11-HSDlB mRNAs suggests that the 11-HSD1 gene uses alternative promoters to express the two forms of the message. Alternatively, the two 11-HSD1 isoforms may be expressed from different genes. However, our study indicates the presence of a single gene in the rat. Although early studies (Agarwal et al., 1989) in the human showed the presence of multiple bands on Southern blot analysis and suggested the presence of several 11-HSD1 genes, a single copy 11-HSD1 gene consisting of six exons has recently been isolated from a human library (Tannin et al., 1991); the fourth intron was of indeterminate size, reminiscent of the large configuration proposed for the rat gene in the present study.
Some genes have been shown to effectively turn off protein production by expression of different forms of the message (Laski et al., 1986). If this were the case for the 11-HSD1 gene a reciprocal relationship may be expected between 11-HSDlA expression and that of 11-HSDlB mRNA. However, the coordinate increase in the 11-HSDlA and 11-HSDlB species in the kidney during development suggests that expression of the 11-HSDlB mRNA is not a means of terminating production of the enzyme.
Other evidence also indicates that 11-HSDlB represents a fully functional enzyme. The NADP+ binding domain of the 11-HSDlB protein shows a high degree of similarity to the NH2-terminal region of the nodG dehydrogenase (Debelle and Sharma, 1986) and 17P-hydroxysteroid dehydrogenase , implying that upstream coding sequences are not critical for cofactor binding or enzyme activity. Alternate splicing, directly adjacent to the NADPH binding site, has been previously reported (Lin et al., 1990) in the human adrenodoxin reductase gene and does not appear to influence the activity of the enzyme.
An analysis of the ontogeny of 11-HSD1 gene expression in the kidney showed a tissue specific pattern of mRNA induction. The large increase in renal 11-HSD1 messages during the second week of life in the neonatal rat is in marked contrast to the more gradual increases observed in the liver and lung over the same period (Krozowski et al., 1990). The exponential rise in renal message parallels the developmental surge in 11-HSD enzyme activity (Ghraf et al., 1975), corticosteroid-binding globulin, and glucocorticoids at this time (Henning, 1978). This suggests that the kidney, but neither the liver nor lung, is dependent on adrenal steroids for the increase in 11-HSD1 gene expression during ontogenesis.
The strongly hydrophobic NH2-terminal domain in 11-HSDlA is strongly suggestive of the presence of a signal peptide. Although there is no consensus sequence as such, the 11-HSDlA leader has all the characteristics of a signal peptide including a charged amino terminus, a hydrophobic core and a more polar carboxyl end (von Heijne, 1986). The length of the hydrophobic leader indicates that it may be sufficient to form a membrane spanning domain (Wickner and Lodish, 1985). 1nspect.ion of the sequence shows no obvious signal peptidase cleavage sites, consistent with the finding of an intact signal peptide in the purified microsomal protein (Agarwal et al., 1989). These data indicate that the hydrophobic NH2-terminal domain of 11-HSDlA forms an uncleaved signal sequence which may also function as a membrane binding domain.
The absence of a signal peptide in 11-HSDlB suggests that the truncated enzyme may be located in a different subcellular compartment to the 11-HSDlA protein.
There are a number of examples where alternative forms of the message generate proteins with different intracellular routing (Caras et al., 1987;Gower et dl., 1988;Hynes, 1985). Renal subcellular fractionation studies have shown the presence of 11-HSD1 enzyme activity in both microsomes and nuclei (Kobayashi et al., 1987). Since 11-HSDlA is present in microsomal membranes, it is possible that the 11-HSDlB protein is localized in the nucleus. The nuclear localization of the 11-HSDlB enzyme in the kidney would be consistent with its role as an autocrine protector of the mineralocorticoid receptor.
It is also likely that the distribution of the alternate forms of the enzyme is cell-specific. The cell-specific expression of an immunologically nonreactive 11-HSDlB enzyme may explain the absence of immunoreactivity in the outer regions of the kidney where the Type I receptor is localized (Rundle et al., 1989). The renal papilla, which displayed a similar pattern of mRNA expression to the liver (Krozowski et al., 1990), may predominantly express the 11-HSDlA protein, whereas the cortex and medulla could contain the majority of renal 11-HSDlB enzyme.
The identification of a kidney-specific 11-HSD1 isoform should help elucidate the mechanism by which this enzyme endows specificity on the Type I and Type I1 adrenocorticoid receptors in mineralocorticoid target tissues. Cell-specific expression and organelle-specific targeting of isoenzymes has the potential to endow renal cells with a complex system for modulating the biological actions of both glucocorticoid and mineralocorticoid hormones.