Congenital Adrenal Hyperplasia: The Molecular Basis of 21-Hydroxylase Deficiency in H-2^aw18 Mice

Abstract

The mouse strain H-2^aw18 shows typical characteristics of 21-hydroxylase deficiency (21-OHD). A deletion of the active Cyp21a1 gene has been postulated; however, the changes on the nucleotide level are still unknown. To investigate whether this animal model, the only one available, is suitable for studying congenital adrenal hyperplasia in man, a detailed analysis of the Cyp21 locus has been performed to ascertain the genetic cause of 21-OHD in H-2^aw18 mice. We demonstrate that 21-OHD is caused by unequal crossing over between the active Cyp21a1 gene and the pseudogene resulting in a hybrid Cyp21a1-Cyp21a2-p gene including a partial deletion of Cyp21a1. Next to several pseudogene-specific point mutations, various novel missense mutations and a nonsense mutation are present. Enzyme activity for each point mutation has been determined in vitro and the structure-function relationship has been studied by sequence conservation analysis and a three-dimensional murine 21-hydroxylase protein (Cyp21) structure model. The mutations are classified in three classes: I, no or minor decrease in enzyme activity: R238Q, P465L, R361K, A362V, P458L; II, loss of enzyme activity caused by inefficient electron flux: R346H, R400C; III, loss of activity due to deficient substrate binding: I462F, L464F. The combination of in vitro protein expression and three-dimensional structure modeling provides a valuable tool to understand the role of the different mutations and polymorphisms on the resulting enzyme activity. The underlying genetic mechanisms are also known to be responsible for 21-OHD in humans, so rodent 21-OHD turns out to be an excellent genetic model for studying the human disease.

ADRENAL STEROIDS IN humans are synthesized by three interconnected enzymatic pathways, leading to the production of mineralocorticoids, glucocorticoids, and adrenal androgens (1). Congenital defects of each enzyme contributing to adrenal steroid biosynthesis have been described. The most frequent cause of congenital adrenal hyperplasia is 21-hydroxylase deficiency (21-OHD), which accounts for 90–95% of cases (2). 21-OHD leads to insufficient cortisol biosynthesis, including a reduced aldosterone biosynthesis in two thirds of cases. Because of the open hypothalamic-pituitary-adrenal feedback loop, the adrenals produce an excess of hormone precursors, which do not require 21-hydroxylation and which are further metabolized to active androgens, leading to various degrees of virilization of the external genitalia. Steroid human 21-hydroxylase protein (CYP21) is a microsomal reduced nicotinamide adenine dinucleotide phosphate-dependent cytochrome P450 enzyme that facilitates the 21-hydroxylation of 17-hydroxyprogesterone and progesterone (P) to 11-deoxycortisol and 11-deoxycorticosterone (DOC), respectively (3). CYP21 consists of 494 amino acids and has a molecular mass of 52 kDa (4). The gene coding for human CYP21 (CYP21A2) is located in the major histocompatibility complex (MHC) on chromosome 6p21.3. A nonfunctional pseudogene (CYP21A1-P) is located approximately 30 kb away. These two genes are arranged in tandem repeat with the genes encoding the fourth serum complement component C4B and C4A (5). CYP21A2 and CYP21A1-P consist of 10 exons each and show a 98% sequence similarity. CYP21A1-P carries several inactivating mutations. Mutations causing 21-OHD are most frequently the result of complex recombination events between CYP21A2 and CYP21A1-P. Possible mechanisms are unequal crossing over during meiosis, leading to complex deletions of the active CYP21A2 gene and apparent gene conversions transferring inactivating mutations from CYP21A1-P to CYP21A2 (6).

Previous studies in the recombinant mouse strain H-2^aw18 revealed elevated P levels, morphological changes of the adrenal glands typical for 21-OHD, activation of the hypothalamic-pituitary-adrenal axis, and insufficiency of adrenomedullary function (7–9). Therefore, this strain has been introduced as rodent model for 21-OHD. In contrast to humans, the adrenal steroid biosynthesis in rodents is characterized by a lack of 17-hydroxylase activity (10). Pregnenolone is catalyzed by Cyp21 via P and DOC to corticosterone as the major glucocorticoid. Corticosterone is further hydroxylated to aldosterone, which is the main active mineralocorticoid (11). Due to the lack of adrenal 17-hydroxylase activity within the mouse adrenals, steroid precursors do not shunt in the androgen pathway in the same way as in humans. Hence, the H-2^aw18 mice do not present with virilization of the external genitalia. However, homozygous H-2^aw18 mice produce no litters (8, 12). Breeding of the recombinant H-2^aw18 mouse haplotype is consistent with a recessive, lethal gene alteration leading to 21-OHD (12). A deletion of the C4 gene and the active 21-hydroxylase gene Cyp21a1 was reported as the underlying molecular mechanism. The murine Cyp21 gene complex lies within a duplicated portion of the class III region of the MHC on chromosome 17 (13). The murine Cyp21a1 gene has a duplicated pseudogene Cyp21a2-p. Cyp21a1 and Cyp21a2-p are situated 3′ of the genes encoding the murine sex-limited protein variant of the fourth component of complement (Slp) and the gene for the fourth component of complement (C4), respectively. The nucleotide sequence of both murine 21-hydroxylase genes in BALB/c mice has been described in detail (14). It has been demonstrated that all 21-hydroxylase mRNA originates from Cyp21a1, as a 215-bp deletion is present in the Cyp21a2-p pseudogene including the second exon. Moreover, the Cyp21a2-p pseudogene carries several nucleotide changes which result in frame shifts and consecutive premature stop codons (14, 15). The murine Cyp21 protein consists of 487 amino acids and has a 71% homology aligned to human 21-hydroxylase (4).

In this study, we present an extensive genetic analysis of the Cyp21 locus and a consecutive functional characterization elucidating the underlying molecular defect of 21-OHD in the congenic H-2^aw18 mouse strain. A complex gene rearrangement due to unequal crossing over which generates a hybrid gene consisting of a truncated active gene and pseudogene is described. Multiple missense mutations and a nonsense point mutation resulting in a premature stop codon are introduced in the hybrid gene neighboring several pseudogene-specific point mutations. To understand the role of each mutation and the impact on the molecular level, enzyme activity was determined in vitro and functional-structural consequences were studied by three-dimensional molecular modeling techniques.

Materials and Methods

Animals

C57Bl/10SnSlc-H-2^aw18 heterozygous mice, kindly provided by Dr. Greti Aguilera (National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, MD), were maintained according to the institution guidelines with a 12-h light, 12-h dark cycle and free access to food and water. Presence of a vaginal plug on the morning after mating was set as d 0.5 of gestation. All dams were treated sc with 5 μg dexamethasone per day from gestational d 20 until delivery, to prevent instant death of homozygous pups at birth. Litters were collected on d 1 after birth for tissue preparation.

DNA and RNA preparation

Genomic DNA was extracted from livers or tail tips using the wizard genomic DNA purification kit from Promega (Mannheim, Germany). Tissue RNA was extracted with TRIzol reagent (Invitrogen, Karlsruhe, Germany) based on the acid guanidinium thiocyanate-phenol-chloroform extraction method, according to the manufacturer’s recommendations.

Cyp21 and C4 Southern blotting, hybridization, and immunological detection

Genomic DNA was digested quantitatively with BglII or HindIII for Cyp21 or C4 analysis. DNA was separated by agarose gel electrophoresis and Southern blotting was performed as previously described (16). A full-length murine Cyp21a1 cDNA was amplified by RT-PCR from wild-type H-2^b animals using adrenal RNA. First-strand cDNA synthesis was performed from 500 ng total adrenal RNA using SuperScriptIII Reverse Transcriptase and Oligo(dT)_12–18 primer (Invitrogen), according to the manufacturer’s recommendations. Cyp21a1 cDNA synthesis was performed by PCR with HotMaster Taq Polymerase (Eppendorf, Hamburg, Germany) with 1 μg of the synthesized total adrenal cDNA as template with the sense primer Cyp21a1-cDNA-S 5′-TGC TGG GCC ATG CTG CTA CCT-3′ and the antisense primer Cyp21a1-cDNA-AS 5′-CAA GGA CGC TCA CCC TGG TC-3′ under the following conditions: 35 cycles, 1 min at 94 C, 1 min at 57.5 C, 4 min at 65 C. The PCR product was cloned in the pGEM-T Easy vector (Promega), and the clones were sequenced to check the insert, using the ABI dye terminator cycle sequencing technique (PerkinElmer, Boston, MA) with a standard protocol. The samples were electrophorized on an ABI PRISM 310 Sequencer and analyzed using the ABI PRISM SeqScape 1.1 software (Applied Biosystems, Foster City, CA). The plasmid was named pGEM-T-Easy-Cyp21a1-WT. The Cyp21 probe, specific for the Cyp21a1 and Cyp21a2-p genes, was amplified from pGEM-T-Easy-Cyp21a1-WT plasmid DNA, using the PCR digoxigenin (DIG) probe synthesis kit (Boehringer, Mannheim, Germany) with primers and conditions described previously (17). The C4 probe, specific for C4 and Slp genes, was amplified from genomic DNA using the PCR DIG probe synthesis kit (Boehringer) with the upstream primer C4/Slp-S 5′-ACA CAT TCC CTC CTC AAT GC-3′ and the downstream primer C4/Slp-AS 5′-TGC TCA CAG CCT TGT CAC TC-3′. PCR conditions for all labeling reactions were as follows: 30 cycles, 1 min at 94 C, 1 min at 53 C, 1 min at 72 C. Hybridization and immunological detection were performed with the DIG luminescent detection kit (Boehringer), using the manufacturer’s protocol with modifications described in detail elsewhere (16).

MHC class II Eb1 microsatellite analysis

A PCR from genomic DNA covering a highly polymorphic microsatellite containing tandem repeats located at the 3′ end of the second intron in the class II Eb1 gene in mice was established for MHC (H-2) characterization, using cycling conditions as described previously (18). Primers were as follows: Eb1-S 5′-CGA CTG TAG AAC CTT AGC CTG-3′, Eb1-AS-FRAG 5′-Fam-TGG AGC TGT CCT CCT TGT AG-3′, Eb1-AS-Seq 5′-GTG GAC ACA ATT CCT GTT TTC-3′. The primers Eb1-S and Eb1-AS-FRAG were used for fluorescence PCR fragment analysis. The samples were separated on an ABI PRISM 310 Sequencer and analyzed using ABI GeneScan 3.1 software (Applied Biosystems). Size determinations were performed using the GeneScan-350 size standard and the local Southern method, which is integrated in the Gene Scan 3.1 software. Primers Eb1-S and Eb1-AS-Seq were used for a second PCR, followed by sequencing of the PCR product as described above.

Cyp21 genotyping by duplex PCR fragment analysis

A duplex semiquantitative allele-specific PCR using 6-carboxyfluorescein (FAM)-labeled primers was performed for Cyp21 analysis. The allele-specific sense PCR primers were Cyp21a1-FRAG-S 5′-GTG AGG CAC GAT GGC TCC TC-3′ for Cyp21a1 and Cyp21a2-p-FRAG-S 5′-GTC CTT TCC CCC TCC TTC TTT-3′ for Cyp21a2-p. The antisense primer for each upstream primer was Cyp21a1/21a2-p-FRAG-AS 5′-Fam-CAG GTC CAA GTC CAT CTT TC-3′. Reactions were carried out using 500 ng of genomic DNA as template and AmpliTaq polymerase (Applied Biosystems) under the following conditions: 21 cycles, 1 min at 94 C, 1 min at 53 C, 1 min at 72 C. PCR conditions were optimized for DNA content and cycle number to ensure linear multiplication of the PCR product. The samples were separated on an ABI PRISM 310 Sequencer (Applied Biosystems). Size determinations were performed as described above. The area under the curve (auc) of each peak was obtained using the GeneScan software to quantify the relative proportion of gene and pseudogene. The ratio pseudogene peak-auc to gene peak-auc was calculated for genotype determination.

Cyp21 nucleotide DNA sequence analysis

To amplify the active Cyp21a1 gene or the inactive Cyp21a2-p pseudogene exclusively, a quantitative restriction of genomic DNA with BglII or NdeI was performed before PCR. PCR contained the primers Cyp21-S 5′-AAG GCT GAT GGG GAC TGT G-3′ and Cyp21-AS 5′-ACA CTC ATG CCA ACC TTG CT-3′, 1 μg of digested genomic DNA, rtTH polymerase (Applied Biosystems) and additives following the manufacturer’s recommendations. Cycling conditions were: 14 cycles, 1 min at 94 C, 1 min at 53 C, 6 min at 72 C followed by 18 cycles, 1 min at 94 C, 1 min at 53 C, 6 min at 72 C with a time increment of 20 sec with each cycle. Undesirable coamplification of either the pseudogene or gene was checked by digesting the PCR product with NdeI or BglII, respectively. Sequencing of the PCR products and subsequent analysis were performed as described above. Primer sequences for sequence analysis of the entire intron and exon nucleotide sequence of the murine Cyp21a1 and Cyp21a2-p genes are available on request. The nucleotide sequence was verified in 6 different homozygotes from different progenitors. The published nucleotide sequence of murine Cyp21a1 (GenBank accession no. M15009) was used as template (14). Restriction analysis of PCR fragments of exons 6, 8, and 10 of the hybrid Cyp21a1 gene, including the respective mutations, was performed where possible with TauI, HpyCH4V, DdeI, NheI, Sau3AI, AluI, CviTI, Tsp509I and BfaI to verify the mutant alleles (data not shown).

Cyp21a1 mRNA analysis

Twenty micrograms of adrenal RNA pooled from five animals with the same genotype were separated by electrophoresis on a 1% agarose, 4-morpholinepropanesulfonic acid (Sigma, Munich, Germany), 0.6 m formaldehyde (37%) gel and transferred in 20× sodium chloride/sodium citrate buffer for 4 h onto a positively charged Hybond N+ Nylon membrane (Amersham, Uppsala, Sweden), using the Turboblotter Rapid Downward Transfer System (Schleicher & Schuell, Dassel, Germany). Northern blots were hybridized overnight at 65 C after RNA cross-linking with UV light with the DIG-labeled Cyp21 cDNA probe used in Southern blot hydridization, as described above. The immunological detection of the DIG-labeled probe was performed with the DIG luminescent detection kit (Boehringer), using the manufacturer’s protocol with the modifications described in detail elsewhere (16). Full-length Cyp21(H-2^aw18) cDNA was amplified by RT-PCR, using pooled adrenal RNA from five homozygous H-2^aw18 animals. First-strand cDNA was synthesized from total adrenal RNA and the Cyp21(H-2^aw18) cDNA PCR was performed using the primers and under the conditions described above. Sequencing of the PCR product and subsequent analysis were performed as described above. Primer sequences used for sequence analysis of the entire Cyp21(H-2^aw18) cDNA nucleotide sequence are available on request.

Construction of plasmids and site-directed mutagenesis

An amino-terminal FLAG tag (italic in primer sequence) was attached to the Cyp21a1 cDNA by PCR for Western blot detection, using the pGEM-T-Easy-Cyp21a1-WT plasmid as template with the sense Cyp21a1-FLAG-S primer 5′-GGC CAT GGA TTA TAA AGA TGA TGA TGA TAA ACT GCT ACC TGG GCT GCT GCT GCT GTT G-3′ including a consensus Kozak sequence (bold in primer sequence) and the antisense primer Cyp21a1-cDNA-AS. The Expand long-template PCR system (Boehringer) was used for PCR amplification under the following conditions: 30 cycles, 30 sec at 94 C, 30 sec at 53 C, and 3 min at 68 C. The PCR product was cloned in the pGEM-T Easy vector (Promega), and the integrity of the construct was checked by sequencing. The plasmid was named pGEM-T-Easy-Cyp21a1-Flag-WT. Mutagenesis was performed from the murine full-length pGEM-T-Easy-Cyp21a1-Flag-WT cDNA, using the QuikChange XL Site-Directed Mutagenesis Kit (Stratagene, Amsterdam, The Netherlands). Primers for mutagenesis are available on request. The introduction of the mutations and clones’ integrity were checked for by sequencing the entire construct. The pGEM-T-Easy-Cyp21a1-WT, the pGEM-T-Easy-Cyp21a1-Flag-WT and the mutagenized pGEM-T-Easy-Cyp21a1-Flag constructs were subsequently subcloned into the NotI site of the pcDNA3 expression vector (Invitrogen), resulting in the pcDNA3-Cyp21a1-WT and pcDNA3-Cyp21a1-Flag-WT constructs. The mutagenized pcDNA3-Cyp21a1-Flag plasmids were named according to the designed mutations. The plasmids were checked for the correct orientation of the insert by sequencing.

In vitro expression and assays of enzyme activity

Approximately 1 × 10⁶ COS-7 cells were transiently transfected, using lipofectamine (Invitrogen) with 1 μg of pcDNA3-cyp21-Flag construct and 1 μg β-galactosidase pSV-Gal vector (Promega). Posttransfection treatment was performed following a standard protocol (Invitrogen) using DMEM supplemented with glutamine, antibiotics, and 10% fetal calf serum (all products for cell cultures from Invitrogen). The determination of 21-hydroxylase activity in intact COS-7 cells was performed 48 h after transfection. The cells were incubated for 90 min at 37 C with 1000 μl DMEM medium containing 0.4 μCi ³H-labeled P (Amersham) and 1 μmol/liter unlabeled P as substrate (Sigma) and 8 mmol/liter reduced nicotinamide adenine dinucleotide phosphate (Sigma). Steroids were extracted from the culture medium with isooctane and ethylacetate (1:1, vol/vol), evaporated to dryness and dissolved in ethanol. The steroids were separated by thin layer chromatography (TLC) with chloroform and acetone (70:30, vol/vol). The radioactivity was measured directly from the TLC plates using the Rita Star TLC-Scanner (Raytest, Straubenhardt, Germany) and analyzed with the Rita TLC analysis version 1.97 software (Raytest). The cells were trypsinized and lysed in reporter lysis buffer (Promega), followed by the measurement of β-galactosidase activity according to the manufacturer’s protocol (Promega) and correction for protein concentration. To determine apparent kinetic constants, intact COS-7 cells were incubated for 90 min with 0.5, 1.0, 2.0, 4.0, or 6.0 μmol/liter unlabeled steroid as described above. Postincubation treatment and analysis were performed as described above. All experiments were performed 6-fold for the WT plasmids and in triplicate for the mutants. The 21-hydroxylase activity was expressed as a percentage of substrate conversion in picomoles per milligram per minute, taking the activity of cells expressing the Cyp21 wild-type protein as 100% percent after correction for total protein and the activity of cells transfected with the empty pcDNA3 plasmid. The ratio of β-galactosidase activity content was measured to ensure the reproducibility of the transformation efficiency. The apparent kinetic constants were calculated from the measurements of 21-hydroxylase activity in intact COS-7 cells at different substrate concentrations. Calculation of enzymatic activity and kinetic constants was performed with the Graph Pad Prism software version 4.0 (GraphPad Software Inc., San Diego, CA). Western blot analysis was performed in a standard protocol using an anti-Flag M2 monoclonal mouse antibody (Sigma) to ensure the expression and translation of the intact Cyp21 wild-type and mutant proteins.

Conservation analysis

To analyze the influence of the mutations, a conservation analysis of 30 sequences from the CYP21A and 17A family was performed. The sequence identity to the murine Cyp21 was between 31–90%. A sequence alignment was calculated using the ClustalX program with default parameters (19). Conservation indices were calculated using the program al2co and the unweighted, not normalized sum of pairs method with the BLOSSUM62 scoring matrix (20). The conservation score depends on the number and kind of residues occurring at the current position. A higher score reveals a higher degree of conservation.

Computer-assisted modeling

A model of the murine Cyp21 protein (GenBank accession no. NM_009995; Ref. 14) was built using three-dimensional (3D) structures of four homologous mammalian P450s. The structures of P450 2C5 from Oryctolagus cuniculus (Protein Data Bank (PDB) code 1N6B) (21), P450 2B4 from O. cuniculus (PDB code 1PO5) (22), P450 2C9 from Homo sapiens (PDB code 1OG5) (23), and P450 2C8 from H. sapiens (PDB code 1PQ2) (24) were selected from the Protein Data Bank (25). The sequence identity between the target sequence and the template sequences was 29% (PDB code 1PO5), 29% (PDB code 1PQ2), 28% (PDB code 1OG5), and 31% (PDB code 1N6B). A multiple sequence alignment was calculated using the Tcoffee web server (26) and, after manual adjustment, it was used for homology modeling with the MODELLER v6.2 program (27). The heme was transferred from its homologous CYP 2C5 (PDB code 1N6B). The PROSA2003 program was used for validation of the model, by analyzing residue interaction energy and the PROSA Z-score (28).

Results

Cyp21 and C4 Southern blotting, hybridization, and immunological detection

As previously shown (12), BglII-digested DNA from the wild-type mice H-2^b showed an 8.8- and 9.8-kb fragment when hybridized with the Cyp21 probe. Heterozygous H-2^b/H-2^aw18 mice exhibited three bands which included an additional 10.5-kb fragment. Homozygous H-2^aw18 mice showed only the 10.5-kb fragment when hybridized with the Cyp21 probe. HindIII-digested DNA from the wild-type mice H-2^b and heterozygous H-2^b/H-2^aw18 showed a 23.0- and 4.0-kb fragment when hybridized with the C4 probe. Homozygous H-2^aw18 mice showed only the Slp-specific 23.0-kb fragment when hybridized with the C4 probe (data not shown).

MHC class II Eb1 microsatellite analysis

The Eb1 microsatellite of the parental B10/H-2^b strain was identified as published before (GenBank accession no. M14231) (29). The H-2^b haplotype corresponds to a single 108-p fragment in the fluorescence PCR fragment analysis. A 108- and 128-bp fragment were detected in H-2^b/H-2^aw18 heterozygotes. H-2^aw18 homozygous mice showed a single 128-bp fragment (Fig. 1). A novel H-2^aw18-specific MHC class II Eb1 microsatellite was sequenced, revealing a repeat number of 14 for the TGGA motif and a repeat number of 4 for the GGCA motif (GenBank accession no. AY611735) (Table 1). The results of MHC II Eb1 microsatellite analysis revealed a strict linkage of the microsatellite marker to the Cyp21 genotype.

Cyp21 genotyping by PCR fragment analysis

A duplex PCR fragment analysis with FAM-labeled primers was established as a rapid and comprehensive method to characterize the Cyp21 genotype. A single 262-bp PCR fragment was amplified from the Cyp21a2-p pseudogene and a single 275-bp PCR fragment was amplified from the active Cyp21a1 gene in one PCR. In the case of a homozygous H-2^b wild-type or a heterozygous H-2^b/H-2^aw18 state, both fragments were detected (Fig. 2). For homozygotes, a gene to pseudogene peak auc ratio of 1.3 ± 0.2 (mean ± 2 sd) was calculated, for the heterozygotes a gene to pseudogene peak auc ratio of 2.4 ± 0.2 (mean ± 2 sd) was measured (n = 100, P < 0.001). All animals could be unequivocally typed as heterozygous or homozygous by either Southern blotting, Eb1 microsatellite, or Cyp21-PCR fragment analysis, as shown in over 100 animals. Interestingly, only the 275-bp peak (predicted functional gene) was detected in the homozygous H-2^aw18 mutant. Because this finding used to be an equivocal explanation for the 21-OHD phenotype of the H-2^aw18 mice, sequencing of Cyp21a1 and Cyp21a2-p was performed.

Cyp21 DNA sequence analysis

No pseudogene-specific PCR product could be amplified after quantitative restriction of genomic DNA with NdeI from homozygous H-2^aw18 mice. This was in accordance with the results from the semiquantitative duplex PCR fragment analysis and the Southern blot analysis, in which either no pseudogene-specific amplicon or only a single Cyp21-specific band were detected in homozygous H-2^aw18 mice. A 3.6-kb Cyp21a1-specific amplicon was detected after PCR from genomic DNA and quantitative BglII digestion. The PCR product was also digested with NdeI to rule out a coamplification of the pseudogene resulting in two fragments (0.5 and 3.1 kb) (data not shown). After the control digestion, Cyp21a1 products were sequenced in 14 overlapping sequencing reactions covering the entire gene. The nucleotide sequence of the Cyp21a1 gene in H-2^aw18 homozygotes is published in GenBank (accession no. AY613781). The pseudogene-specific deletions of exon 2 and the first part of intron 2 are missing. Various point mutations are introduced into the gene. In H-2^aw18 homozygous animals, the following mutations leading to single amino acid changes were detected by direct sequencing and restriction analysis (nucleotide numbering from DNA denoting the first nucleotide of the start codon ATG as nucleotide +1 (30)): exon 6: 1369G>A (R238Q); exon 8: 2047C>A (R346H), 2092G>A (R361K), 2095C>T (A362V); exon 10: 2360C>T (R400C), 2535C>T (P458L), 2546A>T (I462F), 2552C>T (L464F), 2556C>T (P465L), 2570C>T (Q470X) (Fig. 3). The mutations were detected in heterozygotic state in H-2^b/H-2^aw18 animals and were absent in H-2^b/H-2^b mice. The putative protein sequence of the parental strain and of the 21-hydroxylase-deficient H-2^aw18 strain is depicted in Fig 4. Several single nucleotide changes were detected in the intronic sequence. The 3′ untranslated region resembles the pseudogene sequence (data not shown).

Cyp21 mRNA analysis

We examined the transcription of Cyp21a1 mRNA in adrenal glands of wild-type (H-2^b/H-2^b), heterozygous (H-2^b/H-2^aw18), and homozygous (H-2^aw18/H-2^aw18) animals by Northern blotting using the DIG-labeled Cyp21-cDNA probe. Northern blot analysis revealed a single band of about 2.0 kb for each genotype (Fig. 5). Cyp21 mRNA was detectable in all animals. A 1.5-kb amplicon was detected after PCR with cDNA reverse transcribed from adrenal RNA as template representing the Cyp21(H-2^aw18) gene transcript. The Cyp21(H-2^aw18) cDNA PCR product was sequenced in 11 overlapping sequencing reactions from both directions. The identical point mutations (c713G>A, c1037C>A, c1082G>A, c1085C>T, c1198C>T, c1373C>T, c1384A>T, c1390C>T, c1394C>T, c1408C>T) corresponding to the genomic sequence were detected in the cDNA sequence.

In vitro expression and assays of enzyme activity

Initially, we examined the 21-hydroxylase activity of proteins derived from the pcDNA-Cyp21a1-WT and pcDNA-Cyp21a1-FLAG-WT plasmid. Both proteins showed a similar 21-hydroxylase activity (data not shown). The R238Q and P465L nonsynonymous changes can be considered as polymorphisms, because the conversion of P to DOC is of the same magnitude as in the wild-type (98 ± 8% and 93 ± 8%, respectively) (Fig. 6). The apparent Michaelis-Menten constant (K_M) value was significantly higher for R238Q, indicating a lower substrate affinity (P < 0.05) (Fig. 7 and Table 2). The apparent K_M of P465L was similar. The mutations R361K, A362V, P458L, and L464F show residual 21-hydroxylase activity of considerably different degrees. The 21-hydyroxylase activity was 64 ± 18% for R361K, 34 ± 3% for A362V, 63 ± 9% for P458L, and 6 ± 0% for L464F (A362V, P458L, and L464F activities significantly different from wild-type, P < 0.05). The maximum reaction velocity (V_MAX) was significantly higher in A362V compared with wild-type (P < 0.05). R361K and P458L showed a V_MAX value similar to the wild-type. V_MAX of L464F was significantly lower compared with wild-type (P < 0.05). Although the apparent K_M was comparable to the wild-type for A362V and P458L, it was significantly higher in R361K and L464F (P < 0.05). The mutations R346H, R400C, I462F, and Q470X resulted in minor or absent 21-hydroxylase activity (2 ± 1%, 0 ± 0%, 2 ± 1%, 0 ± 0%) and did not achieve saturation under the experimental conditions. Western blot analysis of the wild-type and mutant proteins expressed in vitro demonstrated that all mutations did not affect the translation efficiency (data not shown).

Conservation analysis

The conservation analysis of 30 homologous P450, showed a broad range of conservation: R361 (0.0), P458 (0.43), P465 (0.67), R238 (1.81), A362 (2.3), L464 (2.4), Q470 (2.5), I462 (3.1), R346 (4.32), R400 (4.5). The conservation ranged from eight of 30 sequences (R361) to 29 of 30 sequences (R400).

Structure modeling

The sequence alignment of Cyp21 with the four mammalian P450 of known structure shows that functionally relevant residues of P450 are conserved in the alignment (Fig. 8). PROSA quality check shows that the model of the murine Cyp21 is of good quality; each residue interaction energy with the residual of the protein is negative (Fig. 9) and similar to the energy of the template structures (data not shown). The PROSA Z-score of the model is also similar to that of the templates (data not shown).

Structural position of the mutations

The mutated residues are located in the substrate binding site, in the putative redox partner binding site, or in other unspecified regions of the protein (Fig. 10). Combining the conservation analysis, structural position analysis, and mutagenesis experiments enabled us to classify the mutations into three classes: class I includes putative polymorphisms and mutations with minor decrease of enzyme activity; classes II and III contain mutations which cause a severe loss in activity, whereas class II mutants decrease V_MAX, and class III mutants increase K_M. Q470X is a special mutation, because it introduces a premature stop at a conserved (2.5) position. Therefore, the total loss of activity is caused by the lack of 17 residues of the C-terminal part of the protein.

Class I includes five mutations. The variations R238Q and P465L have almost no influence on the enzyme activity. Residue R238 is located at the C-terminal end of the helix G on the surface of the protein with no contact to functionally vital regions of the protein. Residue P465 is located on the C-terminal side of substrate recognition site 6 (SRS6) (31). The three mutations R361K, A362V, and P458L show a residual enzyme activity of 34–64% (Fig. 6). R361 and A362 are both located in the β-sheet domain on the protein surface with no contact to catalytically relevant residues. P458 is located in SRS6 but points away from the heme, and therefore is not part of the binding site of Cyp21. All five residues show a low conservation score ranging from 0.0 to 2.3, which is consistent with a weak influence on activity. Class II consists of the two mutations R346H and R400C, which show very low activity. R346 is part of the highly conserved E-X-X-R motif in helix K. R400 is located in the meander loop, which is located on the proximal side of the heme, the putative redox partner binding site. Both are part of the ERR triad motif of the 21-hydroxylase protein. The possible role of R346 and R400 in redox partner interaction and electron transfer is consistent with the loss of activity for these mutations based on a decrease of the V_MAX value, indicating a less efficient electron transfer to the heme. Class III includes the two mutations I462F and L464F both showing very low activity. They are located in SRS6. Thus a mutation at this site interferes with substrate binding. This is in accordance with the loss of activity in these mutations most probably being based on a reduction in substrate binding. All class II and III mutations show high scores in the conservation analysis with a conservation index ranging from 2.4–4.5. This is consistent with their position in functionally important regions and a severe decrease in 21-hydroxylase activity.

Discussion

Steroid 21-hydroxylase plays a key role in adrenal steroidogenesis. Insufficient activity of this enzyme caused by various gene alterations is responsible for one of the most frequent fatal inborn errors of metabolism in humans (1).

MHC linkage analysis

The human and murine 21-hydroxylase gene locus show high homology. The class III region of the MHC, two genes encoding components of the complement activation pathway and two structurally homologous genes for 21-hydroxylase are located in the same region in both species (14, 32). Linkage disequilibrium between certain HLA types and congenital adrenal hyperplasia is a well known phenomenon in humans (33, 34). Murine 21-OHD is also linked to a distinct recombinant H-2 haplotype named H-2^aw18 (12). We have identified an informative microsatellite within the murine class II Eb1 gene, allowing easy characterization of the H-2^aw18 haplotype. The microsatellite consists of two tandem repeat alleles with a simple structure of repeats, (TGGA)₁₄ (GGCA)₄, which is typical for wild-derived haplotypes (18). The number of tandem repeats varies in different MHC haplotypes because of unequal meiotic exchanges or slipped-strand mispairing during DNA replication, creating additional alleles (35). The identical mechanism is responsible for the generation of the most frequent inactivating mutations within the 21-hydroxylase gene locus in humans (36) and in mice (12).

Cyp21 genotyping

It was reported that H-2^aw18 mice carry a deletion of the active Cyp21a1 gene (12). In the present study, we were able to redefine this as unequal crossing over within the 21-hydroxylase locus in these animals. Nucleotide sequence analysis revealed a mutant Cyp21a2-p-Cyp21a1 hybrid gene. The typical characteristics of the Cyp21a2-p gene, such as the absence of exon 2 and parts of intron 2, were detectable in the H-2^aw18Cyp21 gene (14). The pseudogene-specific PCR fragment was absent in the homozygous state and the gene to pseudogene ratio was altered in the heterozygous state using our novel semiquantitative PCR strategy. The 5′ part of the pseudogene is deleted in mice with 21-OHD. Taking the intronic sequence and pseudogene-specific point mutations together, we speculate that an apparent conversion between the 5′ part of Cyp21a1 and the 3′ part of Cyp21a2-p has occurred, with a putative recombination within exon 8. An approximately 80-kb stretch of DNA that encompasses the 3′ part of Cyp21a1, C4 and the 5′ part of Cyp21a2-p is deleted, as shown by Southern blotting and direct sequencing (Fig. 3). The mRNA transcript of the fused gene is detectable in the adrenals. Therefore, we conclude that the mutant gene is under the control of the wild-type Cyp21a1 gene promoter. This intergenic recombination perfectly resembles one group of mutations in human 21-OHD. In humans, approximately 20% of CYP21A2 mutations are partial gene deletions due to unequal crossing over during meiosis, whereas the remainder are transfers of deleterious mutations normally present in the pseudogene to the active gene (1). It is remarkable that no intronic splice site mutations, often detected in the human 21-OHD, were found in the mutant murine Cyp21 gene.

Structure modeling

Until now no 3D crystal structure of mammalian Cyp21 enzymes is available. For the human CYP21 protein two models have been reported, based on the structure of bacterial P450cam (37) or bacterial P450 BM-3 (38). However, models of mammalian P450 based on bacterial P450 are of limited use as x-ray structures of mammalian P450s (21–24) revealed significant structural differences. For this reason a 3D model of the murine P450 was built in the present study using structures of mammalian P450. This model was than used to investigate the role of individual mutations on enzyme activity. The fact that residues which are generally expected to be highly conserved in cytochrome P450 enzymes can be found at corresponding positions in the Cyp21 model proves that the alignment, the most crucial step in homology building, was reliable throughout the sequence. As an additional approach to understand the influence of mutations on enzyme activity, conservation analysis of amino acids in multiple sequence alignments has been performed. Proteins within a superfamily usually share the same fold and possess related functions (39), and conservation of structure and function is therefore reflected in the conservation patterns derived from multisequence alignments (40, 41).

Class I mutations

Because glutamine at position 238 is found in Cyp21 of BALB/c mice it is very likely to be a silent polymorphism (14). Although this residue is located in a part of helix G forming the substrate binding pocket, the exchange R238Q seems not to affect the enzyme architecture as the in vitro function of both proteins was similar (42). Both, the pseudogene-specific amino acid exchange R361K and the novel A362V substitution have residual enzyme activity. According to our 3D model, R361 and A362 are located within the β-sheet domain. The β-sheet domain is located on the protein surface in no contact to catalytically important residues (42, 43). The localization of the residues in combination with the minor amino acid changes (from the basic amino acid arginine to the basic amino acid lysine at position 361 and from the short side chained hydrophobic residue alanine to the short side chained hydrophobic valine at position 362) may contribute to the residual activity of the mutant proteins. The single nucleotide changes resulting in P458L and P465L were also found in the pseudogene Cyp21a2-p (14). These two point mutations are both located within the β-turn, including the SRS6, which is involved in substrate specificity (44). In the three dimensional model neither mutation is in contact to SRS6 and, therefore, both display 21-hydroxylase activity. No corresponding variations are described in the human CYP21A2 gene.

Class II mutations

Both the mutations R346H and R400C inactivate protein function because of insufficient electron transfer to the heme. The first alteration arises from the pseudogene, the second mutation is a de novo mutation. Both are located in the ERR triad motif of the 21-hydroxylase protein. The murine R346 residue corresponds to the human R354 residue (45). Two inactivating substitutions consistent with the inactivity of the murine R346H mutation are described in humans at residue R354 (46, 47). The murine amino acid R400 corresponds to the residue R408 in the human Cyp21 protein (45). A substitution of arginine by cysteine (R408C) is described in a heterozygous form in two siblings with the salt-wasting form of 21-OHD (48). The authors deduce from the combination of R408C with an inactivating splice site mutation that R408C leads to a severe impairment of enzyme activity. This finding is concordant with the lack of activity in the murine R400C mutant. Using our novel model, we conclude that R346 and R400 are part of the ERR motif (31). R346 is part of the highly conserved E-X-X-R motif, which forms together with residues from the meander loop a group of charged residues at the bottom of the protein which were identified to form a hydrogen-bond network to stabilize the P450 structure (31). The E-X-X-R motif is conserved in all P450 enzymes. All mutations of one of the ERR motif amino acids described so far in different P450 enzymes lead to complete inactivity (49–53). This is in perfect accordance with our results with regard to the R346H and R400C residues.

Class III mutations

The mutations I462F, L464F, and Q470X appear to be de novo mutations which are not detectable in the pseudogene. These additional mutations are also located in the β-turn region disrupting SRS6. No corresponding variations are described in the human CYP21A2 gene. The high conservation indices of the residues I462, L464, and Q470 underline the importance of SRS6 for the enzyme activity. Variations at the conserved residues I462 and L464 located in the SRS6 seem to inactivate SRS6 completely, whereas changes in less conserved regions, such as P465 in the same region, appear as polymorphisms. The premature stop at the conserved position Q470 within the β-turn motif inactivates the enzyme completely. This is in accordance with various carboxy-terminal stop mutations in the human CYP21 enzyme (54, 55).

Conclusion

In conclusion, we have redefined the mechanisms causing 21-OHD in the H-2^aw18 strain on the nucleotide level. Interestingly, the genomic alteration perfectly resembles one form of the CYP21A2 gene defects reported in humans (1). To understand the role of individual mutations, combining bioinformatics with site-directed mutagenesis is a valuable tool. Our homology model of the murine Cyp21 provides a good basis for further studies of the effect of polymorphisms on the activity of the enzyme. Exact knowledge of the underlying genetic defect specific to the mouse model is of key importance for elucidating the pathophysiological consequences of congenital adrenal hyperplasia in this mammalian model. Further research into the murine 21-OHD model will certainly contribute to a better understanding of the human disease.

Acknowledgments

We are grateful to Dr. G. Aguilera (Section of Endocrine Physiology, Developmental Endocrinology Branch, National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, MD) for providing the mutant H-2^aw18 mice. We appreciate the expert technical assistance of Gisela Hohmann and Brigitte Andresen and thank Joanna Voerste for linguistic help with the manuscript.

This work was supported in part by an institutional grant from the Christian-Albrechts Universität zu Kiel. S.T. acknowledges support by the German Federal Ministry of Education and Research (Project 0313080A).

This work was presented in part at the 42nd Annual Meeting of the European Society for Paediatric Endocrinology, Ljubljana, Slovenia, 2003.

Abbreviations

 
• 3D

  Three dimensional

 
• 21-OHD

  21-hydroxylase deficiency

 
• auc

  area under the curve

 
• CYP21

  human 21-hydroxylase protein

 
• Cyp21

  murine 21-hydroxylase protein

 
• DIG

  digoxygenin

 
• DOC

  11-deoxycorticosterone

 
• FAM

  6-carboxyfluorescein

 
• MHC

  major histocompatibility complex

 
• P

  progesterone

 
• SRS6

  substrate recognition site 6

 
• TLC

  thin layer chromatography

 
• V_MAX

  maximum reaction velocity

References