(Received for publication, December 3, 1996, and in revised form, January 21, 1997)
From the Department of Surgery, University of Pittsburgh,
Pittsburgh, Pennsylvania 15261, § Department of Biochemistry
and Molecular Biology, University of Georgia, Athens, Georgia 30602,
and the ¶ Department of Biochemistry, Osaka Medical College, Osaka
569, Japan
GTP cyclohydrolase I feedback regulatory protein (GFRP) mediates feedback inhibition of GTP cyclohydrolase I activity by tetrahydrobiopterin and also mediates the stimulatory effect of phenylalanine on the enzyme activity. To characterize the molecular structure of GFRP, we have purified it from rat liver using an efficient step of affinity chromatography and isolated cDNA clones, based on partial amino acid sequences of peptides derived from purified GFRP. Comparison between the amino acid sequence deduced from the cDNA and the N-terminal amino acid sequence of purified GFRP showed that the mature form of GFRP consists of 83 amino acid residues with a calculated Mr of 9,542. The isolated GFRP cDNA was expressed in Escherichia coli as a fusion protein with six consecutive histidine residues at its N terminus. The fusion protein was affinity-purified and digested with thrombin to remove the histidine tag. The resulting recombinant GFRP showed kinetic properties similar to those of GFRP purified from rat liver. Cross-linking experiments using dimethyl suberimidate indicated that GFRP was a pentamer of 52 kDa. Sedimentation equilibrium measurements confirmed the pentameric structure of GFRP by giving an average Mr of 49,734, which is 5 times the calculated molecular weight of the recombinant GFRP polypeptide. Based on the pentameric structure of GFRP, we have proposed a model for the quaternary structure of GFRP and GTP cyclohydrolase I complexes.
(6R)-L-erythro-5,6,7,8-Tetrahydrobiopterin (BH4)1 is an essential cofactor for aromatic amino-acid hydroxylases and nitric-oxide synthases (1, 2). The intracellular level of BH4 is subsaturating for these enzyme reactions and thus is thought to be an important regulator of the activities of these enzymes. BH4 is synthesized from GTP by a pathway composed of four enzyme reactions (1). Most important for regulation of BH4 biosynthesis is GTP cyclohydrolase I (EC 3.5.4.16), which catalyzes the first and rate-limiting step of the conversion of GTP to dihydroneopterin triphosphate. GTP cyclohydrolase I is present in many organisms, ranging from bacteria to animal. In bacteria, this enzyme functions as the first enzyme in the biosynthesis of folic acid and not of BH4 (1). Rat GTP cyclohydrolase I is composed of multiple identical subunits and shows positive cooperativity against GTP (3, 4).
We have recently reported the identification of a new regulatory protein (GFRP) which mediates the feedback inhibition of GTP cyclohydrolase I by the end product of this pathway, BH4 (5). GFRP and BH4 inhibit the enzyme activity of GTP cyclohydrolase I by decreasing its maximum velocity while having little effect on the affinity of GTP, indicating noncompetitive inhibition. GFRP exerts its inhibitory effect on GTP cyclohydrolase I by forming a complex with GTP cyclohydrolase I in the presence of BH4 and GTP. Formation of the complex between the two proteins is reversible, depending on the presence of BH4 and GTP. Furthermore, the inhibition of GTP cyclohydrolase I by GFRP and BH4 is specifically reversed by L-phenylalanine. Phenylalanine is the substrate of phenylalanine hydroxylase, which requires BH4 as a cofactor. Since the stimulatory effect of phenylalanine on GTP cyclohydrolase I activity results in an increase in the rate of BH4 biosynthesis and, in turn, BH4 is used for the conversion of phenylalanine to tyrosine, we suggest that this phenomenon be regarded as feed-forward regulation. Although phenylalanine reverses the GFRP-mediated inhibitory effect of BH4 on GTP cyclohydrolase I, the protein complex remains intact in the presence of GTP, BH4, and phenylalanine. Furthermore, phenylalanine alone induces complex formation between GFRP and GTP cyclohydrolase I, and under these conditions, GTP cyclohydrolase I, which shows positive cooperativity against GTP (3), changes its kinetic behavior from sigmoidal to hyperbolic.
The discovery of GFRP and its function related to phenylalanine revealed the molecular mechanism of a mysterious clinical observation that, in patients with phenylketonuria, the plasma levels of biopterin are increased as are those of phenylalanine (6-9). Thus, the high levels of phenylalanine, which accumulate due to a deficiency of phenylalanine hydroxylase, continuously stimulate the rate of BH4 biosynthesis by the stimulatory effect on GTP cyclohydrolase I activity mediated via GFRP. If this mechanism also operates in cells producing catecholamines or nitric oxide, phenylalanine should affect the biosynthesis of these biologically active molecules by regulating the biosynthesis of the rate-limiting cofactor BH4. Thus, GFRP also may be important in these cells.
As a first step toward elucidating the molecular mechanism of GFRP action, we have established procedures for the purification, cDNA cloning, and bacterial expression of GFRP. These procedures and the characterization of the subunit structures of GFRP are presented here. In particular, the information on the quaternary structure of GFRP provides important insights into the mechanism of GFRP-mediated GTP cyclohydrolase I regulation. Another method for purifying and cloning GFRP was recently reported by Milstien et al. (10). These authors proposed naming the protein GTP cyclohydrolase I feedback regulatory protein (GFRP), a name that is similar to feedback regulator protein for GTP cyclohydrolase I, which we originally designated the protein (5). This suggested terminology is adopted here.
GTP was obtained from Yamasa (Chiba, Japan).
BH4 was a generous gift from the Suntory Institute for
Medicinal Research and Development (Gunma, Japan).
[-32P]dCTP was from Amersham Corp. Dimethyl
suberimidate was from Pierce.
GTP cyclohydrolase I activity was assayed as described (3). Briefly, the standard reaction mixture contained 50 mM Tris-HCl (pH 7.2 at 37 °C), 100 mM KCl, 1 mM EDTA, 1 mM DTT, 0.2 mM GTP, the enzyme, GFRP, and BH4 and/or phenylalanine. The recombinant rat GTP cyclohydrolase I obtained from the plasmid construct described below was used for the assay and had the same enzymatic properties as GTP cyclohydrolase I purified from rat liver.2 The reaction was carried out at 37 °C for 15 min. Then the reaction product, dihydroneopterin triphosphate, was converted to neopterin triphosphate by oxidation with I2 and KI and then dephosphorylated to neopterin by alkaline phosphatase. Finally, neopterin was quantitated by HPLC and fluorescence detection (3).
Protein DeterminationThe concentration of protein was
determined using a Bio-Rad dye-binding protein assay kit with bovine
-globulin as a standard (11).
The maltose-binding protein (MBP) vector system (12) was used for the bacterial expression of rat GTP cyclohydrolase I. The NcoI recognition sequence is located around the initiation codon of rat GTP cyclohydrolase I cDNA (4). The plasmid harboring the cDNA (pRGC) was digested with NcoI and then further digested with mung bean nuclease to create a blunt end. Then the plasmid was further digested with HindIII, and the resulting NcoI (blunted)-HindIII fragment was cloned between the SmaI and HindIII sites of a pMALc vector (New England Biolabs Inc.) to generate pMAL-RGC. The pMAL-RGC plasmid contained the entire coding region of GTP cyclohydrolase I except for the initiator methionine, located just after the sequence coding Ile-Glu-Gly-Arg (factor Xa cleavage site), which was attached to the C terminus of MBP.
Escherichia coli strain JM109 containing pMAL-RGC was
cultured overnight at 37 °C. The 5-ml culture was used to inoculate 1 liter of rich medium (New England Biolabs Inc.), which was incubated until A600 reached 0.5. Isopropyl--D-thiogalactopyranoside was added at a
concentration of 0.6 mM, and the E. coli culture
was incubated for an additional 3 h. The bacterial cells were
harvested by centrifugation and stored at
70 °C. Half of the
bacterial pellet was thawed on ice and sonicated in 10 ml of lysis
buffer containing 10 mM sodium phosphate (pH 7), 30 mM NaCl, 0.25% Tween 20, 10 mM EDTA, 10 mM EGTA, 10 mM
-mercaptoethanol, 0.2 mM phenylmethylsulfonyl fluoride, 0.5 µg/ml leupeptin,
and 0.72 µg/ml pepstatin. The lysate was centrifuged at 14,000 rpm
for 10 min. The supernatant solution was recovered, and the precipitate
was sonicated again in 2 ml of lysis buffer and centrifuged. The
supernatants were combined, and the lysate was divided into four parts
(each 3 ml) and stored at
70 °C.
All procedures were done at 4 °C
unless otherwise stated. Frozen rat livers (120 g) obtained from male
Wistar rats were homogenized in 480 ml of 10 mM Tris-HCl
(pH 7.8 at 1 M), 1 mM EDTA, 1 mM
EGTA, 1 mM DTT, 0.2 mM phenylmethylsulfonyl
fluoride, 0.5 µg/ml leupeptin, 0.35 µg/ml pepstatin, and 1 µg/ml
E-64
(N-[N-(L-3-trans-carboxirane-2-carbonyl)-L-leucyl]agmatine; Boehringer Mannheim) by an Ultra-Turrax homogenizer. The homogenate was
centrifuged successively at 8,000 rpm for 20 min and then at 42,000 rpm
for 1 h to obtain the supernatant solution. The solution was
applied to a 250-ml column of DEAE-Toyopearl (Tosoh, Tokyo, Japan)
equilibrated with 10 mM Tris-HCl, 1 mM EDTA,
and 1 mM EGTA at a flow rate of 400 ml/h. The column was
washed with 400 ml of the homogenization buffer and then eluted with
the buffer containing 0.15 M NaCl. The 120-ml eluate was
divided into two parts and stored at 70 °C until use.
The frozen DEAE-Toyopearl eluate was thawed on ice, and the following solution was successively added to the eluate while stirring: 1.2 ml of 0.5 M sodium phosphate buffer (pH 7.2), 0.6 ml of 25% Tween 20, 3 ml of the E. coli lysate containing the MBP-GTP cyclohydrolase I fusion protein, 65 µl of 1 M DTT, 130 µl of GTP, and 13 µl of 0.1 M BH4. The 67-ml solution was mixed with 14.2 ml of the amylose column equilibration buffer (described below) containing 2 M NaCl. Then the solution was applied to a 15-ml column of amylose (New England Biolabs Inc.) equilibrated with an equilibration buffer of 10 mM sodium phosphate (pH 7.2), 500 mM NaCl, 0.22% Tween 20, 1 mM EGTA, 1 mM DTT, 0.2 mM phenylmethylsulfonyl fluoride, 0.2 mM GTP, and 20 µM BH4 (buffer A). The column was washed with 25 ml of buffer A and then with 50 ml of buffer A without Tween 20 and eluted with buffer A containing 10 mM maltose and no Tween 20. The eluate was concentrated to 1.7 ml by ultrafiltration (Amicon Centriflow CF-25).
The following column operations were done at room temperature. The
concentrated solution was applied to a Superose 6 column (1.6 × 47 cm; Pharmacia Biotech Inc.) equilibrated with TKE buffer (50 mM Tris-HCl (pH 7.5), 100 mM KCl, and 1 mM EDTA) containing 1 mM DTT, 0.1 mM GTP, and 20 µM BH4 at a flow
rate of 1 ml/min. The fractions containing the complex between GFRP and
GTP cyclohydrolase I were combined and concentrated to 0.5 ml by
ultrafiltration. The concentrated solution was filtered through a
Sephadex G-25 HR10/10 column (Pharmacia) equilibrated with TKE buffer
containing 1 mM DTT and 0.01% Triton X-100 (Boehringer
Mannheim) at a flow rate of 2 ml/min. The 1.5-ml eluate was applied at
a flow rate of 1 ml/min to the Superose 6 column equilibrated with the
same buffer as that used for Sephadex G-25. The second smaller peak, which eluted after a larger peak of the MBP-GTP cyclohydrolase I fusion
protein, was collected (see Fig. 1A). The fractions were combined and concentrated to 0.4 ml by ultrafiltration. To verify the
homogeneity of the purified preparation of GFRP, 0.1 ml of the final
sample was analyzed by DEAE-nonporous resin (Tosoh) HPLC. The column,
run at a flow rate of 1 ml/min, was washed with 4 ml of 10 mM Tris-HCl (pH 8), 20 mM KCl, and 0.1 mM EDTA and then eluted with a 10-ml linear gradient of KCl
in the equilibration buffer. The single protein peak that eluted at a
KCl concentration of 0.17 M was collected (see Fig.
1B).
SDS-Polyacrylamide Gel Electrophoresis
SDS-polyacrylamide gel electrophoresis was performed using the PhastSystem (Pharmacia). After electrophoresis, protein was stained with Coomassie Brilliant Blue.
Determination of Partial Amino Acid Sequences of GFRPA sample of purified GFRP (3 µg) was directly subjected to amino acid sequencing using an Applied Biosystems Model 470A automated gas-phase protein sequencer equipped with an on-line HPLC apparatus (Model 120A). The purified GFRP (7 µg) also was digested with lysyl endopeptidase (4 pmol; Wako Chemicals, Osaka, Japan) in a solution (150 µl) containing 50 mM Tris-HCl (pH 9.5), 2 M guanidine HCl, and 1 mM DTT at 30 °C for 21 h. The digest was injected onto an ODS column (4.6 × 250 mm; Nacalai Tesque Inc., Kyoto, Japan) and separated with a 100-ml gradient of acetonitrile from 5 to 60% in 0.1% trifluoroacetic acid at a flow rate of 1 ml/min. The eluate was monitored at 220 nm, and the peak fractions were collected and subjected to amino acid sequence analysis.
cDNA Cloning of GFRPFrom the N-terminal and peptide
amino acid sequences of GFRP, we designed two sets of mixtures of
oligonucleotide primers that cover all possible codons for each amino
acid. One set (primers 1 and 3) was used for the first PCR, and the
other set (primers 2 and 4) was used for a second, nested PCR (see Fig.
2). From 10 µg of rat liver poly(A)+ RNA as a template,
first strand cDNA was synthesized from primer 3 (2.5 µM) by Moloney murine leukemia virus reverse
transcriptase (200 units). Out of a total of 20 µl, 1 µl of the
resulting cDNA was PCR-amplified in a 100-µl reaction using
primers 1 (3 µM) and 3 (1 µM). The PCR
conditions were 1 min of denaturation at 94 °C, 2 min of annealing
at 48 °C, and 1 min of extension at 72 °C for 30 cycles. The
amplified products were separated on 3% Nusieve agarose GTG (FMC
Corp.), and DNA products between 100 and 200 bp in size were isolated
because the expected length of cDNA fell in this range. Using the
first PCR product as a template, we performed the second PCR with
primers 2 (1 µM) and 4 (1 µM) under the
same PCR conditions as described above. The amplified ~130-bp band
was isolated and then subcloned into a pCRII vector (Invitrogen). Both
strands of the inserts were sequenced using an ABI PRISM 377 DNA
sequencer.
From the sequence of the GFRP cDNA isolated above, two primers
(primers 5-1 and 5-2) were designed for performing 5-RACE (13) (see
Fig. 2). We used the 5
-AmpliFINDER RACE kit (CLONTECH) following the
manufacturer's instructions. Primer 5-1 was used for cDNA
synthesis, and primer 5-2 was used for PCR amplification with an anchor
primer (CLONTECH). The PCR products were digested with EcoRI
and BamHI and then cloned into a pBluescript
SK
vector (Stratagene) between the EcoRI and
BamHI sites.
From the sequence of the GFRP cDNA isolated using mixed primers,
two other primers (primers 3-1 and 3-2) were designed for performing
3-RACE (13) (see Fig. 2). We used the 3
-RACE system (Life
Technologies, Inc.). First cDNA strand was synthesized with an
adapter primer (Life Technologies, Inc.). Then the first PCR was
carried out with primer 3-1 and a universal adapter primer (Life
Technologies, Inc.). The PCR products were separated on a gel, and
products between 0.2 and 1.0 kilobase pair were isolated. Using the
isolated PCR products as a template, the second PCR was carried out
with primer 3-2 and the universal adapter primer, and the major
products were isolated and cloned into a pCRII vector.
Two PCR primers were designed to construct a pET expression
vector (14) for GFRP. The E-1 primer was a 36-mer
(5-GGGAATTCCCTTACCTGCTCATCAGCACTC-3
), which
contained an NdeI site (underlined) covering the ATG
translation start codon of GFRP. The E-2 primer was a 39-mer
(5
-GGGAATTCAGGTCATTCCTTGTGCAGACACCAC-3
), which
contained a SalI site (underlined) 3 bp downstream of the TGA translation stop codon. The PCR products obtained with the E-1 and
E-2 primers were separated on a gel, and the major 230-bp band was
isolated. After digestion with NdeI and SalI, the
fragment was cloned into plasmid pET-15b (Novagen) between the
NdeI and XhoI sites. The resulting construct
encoded a stretch of six histidine residues (His6 tag)
followed by the complete coding region of GFRP.
E. coli strain BL21(DE3) transformed with the
His6-GFRP construct (pET-His6-GFRP) was induced
with isopropyl--D-thiogalactopyranoside for 4 h,
harvested, and frozen. The cells obtained from a 225-ml culture were
sonicated in 6 ml of lysis buffer containing 20 mM Tris-HCl
(pH 7.9), 0.3 M NaCl, 40 mM imidazole, 0.25%
Tween 20, 5 mM
-mercaptoethanol, 0.2 mM
phenylmethylsulfonyl fluoride, and 0.72 µg/ml pepstatin, and
insoluble materials were removed by centrifugation. The supernatant
solution was loaded onto a 1-ml nickel-nitrilotriacetic acid column
(QIAGEN Inc.) equilibrated with lysis buffer and washed with lysis
buffer and then with an elution solution of 20 mM potassium
phosphate (pH 5.9), 0.3 M NaCl, and 5 mM
-mercaptoethanol containing 20 mM imidazole. The His6-GFRP fusion protein was then eluted from the column
with the elution solution containing 0.5 M imidazole. The
eluate was concentrated by ultrafiltration (Amicon Centriflow CF-25)
and filtered through a Sephadex G-25 HR10/10 column equilibrated with 0.1 M sodium phosphate (pH 6.5).
To remove the N-terminal histidine residues of the purified His6-GFRP fusion protein, the protein was digested with thrombin, which cleaves at its recognition sequence (Leu-Val-Pro-Arg-Gly-Ser) located between the His6 tag and GFRP. Thrombin cleaves peptides at the carboxylic side of the arginine residue. Ten mg of the protein was digested with 6.6 units of thrombin (Boehringer Mannheim) in 2.77 ml of 100 mM sodium phosphate buffer (pH 6.5) containing 2.5 mM CaCl2 at room temperature for 4 days. After the addition of 30 µl of 0.5 M EDTA, insoluble materials were removed by centrifugation. The supernatant was concentrated to 1.3 ml by ultrafiltration (Amicon Centriflow CF-25). Aliquots (200 µl) of the concentrated solution were separated on a Superdex 75 HR column (1 × 30 cm; Pharmacia) equilibrated with 20 mM Hepes-KOH (pH 7.5), 0.2 M KCl, and 1 mM EDTA at a flow rate of 0.5 ml/min.
Cross-linking of GFRPCross-linking reactions were performed in 100 µl of 50 mM Hepes-KOH buffer (pH 8.2) containing 0.2 M KCl, 1 mM EDTA, and 1 mM DTT or 0.2 M triethanolamine HCl buffer (pH 8 or 9) containing 0.1 mM DTT at a GFRP subunit concentration of 2 or 20 µM. The reaction mixture was prepared on ice. To the mixture was added freshly prepared dimethyl suberimidate to concentrations varying from 0.2 to 25 mM. The resulting mixture was incubated at room temperature for 10 min or 1.5 h, and the reaction was quenched by the addition of 1 M lysine. Protein was then precipitated using a deoxycholate/Cl3CCOOH procedure (15). The precipitates were washed with ethyl ether to remove Cl3CCOOH and dissolved in electrophoresis sample buffer. Cross-linked protein was resolved by electrophoresis on 10% (w/v) polyacrylamide gels in the presence of SDS as described by Weber and Osborn (16). After electrophoresis, protein was stained with silver (17).
Sedimentation Equilibrium of GFRPSedimentation equilibrium
measurements were carried out using a Beckman Instruments OptimaTM XLA
analytical ultracentrifuge. The data were fit using the OriginsTM
software supplied with the instrument (18). For the fitting, we
calculated the molar extinction coefficient and partial specific volume
from the composition of recombinant GFRP using the residue values of
Perkins (19); these were 11,237 M1
cm
1 and 0.739 cm3/g, respectively.
The sedimentation equilibrium experiment utilized three double sector cells and was carried out at three protein concentrations (8, 16, and 24 µM initially), a rotor speed of 16,000 rpm, and 20 °C. In each cell, 100 µl of protein solution and 110 µl of buffer (20 mM Hepes-KOH, 0.1 M KCl, and 1 mM EDTA (pH 7.6)) were loaded. The data were collected after 24 h. Corrections for the density of the solvent were based on a value of 1.006 g/ml, calculated from values provided in Ref. 20.
Purification, cDNA Cloning, and Bacterial Expression of Rat GTP Cyclohydrolase I Feedback Regulatory Protein
Purification of GFRPWe affinity-purified GFRP from rat liver by utilizing the fact that it forms a complex with GTP cyclohydrolase I in the presence of GTP and BH4 and dissociates from the complex in the absence of GTP and BH4 (5). Because GFRP also inhibited the activity of a recombinant form of GTP cyclohydrolase I fused to MBP in the presence of BH4, we inferred that GFRP also formed a complex with the MBP-GTP cyclohydrolase I fusion protein. As described under "Experimental Procedures," partially purified rat liver extracts were mixed with E. coli lysate containing the MBP-GTP cyclohydrolase I fusion protein, and then the mixture was supplemented with GTP and BH4. When the mixture was applied to an amylose column, all the enzyme activity and the inhibitory activity were bound to the amylose. Thus, GFRP appeared to bind to the GTP cyclohydrolase I fusion protein, which in turn was bound to amylose through its MBP portion. Since the amylose fraction also contained endogenous E. coli MBP, the fraction was filtered through Superose gel equilibrated with a buffer containing GTP and BH4 to separate the complex from E. coli MBP. The isolated complex of GFRP and the MBP-GTP cyclohydrolase I fusion protein was then filtered through Superose gel equilibrated with a buffer containing neither GTP nor BH4 to separate GFRP from the MBP-GTP cyclohydrolase I fusion protein. As shown in Fig. 1A, a small protein peak was eluted from the column at an elution volume (80 ml) corresponding to an apparent molecular mass of 35 kDa. The peak contained a BH4-dependent inhibitory activity against GTP cyclohydrolase I. When the preparation was analyzed by SDS-polyacrylamide gel electrophoresis, a single band was observed (data not shown). To further verify the homogeneity of the preparation, it was analyzed on a DEAE-nonporous resin HPLC column (Fig. 1B). A single peak was eluted from the DEAE column. The flow-through fraction showed absorbance at 280 nm, but did not contain any protein. The fraction from the single peak contained a BH4-dependent inhibitory activity against GTP cyclohydrolase I as shown by enzyme assay and also contained a protein that migrated on SDS gel at the same position as that observed using the Superose fractions. The overall purification of GFRP is summarized in Table I. Thus, an ~35,000-fold purification in the specific activity of GFRP from rat liver extract was achieved with a yield of 30%.
|
The subunit molecular mass of GFRP was estimated to be 9.4 kDa by SDS gel analysis, while the apparent native molecular mass of GFRP from gel filtration data was 35 kDa. In addition, a single amino acid (proline) was identified as the N-terminal residue for the purified preparation of GFRP as described below. Therefore, we conclude that GFRP is a protein consisting of multiple identical subunits.
The purified preparation of GFRP had the same effect on the kinetic
properties of GTP cyclohydrolase I as we reported using partially
purified GFRP (5), as shown in Fig. 3. The activity of GTP
cyclohydrolase I was inhibited in the presence of GFRP in a
BH4 dose-dependent manner, and the median
effective concentration of BH4 was 2 µM (Fig.
3A). The inhibition of GTP cyclohydrolase I by GFRP at a
BH4 concentration of 7 µM was
dose-dependently reversed by phenylalanine (Fig.
3B). In the presence of phenylalanine and in the absence of
BH4, the substrate velocity curve for GTP cyclohydrolase I
activity changed from a sigmoidal form to a hyperbolic form in the
presence of GFRP (Fig. 3C). The Hill coefficient decreased from 2.4 to 1.2 in 1 mM phenylalanine.
Partial Amino Acid Sequencing and cDNA Cloning of GFRP
To determine the primary structure of and to establish a bacterial expression system for GFRP, we cloned a cDNA encoding GFRP. The N-terminal amino acid sequence was determined from a purified preparation of GFRP to be Pro-Tyr-Leu-Leu-Ile-Xaa-Xaa-Gln-Ile-Xaa-Met (Xaa represents unidentified residues). Then purified GFRP was digested by lysyl endopeptidase, and four fractions were isolated and sequenced. Fraction L3 gave no amino acid sequence. The amino acid sequences obtained from fractions L1, L2, and L4 are shown in Fig. 2. By comparison with sequences from the N terminus of GFRP, it was deduced that the L2 peptide was derived from the N terminus.
Considering the expected paucity of GFRP in rat liver and its
relatively low molecular weight (5), we undertook to clone its cDNA
by reverse transcription-PCR techniques. Fig. 2 shows the primers used
for cloning and the determined nucleotide sequence. The sequence of the
longest of eight 5-RACE clones isolated is shown in Fig. 2. Two other
clones were just 15 bp shorter than the longest clone, probably because
the additional 15-bp region is GC-rich and reverse transcriptase might
not go through it. In the eight clones sequenced, bases C and T were
found equally in number for the nucleotide located
150-bp downstream from the first nucleotide of the ATG
translation start codon (given as T in Fig. 2). Either base that is in
the third codon encodes asparagine, and the nucleotide variation is
probably due to genetic heterogeneity of the rat and not to PCR
misincorporation. Among six 3
-RACE clones isolated and sequenced, the
longest four were connected to the poly(A) tail at the same site. The
whole sequence shown in Fig. 2 contains 676 nucleotides, including a
poly(A) tail. When compared with the recently reported nucleotide
sequence (10), our sequence differed in three positions. Base A was
identified instead of G at position
42 from the first nucleotide of
the ATG translation start codon, and the sequence GA instead of CT at
positions
50 and
49. Moreover, the sequence presented here contains
an additional 17 nucleotides in the 5
-untranslated region.
The open reading frame encodes 84 amino acid residues. All of the peptide sequences obtained from purified GFRP were found in the translated cDNA sequence (Fig. 2). The deduced amino acid sequence is the same as that reported (10). Comparison of the amino acid sequence deduced from the nucleotide sequence and that determined by Edman degradation from purified GFRP revealed that the initial methionine was missing in purified GFRP. Thus, the mature form of GFRP was inferred to consist of 83 amino acid residues with a calculated Mr of 9,542, similar to the value of 9,400 determined by SDS-polyacrylamide gel electrophoresis for the purified sample. In addition, the sequence contained three possible phosphorylation sites that coincided with the amino acid sequence motifs proposed for kinases (21): Ser35-Arg37 for protein kinase C, Glu46-Tyr48 for epidermal growth factor receptor kinase, and Arg66-Ser69 for calmodulin-dependent protein kinase II.
Bacterial Expression and Purification of GFRPTo obtain a large amount of GFRP for structural and functional characterization, we constructed a bacterial expression vector (pET-15b) to connect a stretch of six histidine residues at the N terminus of GFRP and purified the His6-GFRP fusion protein using a nickel-nitrilotriacetic acid column. The purified His6-GFRP fusion protein was digested with thrombin to remove the His6 tag from GFRP. The final preparation of recombinant rat GFRP thus obtained contains four additional amino acid residues (Gly-Ser-His-Met) at the N-terminal proline of native GFRP. Recombinant GFRP showed kinetic effects on the enzyme activity of GTP cyclohydrolase I identical to those shown in Fig. 3 for the GFRP preparation purified from rat liver (data not shown). These data provide definitive evidence that the isolated cDNA encoded GFRP and that recombinant GFRP is functionally active, with features indistinguishable from native GFRP, although it contained four additional amino acid residues at its N terminus. This bacterial expression system yielded ~30 mg of protein from a 1-liter culture of E. coli. We used His6-free recombinant GFRP for further experiments.
Subunit Composition of GTP Cyclohydrolase I Feedback Regulatory Protein
Cross-linking of GFRPTo determine the number of subunits in
GFRP, we performed cross-linking experiments using the bifunctional
reagent dimethyl suberimidate. The sizes of the products were estimated
by the mobilities of the cross-linked proteins as determined by
SDS-polyacrylamide gel electrophoresis (22). Fig.
4A shows the typical result of reacting 20 µM GFRP with 3 mM dimethyl suberimidate in a
Hepes buffer at room temperature for 1.5 h. Five bands of
decreasing intensity and increasing molecular mass were observed. As
shown in Fig. 4B, the cross-linking reaction proceeded
further in a triethanolamine buffer. The degree of cross-linking was
lesser at a shorter incubation time of 10 min or at lower
concentrations of dimethyl suberimidate (data not shown). The pattern
of cross-linking was independent of the concentration of dimethyl
suberimidate in the range of 1-25 mM. Thus, the reaction
conditions appeared to be saturated with respect to the efficiency of
cross-linking with this reagent. Furthermore, as dimethyl suberimidate
is short-lived because of its high reactivity with water, dimethyl
suberimidate was added to the reaction mixture every 30 min during the
1.5-h reaction in an attempt to improve cross-linking efficiency, but similar results were obtained.
The five bands observed were of decreasing intensity and increasing molecular mass. The extensiveness of cross-linking of GFRP is probably limited by the fact that there are only three lysine residues/GFRP subunit (Fig. 2).
When cross-linking was performed at a lower GFRP concentration of 2 µM (Fig. 4A, lane 2), the relative yields of the protein species were similar to those at 20 µM. The cross-linking reaction was thus insensitive to concentration of the protein. This finding indicates that the reaction occurs within GFRP oligomers and not between GFRP oligomers.
By comparison with the relative mobilities of standard marker proteins, the molecular masses of the proteins in the five stained bands were estimated to be approximately 9.7, 21.8, 31, 43, and 52 kDa (Fig. 4C). These molecular masses correlate well with the predicted values of monomer, dimer, trimer, tetramer, and pentamer of the recombinant GFRP polypeptide chain of 9,954 Da. We therefore conclude that GFRP is a pentamer.
Sedimentation Analysis of GFRPTo further confirm our
conclusion, we performed sedimentation equilibrium measurements (Fig.
5). The data from three cells gave good fits, assuming a
single thermodynamically ideal species, to an average
Mr of 49,734, with a standard deviation of the
calculated molecular weights in the three cells of ±2.7%. There was
no dependence of the calculated molecular weight upon initial protein
concentration. Calculations of molecular weight versus
absorbance at 280 nm (18) showed no tendency toward dissociation or
aggregation. The measured average molecular weight is 5.0 times that of
the subunit, so we conclude that the solution form of GFRP is a stable
pentamer.
Gel Filtration Analysis of GFRP
Both the purified preparation
of GFRP from rat liver and that of recombinant GFRP eluted from a
Superdex 75 column at a position corresponding to a molecular mass of
35 kDa (Fig. 6A). The eluting position is the
same as the one we observed using crude extracts from rat liver (5).
The elution position of GFRP on the Superdex 75 column was not
different in the range of the pH of the equilibration buffer from 6 to
9: Mes-KOH buffer was used for pH 6 and 7; Hepes-KOH buffer for pH 7 and 8; and Tris-HCl for pH 7, 8, and 9. No significant difference in
the behavior of GFRP also was observed between the equilibration
buffers containing 0.1 M and 0.2 M KCl or for
buffers with or without 5% glycerol. There was no dependence of
the elution position and recovery of GFRP on the amount of protein
injected ranging from 1 to 50 µg.
A different group recently reported that GFRP eluted from tandemly linked columns of Superose 6 and Zorbax GF-250 at a position corresponding to a molecular mass of 20 kDa (10). These investigators suggested that the higher molecular mass of GFRP that we had observed (5) using crude rat liver extracts was due to a slow dissociation of GFRP from GTP cyclohydrolase I during the column operation. However, this explanation is unlikely because the eluted peak of BH4-dependent inhibitory activity (GFRP) was symmetrical, and no such activity was detected between the fractions containing GTP cyclohydrolase I activity and those containing GFRP (5). The present result using the purified preparation of GFRP further confirmed that GFRP behaved as a 35-kDa protein on Superdex 75.
GFRP behaved as a 35-kDa protein on a Superose 6 column as well as on a Superdex 75 column (5). Therefore, we inferred that the silica-based column of Zorbax GF-250 was the cause of the slower migration of GFRP. As predicted, recombinant GFRP eluted from a TSK G3000SWXL column (Tosoh), another silica-based column, at a position corresponding to a molecular mass of 17 kDa (Fig. 6B).
The fact that the elution position of GFRP relative to the marker proteins from the silica-based resins was later than that from the agarose-based resins suggests the adsorption of GFRP to the silica-based resins. The relative molecular mass of GFRP estimated by gel filtration on the agarose-based resins was still lower than those determined by cross-linking and sedimentation equilibrium experiments. GFRP also may interact with the agarose-based resins. Alternatively, the anomalous behavior of GFRP may reflect a more compact conformation, lower hydration, or both of GFRP relative to ovalbumin and carbonic anhydrase (23).
Conclusions
From the data presented, GFRP is inferred to be a protein of 83 amino acids that is produced by deletion of the initial methionine residue of the primary translated peptide. Recombinant GFRP produced in
E. coli, which contains four additional amino acid residues at the N terminus, is indistinguishable from native GFRP in terms of
its effect on GTP cyclohydrolase I activity and its behavior on gel
filtration. Cross-linking and sedimentation equilibrium experiments
indicate that GFRP is a pentamer of identical subunits. The observation
that GFRP is a pentamer prompts us to speculate on the structure of the
complexes formed between GFRP and GTP cyclohydrolase I. Our current
hypothesis regarding the mechanism of GFRP action is based on the
results of our previous work (5) and this work, as depicted in Fig.
7. By supposing a general symmetrical structure of
protein complexes, we hypothesize that one subunit of GFRP interacts
with one subunit of GTP cyclohydrolase I. As GTP cyclohydrolase I is
probably an oligomer consisting of 10 subunits (3, 4), this suggests
that two pentamers of GFRP interact with one decamer of GTP
cyclohydrolase I. Consistent with this hypothesis is the recent
crystallographic evidence that E. coli GTP cyclohydrolase I
is a decamer formed by face-to-face association of two pentamers (24).
The considerable sequence similarity between E. coli (25)
and rat (4) GTP cyclohydrolases I suggests that rat GTP cyclohydrolase
I has a three-dimensional structure similar to that of the bacterial
enzyme. This possibility is currently under investigation in our
laboratory.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U85512[GenBank].
We thank Drs. Sidney M. Morris, Jr. and David A. Geller for critical reading of the manuscript and Drs. Timothy R. Billiar and Richard L. Simmons for continuous support.