(Received for publication, October 23, 1996, and in revised form, February 5, 1997)
From the Department of Biochemistry, Vanderbilt
University School of Medicine, Nashville, Tennessee 37232 and the
¶ Research Service, Department of Veterans Affairs Medical Center,
Nashville, Tennessee 37212
We expressed the NH2-terminal domain of the multidomain, multifunctional enzyme, 10-formyltetrahydrofolate dehydrogenase (FDH), using a baculovirus expression system in insect cells. Expression of the 203-amino acid NH2-terminal domain (residues 1-203), which is 24-30% identical to a group of glycinamide ribonucleotide transformylases (EC 2.1.2.2), resulted in the appearance of insoluble recombinant protein apparently due to incorrect folding. The longer NH2-terminal recombinant protein (residues 1-310), which shares 32% identity with Escherichia coli L-methionyl-tRNA formyltransferase (EC 2.1.2.9), was expressed as a soluble protein. During expression, this protein was released from cells to the culture medium and was purified from the culture medium by 5-formyltetrahydrofolate-Sepharose affinity chromatography followed by chromatography on a Mono-Q column. We found that the purified NH2-terminal domain bears a folate binding site, possesses 10-formyltetrahydrofolate hydrolase activity, and exists as a monomer. Titration of tryptophan fluorescence showed that native FDH bound both the substrate of the reaction, 10-formyl-5,8-dideazafolate, and the product of the reaction, 5,8-dideazafolate, with the same affinities as its NH2-terminal domain did and that both proteins bound the substrate with a 50-fold higher affinity than the product. Neither the NH2-terminal domain nor its mixture with the previously purified COOH-terminal domain had 10-formyltetrahydrofolate dehydrogenase activity. Formation of complexes between the COOH- and NH2-terminal domains also was not observed. We conclude that the 10-formyltetrahydrofolate dehydrogenase activity of FDH is a result of the action of the aldehyde dehydrogenase catalytic center residing in the COOH-terminal domain on the substrate bound in the NH2-terminal domain and that the intermediate domain is necessary to bring the two functional domains together in the correct orientation.
The amino-terminal sequence (residues 1-203) of the tetrameric multifunctional enzyme, 10-formyltetrahydrofolate dehydrogenase (FDH)1 (EC 1.5.1.6) is 24-30% identical to glycinamide ribonucleotide transformylase (GAR-transformylase) (EC 2.1.2.2) from different species (1). There is also a 32% identity (2) (residue 1-310) to Escherichia coli L-methionyl-tRNA formyltransferase (MFT) (EC 2.1.2.9) (3). The NH2-terminal domain of FDH presumably contains a folate binding site. The sequence HPSLLP (residues 106-111) and a glycine (residue 115) four residues downstream were predicted to be the key motif for the 10-FTHF binding (1). This motif is strictly conserved among a large number of enzymes that use 10-FTHF as a formyl donor. No direct evidence has been obtained, however, that this sequence is responsible for folate binding. The structure of a complex between E. coli GAR-transformylase and a multisubstrate adduct, part of which imitates 10-FTHF, was recently resolved (4). It was found that His108 of this conserved motif in GAR-transformylase (which corresponds to His106 of FDH) might participate in folate binding through formation of a hydrogen bond with the oxygen of the formyl group (4). Later experiments based on site-directed mutagenesis led to the conclusion that this histidine is not absolutely required for catalytic function (5).
In the absence of NADP+, FDH hydrolyses 10-FTHF to THF and
formate (6, 7). Presumably, the NH2-terminal domain of FDH is responsible for this hydrolase activity. An enzyme,
formyltetrahydrofolate hydrolase, which also catalyzes the hydrolysis
of 10-FTHF to THF and formate was recently found in E. coli
(8). This enzyme is activated by methionine and inhibited by glycine
(9). It is composed of 280 amino acid residues; the sequence from
residues 84 to 280 exhibits 27% identity with GAR-transformylase (8) while the sequence from residues 1 to 84 serves as a regulatory domain,
which binds methionine and glycine (8, 9). The enzyme contains the
sequence HHSFLP and a glycine four residues downstream (see Fig. 1)
that is similar to the possible 10-FTHF binding motif mentioned above
with just two substitutions, proline for histidine and leucine for
phenylalanine. Despite their similar catalytic function, identity of
the sequence of formyltetrahydrofolate hydrolase (residues 84-280)
with the NH2-terminal domain of FDH (residue 1-197) does
not exceed 20%, and most of the identity is connected with the
putative 10-FTHF binding motif and sequence adjacent to aspartate 142, which was also predicted to participate in the substrate binding
(Fig. 1). In addition, FDH has no regulatory domain.
Therefore, while the hydrolase can regulate the pools of 10-FTHF and
THF in response to the methionine/glycine ratio (9), such a regulatory
role looks unlikely for FDH.
Using limited proteolytic digestion of rabbit liver FDH, Schirch et al. have shown that the hydrolase activity of the enzyme is associated with a 32-kDa NH2-terminal domain (10). As a result of studies employing fluorescence and isothermal titration calorimetry, these investigators also reported that the enzyme contains only one folate binding site per tetramer. This differs from an earlier report from our laboratory using covalent addition of folate that found two folates bound per monomer or eight per tetramer (11).
Recently, we expressed the COOH-terminal domain of FDH (residues 420-902), which is about 48% identical to a group of aldehyde dehydrogenases (12). This recombinant peptide was a functional protein displaying aldehyde dehydrogenase activity and bearing an NADP+ binding site. It was perhaps not surprising that the COOH-terminal domain could be folded into a functional protein because of the high homology with aldehyde dehydrogenase and the presence of aldehyde dehydrogenase activity in native FDH. In the case of the NH2-terminal domain, it is difficult to predict whether this domain itself can fold into a functional protein as it has significantly less identity with any natural proteins. Nevertheless, expression and study of protein fragments often provide important information about protein structure and function. Therefore, in the present study, we expressed the NH2-terminal domain of FDH to learn whether this domain is able to fold independently of the other two domains into a functional enzyme having formyltetrahydrofolate hydrolase activity and to investigate its properties.
Materials used, protein expression, assay of enzyme activity, fluorescence studies, analysis of kinetic data, and protein molecular size determination are described in the accompanying article (12).
Vector ConstructionTo prepare the construct for expression
of the 203-amino acid NH2-terminal protein, the terminal
codon, TAG, was introduced into the FDH sequence cloned into the
vector, pVL1393, (13) immediately after isoleucine 203. The codon, CAG,
corresponding to glutamine 204 in the sequence of FDH was changed to a
TAG stop codon using the Transformer site-directed mutagenesis kit
(Clontech, Palo Alto, CA). Two oligonucleotides were used, the
mutagenic oligonucleotide 5-CTATGAGGGCATCAAGAAGGAGA-3
for introduction of the mutation in the FDH sequence (stop codon is
underlined) and a selection oligonucleotide
5
-GAATTCCGGAGCTGCAGATC-3
for conversion of one
unique restriction site, XmaIII, to StuI (underlined), which was also unique, in pVL 1393.
To construct the vector for expression of the 310-amino acid residue
NH2-terminal domain, we removed the coding sequence between the NheI and ScaI restriction sites (amino acids
311-902) from the vector used for expression of FDH (13). This was
done in several steps. First, we removed from pVL 1393, with the FDH
sequence inserted, one of the two ScaI restriction sites
(residing in the ampicillin resistance region) by site-directed
mutagenesis. The Transformer site-directed mutagenesis kit, mutagenic
oligonucleotide 5-GACTGGTGCAACCAAG-3
that altered nucleotide T to C (bold) in the ScaI
restriction site (former ScaI restriction site is
underlined), and the selection oligonucleotide was used in the
experiment. The mutated plasmid was cut with NheI
restriction endonuclease (that removed the sequence corresponding to
amino acid residues 310-504 between the two NheI
restriction sites) and was treated for 15 min with Mung bean nuclease
(Promega, Madison, WI) to create blunt ends. Then the linear plasmid
was cut with ScaI restriction endonuclease (that removed the
sequence corresponding to amino acids 505-902) and was ligated through
internal blunt ends.
The mutagenic experiments were carried out according to the Clontech protocol. The sequence of the constructs was confirmed by sequencing on model 373A fluorescence sequenator (Applied Biosystems).
Purification of the Recombinant ProteinAll buffers used in purification steps contained 10 mM 2-ME and 1 mM NaN3. FDH was purified as described earlier (13). The 310-amino acid NH2-terminal domain was purified from the cell-free culture medium by affinity chromatography on a column of 5-formyltetrahydrofolate-Sepharose (14, 15). A column (1.5 × 10 cm) was packed with about 8.0 ml of settled gel and equilibrated with 10 mM Tris-HCl buffer, pH 7.4 (buffer 1). Medium (200 ml), plus 2-ME (10 mM) and NaN3 (1 mM) was applied to the affinity column. The column was then washed with buffer 1 (100 ml). The enzyme was eluted from the column with buffer 1 containing 1 M NaCl. The eluate was passed through a column of Bio-Gel P6-DG (Bio-Rad) equilibrated with buffer 1 to remove salt. The eluate was concentrated to approximately 5 ml using an Amicon, Inc. (Beverly, MA) filtration cell. Additional purification was done on a Mono-Q column with a linear NaCl gradient (0-1.0 M in buffer 1) using an FPLC system (Pharmacia Biotech Inc., Piscataway, NJ).
Two different length peptides were expressed, one containing the first 203 amino acids from the NH2 terminus and the other containing the first 310 amino acids. The expressed 203-amino acid length protein was present only in the cells but not in the medium even in the very late postinfection period (data not shown). This is unlike the results obtained when recombinant FDH (13) and its COOH-terminal domain were expressed (12) where most of the enzyme was found in the medium. Attempts to purify the protein from the cells showed that the expressed 203-amino acid length NH2-terminal peptide was insoluble and was not purified or characterized.
In contrast, the recombinant NH2-terminal 310-amino acid
polypeptide was found to be present in both cells and culture medium (Fig. 2), similar to the full-length enzyme (13) and its
COOH-terminal domain (12). To purify this protein, we used affinity
chromatography on a 5-formyltetrahydrofolate-Sepharose column. This
procedure was effective for isolation of the 310-amino acid
NH2-terminal domain from both the cell lysate (data not
shown) and the culture medium (Fig. 3). The recombinant
protein was bound by the affinity column and retained during washing
with low salt buffer. The protein was completely eluted from the column
by 1.0 M NaCl. Additional elution with 20 mM
folate revealed only trace amounts of the protein. Further purification
was done by chromatography on a Mono-Q column (Fig. 3). The peak
containing the NH2-terminal domain was eluted at 0.27 M NaCl, pooled, and stored at 4 °C.
Properties of the NH2-terminal Domain
Initial
studies showed that the 310-amino acid NH2-terminal domain
was able to hydrolyze both 10-FTHF and 10-FDDF. For further study on
this protein, we used the dideaza analog, which is much more stable and
has been shown to replace the natural substrate in enzymatic reactions
carried by FDH (16). Kinetic parameters of the reaction for the
NH2-terminal domain were similar to those for the native
enzyme with Km values of 7.3 and 5.8 µM and Vmax values of 99 and 80 nmol min1 mg
1 (kcat
of 3.5 and 8.0 min
1/protein monomer), respectively.
Addition of NADP+ did not change the velocity of the
hydrolase reaction, and no increase in absorbance at 340 nm was
observed showing no reduction of NADP+ to NADPH. This
showed that the NH2-terminal domain does not possess dehydrogenase activity. To check whether the NH2-terminal
domain has a binding site for the coenzyme, we titrated the quenching of tryptophan fluorescence with NADP+. No changes in
fluorescence of the NH2-terminal domain were observed in
the presence of NADP+ (data not shown). In contrast,
fluorescence of the full-length enzyme (17) and its COOH-terminal
domain (12) was decreased 25-30% in the presence of
NADP+.
It has been shown that the hydrolase reaction performed by FDH is
dependent on the presence of 2-ME (2, 16). To study whether the
NH2-terminal domain requires the presence of 2-ME, we
measured the velocity of the reaction in the presence of varying concentrations of 2-ME. It was observed that the reaction is strictly dependent on the presence of 2-ME (Fig. 4). Moreover,
the reaction requires relatively high concentrations of 2-ME as the
velocity was still going up at 100 mM of 2-ME (Fig. 4).
The pH-dependence of the hydrolase reaction was measured for both FDH
and its NH2-terminal domain. This was carried out using two
buffer systems, potassium phosphate (pH 4.5-9.0) and acetate (pH
3.0-5.0). No difference in the activity was found between the two
buffer systems in the overlapping pH range. The pH dependence of the
hydrolase reaction was similar for both FDH and its 310-amino acid
NH2-terminal domain with a very broad reaction maximum
(Fig. 5).
Folate Ligand Binding Study
We used titration of tryptophan
fluorescence to investigate differences in ligand binding between FDH
and the 310-amino acid NH2-terminal domain. The fact that
the hydrolase reaction does not proceed in the absence of 2-ME and the
dehydrogenase reaction requires the presence of NADP+
allowed us to specifically measure ligand binding in the absence of
both compounds since no reaction can take place. This was done with the
substrate 10-FDDF and the reaction product DDF. The tryptophan fluorescence of both FDH and the NH2-terminal domain was
decreased by about 50% when either 10-FDDF or DDF was bound (data not
shown). The affinity for the substrate was 50-fold higher than the
affinity for the product (Fig. 6) with dissociation
constants of 6 and 300 nM, respectively. No significant
difference in the affinity between FDH and NH2-terminal
domain was observed (Fig. 6). We also studied the influence of 2-ME on
the binding of 10-FDDF and DDF to the proteins. Because the
Kd for 10-FDDF is about three orders of magnitude
less than the Km in the hydrolase reaction, the
addition of 2-ME did not evoke substrate decomposition in the
concentration range studied, and we were able to measure binding in the
presence of 2-ME. We did not see any differences in affinity of FDH and
the NH2-terminal domain for either 10-FDDF or DDF in the
presence of 2-ME (data not shown).
Analysis of the NH2-terminal Domain Oligomeric Structure
We investigated the oligomeric structure of the
NH2-terminal domains using size-exclusion chromatography on
a Sephacryl S-300 column. For this purpose, medium after expression of
the NH2-terminal domain and without any preceding
purification steps was applied to a Sephacryl S-300 column, and
hydrolase activity was measured. Fig. 7 shows that
maximum hydrolase activity coincided with a protein of about 35 kDa.
SDS-PAGE and immunoblot analysis of the eluted fractions have also
confirmed that the NH2-terminal domain was eluted in a
position coinciding with a protein of about 35 kDa molecular mass (Fig.
7), corresponding to the protein monomer.
Study of Interaction between COOH- and NH2-terminal Domains
To determine whether the COOH- and NH2-terminal domains together can combine to produce 10-FTHF dehydrogenase activity, we measured the activity in the presence of both proteins. No 10-FTHF dehydrogenase activity was found when the two domains were mixed. To investigate whether the two domains can interact in the absence of the intermediate domain, we expressed both the COOH- and NH2-terminal domains together. SDS-PAGE analysis of the culture medium showed that both recombinant proteins were expressed (data not shown). The medium was applied to a Sephacryl S-300 column, and all three FDH enzyme activities were assayed in collected fractions. Aldehyde dehydrogenase activity was found in fractions corresponding to a tetramer of the COOH-terminal domain, and 10-FTHF hydrolase activity was observed in fractions corresponding to the NH2-terminal domain monomer (data not shown). No 10-FTHF dehydrogenase activity was found in any of the collected fractions nor did any fraction contain both aldehyde dehydrogenase and hydrolase activities together. SDS-PAGE analysis confirmed that the COOH- and NH2-terminal domains resided in different fractions. These experiments prove that COOH- and NH2-terminal domains do not interact in the absence of the intermediate domain.
FDH consists of an NH2-terminal domain, an
intermediate domain, and a COOH-terminal domain (1). We have recently
expressed the COOH-terminal domain which has 48% identity with
aldehyde dehydrogenase (12) and showed that this domain bears an
NADP+ binding site and a dehydrogenase catalytic center but
has no folate binding site. The intermediate domain is not believed to be directly involved in catalysis but serves to bring the functional COOH- and NH2-terminal domains together in the correct
orientation. Earlier, we reported that mutation of cysteine 707 of FDH,
which is located in the COOH-terminal domain and thought to be the
dehydrogenase catalytic center nucleophile, to alanine abolished the
dehydrogenase activity, but the hydrolase activity remained (17). Based
upon this information, we predicted that 10-FTHF hydrolase activity resides in the NH2-terminal domain while 10-FTHF
dehydrogenase activity is a result of the action of the aldehyde
dehydrogenase catalytic site on the folate substrate bound to the
NH2-terminal domain. We expressed two
NH2-terminal peptides, one contained 203 amino acids that
corresponds to the length of GAR-transformylase and the other contained
310 amino acids that corresponds to the length of MFT from E. coli. MFT shares higher identity with the NH2-terminal
domain of FDH than GAR-transformylase does, 32 (Fig. 8)
and 27%, respectively. It also has a sequence similar to the putative
10-FTHF binding motif. There is only one conservative change, that of a
proline to a glycine (Fig. 8). MFTs from several other species have
proline in this position, confirming conservation of the sequence. FDH
and MFT also have a higher identity in this region than FDH and
GAR-transformylase (Fig. 8).
We were surprised to find that the expressed 203 amino acid residue NH2-terminal domain was completely insoluble. The sequence does not contain regions that are highly hydrophobic, and it is not a membrane protein. This suggested that the protein was not folded appropriately. In contrast, the expressed 310-amino acid NH2-terminal protein was soluble and possessed properties that were inherent in the entire enzyme; it contained a folate binding site and displayed 10-FDDF hydrolase activity, suggesting it was folded appropriately. This extends the putative NH2-terminal domain to about 310 amino acid residues rather than the 203 residues derived by comparison with GAR-transformylase. The folate binding parameters and kinetics of the hydrolase reaction were similar for both the NH2 terminus and the complete enzyme. The reaction had the same pH dependence and had a strict requirement for 2-ME. Using the quenching of tryptophan fluorescence, we found that the NH2-terminal domain does not have an NADP+ binding site. The NH2-terminal domain exists as a monomer, unlike the E. coli formyltetrahydrofolte hydrolase enzyme, with the same catalytic function that exists as a hexamer (9).
Although our results showed that the folate binding site is present in
the NH2-terminal domain and previously we also showed that
the COOH-terminal domain of FDH has no folate binding site (12), the
possibility could not be excluded that another folate binding site
might be formed from the participation of the two functional domains.
However, results obtained after fluorescence titration using DDF or
10-FDDF revealed that there is no difference in substrate binding
between FDH and its NH2-terminal domain. This supports the
conclusion that only the NH2-terminal domain is involved in
substrate binding while the COOH-terminal domain has neither a folate
binding site nor participates in formation of such a site together with
the other domains. The formyl group of the folate substrate bound to
its binding site in the NH2-terminal domain must be
accessible to the dehydrogenase catalytic center, and this leads to the
conclusion that this group must protrude above the surface of the
NH2-terminal domain or at least be close to the surface.
This assumes that the pteroyl moiety plays the main role in the
interactions of the substrate with the FDH folate binding site. This
suggestion is consistent with data obtained from the crystal structure
of GAR-transformylase complexed with 5-deazatetrahydrofolate that
showed that the pterin part of the folate is well ordered deep in the
hydrophobic cleft between the NH2- and COOH-terminal
domains of the enzyme, the benzoyl ring is less ordered, and glutamate
moiety is at the enzyme surface and is disordered (18). Both FDH (16)
and GAR-transformylase (19) can utilize 10-FDDF instead of 10-FTHF as a
formyl donor, suggesting structural similarity between the
folate-binding sites of the two enzymes. 10-FDDF binds to FDH and its
NH2-terminal domain 50-fold more tightly than DDF, probably
because of some enhancement of binding by the formyl group. Calculation
of the difference between free energies for binding substrate, 10-FDDF, and product, DDF, gives a difference in G of 2.3 kcal/mol. This corresponds to a hydrogen bond energy that is in the range of 2-5
kcal/mol (20) and suggests that the formyl group of the folate
substrate forms a hydrogen bond with the protein in the active site.
Either hydrogen or oxygen of the formyl group could be a candidate for
bond formation. The resolved crystal structure of GAR-transformylase
(4) showed that such a hydrogen bond can be formed with
His108 or Asp144 of GAR-transformylase
(corresponding to His106 and Asp142 for FDH),
which are strictly conserved in proteins using 10-FTHF as a formyl
donor (1). Possible reasons for such hydrogen bonding of the formyl
group in the catalytic mechanism of FDH could be to limit the
flexibility of the group, to create stress in the structure of the
substrate, and to facilitate the nucleophilic attack and intermediate
hemiacetal complex formation during the dehydrogenase/hydrolase
reaction (2).
One discrepancy with previous reports was found from the measurement of ligand binding. Previous studies reported (21) that the product of the reaction catalyzed by FDH, THF, binds more tightly to the enzyme than the substrate, 10-formyl-THF, while we show here that the substrate was bound with 50 times higher affinity than was the product of the reactions. The previous study, however, did not measure the binding constant for the substrate, probably due to the instability of the natural substrate, 10-FTHF, and because of the hydrolase reaction that converts 10-FTHF to THF and formate. In the present study, we avoided these problems by using the stable substrate analogue, 10-FDDF, and performed the analysis in the absence of 2-ME so that the hydrolase reaction did not proceed.
The data obtained concerning the domain structure of FDH suggests the
following topography for the native protein (Fig. 9). The COOH-terminal domain of the enzyme has sites responsible for the
protein oligomerization and bears the NADP+ binding site
and the aldehyde dehydrogenase active site, which acts as a catalytic
center in the dehydrogenase reaction utilizing 10-FTHF when bound to
the adjacent NH2-terminal domain. The
NH2-terminal domain can now be extended to about the first
310 amino acid residues and bears a folate binding site and the
hydrolase catalytic center. The intermediate domain (which now is
shortened to about a hundred amino acid residues) is necessary to bring
the two functional domains together to carry out the 10-FTHF
dehydrogenase reaction. The two functional domains apparently do not
form multiple and close contacts to each other. That was shown by the
absence of interdomain complexes, and only the presence of the
intermediate domain brings them into the correct orientation. We would
like to mention that the intermediate domain (residues 313-403) is enriched with negatively charged amino acid residues, glutamic acid
(17.6 versus 6.5% for combined NH2- and
COOH-terminal domains) and aspartic acid (8.8 versus 4.7%
for NH2 and COOH-terminal domains). Thus, we suggest that
electrostatic interactions provided by the intermediate domain could be
major contributors to the orientation of the NH2- and
COOH-terminal domains in native FDH. Additionally, the fact that the
two functional domains not only can function separately but also can be
folded separately to produce functional proteins supports the idea that
FDH has evolved from a fusion of two different genes, one of which
codes for aldehyde dehydrogenase while the other is not yet identified.
Our study still leaves open the question of whether the intermediate
domain of FDH functions merely as a bridge that brings the two
enzymatic domains together to create a new enzyme function or whether
it also plays a regulatory role to modulate activity in response to
changes in physiological conditions.