Cytosolic FDH (ALDH1L1) is an abundant folate enzyme which is well characterized with regard to its catalytic properties and structure [11
]. An interesting feature of this protein is its chimeric nature. Indeed, the enzyme has three distinctive domains, one of which, the C-terminal, is a homolog of aldehyde dehydrogenases and thus defines the classification of the enzyme [8
]. This complex organization enables three catalytic activities of the enzyme: the 10-fTHF hydrolase, ALDH, and 10-fTHF dehydrogenase. Numerous studies suggest that the 10-fTHF dehydrogenase reaction, believed to be the main biological function of the enzyme, is a combination of the first two reactions, the hydrolase and ALDH. While cytosolic FDH has long been described, the identification of its mitochondrial isoform is a relatively recent discovery [15
]. The catalytic properties of mtFDH were previously characterized using protein purified from pig liver. A lower abundance of the endogenous enzyme compared to the cytosolic isoform, as well as an obvious inability to characterize its different domains independently, prompted us to pursue studies of recombinant mtFDH. In the present work, the catalytic properties of recombinant full-length mtFDH and its different domains, expressed in E. coli
, were assessed.
Due to the high sequence similarity to cytosolic FDH, mtFDH was expected to possess the same catalytic residues and activities as the cytosolic isoform (). Indeed, mtFDH, purified from mammalian liver, was capable of catalyzing the 10-fTHF hydrolase and 10-fTHF dehydrogenase reactions [15
]. With no surprise, recombinant mtFDH demonstrated strong 10-fTHF hydrolase activity. Moreover, while the Km
values for this reaction were similar between the two isoforms, the Vmax
for the mtFDH-catalyzed reaction was about 7-fold higher than that of cytosolic FDH. Of note, one of the products of this reaction is formate. Since it is believed that mitochondrial folate metabolism serves to generate formate for its further utilization in cytosolic folate-dependent biosynthetic reactions [6
], a high hydrolase activity points toward mtFDH as a potential producer of formate from folate-bound one-carbon groups. Similar to cytosolic FDH, this reaction requires high levels of a reducing agent in the in vitro
assay (2-mercaptoethanol or DTT were used in our experiments). While it is not clear at present if this reaction truly occurs in mitochondria, high concentrations of glutathione in this organelle (more than 10 mM [30
]) could potentially support in vivo
hydrolase activity of mtFDH.
Important Catalytic Residues of the Human FDH Isoforms
We have previously shown that cytosolic FDH is post-translationally modified by PPT, which adds a 4-PP prosthetic group to the enzyme at serine 354 [14
]. This modification is necessary for 10-fTHF dehydrogenase catalysis [13
]. Interestingly, mtFDH has two serine residues near the predicted site of 4-PP attachment (Ser374 and Ser375), and it was speculated that either of these residues could be modified. Our results, however, demonstrate that PPT modifies mtFDH exclusively at Ser375. Thus, the modification of the intermediate domain by PPT appears to be highly specific. As in the case of the cytosolic enzyme, this modification enables 10-fTHF dehydrogenase catalysis of mtFDH. In our previous studies, siRNA knockdown of PPT abrogated the ability of cell lysate to modify recombinant FDH [14
]. In the present study, the same effect was seen in our experiments with mtFDH. This data, taken together with the strong antiproliferative effects of PPT knockdown seen in human cell lines [14
], suggest the presence of a single PPT type enzyme in humans. Interestingly, a prior study found that PPT is a cytosolic enzyme [24
]. Thus, these findings suggest that mtFDH should be modified in the cytosol prior to its translocation into mitochondria. Since PPT recognizes the ternary structure of proteins [25
], mtFDH must be at least partially folded prior to modification. Overall, these observations raise important questions about the processing of mtFDH and the mechanism of its translocation. Specifically, what is the oligomeric form in which the enzyme translocates (unfolded monomer or folded tetramer) and whether mtFDH is folded into a fully functional protein in cytosol?
Surprisingly, our studies revealed that recombinant mtFDH is unable to catalyze the ALDH reaction. Since mtFDH still catalyzes the 10-fTHF dehydrogenase reaction and previous studies have implied that ALDH activity is a component of this catalysis, it would be logical to assume that the enzyme is capable of ALDH catalysis. The fact that mtFDH catalyzes the 10-fTHF dehydrogenase reaction eliminates the possibility that the lack of ALDH activity is due to incomplete or incorrect folding of the C-terminal domain. Indeed, when expressed separately, this domain forms tetramers and displays a typical CD-spectrum, both parameters indicative of a folded protein. Nevertheless, Ct
-mtFDH itself does not possess ALDH activity in contrast to the corresponding domain of the cytosolic enzyme. While the lack of ALDH catalysis in mtFDH does not have a definitive explanation at present, several mechanisms for such a phenomenon could be considered. For instance, it is possible that short-chain aldehydes are not substrates for mtFDH. Another explanation would be that the 10-fTHF dehydrogenase mechanism is not a precise replica of the ALDH catalytic mechanism (i.e. active ALDH catalysis, as whole, might not be a prerequisite for 10-fTHF catalysis). Indeed, the two reactions result in different oxidation states of the carbon atom of the substrate: it is oxidized from the level of aldehyde to the level of acid in one case (ALDH) and from the level of acid to the level of carbon dioxide in the other (10-fTHF dehydrogenase reaction). ALDH catalysis includes two steps, acylation and deacylation, with a covalent intermediate [31
]. Taking into consideration the requirement for 4-PP for the 10-fTHF dehydrogenase reaction, the difference between a typical ALDH reaction and the one catalyzed by the dehydrogenase domain of mtFDH may be in the acylation step, which in the latter enzyme probably consists of the transfer of the formyl group from the thiol of 4-PP to the thiol of the FDH catalytic cysteine. Thus, the inability to accomplish canonical acylation step could be accountable for the lack of the ALDH catalysis in mtFDH.
Importantly, it has recently been reported that zebrafish cytosolic FDH (zFDH) also lacks ALDH activity [34
]. Of note, recombinant zFDH, similar to mtFDH, is capable of the 10-fTHF dehydrogenase catalysis (after activation with mammalian cell lysate and CoA). The C-terminal domains of human mitochondrial and zebrafish FDH are highly similar to that of the rat cytosolic enzyme (77% and 81% sequence identity, respectively). In particular, the catalytic cysteine and glutamate (), as well as the residues forming the substrate entrance tunnel, are conserved in all three proteins. The ALDH catalytic centers of the active C-terminal domain of cytosolic FDH and non-active domains of mtFDH and zFDH were analyzed for potential structural alterations using homology modeling. To confirm the utility of our models, they were examined for sterical clashes at the monomer-monomer interface, as well as for other potential hindrances due to amino acid residue replacements. These in silico
experiments suggest a near identical arrangement of the active sites in the examined proteins (). However, it can be speculated that the ALDH catalysis of the inactive enzymes may be influenced by distant residues, not participating directly in substrate binding. For example, it has been suggested that the half-of-the-site reactivity observed in some aldehyde dehydrogenases could be a result of yet to be characterized long-range interactions occurring between the catalytic site and the nucleotide binding site [35
]. Alternatively, insertion of the bulky 4-PP prosthetic group in the substrate entrance tunnel may be necessary for “fine tuning” of the structures of human mitochondrial and zebrafish enzymes in order for optimal substrate binding and orientation for nucleophilic attack of the catalytic cysteine.
Figure 6 Superposition of the active site residues in the crystal structure of rat cytosolic Ct-FDH (PDB ID 2o2p, orange) and a homology model of human mitochondrial Ct-FDH (cyan) generated with the SWISS-MODEL server . Amino acid residues are numbered according (more ...)
Aldehyde dehydrogenases also possess an esterase activity, the hydrolysis of p
]. This activity was found in both isoforms of FDH, in contrast to ALDH activity. It appears, however, that mtFDH displays much lower esterase activity than the cytosolic enzyme. Esterase catalysis relies on the action of the same cysteine which acts as the nucleophile in ALDH catalysis [37
]. Thus, the diminished esterase activity of mtFDH may reflect the apparent lack of the ALDH activity. It is hard to predict whether mtFDH or zFDH would have aldehyde dehydrogenase activity within the cellular environment. In this regard, a recent study demonstrated activation of the low-activity oriental ALDH variant by a small molecule [38
]. The existence of such a compound implies a theoretical possibility for activation of mtFDH ALDH catalysis, but this activity may not have an independent physiologic function for FDH enzymes. Regardless, the main catalytic function of the FDH isoforms as 10-fTHF metabolizing enzymes appears to be competent in the absence of ALDH activity. On an evolutionary note, since the divergence between ALDH1L1 and ALDH1L2 took place at some time before the emergence of bony fish [39
], it is possible that cytosolic FDH acquired the ability to catalyze the ALDH reaction at some distant point after ALDH1L1