cDNA Cloning and Characterization of a Vitamin D3 Hydroxylase-associated Protein*

We previously reported the generation of monoclonal antibodies which immunoprecipitate a fraction of the total chick kidney 1,25-dihydroxyvitamin D3-24R-hy- droxylase activity. These antibodies were used to screen a chick kidney A g t l l cDNA library resulting in the iso- lation of a full-length cDNA encoding a protein that is not the 1,25-dihydroxyvitamin D3-24R-hydroxylase but another protein we term the vitamin DS hydroxylase-associated protein (VDHAP). The deduced amino acid sequence agreed with an NH2-terminal amino acid sequence from the isolated VDHAP. Gene and protein bank search did not identify homology to known se- quences or functional domains in the VDHAP cDNA. VDHAP mRNA levels are not altered by conditions which either induce 1,25-dihydroxyvitamin Ds-24R-hydroxy- lase activity (78-fold) or 25-hydroxyvitamin D3-la-hy-droxylase activity (30-fold). Northern analysis of poly(A)+ RNA from chick tissues revealed VDHAP only in kidney. Cellular fractionation experiments demon-strated that VDHAP and the 25-hydroxyvitamin D3-la- hydroxylase are colocalized in the inner membrane of mitochondria. The VDHAP antibody immunoprecipitates 14% of the total 1,25-dihydroxyvitamin D3-24R-hy- droxylase activity (7-fold over background) using IVC2F10-Sepharose and IVG8C11-Sepharose and the 1,25-(OH)2D3-24R-hydroxylase activity was determined as de- scribed under "Materials and Methods." Vitamin D-deficient soluhilized chick kidney mitochondria was immunoprecipitated using IVC2F10- Sepharose and IVG8C11-Sepharose and the 25-OH-D3-la-hydroxylase activity was determined as described under "Materials and Methods." Activity present in each fraction is compared with the activity obtained from solubilized chick kidney mitochondria in the presence of Sepha- rose CL-4B.

We previously reported the generation of monoclonal antibodies specific to the chick kidney 24-hydroxylase and l-hydroxylase (13). In this report, we present the isolation of a full-length cDNA using these monoclonal antibodies. The cDNA sequence and the absence of regulation by vitamin D3 suggests that these antibodies are not directed against the chick kidney vitamin D3 hydroxylases. Instead they recognize a novel protein that coimmunoprecipitates with the vitamin D3 hydroxylases. This protein, which we have named vitamin D3 hydroxylase-associated protein (VDHAP), is a kidney-specific protein located in the inner membrane of mitochondria.
Animals-One-day-old white Leghorn cockerels were obtained from Sunnyside, Inc. (Beaver Dam, WI) and maintained for 2 weeks on a normal, vitamin D-deficient, or vitamin D-sufficient diet. Normal chicks were maintained on 1.2% calcium, 0.7% phosphorus diet (18) supplemented with 200 IU vitamin DJkg diet. Vitamin D-deficient chicks were maintained on a 0.08% calcium, 0.7% phosphorus diet as described (18) except CaCO, was eliminated and CaHP04 ratio was decreased to 0.18. Vitamin D-sufficient chicks were maintained on a 3% calcium, 0.7% phosphorus diet (19) supplemented with 400 IU of vitamin D a g diet and injected intramuscularly with 1 pg each of and 1,25-(OH),D,, 48 and 24 h prior to sacrifice.
Northern Blot Analysis-Total RNA was isolated by the acid guanidium thiocyanate-phenol-chloroform extraction method (20). Poly(A)+ RNA was purified by oligo(dT1-cellulose chromatography using either a mRNA spun column kit (5 Prime -3 Prime, Inc., West Chester, PA) or a drip column method (21). RNA blotting to nylon and hybridization of random primed 32P-labeled cDNA probes were performed as described (21). The filter was washed with 1 x SSC (0.15 M sodium chloride, 0.015 M sodium citrate), 0.1% SDS at room temperature, three times for 15 min, and 42 "C for 30 min, and with 0.1 x SSC (15 m M sodium chloride, 65 "C for 1 h. The VDHAP probe was a 1.3-kb VDHAP cDNA fragment 1.5 m M sodium citrate), 0.1% SDS at room temperature for 15 min and (nucleotide 295-1551). The chick @actin probe was a full-length cDNA in pBR322. The chick glyceraldehyde-3-phosphate dehydrogenase probe was a 1.0-kb polymerase chain reaction product (accession no. K01458).
The putative chick kidney 24-hydroxylase probe was a 1.0-kb cDNA fragment. 2 Cloning and Sequencing of Full-length cDNA--Total RNA was isolated from vitamin D-deficient and vitamin D-sufficient chick kidneys and poly(A)+ RNA was purified. A vitamin D-deficient and a vitamin D-sufficient chick kidney h g t l l cDNA library was prepared from the poly(A)+ RNA by oligo(dT) and random priming (CLONTECH Laboratories, Inc., Palo Alto, CA). Plaques were screened with a VDHAP monoclonal antibody (EB2C2) (22). Phage DNA was isolated from positive clones and subcloned into Bluescript I1 K S + plasmid (Stratagene, LaJolla, CA) (21). Afull-length clone was obtained by screening a hgtll library with a 32P-labeled cDNA insert (21) and also by reverse transcription of poly(A)+ RNA followed by polymerase chain reaction of cDNA products using an oligo(dT)17 and specific primer (23). Both strands of the plasmid cDNA inserts were sequenced by the enzymatic dideoxy chain termination method (24).
Antibodies-VIIIB6C3, IVC2F10, and EB2C2 monoclonal antibodies were generated to a partially purified 24-hydroxylase fraction as described (13). Ascites fluids were produced in BALB/c mice as previously described (25). Monoclonal antibodies were purified from ascites fluids on goat anti-mouse IgG-Sepharose (Hyclone, Logan, UT). For the preparation of immobilized monoclonal antibodies, Sepharose CL-4B (Pharmacia LKB, Uppsala, Sweden) was activated with cyanogen bromide by the procedure of Kohn and Wilchek (26) using 20 mg of CNBr/g of Sepharose CL-4B. Monoclonal antibodies IVC2FlO and IVG8C11 (pig intestinal 1,25-(OH)2D3 receptor antibody) (27) were coupled to Sepharose CL4B by overnight incubation a t 4 "C of approximately 0.5 mg of antibody/ml of activated Sepharose.
Fractionation of Kidney Homogenate-Kidney homogenate was fractionated by the method of De Duve et al. (29) with slight modifications. Briefly, kidneys were isolated from 2-week-old normal chicks and homogenized with a Teflon pestle using an overhead stirrer in 3 volumes of 0.25 M sucrose, 0.001 M EDTA (SE buffer). All manipulations were a t 4 "C. The homogenate was centrifuged at 1,000 x g for 10 min. The pellet was rehomogenized in SE buffer and centrifuged at 600 x g for 10 min. The pellet was rehomogenized in SE buffer and designated the nuclear fraction. The supernatant was centrifuged at 12,500 x g for 20 min. The pellet was washed twice in SE buffer and designated the mitochondrial fraction. The supernatant was centrifuged at 50,000 x g for 60 min. The pellet was washed twice in SE buffer and designated the microsomal fraction. The supernatant was designated the cytosolic fraction. Tween 20 was added to each fraction to a final concentration of 0.5% and homogenized with a Polytron (Brinkmann Instruments, Westbury, N Y ) to disrupt membranes. Fractions were stored at 4 "C.
Isolation and Fractionation ofMitochondria-Kidneys were removed from 4-week-old chicks fed a normal diet without vitamin D. Kidneys were hand homogenized with a Teflon pestle (six strokes) in 3 volumes of 0.25 M sucrose, 1 n m EDTA, 3 m~ Tris-HC1, pH 7.4 (SET buffer). All manipulations were a t 4 "C. The homogenate was centrifuged at 450 x R. Ismail and H. F. DeLuca, unpublished results. g for 10 min. The supernatant was centrifuged at 5,500 x g for 20 min. The mitochondrial pellet was washed three times in SET buffer.
Mitochondria were fractionated by the method of Schnaitman and Greenawalt with modifications (30,31). Digitonin (20 mg/ml) was added dropwise to mitochondria to a final concentration of 2 mg of digitonin per 10 mg of protein and stirred for 15 min on ice. The sample was diluted with 1 volume of SET buffer and centrifuged a t 9,000 x g for 10 min. The pellet (inner membrane + matrix) was washed once and resuspended in SET buffer. Lubrol FX (20 mg/ml) was added dropwise to a final concentration of 1 mg of Lubrol per 10 mg of protein and stirred for 15 min on ice. The Lubrol-treated sample was centrifuged at 144,000 x g, 1 h to isolate an inner membrane (pellet) and matrix (supernatant) fraction. The supernatant (outer membrane + intermembrane space) from the low speed centrifugation was centrifuged at 105,000 x g, 1 h to isolate an outer membrane (pellet) and intermembrane space (supernatant) fraction. Fractions were stored at 4 "C.
Electrophoresis and Immunoblotting-Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed according to Laemmli (37). The gels consisted of a 4.75% stacking gel and a 9% separating gel and were run using an SE 250 electrophoresis unit (Hoefer Scientific Instruments, San Francisco). Proteins were transferred electrophoretically to polyvinylidene difluoride (PVDF) membrane (Immobilon Transfer Membranes, Millipore Corp., Bedford, MA). The membrane was blocked with 5% nonfat dry milk, incubated with hybridoma supernatants as primary antibodies followed by alkaline phosphatase-conjugated goat anti-mouse IgG (Promega Corp., Madison, WI). Color development system contained nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate (Promega Corp.) GenerabSolubilized chick kidney mitochondria was isolated as described (13). Protein concentrations were determined using the Bio-Rad protein microassay with bovine serum albumin as standard (38). Fig. 1 shows the nucleotide sequence of the VDHAP cDNA as determined from the clones isolated by antibody and nucleotide screening. The deduced amino acid sequence resulted in an open reading frame of 464 amino acids with a calculated molecular weight of 50,800. The size of VDHAP as determined by gel electrophoresis is 59 kDa. The NH2-terminal amino acid sequence of the immunopurified protein (13) is present in the deduced amino acid sequence except for amino acid residues 5,6, and 13 which differ from both reported NH,-terminal amino acid sequences. These results suggest that the cDNA isolated is the protein which the antibodies recognize and not a cloning artifact.

Isolation of a Full-length VDHAP cDNA-
The VDHAP cDNA sequence was analyzed using the GCG sequence analysis software package (Madison, WI). Comparison to gene and protein sequence banks (EMBL, GenBank, and NBRF) did not identify any homologous DNA or protein sequences. A Profilescan program (39) which uses a database of profiles to find structural motifs in protein sequences did not identify any known structural motifs including the cytochrome P450 heme consensus sequence. These results suggest that VDHAP is not the 1-hydroxylase or the 24-hydroxylase but is a novel protein.
The full-length cDNAs isolated from the vitamin D-sufficient and vitamin D-deficient chick kidney h g t l l libraries were identical with the exception of one nucleotide in the 3'-untranslated region of the clone. This result also suggests that VDHAP is not the 1-hydroxylase or the 24-hydroxylase.
Effect of Dietary Vitamin 0 3 on VDHAP mRNA and Protein--To provide additional evidence that the cDNA isolated The putative amino-terminal processed methionine is assigned -1, and those amino acids in the mature protein are assigned positive numbers. Amino acids determined by protein sequence analysis are underlined. A potential polyadenylation signal is underlined in the 3'-untranslated region.
is not a 24-hydroxylase or 1-hydroxylase cDNA, the effect of dietary vitamin D3 on VDHAP mRNA was examined. Table I shows 24-hydroxylase and 1-hydroxylase activity in solubilized chick kidney mitochondria from vitamin D-sufficient, normal and vitamin D-deficient chicks. This result shows that the 24hydroxylase activity is increased 78-fold in vitamin D-sufficient chicks, and 1-hydroxylase activity is increased 30-fold in vitamin D-deficient chicks with respect to normal chicks. RNA samples isolated from chicks raised on the same diets show that VDHAP mRNA levels remain constant relative to p-actin (Fig. 2, A and C ) while a putative 24-hydroxylase mRNA (Fig.  2 B ) is present only in the vitamin D-sufficient RNA samples (lunes 1 and 4). Solubilized kidney mitochondria from vitamin D-sufficient, normal, and vitamin D-deficient chicks analyzed by immunoblotting using a VDHAF' monoclonal antibody showed no variation between diet groups in levels of VDHAP  Poly(A)+ RNA from chick kidney, bone, brain, crop, heart, large intestine, liver, lungs, muscle, skin, small intestine, and spleen was isolated. A single 2.3-kb band was detected only in kidney (Fig. 3A). Probing the RNA samples with chick B-actin and glyceraldehyde-3-phosphate dehydrogenase showed that RNA samples were intact (Fig. 3, B and C ) . The levels of actin and glyceraldehyde-3-phosphate dehydrogenase show a different variation between tissues indicating that this internal control is not useful for quantitative purposes when comparing tissues. Subcellular Location of VDHAP-Figs. 4 and 5 show the distribution of VDHAP as compared to that of markers in different fractions isolated from homogenates of chick kidney. The subcellular distribution patterns of nuclear (DNA), lysosomal (0-glucuronidase), mitochondrial (succinate cytochrome c reductase), cytosolic (lactic dehydrogenase), and microsomal (NADPH cytochrome c reductase) markers (Fig. 4) correspond to those observed by other investigators (29,401. 0-Glucuronidase is found in both mitochondria and microsomes and clearly distinguishes a lysosomal marker from a mitochondrial or microsomal marker. The majority of the total activity was recovered for each marker: DNA (86.7%), succinate cytochrome c reductase (96.2%), NADPH cytochrome c reductase (89.6%), lactic dehydrogenase (94.2%), and 0-glucuronidase (88.8%). The subcellular distribution of VDHAP (Fig. 5) is similar to that of succinate cytochrome c reductase. The strongest blotting bands are observed in the mitochondrial fraction (lanes 7-9 ).  outer membrane (rotenone-insensitive NADH cytochrome c reductase), intermembrane space (adenylate kinase), inner membrane (succinate cytochrome c reductase), and matrix (glutamate dehydrogenase) markers (Fig. 6) correspond to those observed by other investigators (30,31). The majority of the total activity was recovered for each marker: rotenone-insensitive NADH cytochrome c reductase (82.1%), adenylate kinase (74.8%), succinate cytochrome c reductase (115.8%), and glutamate dehydrogenase (88.5%). Adenylate kinase was distributed between the outer membrane and intermembrane space fraction possibly due to incomplete extraction of soluble proteins in the high speed centrifugation. Adenylate kinase has been reported in the intermembrane space fraction but is also quite labile at low protein concentrations (31). The adenylate kinase in the intermembrane space fraction (0.40 mg/ml) may have lost activity relative to the mitochondrial fraction (9.2 mg/ml). Glutamate dehydrogenase, the matrix marker, was also found in the inner membrane fraction, possibly due to incomplete extraction and separation with Lubrol and centrifugation. The subcellular distribution of l-hydroxylase (Fig. 6) is similar to succinate cytochrome c reductase. 122.3% of the total activity was recovered for the l-hydroxylase. An inner membrane location for the l-hydroxylase, a cytochrome P450 (31, is in agreement with the location of other mitochondrial cytochrome P450s (41,42). Previously, the guinea pig 1-hydroxylase was localized to mitoplasts (inner membrane-matrix fraction) (43). Thesubcellular distribution of VDHAP (Fig. 7) is also similar to succinate cytochrome c reductase. A single band is observed in the inner membrane fraction (lanes 4 and 5) and whole mitochondria (lane 1 ).

Immunoprecipitation of 24-Hydroxylase and 1-Hydroxylase
Activity by VD€€AP Antibody-Previously we showed that 25-30% of the total 24-hydroxylase activity (5-fold over background) was immunoprecipitated by the VDHAP monoclonal antibodies (IVC2H2 and VIIIB6C3). The protein that was blotted by the VDHAP monoclonal antibody was found primarily in the pellet fraction (13). Table I1 shows immunoprecipitation of 24-hydroxylase and 1-hydroxylase activity from vitamin D-sufficient and vitamin D-deficient solubilized chick kidney mitochondria by VDHAP monoclonal antibody (IVC2F10). There is 7-fold more 24-hydroxylase and 2-fold more 1-hydroxylase activity in the pellet for the VDHAP monoclonal antibody (lVC2F10) than for the 1,25-(OH)2D3 receptor monoclonal antibody (lVG8C11). There is also less 24-hydroxylase and less 1-hydroxylase activity in the supernatant for IVC2F10 than for WG8C11. The total activity recovered for lVC2F10 and lVG8C11 for both the 24-hydroxylase and 1-hydroxylase is not significantly different. This result suggests that VDHAP binds 24-hydroxylase from vitamin D-sufficient chicks and the l-hydroxylase from vitamin D-deficient chicks.

DISCUSSION
The 1-hydroxylase is responsible for production of the active form of vitamin D3, 1,25-(OH)2D3 (11). The 24-hydroxylase plays a role in the inactivation of 1,25-(OH)2D3 (44). A molecular mechanism of regulation for these important enzymes of vitamin D3 metabolism is not well understood. In our study of the molecular mechanism of regulation of the 24-hydroxylase and 1-hydroxylase (13), we discovered a novel protein, the vitamin D3 hydroxylase-associated protein.
Our conclusion that the monoclonal antibodies designated anti-24-hydroxylase antibodies (13) are instead anti-VDHAP antibodies is supported by two lines of evidence. First, a cDNA was isolated that does not have homology to cytochrome P450 enzymes or contain a heme consensus sequence. The heme consensus sequence is obtained through the characterization of a large family of cytochrome P450 cDNAs and genes (45,46). Second, the mRNA is not regulated by vitamin D3 status. It has been shown that the rat 24-hydroxylase mRNA is regulated by vitamin D3 status (4) and more specifically by 1,25-(OH)2D3

Immunoprecipitation of 24-hydroxylase and 1-hydroxylase activities by VDHAP antibody
Vitamin D-sufficient solubilized chick kidney mitochondria was immunoprecipitated using IVC2F10-Sepharose and IVG8C11-Sepharose and the 1,25-(OH)2D3-24R-hydroxylase activity was determined as described under "Materials and Methods." Vitamin D-deficient soluhilized chick kidney mitochondria was immunoprecipitated using IVC2F10-Sepharose and IVG8C11-Sepharose and the 25-OH-D3-la-hydroxylase activity was determined as described under "Materials and Methods." Activity present in each fraction is compared with the activity obtained from solubilized chick kidney mitochondria in the presence of Sepharose CL-4B. The amino-terminal sequence and amino acid composition data generated by analysis of the immunopurified protein (13) are not in complete agreement with the deduced amino acid sequence. The immunopurified protein previously designated as 1-hydroxylase, from vitamin D-deficient chicks, and 24-hydroxylase, from vitamin D-sufficient chicks, is thought to be VDI" based on VDHAP antibody recognition. It is possible that different forms of VDHAP exist in vitamin D-deficient and vitamin D-sufficient chicks. Alternatively, amino acid differences may be due to protein sequencing error. The isolation of identical clones from a vitamin D-deficient and vitamin Dsufficient cDNA library supports the hypothesis that these differences are due to protein sequencing error, making it reasonable to suggest that the differences between protein and nucleotide data is also due to protein sequencing error. Cellular fractionation and mitochondrial fractionation experiments clearly indicate that VDHAP is an inner membrane protein of mitochondria. This result was not surprising since the preparation that was used to immunize mice for antibodies was partially purified from solubilized chick kidney mitochondria. However, the deduced amino acid sequence did not reveal the classical amino-terminal mitochondrial signal peptide. Recent studies on protein sorting have shown that a cleavable presequence is not an essential feature of mitochondrial proteins, e.g. apocytochrome c of the intermembrane space, adenine nucleotide translocase of the inner membrane, and isopropylmalate synthase of the matrix (48). It is thought that these proteins contain targeting information within the mature protein. In the case of adenine nucleotide translocase there is evidence for a mitochondrial targeting signal at the COOH terminus (49). The methionine assigned to -1 relative to the amino-terminal sequence of the purified protein is likely removed due to processing by a methionine aminopeptidase (50). There is general agreement that the NH2-terminal methionine is removed from proteins in which the penultimate amino acid has a small aliphatic side chain (51)(52)(53). The presence of threonine as the penultimate amino acid in VDI" suggests a posttranslational amino-terminal processing event for V D W .
VDHAP monoclonal antibody linked to Sepharose CL-4B immunoprecipitates 13.7 2 2.3% of the total 24-hydroxylase ac-tivity which is 7-fold over background. VDHAP monoclonal antibody also immunoprecipitates 20.8 2 2.1% of the total 1-hydroxylase activity which is 2-fold over background. Coimmunoprecipitation of vitamin D3 hydroxylases and VDHAP suggests that a fraction of 24-hydroxylase and 1-hydroxylase are bound to VDHAP. It is not clear why the nonspecific binding for the 1-hydroxylase (11.8 +: 1.6%) is higher then the nonspecific binding for the 24-hydroxylase (2.0 2 2.9%). However, the 24-hydroxylase and 1-hydroxylase experiments differ in protein sample, substrate, and method for measuring product formation. Based on fold immunoprecipitated over background, it would seem that 24-hydroxylase associates more tightly with VDHAP than does the 1-hydroxylase. Different association constants for the 24-hydroxylase and 1-hydroxylase to VDHAP is supported by a n alternative line of experimentation. A strong anion-exchange column (Mono Q ) separates 24-hydroxylase activity into two components in which one peak coelutes with VDHAP (54). This same column did not separate 1-hydroxylase activity into two peaks but instead completely separated l-hydroxylase and V D H A P 3 It is not known why only a fraction of 24-hydroxylase and 1-hydroxylase immunoprecipitates with VDHAP. It is possible that VDHAP is simply less abundant then the hydroxylase proteins in the solubilized preparation. Alternatively, only a fraction of the hydroxylases may be in the correct form to associate with VDHAP or equilibrium binding may be set at 13.7 2 2.3% and 20.8 -c 2.1% under the experimental conditions.
The tissue distribution and location of VDHAP in conjunction with the immunoprecipitation of vitamin D3 hydroxylase activity suggest that VDHAP may have a function specific for the kidney vitamin D3 hydroxylases. VDHAP may carry out its function through a protein-protein interaction with the vitamin D3 hydroxylases since VDHAP and the 1-hydroxylase colocalize to the inner membrane of mitochondria and the vitamin D3 hydroxylases coimmunoprecipitate with VDHAP. The function of VDHAP is most likely not as a structural protein of membranes since VDHAP mRNA is located specifically in kidney. The 1-hydroxylase is located exclusively in kidney whereas the 24-hydroxylase has been detected in kidney, intestine, and cartilage suggesting that VDHAP is specific to the vitamin D3 hydroxylases of the kidney. VDHAP presumably is not required for hydroxylase activity since the vitamin D3 hydroxylases are active in the supernatant after immunoprecipitation of VD-HAP. Collectively, these results point to a vitamin D3 hydroxylase regulatory function for VDHAP possibly through a proteinprotein interaction.
As antibodies become available to study the chick kidney 24and 1-hydroxylase, it may be possible to find evidence for a role for VDHAP in the regulation of the enzymes of vitamin D3 metabolism. antibody preparation and Mary Phelps for synthesis of 25-OH-[la-