(Received for publication, November 15, 1996, and in revised form, January 9, 1997)
From the Biotechnology Centre of Oslo and The Animal
Department, The National Hospital, University of Oslo, P.O. Box
1125, Blindern, N-0317 Oslo, Norway, ¶ Norsk Hydro, Research
Center, 3901 Porsgrunn, Norway, and the Medical Department,
The National Hospital of Norway, N-0027 Oslo, Norway
The gene of a novel chymotrypsin-like serine protease has been cloned from human pancreas. The chymotrypsin-like enzyme-1 gene is located on chromosome 16q22.1 in a tight cluster with four unrelated genes. The gene has seven exons with the signal and activation peptide and the three main catalytic residues forming the active site encoded by separate exons. Northern blots of pancreatic mRNA showed a major transcript of 1.0 kilobases and a minor transcript of 1.3 kilobases due to alternative polyadenylation. No transcript was found in other tissues. Its presence in pancreatic tissue, duodenal juice, and urine was demonstrated with antisera raised against synthetic peptides from the derived amino acid sequence of the gene. The peptide sequences were chosen for being most dissimilar to chymotrypsin, and the antisera obtained did not react with purified human chymotrypsin. The proteolytically active CTRL-1 has been identified in pancreatic homogenate, duodenal juice, and urine, and a recombinant CTRL-1 has been characterized. Increased pancreatic secretion of CTRL-1 was induced by protease inhibitors indicating that the enzyme is secreted from pancreas upon feedback stimulation. Both native and recombinant CTRL-1 displayed chymotrypsin- and elastase-2-like activities and hydrolyzed the amide bonds of substrates having tyrosine, phenylalanine, or leucine residues at the P1 position. The enzyme was active over a broad pH range (6.5-9.0), with a maximum at pH 8.0-8.5. CTRL-1 was produced as a zymogen of 264 amino acids as deduced from the gene sequence, with a sequence identity of 54% with human chymotrypsin B. The number and location of intron/exon junctions as well as the sequence similarity to chymotrypsin both at the DNA and protein level and the presence in duodenal juice indicate that this is a novel digestive enzyme of the chymotrypsin superfamily, albeit one with distinct physiological and biochemical features.
Serine proteases are enzymes utilizing an activated serine residue in their substrate binding site to catalyze the hydrolysis of certain peptide bonds. These proteases are encoded by related genes that evolved from a common ancestral protease through gene duplication (1, 2). The pancreatic subfamily of proteases comprises trypsins, chymotrypsins, kallikrein, and elastases. Their structural genes encode related enzymes of similar size, structure, and function but with different peptide bond specificities. The catalytic activity of chymotrypsin (EC 3.4.21.1) is directed at peptide bonds involving aromatic, and to a lesser degree, aliphatic amino acids (3, 4). Chymotrypsin is synthesized by pancreatic acinar cells as an inactive precursor (chymotrypsinogen) secreted to the duodenum and activated by tryptic cleavage. Two or three forms of chymotrypsin (5-7) as well as one major and one minor form of chymotrypsinogen have been reported in humans (8). Chymotrypsins are the most abundant among the pancreatic proteases and may represent 10-20% of the total protein synthesized by the exocrine pancreas (3, 4).
We describe here a novel human pancreatic serine protease (chromosome 16q22.1) with a 54% sequence identity to human chymotrypsin B (chromosome 16q22.3). A comparison with human elastases gave 45% identical residues between the new enzyme and elastase 2A. The enzyme was provisionally named chymotrypsin-like enzyme-1 (CTRL-1).1 Its expression was enhanced by duodenal instillation of protease inhibitors, and it may consequently be part of a pancreatic protease backup system.
Kunitz trypsin inhibitor, Bowman-Birk inhibitor,
and diaminobenzidine were obtained from Sigma;
proteinase K was from Boehringer Mannheim; camostate (FOY-305) was a
gift from Dr. H. Babst, Sanol-Schwarz, Monheim, Germany; Dynabeads
oligo(dT) and anti-rabbit IgG were from Dynal, Inc.; Genescreen
membranes were from DuPont NEN; restriction enzymes were from New
England Biolabs Inc.; HL1163b gt11 library and 5
-RACE kit were from
Clontech; Magic Miniprep was from Promega; Bluescript vectors were from
Stratagene; T7 polymerase, precast 8-18 and 10-15%
SDS-polyacrylamide gels, Sephadex G100 HiLoad 16/60, Superdex prep
grade 75, CNBr-activated Sepharose, and the Sephasil C18 SuperPac
Cartridge Pep-S column were from Pharmacia Biotech Inc.; Microsep 10 K
was from Filtron Scandinavia; horseradish peroxidase-labeled goat
anti-rabbit sera were from Southern Biotechnology; human chymotrypsin,
human trypsin, anti-human chymotrypsin, and anti-human trypsin were
from Calbiochem; low molecular weight SDS-PAGE standards were from
Bio-Rad; chromogenic substrates
-OMe-succinyl-Arg-Pro-Tyr-pNA (S-2586),
benzoyl-Arg-Val-Trp-pNA (S-2532),
benzoyl-Lys-Val-Tyr-pNA (S-2504),
D-Ser-Leu-Met-pNA (S-2327), and
D-Lys-Val-Phe-pNA (S-2538) were from
Chromogenix, and glutaryl-Ala-Ala-Pro-Leu-pNA was from Peptide Institute; minimal essential medium (MEM) and Glasgow MEM were
from Life Technologies, Inc. Patient material was obtained from
the Transplantation Unit, Surgical Department,
Rikshospitalet, Oslo, Norway. Two lines of BHK cells and the
Semliki Forest virus (SFV) expression system were kind gifts from Dr.
Henrik Garoff (Center of Biotechnology, Huddinge, Sweden). The pXT
vector was kindly donated by Dr. S. J. Clark (CSIRO Division of
Biomolecular Engineering, Sydney, Australia).
Human pancreas,
spleen, liver, and adrenals removed surgically were immediately frozen
and stored at 70 °C until use. Total cellular RNA was isolated
(9), and poly(A) mRNA was purified with Dynabeads
oligo(dT)25. Northern blots were prepared (9) using
GeneScreen membranes and then hybridized and washed in 1 × SSC
(0.15 M NaCl, 0.015 M sodium citrate, pH 7.0)
at 68-70 °C (10). DNA probes were labeled with 32P by
random priming (11). The blots used for Fig. 2B were from Clontech.
Screening for cDNA Clones and Sequencing
A gt11
cDNA library (HL1163b) was screened with a 32P-labeled
780-base pair genomic fragment (EcoRI and NsiI)
covering three putative exons. Screening and rescreening were performed
as described (12). DNA was isolated from overnight cultures of
candidate clones by ultracentrifugation of phages and phenol extraction (9). Inserts of cDNA were cloned in Bluescript vectors for sequencing. Plasmids for sequencing were prepared using Magic Miniprep.
Sequencing was done with T7 polymerase and a fluorescent primer using
an automated instrument (13, 14). The 5
end of the transcription unit
was determined by a 5
-RACE kit and by sequencing cDNA clones.
Two peptides were synthesized from the derived amino acid sequence (peptide A, EYDRSSNAEPLQVLSVSR, 18-mer, residues 89-106 in the prepro-sequence and peptide B, NVRAPAVYTRVSKFSTWINQ, 20-mer, residues 240-259), having been selected for being the most different from the corresponding sequence of chymotrypsin B and for potential antigenicity by the Jameson and Wolf algorithm (15). They were created using the Fmoc/t-butyl protecting group strategy in continuous flow solid phase peptide synthesis (16) performed by an automated Biolynx 4170 synthesizer (Pharmacia). 4-Methoxy-2,3,6-trimethylbenzenesulfonyl was used as a protector of arginine side chains (17). The Fmoc amino acids were activated for coupling as pentafluorophenyl esters (18) except for serine and threonine, which were activated as 3,4-dihydro-4-oxo-1,2,3-benzotriazin-3-yl esters (17). Four equivalents of activated amino acid and 1-hydroxybenzotriazole were used in each coupling step, which was performed in dimethylformamide (DMF). Novasyn PA 500 (Novabiochem USA) was used as solid support, and Fmoc cleavage was done in 20% piperidine in dimethylformamide, 5% thioanisole in trifluoroacetic acid was used to cleave peptides from the resin and to remove side chain protecting groups. The peptides were purified and analyzed by reverse phase (C18) HPLC (Shimadzu LC8A).
Antisera were raised in rabbits against these two peptides by
immunizing every 3rd week with 100 µg of peptide coupled to keyhole
limpet hemocyanin and Freund's complete or incomplete adjuvant. Their
antibody responses were monitored in an ELISA using the respective
peptides as antigens and horseradish peroxidase-labeled goat
anti-rabbit Ig (1:5000) as the second antibody. The 3,3- diaminobenzidine (tetrahydrochloride)-based horseradish peroxidase reaction product was visualized using 0.03%
H2O2, 0.3 g/liter diaminobenzidine, and 0.5 g/liter NiCl2 in 50 mM
(NH4)HCO3 (19). Anti-peptide A (APA) sera from
two rabbits were pooled, as were two anti-peptide B sera (APB). In an
ELISA assay, APA and APB were found not to react with
immunoprecipitated chymotrypsin or commercially available chymotrypsin.
Anti-chymotrypsin, however, reacted with CTRL-1 immunoprecipitated by
APA or APB.
SDS-PAGE) with and without reducing agents was performed on precast gels (8-18%). The SDS-PAGE-separated proteins were electroblotted onto nitrocellulose membranes (20). Proteins from different sources of pancreatic enzymes were analyzed and compared with prestained SDS-PAGE standards in the low molecular weight range. Three sources of pancreatic enzymes were utilized: clamp frozen slices of pancreatic tissue homogenized (25 mg/ml) and diluted in 200 mM Tris-HCl, 50 mM CaCl2, pH 8.0, at 4 °C for 24 h and activated by the addition of 1 mM bovine trypsin, duodenal juice from healthy volunteers collected every 10 min during stimulation of pancreatic secretion by intraduodenal instillation of protease inhibitors (20), and urine from pancreas-transplanted patients. Excess binding capacity of nitrocellulose membranes was blocked by incubation in 1% casein for 1 h at room temperature before the antiserum solution was added. Anti-chymotrypsin was used in a 1:3000 dilution in PBS, while APA and APB were diluted 1:1000. After incubation overnight at 4 °C, the membranes were washed 5 times in PBS before a further 2-h incubation in a 1:3000 dilution of horseradish peroxidase-labeled goat anti-rabbit immunoglobulin and visualized as described above (19).
Substrate specificity of immunoprecipitated enzymes was tested using
Nunc-Immuno Plate Maxi Sorb (Nunc, Roskilde, Denmark) coated with APA,
APB, or antisera against human chymotrypsin (10 µg/ml overnight at
4 °C). Excess protein binding capacity was blocked by 1% casein in
PBS for 1 h at room temperature. The source of pancreatic enzyme
(diluted duodenal juice, urine from transplanted patients, or
homogenate of pancreatic tissue) was added in 10-fold dilution, and the
plates were incubated at 4 °C for 4 h to avoid autodigestion.
After washing, the substrates were added at a final concentration of
2-5 mM in 200 mM Tris-HCl, pH 8.3, and 20 mM CaCl2. The substrate reaction was monitored
in an ELISA reader (Titertek Multiskan MK II) at 405 nm after
incubation for 24 h at room temperature. Optical density was
recorded against the appropriate blanks. The substrates used were
S-2532, S-2504, S-2327, S-2538, S-2586, and
glutaryl-Ala-Ala-Pro-Leu-pNA. Results of solid phase bound
enzyme activity are given as nmol·h1. When in solution
the enzyme activity curve was linear at least up to 10 min and the
results are given as nmol·10 min
1.
APA and APB (130 g) were coupled to CNBr-activated Sepharose (5 g). Urine from four patients who had undergone pancreatic transplantation with anastomosis to the ureter was collected on postoperative day 2, pooled, dialyzed overnight against 0.05 M NH4HCO3, pH 7.5 (at 4 °C), and freeze-dried. Aliquots of 50 mg of freeze-dried material were dissolved in 1 ml of PBS and affinity-purified on the APA-APB column. The eluted fractions were monitored by immune detection on dot blots and by hydrolysis of the substrate S-2586. Fractions showing peptidolytic activity and reactivity with the antisera in dot blots were pooled, freeze-dried, and purified further on reverse phase HPLC (Spectra-Physics) using a C18 column. The eluant was a linear gradient of 70% acetonitrile in 0.1% trifluoroacetic acid. N-terminal amino acid sequences were analyzed with an Applied Biosystems 470A instrument.
Recombinant CTRL-1CTRL-1 cDNA cloned from the HL1163b
gt11 library was digested with EcoRI and cloned into
Bluescript SK. Its transcription start was found by RACE, and the PCR
product was cloned into the pXT vector (21). Both the cDNA and the
PCR product were digested with BlpI and BamHI,
and the complete 5
end of the cDNA was replaced by the PCR
fragment. After digestion with BamHI and EcoRV
and blunting of the BamHI end, an 865-base pair fragment was
cloned into the SmaI site of pSFVI to create pSFV-CTRL. RNA
made in vitro from pSFV-CTRL and helper vector 1 was
introduced into BHK cells by electroporation (22). Virus particles from
the supernatant were stored in aliquots at
80 °C.
CTRL-1 SFV particles diluted 1:10 in MEM with 0.2% bovine serum
albumin were added to subconfluent BHK cells for 1 h. The cells
were washed with PBS before incubation for 24 h in Glasgow MEM
containing 20 mM Hepes (pH 7.4) and 2 mM
glutamine. Expression of CTRL-1 was monitored by fluorescence
microscopy of methanol-fixed cells stained with a mixture of APA and
APB (diluted 1:500 in PBS) and goat anti-rabbit fluorescein
isothiocyanate (diluted 1:500 in PBS). Infected cells were lysed in 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 2%
Nonidet P-40 and were left on ice for 30 min. The lysed cells were
centrifuged at 10,000 rpm for 15 min, and the supernatant was stored at
80 °C prior to purification of CTRL-1.
Extracts of infected cells were separated on a Sephadex G-100 column. The catalytic activity of the fractions was monitored using S-2586 and glutaryl-Ala-Ala-Pro-Leu-pNA as substrates. Fractions constituting two partially overlapping peaks of activity were pooled and concentrated by centrifugation at 7500 rpm and 4 °C for 1 h using Microsep 10-kDa filters. The pooled fractions were further separated on a Superdex 75 prep grade column, giving one distinct peak of proteolytic activity. The fractions constituting this peak were pooled and concentrated using the 10-kDa cutoff filter tubes. Enzyme specificity and Mr of the protein in the pooled and concentrated fractions were analyzed. The yield from one 180-cm2 flask was determined to be about 6 µg of purified protein. The enzyme from one flask was used for 20 activity measurements with the various substrates.
Removal of CTRL-1 Activity by APA/APB-coupled Sheep Anti-rabbit DynabeadsFractions with catalytic activity after Sephadex G-100 and HiLoad Superdex 75 were incubated with APA/APB-coupled M-280 Dynabeads rotating at 4 °C. The catalytic activity was tested after removal of the Dynabeads using S-2586 and glutaryl-Ala-Ala-Pro-Leu-pNA as substrates.
The CTRL-1 gene has seven exons with an open
reading frame of 264 amino acids (Fig. 1). By screening
a human cDNA library from the pancreas, 50 clones were obtained, 11 of which were sequenced. Seven of the sequenced clones were complete at
the 3 end including a poly(A) tail. There was a difference in the 3
end of the cDNA caused by alternative polyadenylation. Three
putative polyadenylation signals (A, B, and C) were identified in the
genomic sequence (Fig. 1). The use of signal A (AUUAAA) results in a
minimum transcript of 861 bases plus the poly(A) tail, in accordance
with the 1.0-kb transcript seen on Northern blots (Fig.
2). On the basis of the sequenced cDNA clones (5 out
of 7) and Northern blots, this is the major signal for polyadenylation.
The actual polyadenylation site varied. Poly(A) tails were added 15, 21, and 26 bases downstream of the polyadenylation signal A, with the
latter probably as the major site (3 out of 5). Two minor
polyadenylation signals, B (AAUACA) and C (AAUAAA), were probably used
in two of the sequenced cDNA clones with the poly(A) tail added 17 and 21 bases downstream of the signals, respectively (Fig. 1). The use
of the latter signal creates minimum transcripts of 1129 and 1167 bases
plus poly(A) tails corresponding to the 1.3-kb transcripts seen on the
Northern blots (Fig. 2). A probe from the 3
-untranslated region
identified the larger transcript (Fig. 2). Only three clones were close
to full-length cDNA covering the translation start, and two
cDNA clones had a 5
-untranslated sequence of 10 bases. This was
confirmed by sequencing of 5
-RACE PCR-amplified fragments, which
showed a sequence of 10 bases upstream of the putative translation
start. Thus, the transcription unit starts 10 bases above the
translation start codon.
No transcript was found in tissues other than pancreas (Fig.
2B). By alignment of the deduced CTRL-1 protein sequence to
the sequences of five chymotrypsins (Fig. 3) the
similarities were evident. Putative leader and activation peptides,
active site residues, and disulfide bridges show perfect alignment in
spite of only 54% identical residues (Fig. 3). Among themselves the chymotrypsins were 82-88% identical. By analogy to other serine proteases CTRL-1 is probably synthesized as a prepro-protein with a
signal peptide consisting of 18 amino acids encoded by exon 1 and the
first two bases of exon 2 (Fig. 1). The zymogen must be cleaved,
probably between Arg33 and Ile34, to form an
active protease. The 15 residues of this activation peptide as well as
19 residues of the active enzyme were encoded by exon 2. The three
active site residues His75, Asp121, and
Ser214 were encoded by exons 3, 5, and 7 as in chymotrypsin
B. The CTRL-1 zymogen of 246 residues had a calculated
Mr of 26,100 and a pI of 8.1 in the absence of
glycosylation. A putative N-glycosylation site was the
asparagine at residue 114.
Identification of an Exocrine Enzyme
APA sera from two
rabbits were pooled, as were APB sera. None of the antisera reacted
with purified human chymotrypsin A (Calbiochem). Under non-reducing
conditions both APA and APB sera reacted in Western blotting with a
protein of an apparent Mr of 25,000-27,000 in
pancreatic homogenates (Fig. 4). A marked band at about
55-60 kDa and a more diffuse band above 106 kDa were seen (Fig. 4,
lane 2). The band above 106 kDa was caused by direct
nonspecific binding of the secondary antibody (goat anti-rabbit). The
55-60-kDa band was seen both with APA and APB in unreduced gels of
pancreatic homogenates. Its origin remains unexplained, but a
disulfide-linked dimer is a distinct possibility. Very low levels of
CTRL-1 protein (25-27 kDa) were seen in duodenal juice under basal
conditions (Fig. 4, lane 3 and 4). In a period of
20-30 min after inhibitor-induced stimulation of pancreatic secretion,
proteins of Mr 23,000-24,000 and 21,000 that
reacted with both APA and APB in Western blots appeared in the duodenal
juice (Fig. 4, lane 5). CTRL-1 secretion ended when the
stimulatory signal was terminated by intraduodenal instillation of
saline 60 min after instillation of stimulant. The 25-27-kDa band had
a pI of 6.7, and human chymotrypsin had a pI of 6.9 in the same system.
The apparent relative molecular masses of these antigens (21, 23, 24,
25-27) are consistent with the values predicted from the amino acid
sequence of CTRL-1.
Purification of CTRL-1
Direct immunoprecipitation of CTRL-1 using APA/APB antibodies revealed the presence of a specifically bound peptidolytic activity in urine from pancreas-transplanted patients (Table I). Partial purification of CTRL-1 using urine from such patients as well as APA and APB resulted in a mixture of two main proteins (Fig. 5, lane 3). The N-terminal amino acid sequence of the protein with the lower Mr was found to start with IS-(C?)-PEGTNAYYS. Search in the Swissprot data base gave a 91% identity with human lithostathine precursor (pancreatic stone protein, ISCPEGTNAYRS). The protein with Mr of 24,000-25,000 was identified by Western blotting to be CTRL-1 (Fig. 5, lane 4). Further purification of this material was not achieved. We therefore decided to express CTRL-1 in the Semliki Forest virus system using BHK cells.
|
Purification of the recombinant CTRL-1 resulted in a single
protein band with Mr 25,000 visualized by silver
staining and a peptidolytic activity (Fig. 6, lane
6, and Table II). The peptidolytic activity of the
purified fractions of recombinant CTRL-1 was removed by APA/APB bound
to M280 Dynabeads, identifying CTRL-1 as the only protein with
catalytic acitvity in the samples.
|
The specificity of immunoprecipitated CTRL-1 toward a number of elastase and chymotrypsin substrates was compared with that of chymotrypsin immunoprecipitated from the same source (duodenal juice, homogenates of pancreatic tissue, or urine from pancreas-transplanted patients). The specificity of CTRL-1 was significantly different from that of chymotrypsin (Table I), but the differences were not dramatic and indicated proteolytic activity at amino acids with aromatic side chains (Tyr, Phe, and Trp) and with aliphatic (Leu) and sulfur-containing (Met) side chains. The substrate specificity of recombinant CTRL-1 and immunoprecipitated CTRL-1 was compared with that of human chymotrypsin (Tables I and II). The pH optimum for CTRL-1 was tested from pH 5 to 9 and found to have a peak at 8.0-8.5, similar to chymotrypsin in the same assay (results not shown).
CTRL-1 is a novel human exocrine pancreatic enzyme with substrate specificity similar to that of chymotrypsin. It contains the conserved sequences in the vicinity of the active site serine, aspartic acid, and histidine residues found in this family of serine proteases (25). The pancreatic secretion of this proteinase is feedback-regulated.
Previously, the genes for human and rat chymotrypsinogen B have been cloned and their amino acid sequences derived (26, 27). The amino acid sequences of both bovine chymotrypsin A and B (28, 29) and dog chymotrypsin II (30) have been determined. The CTRL-1 gene is localized on chromosome 16q22.1 in a cluster with four unrelated genes (23). The closest similarities (about 54% identity at the amino acid level) are found with the human chymotrypsin B, compared with 82-88% identity among the chymotrypsins themselves (26).
The antibodies raised against peptides derived from the cDNA sequence allowed detection of the expressed gene product in pancreas as well as in duodenal juice after inhibitor instillation. The inhibitor mixture used is known from previous experiments to prevent the presence of intraduodenal activities of both trypsin and chymotrypsin and to induce an enhanced secretion of pancreatic proteolytic enzymes (25). This suggests that the enzyme is part of an inhibitor-induced pancreatic backup system of proteases. The antigen had the expected Mr and peptidolytic activity. The antibodies raised against the peptides from the derived amino acid sequence did not react with human chymotrypsin. The antigen was absent or present only at very low levels in duodenal juice harvested under basal conditions.
We have shown that the CTRL-1 protein has peptidolytic activity that is significantly different from that of chymotrypsin. The activity ratio of chymotrypsin:CTRL-1 varies with the substrate used. CTRL-1 is clearly more active against peptides with tyrosine, leucine, or phenylalanine in the P1 position, apparently further enhanced by proline in the P2 position. The differences between chymotrypson and CTRL-1 are most pronounced when benzoyl-Arg-Val-Trp-pNA (tryptophan in the P1 position) or benzoyl-Lys-Val-Tyr-pNA (valine in the P2 position) is the substrate. CTRL-1 seems to prefer leucine in P1 or the combination of tyrosine in P1 and proline in P2. The results of the peptidolytic assays with enzyme in solution or those that are solid phase bound are not really comparable. In the solid phase assays, the endogenous anti-chymotrypsin-reactive material of each patient was used, whereas the enzyme in solution was commercial chymotrypsin B. The antibody and the solid phase state of the enzyme may affect CTRL-1 and the various chymotrypsins differently. It is therefore not surprising that the activity ratios show differences.
Because it is being expressed in pancreas and is present in intestinal fluid after stimulation of pancreatic secretion, it is highly likely that the CTRL-1 protein is a novel digestive enzyme. It is not the human homologue of bovine chymotrypsinogen A. The enzyme reported here may represent a new kind of pancreatic chymotrypsin-like enzymes. The Northern analysis and the number of positive clones in the cDNA library (0.2%) suggest the presence of abundant transcripts of the gene. The gene organization, similar to that of rat chymotrypsin B, as well as the fact that the protein is produced solely in pancreas and that the deduced protein sequence shows highest similarity to the chymotrypsins make it reasonable to classify this as a novel chymotrypsin-like enzyme. Accordingly, the name chymotrypsin-like enzyme-1 and the symbol CTRL-1 have been assigned to it.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) X71874[GenBank].
We thank Dr. Tom Kristensen and Dr. Marianne Wright for sequencing the cDNA clones and Dr. Jarle Nordheim for the tissue samples.