Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.enzyme.chem.msu.ru/~tishkov/Publications/Proteins_19_FDH.pdf Äàòà èçìåíåíèÿ: Wed Jan 5 21:32:38 2005 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 23:10:08 2012 Êîäèðîâêà:

Research Signpost 37/661 (2), Fort P.O., Trivandrum-695 023, Kerala, India

.
Protein Structures: Kaleidoscope of Structural Properties and Functions, 2003: 441-443 ISBN: 81-7736-177-5

Editor: Vladimir N. Uversky

19

NAD+-dependent formate dehydrogenase. From a model enzyme to a versatile biocatalyst
Vladimir O.Popov1 and Vladimir I.Tishkov2 1 A.N.Bakh Institute of Biochemistry, Russian Academy of Sciences Leninskiy pr., 33 119071 Moscow, Russia. E-mail: vpopov@inbi.ras.ru; 2 Department of Chemical Enzymology Chemistry Faculty, M.V.Lomonosov Moscow State University Leninskie Gory, 119992 Moscow, Russia. E-mail vit@enz.chem.msu.ru

I. Introduction
Very few enzymes enjoyed such a close attention from various sectors of life sciences community ranging from quantum chemists to biotechnologists as NAD+-dependent formate dehydrogenase (EC 1.2.1.2; FDH). First discovered in 1950-1951 [1,2] it has not attracted much interest until the middle of seventies of the last century when a problem of cofactor regeneration was brought on the agenda due to the use of redox enzymes for synthesis of high value added organic chemicals. Quite soon it was realized also that FDH is one of the most suitable models for investigating the general mechanism of catalysis involving hydride ion transfer. The chemical reaction catalyzed by FDH is devoid of proton release or abstraction steps and entails cleavage of a single
C o rre s pondence/R e print re ques t: Prof. Vladimir O. Popov, La boratory of Enzyme Engineering, A. N. Ba kh Ins t it ut e of Bioche mis t ry, R us s ian Academy of Scie nces , Le nins kiy pr. , 33, 119071 M o s c ow. E-ma il: vpopov@inbi. ra s . ru


442

Vladimir O.Popov & Vladimir I.Tishkov

carbon-hydrogen bond in the substrate and formation of a single new one in the product. This fortunate combination of advantageous intrinsic enzymatic and physico-chemical properties alongside with a high biotechnological potential brought FDH into the focus of research in the field of NAD+-dependent dehydrogenation. By now NAD+-dependent FDH is one of the most extensively studied among NAD+dependent dehydrogenases and can be regarded as a "text-book" enzyme along with such well-known examples as alcohol dehydrogenase or lactate dehydrogenase. State of the art in FDH research was last comprehensively reviewed in 1994 [3]. The present review provides an update to the previous one, is confined to the recent developments only and focuses on the modern understanding of the structure and mechanism of action of NAD+dependent FDH and issues relevant to its biotechnological applications.

II. Structure
Sequence
The past decade and especially several recent years due to the progress in the field of genomics resulted in the outburst of the FDH sequences form various sources. Genes of FDH were found in uncultured proteobacterium EBAC31A08 (EMBL Accession AF279106); bacteria - Pseudomonas sp. 101 [4], Moraxella C-1 (EMBL Y13245), Paracoccus sp. 12A [5], Mycobacterium vaccae N10 [6], Hyphomicrobium sp. JC17 [7], in pSymA megaplasmid of symbiotic nitrogen-fixing bacterium Sinorhizobium meliloti [8]; methylotrophic yeast - Pichia angusta (previously named Hansenula polymorpha) [9], Candida methylica [10], Candida boidinii [11,12], baker's yeast Saccharomyces cerevisiae (EMBL Accession Z75296); fungi - Aspergillus nidulans [13], Neurospora crassa [14], Magnaporthe grisea (EMBL AA415108); higher plants ­ potato Solanum tuberosum [15], barley Hordeum vulgare [16], rice Oryza sativa (EMBL AB019533), Arabidopsis thaliana (EMBL AB023897) and even mammals ­ N-terminus of FDH from mouse Mus musculus (translated from partial cDNA, EMBL AI505623). By now more than 25 full and partial sequences of NAD+-dependent FDHs or ORFs attributed to FDH are documented compared to 4 known in 1994. Some of them representing various classes of organisms are shown in Fig. 1 (numbering of the amino acid residues throughout the paper refers to that in PseFDH). All the FDHs sequenced to date show strong similarity in primary structures. Identity within a group of organisms (bacteria, yeast, fungi, plants) is usually around 80+ %. However even the distantly related organisms such as FDHs from bacteria and higher plants reveal more than 50 % identity. A specific feature of bacterial sequences is a long N-terminal irregular loop (amino acid residues 12-46) present also in the mouse sequence but substituted by a short 6 residues segment in yeasts, fungi and plants, Fig.1. All the amino acid residues critical for catalysis or coenzyme and substrate binding are highly conserved, Fig. 1. The extent of similarity in and around the active site region is as high as 95 %, all the few substitutions compared to bacteria are for homologous amino acid residues and specific to a group of organisms (e.g. I122V substitution in yeasts or T282N in plants). A notable exception is SceFDH where two conservative prolines, Pro312 and Pro332 are substituted for lysine and valine respectively. FDH falls into the family of D-specific 2-hydroxyacid dehydrogenases acting on Dstereoisomers of respective substrates. Significant sequence similarity has been detected within this family [3,17]. Recent structural studies of D-specific dehydrogenases -


Formate dehydrogenase
PseFDH MorFDH ParFDH MmuFDH PotFDH BarFDH RicFDH SceFDH CboFDH HanFDH MagFDH NeuFDH PseFDH MorFDH ParFDH MmuFDH PotFDH BarFDH RicFDH SceFDH CboFDH HanFDH MagFDH NeuFDH PseFDH MorFDH ParFDH PotFDH BarFDH RicFDH SceFDH CboFDH HanFDH MagFDH NeuFDH LheDLD LbuDHD HmeDGD EcDPGD EfVanH PseFDH MorFDH ParFDH PotFDH BarFDH RicFDH SceFDH CboFDH HanFDH MagFDH NeuFDH PseFDH MorFDH ParFDH PotFDH BarFDH RicFDH SceFDH CboFDH HanFDH MagFDH NeuFDH ..... ..... ..... ..... ............... ............... ............... ............... 1 ..........<---------->..'.........'.........'... ..........AKVLCVLYDDPVDGYPKTYARDDLPKIDHYPGG ..........AKVVCVLYDDPINGYPTSYARDDLPRIDKYPDG ..........AKVVCVLYDDPVDGYPTSYARDSLPVIERYPDG AKILCVLYPDPVDGYPPVYARDSIPYIGGYPDG TRELQASPGPKKIVGVFYKAN------EYAEMN---------SRAAHTSAGSKKIVGVFYQAG------EYADKN---------SRAAHTSAGSKKIVGVFYKGG------EYATKN---------SKGKVLLVLYEGG------KHAEEQ---------...........KIVLVLYDAG------KHAADE---------...........KVVLVLYDAG------KHAQDE---------LTTQREKVKVLLVLYDGG------QHAKDV---------..........VKVLAVLYDGG------KHGEEV---------......'....... QTLPTPKAIDFTPG QTLPTPKAIDFTPG QTLPTPKAIDFVPG QSLATPSAIDFTPG -------------P -------------P -------------P -------------E -------------E -------------E -------------P -------------P 3/10-1A 1 ..'....<------><------------>.68 QLLGSVSGELGLRKYLESNGH ALLGSVSGELGLRKYLESQGH SLLGSVSGELGLRNYLEAQGH ELLGCVSGELGLRxYLEAQGH NFLGCAENALGIREWLESKGH NFVGCVEGALGIRDWLESKGH NFVGCVEGALGIREWLESKGH KLLGCIENELGIRNFIEEQGY KLYGCTENKLGIANWLKDQGH RLYGCTENALGIRDWLEKQGH ELLGTTENELGIRKWLEDQGH ELLGTIQNELGLRKWLEDQGH

443

.SRVASTAARAITSPSSLVF ...AAMWRAAARQLVDRAVG AMWRAAAGHLLGRALG .................... .................... .................... 4 2 <------>........<----------> TLVVTSDKDG-PDSVFEREL ELVVTSSKDG-PDSELEKHL ELVVTSSKDG-PDSELEKHL ELVVTSDKDG-PDSVFEKEL QYIVTPDKEG-PDCELEKHI HYIVTDDKEG-FNSELEKHI HYIVTDDKEG-LNSELEKHI ELVTTIDKDPEPTSTVDREL ELITTSDKEGE-TSELDKHI DVVVTSDKEGQ-NSVLEKNI TLVTTSDKDGE-NSTFDKEL TLVTTCDKDGE-NSTFDKEL 5 6 .<------------------>..<---------- NYLPSHEWARKGGWNIADCV NYIPSHDWARNGGWNIADCV NYTPSHDWAVKGGWNIADCV NFLPGHHQVINGEWNVAAIA NFLPGYQQVVKGEWNVAGIA NFLPGYQQVVHGEWNVAGIA NYNGGHQQAINGEWDIAGVA NFVPAHEQIINHDWEVAAIA NFVPAHEQIISGGWNVAEIA NFVPAHEMIQAGEWDVAGAA NFVPAHEQIQEGRWDVAEAA

3 7 4 8 A 5 ...<------>.....'..<------------>...<-------->'......<------------------>.<-->.....<-------------------------------->163 VDADVVISQPFWPAYLTPERIAKAKNLKLALTAGIGSDHVDLQSAID--RNVTVAEVTYCNSISVAEHVVMMILSLVR HDAEVIISQPFWPAYLTAERIAKAPKLKLALTAGIGSDHVDLQAAID--NNITVAEVTYCNSNSVAEHVVMMVLGLVR HDAEVVISQPFWPAYLTAERIAKAPKLKLALTAGIGSDHVDLQAAID--RGITVAEVTFCNSISVSEHVVMTALNLVR HDADVVISQPF PDLHVLISTPFHPAYVTAERIKKAKNLQLLLTAGIGSDHVDLKAAAA--AGLTVAEVTGSNTVSVAEDELMRILILVR EDMHVLITTPFHPAYVTAEKIKKAKTPELLLTAGIGSDHIDLPAAAA--AGLTVARVTGSNTVSVAEDELMRILILLR EDMHVLITTPFHPAYVSAERIKKAKNLELLLTAGIGSDHIDLPAAAA--AGLTVAEVTGSNTVSVAEDELMRILILLR KDAEIVITTPFFPAYISRNRIAEAPNLKLCVTAGVGSDHVDLEAANE---RKITVTEVTGSNVVSVAEHVMATILVLIR PDADIIITTPFHPAYITKERLDKAKNLKSVVVAGVGSDHIDLDYINQTGKKISVLEVTGSNVVSVAEHVVMTMLVLVR SDADVIISTPFHPAYITKERIDKAKKLKLLVVAGVGSDHIDLDYINQSGREISVLEVTGSNVVSVAEHVVMTMLVLVR EDAEIIITTPFHPGYLSAERLARAKKLKLAVTAGIGSDHVDLNAANKTNGGITVAEVTGSNVVSVAEHVLMTILVLVR EDAEIIITTPFHPGYLTAERLARAKKLKLAVTAGIGSDHVDLNAANKTNGGITVAEVTGSNVVSVAEHVLMTILVLVR A >....'..<---------- SHAYDLEAMHVGTV ARSYDVEGMHVGTV TRSYDIEGMHVGTV HRAYDLEGKTVGTV HRAYDLEGKTVGTV YRAYDLEGKTVGTV KNEYDLEDKIISTV KDAYDIEGKTIATI KDSFDIEGKVIATI KNEYDLEGKVVGTV KNEFDLEGKVVGTV 152 154 153 157 152

AAG AAG AAG GAG GAG GAG GAG GAG GAG AVG GVG GTG GVG GFG GYG GTG

C C 7 B B >. <-------------------------->..<--------> ....<-------------->............<-->..'<------------>248 RIGLAVLRRLAPFDV-HLHYTDRHRLPESVEKELN------------LTWHATREDMYPVC RIGLRVLRLLAPFDM-HLHYTDRHRLPEAVEKELN------------LTWHATREDMYGAC RIGLAVLRRFKPFGM-HLHYTDRHRLPREVELELD------------LTWHESPKDMFPAC RIGRLLLQRLKPFNC-NLLYHDRLKMDSELENQIG------------AKFEEDLDKMLSKC RYGRLLLQRLKPFNC-NLLYHDRLQINPELEKEIG------------AKFEEDLDAMLPKC RIGRLLLQRLKPFNC-NLLYHDRLKIDPELEKEIG------------AKYEEDLDAMLPKC RIGYRVLERLVAFNPKKLLYYDYQELPAEAINRLNEASKLFNGRGDIVQRVEKLEDMVAQS RIGYRVLERLLPFNPKELLYYDYQALPKEAEEKVG------------ARRVENIEELVAQA RIGYRVLERLVAFNPKELLYYDYQSLSKEAEEKVG------------ARRVHDIKELVAQA RIGERVLRRLKPFDCKELLYYDYQPLAPEVEKEIG------------CRRVDNLEEMLAQW RIGERVLRRLKPFDCKELLYYDYQPLSAEKEAEIG------------CRRVADLEEMLAQC HIGQVFMRIMEGFGA-KVIAYDIFKN 179 HIGSGLAEIFSAMGA-KVIAYDVAYN 181 SIGQALAKRAQGFDM-DIDYFDTHRA 180 HIGTQLGILAESLGM-YVYFYDIENK 184 QIGKAVIERLRGFGC-KVLAYSRSRS 179 8 ... '<------340 ISGTTLTA ISGTSLSA ISGTSLSA ISGTTIDA ISGTTIDA ISGTTIDA ISGTSLDA YSGTTLDA YSGSVIDA MSGTSLDA MSGTSLDA

D <-------- DVVTL DVVTL DVVTL DIVVI DVVVI DVIVI DVVTI DIVTV DIVTI EVVTI DVVTI

D >...<------>...<---- NCPLHPETEHMINDE NCPLHPETEHMINDE NCPLHPETEHMVNDE NTPLTEKTKGMFDKE NTPLTEKTRGMFNKE NTPLTEKTRGMFNKE NCPLHKDSRGLFNKK NAPLHAGTKGLINKE NCPLHAGSKGLVNAE NCPLHEKTRGLFNKD NCPLHEKTQGLFNKE

E E 3/10FA F F G G -------->....<---->. . <------><-------------------->'..<---->. ..... . ....<-------->.<---------------->. . TLKLFKRGAYIVNTARGKLCDRDAVARALESGRLAGYAGDVWFPQPAPKDHPWRTMPY-----NGMTPH TLKLFKRGAYLVNTARGKLCDRDAIVRALESGRLAGYAGDVWFPQPAPNDHPWRTMPH-----NGMTPH TLKLFKRGAYLVNTARGKLCDRDAVARALESGQLAGYGGDVWFPQPAPQDHPWRTMPH-----NAMTPH RIAKLKKGVLIVNNARGAIMDTQAVVDACNSGHIAGYSGDVWYPQPAPKDHPWRYMPN-----QAMTPH KIAKMKKGVIIVNNARGAIMDTQAVADACSSGHIAGYGGDVWFPQPAPKDHPWRYMPN-----HAMTPH RIAKMKKGVIIVNNARGAIMDTQAVADACSSGQVAGYGGDVWFPQPAPKDHPWRYMPN-----HAMTPH LISHMKDGAYLVNTARGAICVAEDVAEAVKSGKLAGYGGDVWDKQPAPKDHPWRTMDNKDHVGNAMTVH LLSKFKKGAWLVNTARGAICVAEDVAAALESGQLRGYGGDVWFPQPAPKDHPWRDMRNKYGAGNAMTPH LLKHFKKGAWLVNTARGAICVAEDVAAAVKSGQLRGYGGDVWFPQPAPKDHPWRSMANKYGAGNAMTPH LISKMKKGSWLVNTARGAIVVKEDVAEALKTGHLRGYGGDVWFPQPAPKDHPLRYAKNPFGGGNAMVPH LISKMKKGSWLVNTARGAIVVKEDVAEALKSGHLRGYGGDVWFPQPAPQDHPLRYAKNPFGGGNAMVPH 3/10-9A 9 10 ..<------><-->..<-- PIRDEYLIVQGGAL PIRDEYLIVQGGGL PIRDEYLIVQGGSL DFPAENYIVKDGEL EFPVENYIVKEGEL DFPVQDYIVKEGQL DYRPQDIIVQNGSY DYRPQDIILLNGEY DYRPQDIILLNGKY DYRPQDLIVHAGDY DYSPEDLIVYGGDY 9 ><------------>.................400 AGTGAHSYSKGNATGGSEEAAKFKKAV AGVGAHSYSKGNATGGSEEAAKYEKLDA AGVGAHSYSKGNATGGSEEAAKFKR AP----QYR AS----QYK AS----QYQ AT---RAYGQKK VT---KAYGKHDKK KT---KSYGADK AT---KAYGERAKITKA AT---KSYGERERAKAAAAAAKSA

8 ---------------------------------->.. QARYAAGTREILECFFEGRQTRYAAGTREILECYFEGRQARYAAGTREILECHFEGRQLRYAAGTKDMLDRYFKGEQLRYAAGVKDMLDRYFKGEQLRYAAGVKDMLDRYFKGEQKRYAQGVKNILNSYFSKKF QTRYAEGTKNILESFFTGKF QVRYAQGTKNILESFFTQKF QKRYADGTKAILESYLSGKL QKRYAAGTKAIIESYLSGKH

Figure 1. The alignment of amino acid sequences of formate dehydrogenases from bacteria (marked in green) Pseudomonas sp.101 (PseFDH, SWISS-PROT:FDH_PSESR), Moraxella sp. C1 (MorFDH, EMBL Y13245) and Paracoccus sp. 12-A (ParFDH, EMBL AB071373); mouse (light red) Mus musculus (N-terminal sequence translated from partial cDNA, EMBL AI505623); plants (cyan) ­ potato Solanum tuberosum (PotFDH, EMBL Z21493), barley Hordeum vulgare (BarFDH, EMBL D88272) and rice Oryza sativa (RicFDH, EMBL AB019533); yeasts (yellow) S.cerevisiae (SceFDH, EMBL Z75296), Candia boidinii (CboFDH, EMBL AF004096) and Pichia angusta (HanFDH, former Hansenula polymorpha, EMBL P33677); fungi (magenta) ­ Neurospora crassa (NeuFDH, EMBL L13964) and Magnaporthe grisea (MagFDH, EMBL AA415108). The amino acid sequences of some other D-specific 2-hydroxy acid dehydrogenases in the regions of coenzyme binding domains - D-lactate dehydrogenases from Lactobacillus helveticus (LheDLD, EMBL U07604) and Lactobacillus delbrueckii subsp. bulgaricus (LbuDLD, EMBL X60220), glycerate dehydrogenase from Hyphomicrobium methylovorum (HmeDGD, SWISS-PROT:DHGY_HYPME), 3-phosphoglycerate dehydrogenase from Escherishia coli (EcDPGD, EMBL L29397) and vancomycin resistance protein VanH from Enterococcus faecium (EfVanH, GeneBank M64304) are also shown. Numeration of residues and structural elements correspond to FDH from Pseudomonas sp.101. Residues of FDH active site are shown in bold red. Residues delineating substrate channel are marked in yellow background shading. Blue background shading shows "fingerprint" region and conservative Asp residue in the coenzyme binding domain. Magenta background shading shows Cys residues in PseFDH and CboFDH subjected to mutagenesis to increase chemical stability. Green background shading emphasizes active site residues in the specific protein deviating from a consensus sequence.


444

Vladimir O.Popov & Vladimir I.Tishkov

D-phosphoglycerate dehydrogenase [18], D-glycerate dehydrogenase [19], D-lactate dehydrogenases [20,21] and D-hydroxy-isocaproate dehydrogenase [22] confirmed earlier assumptions and revealed strong similarities in three-dimensional fold between FDH and these proteins, Table 1 (see below section FDH internal symmetry).

3-D structure
Available structures Several high resolution structures of PseFDH are available to date: the apo-enzyme (resolution 1.80 å) [17], the ternary complex of enzyme with NAD+ and azide mimicking putative transition state (2.05 å) [17], and a complex with ADPR (1.50 å) [23]. Several other binary complexes of PseFDH (PseFDH-formate, PseFDH-NAD+, PseFDH-NADH) have been crystallized and preliminary X-ray data obtained. All the complexes had the conformation identical to the structure of apo FDH solved earlier [17]. Neither NAD+, no NADH were detected in the active centers of the solved structures suggesting that for some reasons coenzyme could not bind to the FDH active center in these crystal forms. The nature of this phenomenon is not known, one probable explanation being enzyme inactivation, e.g. oxidation of essential thiol residues that prevents entering of the coenzyme molecule into the FDH active site [3]. Structure of one of the complexes, PseFDH-formate (resolution 2.2 å), was refined to final Rf of 18.4 % [K.Polyakov, personal communication]. In the NAD+-binding pocket of FDH the formate ion was detected near the position occupied by the pyrophosphate moiety of NAD+ in the structure of holo FDH. One of the formate oxygens formed H-bond with the main chain nitrogen of Gly200. A distance between the formate carbon atom and the main chain carbonyl of Asn254 was 3.25 å, suggesting that an H-bond could be formed between formate hydrogen and this residue. No formate ions were detected neither in the putative formate binding site, no in the substrate channel (see below) suggesting that accomplished conformational change and active site rearrangement is required for productive substrate binding. The only structure of FDH from the other source found in the Protein Data Bank is the structure of a putative FDH from hyperthermophilic archaebacterium Pyrobaculum aerophilum (resolution 2.80 å) [1QP8 PDB]. This protein has been expressed in E.coli, had all the methionines substituted to selenomethionines to enable MAD technique for structure solving to be applied and carried a terminal His-tag to enable one-step purification of the protein. However there are serious doubts that the protein in question is a genuine FDH (see below). All the attempts to obtain crystals of FDH from the methylotrophic yeasts, e.g. C.boidinii, suitable for high-resolution structural analysis were up to date unsuccessful. The crystals so far obtained were of poor quality and were either twins that precluded solving of the respective structures or diffracted well above 3 å resolution range. Thus apo and holo high resolution structures of PseFDH obtained earlier remain the structural basis of our present understanding of the enzyme. FDH structure overview Details of FDH structural organisation have been extensively discussed in 1994 review [3]. Here we present the most essential features of FDH structural organization that would be required for further discussions as well as some new findings.


Formate dehydrogenase
+

445

FDH is a typical NAD -dependent dehydrogenase composed of two identical subunits each comprising two domains: a coenzyme binding domain and a substrate binding domain based on Rossmann folds, Figure 2. The two domains are connected via two long a-helices, A and 8. The active center is situated at the domain interface and is formed by residues from only one subunit. FDH undergoes considerable conformational change on cofactor binding, as revealed by a structure of the FDH-NAD+-azide ternary complex, Figure 3. The conformational transition is accomplished via a rotation of a peripheral catalytic domains at an angle of 7.5° around hinges connecting residues 146-147 and 340-341 located in the A and 8 helices respectively. The FDH-ADPR binary complex reveals the same overall conformation as the apo FDH (r.m.s. 0.2 å) with only one minor difference in the region of a short loop Ile122Asp125, where the atoms move more than 1 å [23]. The loop advances towards the

A

B

Figure 2. Structure of PseFDH. A) FDH ternary complex with NAD+ (magenta) and formate (blue) occupying azide binding site. -helices are depicted as red cylinders (left subunit) or helices (right subunit) while -strands as cyan arrows (left) or strips (right). A long loop comprising a/a residues 12-47 present in bacterial FDHs but absent in the enzymes from other species is shown in yellow. B) Representation of the structure of the FDH subunit. Numbering of structural elements from [17].

Figure 3. Displacement of the active site residues upon the transition from apo (yellow) to holo state. Residues in the holo state are colored according to their charge: magenta ­ hydrophobic walls and His332; red ­ negatively charged; blue ­ positively charged and Ile122.


446

Vladimir O.Popov & Vladimir I.Tishkov

enzyme active site in the transition state enabling Ile122 to be implicated in the substrate binding (see below). Thus revealed flexibility of the loop forms an important structural foundation for FDH catalysis. However ADPR does not induce gross structural changes comparable to those found in PseFDH-NAD+-azide ternary complex Figure 3. This suggests that the nicotinamide moiety of NAD+ is the main driving force of the conformational change giving rise to apo-holo transition and essential for transition state formation [23]. FDH internal symmetry. Conserved supersecondary structural motif in FDH and dehydrogenases of D-hydroxyacids It has long been recognized that FDH forms a highly symmetrical structure [3]. Figure 4 presents a stereo view of the two domains of FDH (coenzyme binding and catalytic) superimposed on one another. Both domains of FDH, Figure 2, have a Rossmann fold as a basic structural unit. The topology of the coenzyme binding domain of FDH - A-5-6- A-B- B-C-C-7- D-D-E- E-F- F-G-G (equivalent fragments in both domains are marked in bold underlined) is close to the classical ones found in other NAD+-dependent dehydrogenases, while the core of the catalytic domain 1-1- 4-2- 5-3- 7-4- 8-(insert of coenzyme binding domain)-8 may be regarded as a truncated copy of the coenzyme binding one. The alignment comprises vast stretches of the amino acid sequence and includes the entire -sheet of the FDH domains as well as flanking (A/8) and internal (7/2) helices. Superposition reveals 52 structurally equivalent pairs of C atoms with the r.m.s. deviation of about 1.1 å.

Figure 4. Stereo view of the superposition of the catalytic (thin line) and the coenzyme binding (thick line) domains of FDH.

When structures of the other members of the family of D-specific dehydrogenases of 2-hydroxy acids were made available it became evident (Table 1) that such a symmetrical organization is an intrinsic property of the proteins comprising the family. The catalytic domain in D-specific dehydrogenases shows a strong structural homology to the coenzyme binding domain. A topologically conserved part within D-dehydrogenase superfamily reveals a supersecondary structural motif comprising the 5-stranded lefthandedly twisted parallel -sheet with one complete and one partialRossmann fold units and two -helices, (A/8) and (7/2).


Formate dehydrogenase

447

Table 1. Superposition of the subunits of D-specific dehydrogenases (number of equivalent pairs of C atoms/r.m.s in å) (adapted from [24])
FDH FDH DGDH PGDH * DGDH 223/2.3 * DPGDH 192/2.2 232/1.4 * DLDH /~6.0 216/1.7 (186/1.3) 201/1.5 (151/0.9)

Figures in parenthesis present the alignment with the least r.m.s.

To quantitate structural similarity within the Rossmann-fold domains the parameter of the r.m.s. per aligned pair (R/N) has been suggested (Table 2) [24]. The lower the value of this parameter, i.e. the lower the r.m.s. or the higher the number of equivalent pairs - the closer is the structural relation between the proteins being compared. For very closely related proteins, e.g. apo and holo variants of the same protein the parameter is close to zero (R/N<0.005 å for the alignment of the coenzyme binding domains in apo and holo forms of FDH), while for distantly related structures (r.m.s.>2-3 å, N<20-30) the parameter R/N exceeds 0.1 å and may be considered as a cut-off above which topological similarity becomes questionable. Based on the values of R/N the Rossmann-fold domains comprising NAD+dependent dehydrogenases can be subdivided into at least three structural groups or subclasses of loosening structural similarity within the group: coenzyme binding domains of D-specific dehydrogenases - catalytic domains of D-specific dehydrogenases coenzyme binding domains of L-specific dehydrogenases, Table 2.
Table 2. Superposition of NAD(P)-binding Rossmann-fold domains (adapted from [24])
Structural Domain/ Comparison Coenzyme binding domains of D-specific dehydrogenases (CoE-D) Catalytic domains of Dspecific dehydrogenases (Cat-D) Coenzyme binding domains of L-specific dehydrogenases (CoE-L) Coenzyme binding or catalytic domains of FDH vs. reductases and flafodoxin Number of Equivalent Pairs of C Atoms, N 138±10 r.m.s., å r.m.s./N, å

0.79±0.06

0.006

70±13

1.3±0.3

0.019

53±9

1.3±0.5

0.025

28±6

1.4±0.3

0.051


448

Vladimir O.Popov & Vladimir I.Tishkov

It has been also noted that essential active site residues of NAD+-dependent dehydrogenases acting on L- and D- substrates ­ arginine (Arg284 in PseFDH), histidine (His332), glutamate/aspartate (Gln313) as well as C4 position of the nicotinamide moiety of the coenzyme occupy conservative spatial positions but are interrelated by a symmetry operation reflecting difference in the configuration of a substrate molecule [25]. This observation proves that both D- and L- dehydrogenases use the same type of catalytic machinery however finely tuned to the chirality of the substrate being processed. The only FDH from the other source which structure has been reported to date, the enzyme from archaebacterium P.aerophilum, shows structural organization rather different both from PseFDH, and from the other members of D-specific dehydrogenases family. Alignment of amino acid sequences of P.aerophilum with FDHs from various sources shows only about 20 % of homology contrary to 50 % expected for distantly related FDHs. Also neither a whole molecule, no individual subunits of this protein could be superimposed over the PseFDH (r.m.s > 4 å). Moreover, the two domains comprising P.aerophilum protein did not show any structural similarity as expected among FDH/D-specific dehydrogenases subfamily. Taken altogether these facts cast a serious doubt that the P.aerophilum protein represents an FDH or even a member of the FDH family. Substrate channel The PseFDH active site is deeply buried ~15 å inside the FDH subunit and is accessible to formate anion either through the NAD+ binding site if NAD+ is absent or through a wide channel running from the active center to the surface. The amino acid residues, delineating the substrate channel are presented on Figure 1. They are highly conserved among the FDHs that emphasizes importance of the channel for proper enzyme functioning. Part of the residues comprising the channel are implicated either in substrate binding or catalysis. The channel has a barrel-like form. It is wider in the middle, up to 10-13 å, and more narrow (5-6 A) at the exits. Three side chains comprising part of the FDH active site, Arg284, His332 and to some extent Pro97 form the inner neck of the channel which makes the gateway for the substrate to proceed to the active center. The outer neck opening into the solvent is composed of Lys286, Leu257 and Tyr102. On transition to holo FDH the channel closes, the exit of the channel is additionally shielded from the solvent by the C-terminal loop Asn385-Ser390 and the inner neck separating the interior of the active center from the channel becomes narrower. Thus the interior of the active center appears to be effectively shielded from the bulk. To evaluate a mode of delivery of formate ion to the FDH active site putative substrate pathways were studied by molecular modeling using optimal biased Monte Carlo minimization algorithm (ICM software package). Regularization of the protein structure was performed (hydrogen atoms inserted and geometry of the protein molecule optimized) succeeded by mapping of the substrate- and coenzyme channels. Each channel was divided into separate sections by intersecting spheres 8 å in diameter and a search for local energy minimum for substrate-protein interactions was performed for each section of each channel by an iterative docking procedure. Position of the substrate at the enzyme active site defined as a position of azide in the structure of the holo FDH


Formate dehydrogenase

449

and verified by docking of the formate ion into the FDH active center was regarded as the end-point for both channels. Figure 5 shows energy profiles for substrate-protein interactions for each section of both channels. The benefit of energy for substrate-protein interactions is observed as the substrate advances towards the enzyme active site. For both channels the global energy minimum corresponds to the position of the formate ion in the FDH active center.

Figure 5. Substrate-protein interaction energy profiles for the coenzyme and substrate channels. The bars correspond to the consequent local energy minima of the formate anion while it advances from the bulk into the FDH active center. S marks the position corresponding to location of the formate in the enzyme active site.

In the absence of coenzyme the coenzyme channel may be used by formate to reach the enzyme active site. There are no evident steric obstacles that hinder substrate access to the active center. Simulation performed proved that there are no restrictions from molecule force field as well. Low values of local minima point to easy accessibility of this pathway. However when the coenzyme channel is occupied by NAD+ the only way for formate ion to reach the FDH active center is through the substrate channel. Figure 6 shows consecutive positions occupied by formate ion while traveling into the FDH active site. Two local energy maxima for the putative substrate channel are observed corresponding to the inner and outer necks and one profound local minimum in the middle part (less constraint part) of the channel. The main energy barrier is formed by a gate made of His332 and Arg284 ­ two of the three residues implicated in formate binding (see below). Thus flexibility of the gate and proper local conformation around this site seems to be crucial for enzyme functioning and should be accounted for when interpreting site directed mutagenesis experiments. Structural basis of enzyme specificity FDHs are strictly specific for formate. Geometry of the enzyme active center and especially of the substrate channel precludes access of any bigger molecule to the


450

Vladimir O.Popov & Vladimir I.Tishkov

Figure 6. Trajectories of formate ion movement in the coenzyme and substrate channels. The position of formate with the lowest energy is shown in red. This position coincides with the location of azide anion in the FDH active site.

enzyme active site. As discussed earlier only linear and planar inorganic anions resembling either the substrate (formate) or the product (CO2) are efficient inhibitors of FDHs [3]. Even minor changes in the structure of the substrate molecule, e.g. substitution of the oxygen atom for sulfur (thioformate) results in some cases in the loss of enzymatic activity. Thioformate is a substrate for the yeast enzymes: CboFDH (Vmax ­ 18 % from the activity with formate, Km ­ 2 mM) and HpoFDH (23 % and 11 mM), while is an efficient competitive inhibitor (Ki ­ 1.5 mM) of PseFDH [26]. All studied to date FDHs have a strong preference for NAD+ as the coenzyme. The + + value of coenzyme specificity quantified as (kcat/K)NAD /(kcat/K)NADP varies from >3.109 for SceFDH (nearly absolute specificity) to 2.4.103 for PseFDH (measurable activity with high NADP+ concentrations) (see below Section V Engineering of FDH coenzyme specificity).

III. Solution studies
Main emphasis in the recent studies of FDH was put on modeling, structural characterization and genetic engineering. However some important experiments in solution have been accomplished. Two yeast FDHs from Hansenula polymorpha and Saccharomyces cerevisiae have been purified to homogeneity over the past decade and their kinetic and physicochemical properties studied in detail [27,28]. Studies in solution performed with the enzymes from these species generally confirmed the conclusions drawn earlier for other


Formate dehydrogenase

451

FDHs. Both HpoFDH and SceFDH enzymes operate via ordered Bi-Bi kinetic mechanism with NAD+ binding first [27,28]. Thus FDH from Pseudomonas sp. (PseFDH) remains the only enzyme which inhibition patterns differ from the rest of FDHs and which according to steady-state kinetic analysis operates following random addition of substrates [3]. The origin of pH-dependence of FDH kinetic parameters has been clarified using PseFDH as a sample enzyme [29]. The value of Vmax, rate-limiting hydride transfer, was nearly constant throughout the pH-range of enzyme stability (6.0 to 11.2) while the Km values for both substrates remained constant within the pH range 6 to 10. At pH values below 6 (for the coenzyme) and above 10 (for both substrate and coenzyme) the Michaelis constants increased. In the acidic range the change is attributed to the ionization of two carboxylic groups (pK~5.5-6.0) located at the NAD+-binding site of the enzyme active center. The pH transition in the basic region (pK of 10.5±0.2) has a co-operative conformational origin and affects the enzyme affinity towards substrates and anion inhibitors. Similar transition has been observed in the same work for CboFDH and HpoFDH. Labrou e.a. [30] also attributed pH transition around 10.4 observed for CboFDH to a change in overall protein conformation. The same authors reported another pH-transition around 8.3 for azide binding in the ternary FDH-NAD+-azide complex which was rather close to the pK of 8.2 revealed on kcat/KmHCOO- profile. It is speculated that pK of 8.2-8.3 results from conformational rearrangement of the protein due to formate/azide binding. All the known NAD+-dependent FDHs are homodimers and do not dissociate into constituent subunits under normal conditions. For a long time it was debated if individual FDH subunits could be enzymatically active. Recently it was shown that FDH subunits can be obtained and stabilized in the system of reversed micelles where they display enzymatic activity [31,32]. This observation is fully compliant with the FDH structural model which implies non-interacting independent enzyme active centers and strong-interacting subunit's interface [17]. Conformational mobility and temperature inactivation of PseFDH was studied using CD spectra recorded within 5-90 oC temperature range. FDH revealed regions of enhanced conformational mobility at the predenaturing temperatures (30-55 oC) associated with a change of enzyme kinetic parameters and a co-operative transition around 55-70 oC which was followed by the loss of enzyme activity. Deconvolution of obtained CD spectra showed that the co-operative transition at 55-70 oC in the FDH protein globule was triggered by destruction of the protein -helices which constitute up to 44 % of the enzyme secondary structure [33]. Inactivation mechanism of SceFDH has been studied in detail [34]. It was shown that contrary to majority of other NAD+-dependent FDHs SceFDH is rather temperature labile, inactivates via a two-step mechanism and can be stabilized by high salt concentration or coenzyme binding [34]. Deuterium kinetic isotope effect of reaction catalysed by CboFDH has been determined in [35]. Effect of pressure on deuterium isotope effect has been studied for CboFDH in [36]. It was shown that formation of transition state results in compacting of the protein globule


452

Vladimir O.Popov & Vladimir I.Tishkov

IV. Molecular mechanism
Architecture of the FDH active center as revealed by structural studies is presented in Figure 7. Main features of the FDH catalysis as formulated on this basis several years ago in [3,17] could be summarized as follows:
-

-

-

-

formate ion in the FDH active center is held in place through interactions with at least three residues: it forms a double H-bond with guanidinium group of Arg284 and H-bonds with amide group of Asn146 and backbone amide of Ile122; C4N position of NAD+ is properly positioned and activated through a "twist" and out of plane movement of the carboxamide group which is fixed in trans position (O-atom facing C4N) by multiple H-bonds with Thr282, Asp308, Ser334, Gly335; His332 can participate in substrate binding and is trapped in a non-protonated state by a strong H-bond with amide of Gln313 which is flanked by two conservative prolines fixing its orientation; two hydrophobic walls in the FDH active center are pressed over the reactants as a result of the conformational change in the course of the reaction: one composed of Val150 and Ile202 provides hydrophobic environment for one face of NAD+ pyridine ring, while the other comprising Pro97-Phe98 delineates the substrate-binding pocket.

Several interactions leading either to destabilization of the ground state or stabilization of the transition state were suggested as possible driving forces of FDH catalytic mechanism:

Figure 7. Scheme of the FDH active center. H-bonds are indicated by dashed lines.


Formate dehydrogenase -

453

perturbation of the ground state of both reactants by hydrophobic walls and stabilization of neutral NADH by Val150-Ile202; perturbation by the positive charge of Arg284 of the positively charged nicotinamide moiety of NAD+ in the ground state; "twist" of the carboxamide group of the coenzyme (see above); stabilization of the transition state by stabilizing the migrating negative charge of hydride by Arg284 and favorable interactions of the formate carbon with carbonyl of Ile122.

Since then FDH mechanistic studies mainly concentrated on verification of the suggested molecular mechanism of action either by its probing by site directed mutagenesis of active site residues or by molecular dynamics simulations of ground and transition states.

Molecular modeling
Detailed molecular dynamics simulations of the ground and transition states of the FDH over 2 ns time range using high resolution PseFDH structures as starting points [37,38] generally confirmed the catalytic mechanism of action formulated earlier and enabled better understanding of the possible role of specific amino acid residues comprising enzyme active site in reactants binding and catalysis. Calculations show that in the gas phase as well as in aprotic solvents formation of an ester adduct via nucleophilic addition of formate carboxyl to C4 of NAD+ is favored compared to formation of 1,4-dihydropyridine via hydride transfer. Thus proper positioning of the formate ion with hydrogen atom pointing to C4 of nicotinamide moiety of the coenzyme and preventing formation of unproductive binding of the substrate becomes one of the major tasks to be fulfilled by the FDH active center [37]. All the interactions in the enzyme active center shown in Fig.7 and implicating both the formate ion and the nicotinamide moiety of NAD+ were confirmed. The authors examined the ground state in order to determine the probability of formation of the so called near attack conformations (NACs) of the reactants. NAC is assumed to be a conformation where the distance between C4N of the NAD+ and formate hydrogen is less than 3 å, while attack angles between a plane of nicotinamide and C-H bond in formate are between 132-180 degrees. It was shown that FDH active center acquires NAC in 1.5 % of time throughout the simulation while nicotinamide of NAD+ remains planar in support of the earlier conclusions drawn from the secondary isotope effects [39]. Simulations of the ground state showed that His332 formed persistent H-bonds to the carboxamide group of NAD+ and occasional H-bonds with the formate hydrogen and formate oxygen. In NAC formate hydrogen with a partial negative charge of ­0.1945 a.u. and HE2 of His332 (0.3324 a.u.) approached one another to an average distance of 3.02±0.71 å (minimum distance of 1.46 å). Simulations also confirmed crucial role of Gln313 which is assumed to assist positioning of His332 in a conformation suitable for formate binding and of the hydrophobic wall composed of Pro97-Phe98 which compressive motions towards


454

Vladimir O.Popov & Vladimir I.Tishkov

formate oxygen facilitated attending of NAC and restricted freedom of formate movement in the active center. Simulation gave no evidence of Arg284 interaction neither with formate hydrogen (possible stabilizing role) nor with nicotinamide moiety of the coenzyme (destabilizing role). Examination of the interactions occurring in the transition state showed that they were very much the same as with the ground state. Simulation revealed that on advancing from the ground to transition state a lengthening of the hydrogen bonds between Arg284 and formate (from 1.98-1.86 to 2.32-2.02 å) is coupled with a shortening of the hydrogen bond between His332 and the carboxamide oxygen of NAD+ (from 2.25±0.35 to 1.99±0.19 å). The major events occurring in the FDH active center on transition from ground to transition state as revealed by molecular dynamics simulations are presented in Fig.8. It is assumed that Arg284, Asn146 and Ile122 properly orient formate by electrostatic interactions, while HE2 of His332 can both interact with the formate oxygen and form a tight H-bond with O7N of carboxamide group. On advancing to the transition state the interactions with Arg284 are thus decreased, while with His332 are increased. Simulation also suggested that bulky Phe98 could be a reason for hydride transfer activation barrier in FDH catalyzed reaction. Simulation produced no evidence that the enzyme active center binds reactants in the transition state more efficiently than in the ground state.

Figure 8. Scheme of the putative FDH transition state (adopted from [37,38]).

Site-directed mutagenesis
Site directed mutagenesis experiments were performed on the species and are summarized in Table 3. The aim of the studies was enzyme catalytic mechanism or to increase the enzyme stability coenzyme specificity. The latter two subjects will be dealt in the review, while here only the results essential for interpretation mechanism are discussed. FDH from various either to probe the and/or change its second part of the of FDH catalytic


Formate dehydrogenase

455

Table 3. Mutations performed on formate dehydrogenases from bacteria and yeast.


456

Vladimir O.Popov & Vladimir I.Tishkov

Table 3 continued

Arg284 Arg284 plays a central role in the FDH active center. It is essential both for substrate binding and transition state formation and is important for conservation of the integrity of the enzyme active center. These roles were fully confirmed both for PseFDH [32] and CboFDH [30]. Substitution of arginine for glutamine (a polar residue, about 2 å shorter than arginine, capable of H-bonding) in PseFDH results in the partially active mutant with a diminished formate binding properties. Arg284Gln mutation produces more profound effects on the FDH catalytic constant and azide (transition state analogue) binding than on substrate affinity that could be interpreted as a direct proof of implication of Arg284 in transition state stabilization. Mutation to alanine (small residue, hydrophobic side chain, unable to form H-bonds with the substrate) in PseFDh and CboFDH results in a completely inactive enzyme form. Loss of the enzyme activity on mutation to alanine is attributed to the lack of substrate binding as the ability to bind the cofactor molecule is still retained. Moreover Arg284Ala showed even a 6-fold better binding of NAD+ compared to the wild type PseFDH in accord with the proposed role of Arg284 in destabilization of the enzyme ground state [3]. Several physico-chemical and spectroscopic techniques demonstrated considerable rearrangement of the PseFDH conformation on Arg to Ala substitution emphasizing key role of the residue in maintaining overall integrity of the enzyme active center [33]. Asn146 The role of Asn146 in FDH is ascribed solely to substrate binding both by the structural studies and dynamic simulations. This assumption is fully consistent with the results of PseFDH mutagenesis (Table 3) [40]. Vmax reflecting the rate of hydride transfer is nearly constant (decreases 2-fold), while substrate binding properties are considerably


Formate dehydrogenase

457

impaired. Affinity for the coenzyme decreases 2.2-fold for Asn/Cys and 5.8-fold for Asn/Ala but increases 1.25-fold for Asn/Ser. These effects could be probably attributed to a change of local conformation that affects H-bonding pattern of the coenzyme. Very different results have been obtained for CboFDH. Asn/His substitution exerted dramatic effects both on the enzyme activity and affinity for both reactants. Only trace + catalytic activity was detected (1000-fold decrease), while KmNAD was much more sensitive to mutation than Kmformate. We have to conclude that either extensive conformational rearrangement of the enzyme active center has occurred or there is some intrinsic property(ies) that differ active centers of PseFDH and CboFDH from each other. As pointed out in the previous Sections considerable differences between PseFDH and CboFDH are observed in the kinetic properties including values of the maximal reaction rates, activity towards thioformate and order of substrate binding. It should be remembered that dynamic simulations were performed on PseFDH and if substantial structural differences in organization of the active centers of the enzymes from various sources do occur than the conclusions drawn for PseFDH are not necessarily valid for other FDHs. His332 Mutation of His332 to Phe or Ala results in inactive enzyme in the case of PseFDH [41] while considerable enzyme activity (7 %) is observed for His332Gln mutant in the case of CboFDH (Table 3). The latter mutation also impairs formate binding 10-fold while leaves NAD+ binding properties unaffected. Even higher activity up to 64 % is observed in the CboFDH double mutant Gln313Glu/His332Gln [30]. These observations are in full agreement with the proposed role of essential His in formate binding and transitions state stabilization. Amino acid residues incapable of Hbond formation (Phe, Ala) fail to support these functions, while Gln may still form required H-bonds (Fig. 8) and thus partially substitute His. Pro312-Gln313-Pro314 Glutamine in position 313 substitutes in FDHs a carboxylic acid (Asp in L- and Glu in D-specific enzymes) found in other dehydrogenases acting on 2-hydroxo acids and forming a part of the H+ transfer chain comprising imidazole of a histidine residue and a carboxylic moiety of Asp/Glu. PQPAP is a highly conserved pattern in all the FDHs investigated so far with the exception of SceFDH where Pro312 is substituted for lysine. The role of the prolines flanking essential Gln is to position it in the conformation where amide and not carbonyl is facing His332 residue. This scaffolding role is further emphasized by the fact that Pro312 and Pro314 are the only prolines in FDH present in cys conformation [17]. In FDHs Gln313 forms a tight H-bond with NE2 of His332 and shifts the pK of the essential histidine to the acidic side beyond the pH-staibility range of FDH [41]. A partial proton transfer from glutamine to histidine induces a positive charge on the imidazole and facilitates negatively charged formate binding. The stronger the partial charge, the stronger formate binding should be. Fully protonated His332 (HisH+/Glu- ion pair) would be the best ligand for formate. Absence of a (partial) positive charge would hinder formate binding.


458

Vladimir O.Popov & Vladimir I.Tishkov

Mutation of Gln 313 to glutamate releases the "protonation lock" imposed onto His332. Both in PseFDH and CboFDH Gln/Glu mutation results in the appearance of a new pK of 7.5-7.6 expected for a His-Glu pair, while catalytic constant remains unaffected in a broad pH-range. The affinity of the Gln313Glu mutant towards formate is slightly superior to the wild type at neutral pH but rapidly decreases at basic pH Thus the functional role suggested for Gln313 [40] is to broaden the pH-optimum of the FDH providing a low pK and a partial positive charge on its counterpart, the His332 imidazole over the whole pH-range of enzyme stability (5.5 to 10.5). A change in the precise positioning of Gln313 drastically affected the enzyme activity of CboFDH but had little effect on PseFDH and SceFDH. A double mutation of Pro312Thr/Gln313Glu in CboFDH abolished 97 % of the activity (a single Gln/Glu mutation preserves 87 % of the activity), while single mutations of Pro312 in PseFDH and SceFDH were not essential for enzyme activity. Hydrophobic walls Role of mutations of the residues forming hydrophobic walls of the enzyme active center was probed for CboFDH [29]. As expected Ile202Ala (a wall facing NAD+) mutation mainly influenced the coenzyme, while Phe98Ala mutation (a wall pressing over formate ion) ­ the formate binding leaving coenzyme affinity unaffected (Table 3). According to modeling studies Phe98 could be crucial for determining the rate of hydride ion transfer in FDH (see above Section Molecular modeling). However both mutations resulted in 10-12.5-fold less active enzyme forms and a change of the bulky Phe to a small Ala did not result in improvement of the catalytic properties.

Comparison between L- and D-dehydrogenases
Essential arginine residues have been mutated both in L- [42-45] and D-specific dehydrogenases [46] catalyzing oxidation of 2-hydroxy acids. Results obtained for LLDH from B.stearothermophilus [43-45] unambiguously support conclusions drawn from crystallographic studies and ascribe Arg171 (B.stearothermophilus numbering) a role as a substrate «anchor». Arg171 mutations to lysine, tryptophan or tyrosine resulted in a dramatic (1000-3000-fold) decrease in substrate Km while exerting less profound effects on the turnover rate (only 3-10-fold reduction). Arg109 is another essential arginine of the L-LDH active center. It is located on the mobile loop that closes the enzyme active center on formation of the productive ternary complex, thus shielding it from the solvent [46]. This «mobile arginine» plays a crucial role in the catalytic mechanism of L-LDH and related enzymes through additional polarization of the substrate carbonyl bond [47]. Its role in stabilization of the resulting transition state, originally deduced from structural studies, has been confirmed by sitedirected mutagenesis [42]. D-specific dehydrogenases of 2-hydroxyacids possess only one arginine in their active centers (equivalent of PseFDH Arg284) that can be structurally aligned with the «anchoring» arginine of L-specific enzymes [3,48-51]. To compensate for the absence of the «mobile» arginine while maintaining the basics of the catalytic mechanism formulated for L-specific enzymes, additional functions have been ascribed to this residue. It is suggested that in D-specific dehydrogenases the active site «anchor» arginine assumes a different conformation from that in L-specific enzymes and, in


Formate dehydrogenase

19

addition to binding the substrate, participates in catalysis through polarization of the carbonyl group of the substrate [46,50,51]. Substrate binding functions in D-specific enzymes could also be performed, in part, by a segment of the main chain containing the conserved glycine residue (Gly123 in FDH). In line with these assumptions on a dual role of essential arginine in D-specific dehydrogenases substitution of arginine for lysine or glutamine in D-LDH from L.pentosus results in a substantial (100-600-fold) decrease in both substrate affinity and turnover rate [46]. Several features discriminate FDH among other members of the D-dehydrogenase family. In FDH where no proton transport is required during catalytic turnover the active site histidine (His332) is blocked in only one protonation state through a strong interaction with the amide group of Gln313 which substitutes glutamate found in the active centers of D-specific dehydrogenases [52]. This results in stabilization of a partial positive charge on the imidasole moiety and makes His332 a plausible candidate for substrate co-ordination. Due to the specific structure of the substrate (hydride is abstracted from C1 and not C2 position as in other dehydrogenases of 2-hydroxy acids) FDH should prevent bidentat binding of formate carboxyl to an anion binding center and orient it in such a way that it is ligated by at least two juxtapositioned and opposing each other amino acid residues. These interactions should not be however too strong as excessive binding will result in the elevation of the reaction barrier. If we assume that the chemical foundations of FDH catalyzed reaction are similar to that in other L- and D-specific dehydrogenases of 2-hydroxy acids and additional stabilization of the transition state is required ("mobile" arginine in L-specific and "anchor" arginine in D-specific enzymes) than a candidate for such a role should be found. Molecular dynamics and site directed mutagenesis suggests that in FDH a role of transition state stabilizer may exert His332. This proposition is also in line with the observations that the effects of Arg to Gln substitution in FDH catalyzed reaction (10-30-fold) are much smaller than those in the case of D-LDH catalyzed reaction (600-300-fold). The relative insensitivity of FDH to this mutation compared to D-LDH may result from significant contributions from the other amino acid residues of the FDH active center in particular His332 or the mobile segment comprising Ile122-Gly123, to substrate binding.

Conclusions on the mechanism
Considerable progress in understanding of the FDH molecular mechanism has been made over the past few years. Initially suggested interpretation of the structural and biochemical data and originally proposed molecular mechanism of FDH action has been verified and made more accurate. Systematic application of site-directed mutagenesis methodology to probe various participants of the enzyme catalytic machinery supplemented by molecular dynamic calculations resulted in a concerted picture of FDH catalysis with the fine details that are unavailable for any other NAD+-dependent dehydrogenase to date.

V. Practical application of formate dehydrogenase
The contribution of drugs based on optically pure enantiomers shows a tendency to rise compared to those based on racemic and non-chiral chemicals. Among 500 best


460

Vladimir O.Popov & Vladimir I.Tishkov

selling medications in 2000, the contribution of single enantiomers reached 58% with the sale volume of 107.1 billion dollars [53]. The predicted growth of the chiral drugs market in the next 3 years is estimated as 130 to 172 billion dollars. All dehydrogenases are characterized by the high specificity of hydride transfer from the coenzyme to a substrate and thus, can be successfully used for synthesis of chiral compounds. Fixed orientation of the organic substrate against the nicotinamide ring of the cofactor in the active center of dehydrogenases ensures the stereo-specific hydrideion transfer with extremely high accuracy. For instance, pyruvate reduction catalyzed by L-LDH from porcine muscles yields 1 molecule of D-lactate per 107 molecules of L-lactate [54]. However, dehydrogenase applications based on the use of reduced cofactors, NADH or NADPH, is commercially unfair because of the high price of these reagents. The current bulk price (purchase of more than 1 kg) for 1 mol NADH (709 g) and NADPH (833 g) is $5000 and $39000, respectively. Accounting for a low molecular mass of an optically active product (usually 200-350 Da), the synthesis of 1 kg of the target product will require 3-4 kg of the reduced cofactor. Thus, the production cost of 1 kg of the target product will reach dozens thousand of US $. The problem solution is thought to be in the introduction of an additional enzyme responsible for NAD(P)+ regeneration in situ [55,56]. The general scheme for the synthetic process including the cofactor regeneration system is presented in Figure 9. Dehydrogenase 1 catalyzes the basic reaction of the target optically active product synthesis while dehydrogenase 2 (which in some cases could be the same as dehydrogenase 1) catalyzes NAD(P)+ reduction yielding NAD(P)H.

A

B
Figure 9. Schemes of chiral synthesis using dehydrogenases and coenzyme regeneration system. A ­ general scheme. Main reaction and coenzyme regeneration are catalyzed by different dehydrogenases and B ­ coenzyme regeneration using FDH and formate.


Formate dehydrogenase

461

Various cofactor regeneration enzyme-substrate systems like alcohol dehydrogenase-propanol, glucose dehydrogenase-glucose, etc., were probed for the purposes of chiral synthesis over the past three decades. The summary of this tremendous work can be found in reviews [57,58]. The comparison of various regeneration systems unequivocally demonstrated the superiority of NAD+-dependent FDH from methylotrophic microorganisms (Figure 9B). Only this enzyme meets all the criteria for the universal catalyst of NAD(P)H regeneration: 1. Wide pH-optimum for catalytic activity. FDH activity is unchanged within the range of 5.5-11.0, and Michaelis constants for NAD+ and formate are constant in the range of 6.0-9.5 [29]. It makes FDH applicable for any dehydrogenase-based synthesis. All other dehydrogenases exhibit a narrow pHoptimum for the catalytic activity and cannot be used as a universal catalyst for NAD(P)H regeneration. Providing the maximum yield of a target product. The reaction catalyzed by FDH is irreversible (Figure 9B), and provides the conversion degree of 98100% in all cases studied [57]. Low cost of a substrate for NADP(H) regeneration, the absence of substrate and product inhibition, simplicity of substrate and product removal while purifying the target product. Sodium and ammonium formate are cheap and do not inhibit dehydrogenases catalyzing the basic synthetic reaction. Only one enzyme, i.e. xylithol reductase, is currently known to be inhibited by formate with the constant (Ki 182 mM) comparable to the formate concentrations used in practice [59]. Carbon dioxide, the product of FDH-catalyzed reaction, has no inhibition effect on majority of dehydrogenases. It can be easily removed at lowered pressure and does not interfere with the target product purification. Affordable price and availability of the regeneration enzyme. FDH sources, i.e. methanol-utilizing bacteria and yeast, can be produced in large quantities with methanol as the only source of carbon. The enzyme content under the optimal conditions of cultivation reaches 15-20% of the total soluble cellular protein [60,61]. Enzyme high stability and regeneration after the processing. Yeast and especially bacterial FDHs, (see below) are highly stable and can function in continuous flow membrane reactors for weeks and months [62].

2. 3.

4.

5.

The listed above factors position FDH as almost an ideal candidate for the regeneration of the reduced cofactor. The enzyme disadvantage is a comparatively low specific activity, i.e. 6-7 to 10 U per mg of protein for the yeast [12,63] and bacterial [64] FDHs, respectively. Another drawback is the limited coenzyme specificity of FDH. Unfortunately, there are no NADP+-specific FDH found in Nature so far. The superiority of FDH over the other dehydrogenases ensured its introduction into practice. Currently it is used in a number of large-scale production processes (dozens and hundreds of tons) of synthetic chiral compounds, like the Degussa process of L-tertleucine production [65]). The mostly used is the CboFDH. The cultivation of the original yeast strain has been optimized to give the maximum yield of the biomass and the highest enzyme content [61,66]; the scale-up FDH purification and production method has been developed up to the range of million units [66]. However, the production cost


462

Vladimir O.Popov & Vladimir I.Tishkov

of CboFDH in accordance with the above method is still rather high and limits the enzyme application for chiral synthesis. In this context, we have developed the process of production of NAD(P)H regeneration biocatalysts based on mutant forms of recombinant FDH from Pseudomonas sp.101 expressed in E.coli. The following list of tasks has been solved: 1) 2) 3) 4) Enzyme time/space yield has been increased under the optimized cultivation conditions. Simplified scale-up protocol for the enzyme purification has been developed. Kinetic properties of FDH and its stability toward elevated temperatures and chemical denaturants has been improved. FDH specific to NADP+ has been constructed using protein engineering methods.

All tasks could be solved only in tight connection to each other. For instance, to increase the enzyme yield in the course of cultivation (Task 1) one has to use recombinant E.coli strains providing the production of the target protein at the level of 40-50% of the total soluble protein, i.e. the level that can never be reached with the use of natural strains. The increase in the target enzyme content in the biomass plays an important role for the lowering the purification costs (Task 2). On the other hand, to get the high enzyme content in the biomass as an active protein one needs to enhance its stability (Task 3). The production of recombinant mutant FDH with high thermal stability allowed us to introduce a step of heat treatment of cell-free extract at temperatures >55 into the purification process to remove impurities of E.coli proteins.

FDH source selection
As noted above, all practical applications of FDH for the NADH regeneration purposes were done using CboFDH. To construct a biocatalyst for NADP(H) regeneration we selected PseFDH. The key properties of these two enzymes are summarized in Table 4. Both enzymes exhibit close values of Michaelis constants for formate and NAD+. However, the bacterial enzyme has several important advantages. Firstly, its specific activity is 1.5-1.6-fold higher than that of the yeast enzyme. Secondly, the bacterial enzyme is far more stable than the yeast ortholog (Table 4). Thirdly, PseFDH is stable toward proteases, and this will provide the high yield of the active enzyme under prolonged cultivation of recombinant E.coli strains producing the enzyme. Last but not least, PseFDH is able to catalyze the reaction with NADP+ as a substrate, however with a very low efficiency. This particular property makes PseFDH the better candidate than CboFDH which shows no activity towards phosphorilated coenzyme for the protein design experiments aimed to construct a mutant with the substrate specificity changed from NAD+ to NADP+.

Development of recombinant E.coli strains - producers of FDH
As noted above, wild type strains are unable to provide the high level synthesis of FDH. Therefore, the experiments on FDH gene cloning and construction of genetically engineered strains-producers of recombinant FDHs have been launched.


Formate dehydrogenase

463

Table 4. Comparison of kinetic properties and stability of formate dehydrogenases from yeast C.boidinii and bacterium Pseudomonas sp.101

The first FDH gene cloning was reported for HpoFDH [9], however, the gene had not been expressed in E.coli. In 1991 the PseFDH gene was cloned in our laboratory [6]. The gene was expressed in E.coli cells under the control of lac-promoter [67], however, the expression level was no more than 5-7 %. To increase the expression level, the tandem of two powerful promoters, lac- and tac-, was used [68]. In addition, a number of PseFDH codons non-optimal for E.coli was changed for optimal ones, and the sequence of the ribosome-binding site upstream the FDH gene was optimized as well [69]. As a result, the expression level of active and soluble recombinant PseFDH protein reached the level of 45-50% of the total soluble cellular protein [69]. The cultivation conditions were studied, and the medium was optimized to get the maximum expression level of PseFDH. The process of recombinant enzyme production was scaled-up to the volume of hundreds of liters. The resulting yield of PseFDH under the conditions of large-scale high-density regime cultivation reached at least 10,000 units of activity per L per day [69]. Currently, the enzyme yield was improved even more, up to 30,000 U/L/day. The CmeFDH gene was cloned in 1995 [10]. To express the gene in E.coli cells, tacpromoter was used. The expression level was ca. 15 % of the soluble E.coli proteins. The cloning of CboFDH gene was first reported by Sakai et al in 1997 [11]. Its amino acid sequence differs from that of the CmeFDH in two amino acids only. The authors did not try to express the CboFDH gene in E.coli. Later, Slusarczyk et al. [12] and Labrou and Ridgen [63] in 2000 succeeded in CboFDH expression in E.coli under control of tacpromoter at the level of 20-25% of the total E.coli protein. The enzyme yield was restricted to 1,200-1,500 U per L because of the comparatively low biomass yield, i.e. 4 g per L. The high cell density cultivation of E.coli strain-producer of recombinant CboFDH was reported [70], but no data on the expression level and biomass yield have been presented. The analysis of the electrophoretic data of the cell-free extract shows that the expression level of CboFDH did not exceed 15% of the total protein.


464

Vladimir O.Popov & Vladimir I.Tishkov

The SceFDH was cloned and expressed in E.coli in 2002 by Serov et al. [34,71]. The enzyme was expressed in a soluble active form at the level of 25-30 % of E.coli soluble proteins. The preliminary optimization of cultivation conditions resulted in the enzyme yield of 7,000 ­ 9,000 U per L.

Purification of recombinant FDH expressed in E.coli
The scale-up production protocol for CboFDH purification is based on the method of two-step extractions in two-phase systems comprising water, polyethyleneglycol (PEG) and salts [66]. This approach usually does not result in the enzyme purification in terms of specific activity, if no special affinity reagents are added. The method mainly removes the cell debris and other non-protein cell components. To complete the CboFDH purification, two additional steps, i.e. dialysis and DEAE-chromatography, are needed after the the phase separation steps. The CboFDH activity yield and the purity of preparations were no more than 50-55%. The similar purification protocol has been developed for the recombinant PseFDH expressed in E.coli [69]. To replace the dialysis step, we introduced a hydrophobic chromatography. The activity yield was 70-80% with the purity of 85-95%. This procedure was successfully used to purify 1 million units of recombinant PseFDH (Figure 10).

Figure 10. Purification of 1 million units of recombinant FDH from Pseudomonas sp.101 expressed in E.coli in two phase system.

The CboFDH purification costs are mainly based on the costly polyethylene glycols with different molecular masses. To make the purification protocol cheaper and simpler, we decided to exclude the phase separation steps. Since we have constructed mutant PseFDH with improved thermal stability (see below), to replace two-phase separation step, we introduced one thermal treatment step for the disrupted cell suspension at 62-63 . Figure 11 shows data of SDS-electrophoresis for the original cell-free extract before


Formate dehydrogenase

465

Figure 11. SDS-electrophoresis in 14 % polyacrylamide gel of cell-free extract of E.coli with mutant thermostable FDH from Pseudomonas sp.101 before and after heat treatment at 63 oC at different time intervals. Main band ­ recombinant FDH.

(left lane) and after the heat treatment at 63 . Heat treatment for 20-30 min is sufficient to increase the enzyme purity from 45 to 85-90%. To remove DNA, polysaccharides and pigments, the enzyme preparation was subjected to hydrophobic chromatography. The resultant PseFDH could be stored at +4 at least for 12 months without activity changes. Purification procedure of wild type and recombinant CboFDH in the two phase systems also includes preliminary heat treatment of disrupted biomass, but it is carried out only for 1 min at 55 oC [12,66,70]. For the recombinant CboFDH, affinity chromatography on Procion Red HE3B [70] and Cibacron Blue 3GA [72,73] has been developed. These are high-efficient methods for pure CboFDH production, however, they incur too high costs in the large-scale purification, because the carriers are expensive and elution is performed by NAD+.

Improvement of bacterial FDH stability and kinetic properties
FDH inactivates via two different mechanisms [3]. At temperatures higher than 4045 , thermal denaturation is the main inactivation cause. At temperatures lower than 45 , the loss in enzyme activity proceeds through the chemical modification and oxidation of cysteine residues [74,75]. The nature of inactivation processes dictates two strategies for the improvement of FDH stability, i.e. (1) replacement of essential cysteines for alanine and serine, and (2) site-directed mutagenesis of critical residues to improve the enzyme thermal stability. Improvement of All FDHs of essential for the which only two, globule. Cys255 in the more than FDH chemical stability different sources (bacteria, yeast, plants, etc.) contain cysteine residues catalytic activity. PseFDH contains 7 Cys residues per subunit, among i.e. Cys255 and Cys354, are exposed onto the surface of the protein was replaced for Ser and Met residues [64]. The replacements resulted 200-fold increase of the enzyme chemical stability, however, for the


466

Vladimir O.Popov & Vladimir I.Tishkov

coenzyme increased 3 and 7-fold, respectively. For the PseFDH C255A mutant, the kinetic parameters remained unchanged compared to the wild-type enzyme [76]. The analysis of stability of PseFDH mutants indicated the presence of additional Cys residue in the enzyme molecule [14]. The corresponding mutants C354R, C354S and C354A have been constructed and their properties analyzed [76]. The most appropriate mutation was found to be C354S. This replacement was combined with the C255A mutation. The preliminary data prove that this double mutant of PseFDH exhibits the chemical stability at least 1,000-fold higher than the wild-type enzyme. Site-directed mutagenesis of cysteine residues was also performed for MycFDH [77]. which differs from PseFDH by two amino acid residues only [6]. The authors replaced Cys255 for Ala, Ser and Val residues. In addition, they replaced Cys145 with Ser. The kinetic properties and thermal stability of the obtained mutants have not been studied, however, it has been shown that the mutants are much more stable toward chlorinated compounds [77]. Replacements of cysteine residues in CboF DH, i.e. C23S and Cys262A, also resulted in the improved chemical stability [12]. However, these mutations had worsened the enzyme thermal stability: the thermal inactivation rate constant for the double mutant CboFDH C23S/Cys262A at 50 was 76 times higher than the corresponding rate constant for the wild-type CboFDH. In the case of PseFDH, the double mutation C255A/C354S decreased the thermal stability of the enzyme by 8-fold only. This drop in thermal stability was compensated by additional mutations improving PseFDH thermal stability at elevated temperatures (see below). Increase in PseFDH thermal stability PseFDH is the most stable among all known FDHs so far. Therefore, we could not apply the approach of thermal stability improvement based on the comparison of amino acid sequences for mesophylic and thermophylic proteins. We have checked general recommendations to improve the protein thermal stability as follows: 1) 2) 3) hydrophobization of alpha-helices (replacement of Ser for Ala and Tyr for Phe), introduction of additional Pro residues, optimization of electrostatic interactions.

PseFDH has 5 non-conservative Ser residues in alpha-helices. Single replacements of Ser for Ala in positions 131, 160, 184 and 228 resulted in 13-25% increase of PseFDH thermal stability. Mutant Ser168Ala was 1.7 time less stable than wt-PseFDH. Combination of positive mutations into double Ser(131,160)Ala, Ser(184,228)Ala and four-points Ser(131,160,184,228)Ala mutants showed additive effect in increase of the enzyme thermal stability. Four-points mutant PseFDH had 1.5 times higher stability compared to the wild-type enzyme [78]. Tyr62 (1-helix) and Tyr165 (5-helix) were replaced by Phe. Mutation Tyr62Phe did not influence the enzyme stability. Inactivation rate constant of the mutant Tyr165Phe was 17.6 times higher than one for the wtPseFDH [79]. Proline residue has a rigid conformation and addition of a new Pro can fix protein structure. PseFDH, SceFDH and CboFDH have 25, 12 and 11 Pro residues per subunit, respectively (Figure 1). The most interesting fact is that 7 Pro residues are located in the


Formate dehydrogenase

467

N-terminal loop which is absent in eukaryotic FDHs (Figures 1 and 2). Usually, Pro residues are found in different types of -turns or in the first positions of -helices. Higher Pro content in bacterial FDH is in agreement with its higher thermal stability compared to eucaryotic FDHs. However there is not direct correlation between number of Pro residues and protein resistance against thermal denaturation. For example, all known bacterial FDHs except PseFDH have Pro residue in position 112 (Figure 1). PseFDH has in this position Lys instead of Pro. Mutation Lys112Pro in PseFDH resulted in a decrease of thermal stability by 60%. Increase of rigidity of the protein chain did not compensate loss of the hydrogen bond between Lys112 and Ala109 [80]. SceFDH has glycine in position 141 while all other FDHs have in this position a proline residue (Pro167 in PseFDH) (Figure 1). Substitution of Gly141 by Pro did not result in the increase of the yeast enzyme stability [80]. In this case, one can suppose that there are some other residues responsible for the low SceFDH thermal stability. Slusarczyk et al. reported 10-fold improvement of thermal stability of CboFDH using directed evolution, however mutations resulting in such an effect were not described [81]. Mutation Pro288Thr in CboFDH (this residue corresponds to Pro314 in PseFDH and is located after catalytically important Gln287 residue) produced 18-fold increase of inactivation rate constant [Dr.N.Labrou, personal communication]. At the same time mutation Gln287Glu in CboFDH did not affect the enzyme thermal stability. Mutation His311Gln (His332 in PseFDH) resulted in 1.6-fold improvement of CboFDH stability at 55 oC [Dr.N.Labrou, personal communication]. Electrostatic interactions play important role in thermal stability of PseFDH. Content of charged residues (Arg, Lys, His, Glu and Asp) in this enzyme is more than 30%. These residues produce a complicated network in which one residue participates in interactions with a few other charged residues. Role of the electrostatic interactions in stability of bacterial FDHs was demonstrated in mutation experiments of the residue in position 61 in PseFDH and MycFDH [6, 82]. The enzymes differ in only two amino acid residues among of total 400. Ile35 and Glu61 in MycFDH substitute for Thr35 and Lys61 as in PseFDH [6]. In PseFDH Lys61 interacts with Asp43, while presence of a Glu residue in position 61 results in repulsion between negatively charged Asp43 and Glu61. Different modes of interaction between residues 43 and 61 result in 6-fold higher stability of PseFDH compared to MycFDH. Substitution of Glu61 in MycFDH for Lys provided mutant enzyme with stability similar to PseFDH [81]. Mutation Lys61Arg in PseFDH changed the dependence of the thermal inactivation rate constant Lys61Arg PseFDH mutant was more stable at temperatures below 61 oC, while wt-PseFDH was more stable at temperatures higher 61 oC [82]. Totally more than 60 different mutants of PseFDH have been obtained and studied [83]. Seven of them, providing the positive effect on stability and not affecting kinetic properties, were combined in one multi-site mutant PseFDH 7. The mutant was 20 times more stable than the wild-type enzyme. In addition, some of the mutations increased the enzyme affinity towards the coenzyme. The final mutant, combining 7 "thermal" and 2 "chemical" mutations, PseFDH 72, was significantly superior over the wild-type enzyme both in thermal and in chemical stability. This newly engineered properties of the mutant PseFDH forms were used for the improvement of enzyme purification protocol from E.coli cells by introducing thermal treatment at 62-63 (see above).


468

Vladimir O.Popov & Vladimir I.Tishkov

Wild type FDH is a NAD+-dependent enzyme, however, its specificity for NAD+ over NADP+ strongly depends on the enzyme source (Table 5). The most specific to NAD+ enzyme is SceFDH. The value of the coenzyme specificity of SceFDH for NAD+ + + over NADP+ expressed as a ratio of (k cat/K)NAD /(kcat/K)NADP is more than 3.109 [71] The values of coenzyme specificity for CmeFDH and PseFDH are 250,000 [84] and 2,400, respectively [71]. Thus, PseFDH is the least specific for NAD+ enzyme among all other FDHs. In our laboratory, the experiments to change the coenzyme specificity of PseFDH and SceFDH have been performed. In both cases, the enzymes more specific to NADP+, than to NAD+ have been constructed [71]. For SceFDH and PseFDH, the increase in coenzyme specificity for NADP+ was 9.109 and 104 -fold, respectively [71]. The experiments aimed to change the coenzyme specificity of CmeFDH resulted in a decrease of the enzyme preferency for NAD+ over NADP+ by 600-fold. However the mutant obtained still remained a NAD+-specific dehydrogenase with coenzyme preference for NAD+ by a factor of 410 [84], Table 5.
Table 5. Kinetic properties of recombinant wild-type and mutant formate dehydrogenases from yeasts Saccharomyces cerevisiae and Candida methylica and bacterium Pseudomonas sp.101 (reproduced with small additions from [71])

Engineering of FDH coenzyme specificity

The obtained mutant NADP+-dependent PseFDH exhibited the kinetic parameters very close to those for the wild-type enzyme with NAD+ (Table 5) [71]. This is a rare example of successful changes of the coenzyme specificity of a dehydrogenase. A few successful examples of such coenzyme specificity reversal from NAD+ to NADP+ known in the literature are E.coli dihydrolipoamide dehydrogenase [85], Drosophila alcohol dehydrogenase [86,87], and Thermus thermophilus 3-isopropylmalate dehydrogenase [88]. First mutant variants of NADP+-specific PseFDH exhibited high affinity for the coenzyme only in the pH range of 6.0-7.3. Further experiments allowed us to get the second generation mutants of NADP+-dependent PseFDH with the 6-9 -optimum for the NADP+ binding. The third generation of NADP+-specific PseFDH has constant


Formate dehydrogenase
+

469

for NADP within the pH range of 6-10, which totally covers the optimal pH range for NADP+-specific dehydrogenases used in organic synthesis.

FDH applications for organic synthesis
To date, many processes of enzymatic synthesis of optically active compounds employing NADH regeneration with FDH are described. The details of the processes are not subject of the present work and can be found in reviews and original papers [57,58, 89-98]. Just to mention, the largest commercially realized technological process employing dehydrogenase catalysis, L-tert-leucine synthesis, uses FDH as a cofactor regeneration catalyst [65, 99]. The process is performed in a flow membrane reactor with the in- and out-membranes holding leucine dehydrogenase and FDH inside the reactor. To prevent NAD+ removal from the reactor, the coenzyme was immobilized on the PEG 40,000. To enhance the process efficiency and to reach 100% substrate into product conversion, the cascade of two reactors is used [65,99]. Mutant NADP+-dependent PseFDH of various generations was successfully used for the production of optically active alcohols with different alcohol dehydrogenases [100,101] and in the reaction of Bayer-Villiger for the synthesis of chiral lactones with the use of cyclohexanone monooxygenase [102-104]. Since NADP+-dependent alcohol dehydrogenase and cyclohexanone monooxygenase in particular are not very stable, the further work will be focused on construction of E.coli strains providing simultaneous expression of these enzymes with NADP+-specific FDH as it has been done for the NAD+-dependent enzyme [106,107].

FDH has emerged as one of the most extensively studied enzymes among NAD+dependent dehydrogenases. Comprehensive biochemical information, availability of high resolution structures, cloned and expressed structural genes of a number of FDHs from various organisms, wide potential range of biotechnological applications provide a sound basis for FDH to become a model enzyme for mechanistic studies and practical use. The priorities in studies of FDH could be summarized as follows: To design FDH with improved kcat basing on the understanding of the architecture of its active site and proposed mechanism of action. Crystallization and high-resolution (~1 A) structural characterization of various binary and ternary complexes of FDH from various organisms as well as their genetically engineered mutants in order to understand how subtle changes in FDH structure and organization of its active center affect its catalytic and physico-chemical properties.

VI. Future prospects and challenges

VII. Acknowledgments
Invaluable long-term contribution of Prof.M.-R.Kula in promoting FDH into present day biotechnological practice is acknowledged. The authors would like to thank Dr.Victor Lamzin from EMBL Hamburg Outstation for his continuous co-operation in FDH structural studies, Dr.Nicolaus Labrou from


470

Vladimir O.Popov & Vladimir I.Tishkov

Athens Agricultural University for providing data on CboFDH mutants thermostability prior to publication, Dr.Irene Gazaryan of MSU for valuable discussion and help in manuscript preparation, K.Boyko of A.N.Bakh Institute of Biochemistry for assistance in figure plotting. VOP and VIT thank all former and present colleagues from their laboratories in the A.N.Bakh Institute of Biochemistry and M.V.Lomonosov Moscow State University for their contribution. The work has been supported in part by the following grants: INTAS Grant 94-1309; NATO Linkage Collaboration Grant NATO LST.CLG 977839; Grants from Russian Foundation for Basic Research RFBR -96-04-49927, RFBR -99-04-49156, RFBR 02-04-49415, Grant of The President of Russian Federation RFBR m-96-15-97054; VIT Visiting Professor Grant from Japan Society for the Promotion of Science (JSPS) and VIT Fellowship from Alexander von Humboldt Foundation.

Abbreviations
FDH - formate dehydrogenase, PseFDH - formate dehydrogenase from Pseudomonas sp. 101, MycFDH - formate dehydrogenase from Mycobacterium vaccae N10, CboFDH - formate dehydrogenase from Candida boidinii, SceFDH - formate dehydrogenase from Saccharomyces cerevisiae, HpoFDH - formate dehydrogenase from Pichia angusta (former name Hansenula polymorpha), CmeFDH - formate dehydrogenase from Candida methylica, DPGDH - D-phosphoglycerate dehydrogenase, DGDH - D-glycerate dehydrogenase, DLDH - D-lactate dehydrogenase, L-LDH - Llactate dehydrogenase, ADPR - adenosine diphosphoribose, wt ­ wild type

VIII. References
1. 2. 3. 4. 5. 6. 7. 8. Mathews, M.B., and Vennesland, B. 1950, J.Biol.Chem., 186, 667. Davison, D.C. 1951, Biochem.J., 49, 520 Popov, V.O., and Lamzin, V.S. 1994, Biochem. J., 301, 625 Tishkov, V. I., Galkin, A. G., Marchenko, G. N., Tsygankov, Y. D., and Egorov, A. M. 1993, Biotechnol. Appl. Biochem., 18, 201 Shinoda, T., Satoh, T., Mineki, S., Iida, M., and Taguchi, H. 2002, Biosci. Biotech. Biochem., 66, 271 Galkin, A., Kulakova, L., Tishkov, V., Esaki, N., and Soda, K. 1995, Appl. Microbiol. Biotechnol., 44, 479. Mitsunaga, T., Tanaka, Y., Yoshida, T., and Watanabe, K. 2000, Japan Patent Application JP245471A2 Barnett, M. J., Fisher, R. F., Jones, T., Komp, C., Abola, A. P., Barloy-Hubler, F., Bowser, L., Capela, D., Galibert, F., Gouzy, J., Gurjal, M., Hong, A., Huizar, L., Hyman, R. W., Kahn, D., Kahn, M. L., Kalman, S., Keating, D. H., Palm, C., Peck, M. C., Surzycki, R., Wells, D. H., Yeh, K.-C., Davis, R. W., Federspiel, N. A., and Long, S. R. 2001, Proc. Natl. Acad. Sci. USA, 98, 9883. Hollenberg, C. P., and Janowicz, Z. 1989, European Patent EP1987000110417, Bulletin 89/03. Allen, S. J., and Holbrook, J. J. 1995, Gene, 162, 99 Sakai, Y., Murdanoto, A. P., Konishi, T., Iwamatsu, A., and Kato, N. 1997, J. Bacteriol., 179, 4480. Slusarczyk, H., Felber, S., Kula, M. R., and Pohl, M. 2000, Eur. J. Biochem., 267, 1280. Saleeba, J. A., Cobbett, C. S., and Hynes, M. J. 1992, Mol. Gen. Genet., 235, 349.

9. 10. 11. 12. 13.


Formate dehydrogenase

471

14. Chow, C. M., and RajBhandary, U. L. 1993, J. Bacteriol., 175, 3703 15. Colas des Francs-Small, C., Ambard-Bretteville, F., Small, I. D., and Remy, R. 1993, Plant Physiol., 102, 1171. 16. Suzuki, K., Itai, R., Suzuki, K., Nakanishi, H., Nishizawa, N. K., Yoshimura, E., and Mori, S. 1998, Plant Physiol., 116, 725 17. Lamzin, V. S., Dauter, Z., Popov, V. O., Harutyunyan, E. H., and Wilson, K. S. 1994, J.Mol.Biol.,. 236, 759. 18. Schuller, D. J., Grant, G. A., and Banaszak, L. J. 1995, Nat.Struct.Biol., 2, 69 19. Goldberg, J. D., Yoshida, T., and Brick, P. 1994, J.Mol.Biol. 236, 1123. 20. Stoll, V.S., Kimber, M.S., and Pai E.F. 1996, Structure, 4, 437. 21. Razeto ,A., Kochhar, S., Hottinger, H., Dauter, M., Wilson, K.S., and Lamzin, V.S. 2002, J.Mol.Biol., 318, 109. 22. Dengler, U., Niefind, K., Kiess, M., and Schomburg, D. 1997, J.Mol.Biol. 267, 640. 23. Lamzin, V.S., Popov, V.O., and Wilson, K.S. 1993, Hamburger Synchrotronstrahlungslabor HASYLAB am Deutschen Electronen-Synchrotron DESY, Jahresbericht, 819. 24. Kutzenko, A.S., Lamzin, V.S., and Popov, V.O. 1998, FEBS Lett., 423, 105. 25. Lamzin, V.S., Dauter, Z., and Wilson, K.S. 1994, Nat. Struct. Biol., 1, 281 26. Mezentsev, A.V. 1996, PhD Thesis, A.N.Bach Institute of Biochemistry, Russian Academy of Sciences, Moscow 27. Mesentsev,A.V., Ustinnikova,T.B., Tikhonova, T.V., and Popov,V.O. 1996, Appl.BiochemMicrobiol., 32, 589(Rus). 28. Serov A.E., Popova A.S., and Tishkov V.I. 2002, Doklady Biochem.Biophys., 382(1-3), 26). 29. Mesentsev, A.V., Lamzin, V.S., Tishkov, V.I., Ustinnikova, T.B., and Popov, V.O. 1997, Biochem. J., 321, 475. 30. Labrou, N.E., and Rigden, D.J. 2001, Biochem.J. 354, 455. 31. Klyachko, N.L., Vakula, S.V., Gladyshev, V.N., Tishkov, V.I., and Levashov, A.V. 1997, Biochemistry(Moscow), 62, 1683(Engl.). 32. Trofimova, D.N., and Levashov, A.V. 2002, Bioorg.Khim., 28, 434. 33. Galkin, A.G., Kutsenko, A.S., Bajulina, N.P., Esipova, N.G., Lamzin, V.S., Mezentzev, A.V., Shelukho, D.V., Tikhonova, T.V., Tishkov, V.I., Ustinnikova, T.B., and Popov, V.O. 2002, Biochim.Biophys.Acta, 1594, 136. 34. Serov, A.E. 2002, PhD Thesis, Moscow State University. 35. Xue,H.; Wu,X., and Huskey,W.P. 1996, J.Am.Chem.Soc., 118, 5804. 36. Quirk, D.J., and Northrop, D.B. 2001, Biochemistry, 40, 847 37. Schiott, B., Zheng, Y.-J., and Bruice, T.C. 1998, J.Am.Chem.Soc., 120, 7192 38. Torres, R.A., Schiott, B., and Bruice, T.C. 1999, J.Am.Chem.Soc., 121, 8164 39. Rotberg, N.S., and Cleland, W.W. 1991, Biochemistry, 30, 4068. 40. Matorin, A.D. and Tishkov, V.I. 1998, Abstr. Intern. Conf. "Biocatalysis-98. Fundamentals and Applications", June 1998, Puschino-on-Oka, Russian Federation, 45 41. Tishkov, V.I., Matorin, A.D., Rojkova, A.M., Fedorchuk, V.V. Savitsky, A.P., Dementieva, L.A., Lamzin, V.S., Mezentzev, A.V., and Popov, V.O. 1996, FEBS Letters, 390, 104 42. Clarke, A.R., Wilks, H.M., Barstow, D.A., Atkinson, T., Chia, W.N., and Holbrook, J.J. 1986, Nature, 324, 699 43. Hart, K.W., Clarke, A.R., Wigley, D.B., Waldman, A.D.B., Chia, W.N., Barstow, D.A., Atkinson, T., Jones, J.B., and Holbrook, J.J. 1987, Biochim. Biophys. Acta, 914, 294. 44. Hart, K.W., Clarke, A.R., Wigley, D.B., Chia, W.N., Barstow, D.A., Atkinson, T., and Holbrook, J.J. 1987, Biochem. Biophys. Res. Commun., 146, 346. 45. Luyten, M.A., Gold, M., Friesen, J.D., and Jones, J.B. 1989, Biochemistry, 28, 6605. 46. Holbrook, J., Lijas, A., Steindek, S.J., and Rossman, M.G. 1975, Enzymes, vol. 11 (3rd edn.), P.D. Boyer (Ed.), Academic Press, New York, 191.


472

Vladimir O.Popov & Vladimir I.Tishkov

47. Grau, U.M. 1982, Pyridine Nucleotide Coenzymes, J. Everse, B. Anderson, and K.-S. You (Eds.), Academic Press, New York, 135. 48. Schuller, D.J., Grant, G.A., and Banaszak, L.J. 1995, Nat.Struct.Biol., 2, 69. 49. Goldberg, J.D., Yoshida, T., and Brick, P. 1994, J.Mol.Biol., 236, 1123. 50. Stoll, V.S., Kimber, M.S., and Pai, E. 1996, Structure, 4, 43. 51. Dengler, U., Nie¢nd, K., Kiess, M., and Schomburg, D., 1997, J.Mol.Biol., 267, 640. 52. Blanchard, J.S., and Cleland, W.W. 1980, Biochemistry, 19, 3543 53. Erb,S.E. 2002, Genetic and Engineering News, 22, 47. 54. LaReau, R.D., and Anderson, V.E. 1989, J. Biol. Chem., 264, 15338. 55. Shaked, Z., and Whitesides, G.M. 1980, J. Am. Chem. Soc., 102, 7104. 56. Kula, M.R., and Wandrey, C. 1987, Methods in Enzymology, 136, 9. 57. Hummel, W. and Kula, M.R. 1989) Dehydrogenases for the synthesis of chiral compounds. Eur. J. Biochem., 184, 1 58. Leonida,M.D. 2001, Cur.Medicin.Chem., 8, 345. 59. Neuhauser, W., Steininger, M., Haltrich, D., Kulbe, K.D., and Nidetzky, B. 1998, Biotechnol. Bioeng., 60, 277. 60. Berezin, I.V., Tishkov, V.I., Karzanov, V.V., Avilova, T.V., Egorov, A.M., Petkyavichene, R.I., Vaitkyavichus, R.K., and Glemgha, A.A. 1988. Russian Patent N1479513. 61. Weusterbotz, D., and Wandrey, C. 1995, Process Biochemistry, 30, 563. 62. Wichmann, R., Wandrey, C., Buckmann, A.F., and Kula, M.R. 1981, Biotechnol.Bioeng., 23, 2789. 63. Labrou, N.E., Rigden, D.J., and Clonis,Y.D. 2000, Eur.J.Biochem., 267, 6657. 64. Tishkov, V.I., Galkin, A.G., Marchenko, G.M., Egorova, O.A., Sheluho, D.V., Kulakova, L.B., Dementieva, L.A., and Egorov, A.M. 1993, Biochem.Biophys.Res.Commun., 192, 976. 65. Bommarius,A.S., Schwarm,M., Stingl,K., Kottenhahn,M., Huthmacher,K., & Drauz,K. 1995,) Tetrahedron:Asymmetry, 6, 2851. 66. Weuster-Botz, D., Paschold, H., Striegel, B., Gieren, H., Kula, M.-R., and Wandrey, C. 1994, Chem.Eng.Technol., 17, 131. 67. Tishkov, V.I., Galkin, A.G., and Egorov, A.M. 1991, Dokl. Akad. Nauk SSSR, 317, 745(Engl.). 68. Tishkov, V., Galkin, A.G., Gladyshev, V.N., Karzanov, V.V., and Egorov,A.M. 1992, Biotechnology(Moscow), 52. 69. Tishkov, V.I., Galkin, A.G., Fedorchuk, V.V., Savitsky, P.A., Rojkova, A.M., Gieren, H., and Kula, M.R. 1999,. Biotechnol. Bioeng., 64, 187. 70. Reichert, U., Knieps, E., Slusarczyk, H., Kula, M.R., and Thommes, J. 2001, J.Biochem.Biophys.Methods, 49, 533 71. Serov, A.E., Popova, A.S., Fedorchuk, V.V., and Tishkov, V.I. 2002,) Engineering of coenzyme specificity of formate dehydrogenase from Saccharomyces cerevisiae. Biochem. J., 367, 841-847. 72. Clonis, Y.D., Labrou, N.E., Kotsira, V.P., Mazitsos, C., Melissis, S., and Gogolas, G. 2000, J.Chromatogr. A, 891, 33. 73. Labrou, N.E. 2000, Bioseparation, 9, 99. 74. Popov, V.O. and Egorov,A.M. 1979, Biokhimiia., 44, 207. 75. Dikov, M.M., Osipov, .P., and Egorov, A.M. 1980, Biokhimiia., 45, 1554. 76. Odintseva, E.R., Popova, A.S., Rojkova, A.M., and Tishkov, V.I. 2002, Bulletin of Moscow University, Ser. 2 Chemistry, 43, 356(Rus). 77. Mitsuhashi, K., Yamamoto, H., and Kimoto,N. 2002, European Patent Application EP1211316A1. 78. Rojkova, A.M., Galkin, A.G., Kulakova, L.B., Serov, A.E., Savitsky, P.A., Fedorchuk, V.V., and Tishkov,V.I. 1999, FEBS Lett., 445, 183.


Formate dehydrogenase

473

79. Serov, A.E., Rojkova, A.M., Tishkov, V.I. 2000, Bulletin of Moscow University, Ser. 2 ­ Chemistry, 41, 358. 80. Serov, A.E. and Tishkov, V.I. 2002, Bulletin of Moscow University, Ser. 2 Chemistry, 43, 345. 81. Slusarczyk, H., Pohl, M. and Kula,M.R. 2001, Abstr. Of Enzyme Engineering Conf., October 2001, Potsdam, Germany 82. Fedorchuk, V.V., Galkin, A.G., Yasny, I.E., Kulakova, L.B., Rojkova, A.M., Filippova, A.A., and Tishkov. V.I. 2002, Biochemistry(Moscow), 67, 1145. 83. Tishkov, V.I. 2002, Bulletin of Moscow University, Ser. 2 Chemistry, 43, 380. 84. Gul-Karaguler, N., Sessions, R.B., Clarke, A.R., and Holbrook, J. 2001, Biotechnol. Lett., 23, 283. 85. Bocanegra, J.A., Scrutton, N.S., and Perham, R.N. 1993, Biochemistry, 32, 2737. 86. Chen, Z., Lee, W.R., and Chang, S.H. 1991,. Eur. J. Biochem., 202, 263. 87. Chen, Z., Tsigelny, I., Lee, W.R., Baker, M.E., and Chang, S.H. 1994, FEBS Lett., 356, 81. 88. Chen, R., Greer, A., and Dean, A.M. 1996, Proc. Natl. Acad. Sci. U. S. A , 93, 12171. 89. Kragl, U., Kruse, W., Hummel, W., and Wandrey,C. 1996,. Biotechnol.Bioeng., 52, 309. 90. Patel, R.N. 1999,. J.Am.Oil Che.Soc., 76, 1275. 91. Bae, H.S., Lee, S.G., Hong, .P., Kwak, M.S., Esaki, N., Soda, K., and Sung, M.H. 1999, J. Mol.Cat. B-Enzym., 6, 241. 92. Hummel, W. 1999, Trends Biotechnol., 17, 487. 93. Liese, A. and Filho, M.V. 1999, Cur.Opin.Biotechnol., 10, 595. 94. Orlich, B., Berger, H., Lade, M., and Schomacker, R. 2000, Biotechnol. Bioeng., 70, 638. 95. Patel, R.N. 2000, Adv.Appl.Microbiol., 47, 33/ 96. Bornscheuer, U.T. and Pohl, M. 2001, Cur.Opin.Chem.Biol., 5, 137. 97. Zaks, A. 2001, Curr.Opin.Chem.Biol., 5, 130. 98. Carrea, G. and Ottolina, G. 2002, Biocat.Biotransf., 18, 119. 99. Kragl, U., VasicRacki, D., and Wandrey, C. 1996, Bioproc. Engineering, 14, 291. 100. Seelbach, K., Riebel, B., Hummel, W., Kula, M.R., Tishkov, V.I., Egorov, A.M., Wandrey, C., and Kragl,U. 1996, Tetrahedron Lett., 37, 1377. 101. Kihumbu, D., Stillger, T., Hummel, W.and Liese, A. 2002, Tetrahedron-Asymmetry 13, 1069. 102. Rissom, S., SchwarzLinek, U., Vogel, M., Tishkov, V.I., and Kragl, U. 1997, TetrahedronAsymmetry, 8, 2523. 103. Schwarz-Linek, U., Krodel, A., Ludwig, F.A., Schulze, A., Rissom, S., Kragl, U., Tishkov, V.I., and Vogel,M. 2001, Synthesis-Stuttgart, 33, 947. 104. Zambianchi, F., Pasta, P., Carrea, G., Colonna, S., Gaggero, N., & Woodley, J.M. 2002, Biotechnol.Bioeng., 78, 489. 105. Galkin, A., Kulakova, L., Yoshimura, T., Soda, K., and Esaki, N. 1997, Appl. Env.Microbiol., 63, 4651. 106. Galkin, A., Kulakova, L., Yamamoto, H., Tanizawa, K., Tanaka, H., Esaki, N., and Soda,K. 1997, J.Ferment.Bioeng., 83, 299.