Crystaiiographic studies of the formimino tramferase domain from the bifunctional enzyme formimhotransferase-cyclodeaminase. By Darcy John Reinard KohIs Department of Bioctiemistry McGill University Moatréai Short Title: Structurai studies of the formiminotransferase domain. A thesis submitted to the Faculty of Graduate Studies and Research in partid fdfillment of the requirements for the degree of Master of Science, Q Darcy IR KohIs, August 1999
136
Embed
Crystaiiographic studies the formimino domain fromcollectionscanada.gc.ca/obj/s4/f2/dsk1/tape2/PQDD_0034/... · 2005-02-03 · Crystaiiographic studies of the formimino tramferase
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Crystaiiographic studies of the formimino tramferase domain from the
Short Title: Structurai studies of the formiminotransferase domain.
A thesis submitted to the Faculty of Graduate Studies and Research in partid
fdfillment of the requirements for the degree of Master of Science,
Q Darcy IR KohIs, August 1999
Acquisitions and Acquisibans et Bibiiographic Services services bibiiographiques
The author has granteci a non- exclusive licence allowing the National Liirary of Canada to reproduce, Ioan, distnie or seii copies of this thesis in microform, paper or elecironic formats.
The author retains ownersbip of the copyright in this thesis. Neither the thesis nor substantiai extracts fiom it may be printed or otherwise reproduced without the author's permission.
L'auteur a accordé une licence non exclusive permettant a la Bibliothéque nationale du Canada de reproduire, prêter, disiri'buer ou vendre des copies de cette thèse sous la fonne de microfichelfilm, de reproduction sur papier ou sur format électronique.
L'auteur conserve la propriété du droit d'auteur qui protège cette thèse. Ni la thèse ni des extraits substantiels de celle-ci ne doivent être imprimés ou autrement reproduits sans son autodatio~~
Abstract
The strucnire of the formiminotramferase (FT) domain of
Formiminotransferase-cycIodeaminase (EC 2.12.5) - (EC 4.3.1.4) has been solved in
cornplex with the product analogue, lolinic acid, and in comptex with the substrate
analogue, C5,CIû-dideazatetrahydrofolate. The protein is arranged as a homodimer,
with each subunit comprishg two a /p subdomaius which adopt a novef protein fold.
Wittiin each subunit, an electrostatic tunnel which traverses the width of the molecule
is observeci and comprises the ligand binding sites. The distnïution of charged
residues in the tunnel enables us to propose the mode of binding for the natural
substrate, te~ydropteroylpolyglutamate. ModeIing studies indicate that a y-Iinked
triglutamate form of the tetrahydrofolate substrate can be accommodated thcough the
length of the tunnel. The electron density also indicates ttiat a single molecule of
glyceroI is bound to each protomer at the base of a second tunnei contacthg the main
electrostatic tunnel. This second tunneI is the expected entrance for
formiminoglutamate to the active site of the enzyme. The structure has enabled us to
propose that the residue His82 may be important in ihe cataiytic mechanism of the
transferase reaction,
Résumé
La structure du domaine formiminotrausfera~e (FT) de la
formiminotransferase-cyclodeaminase (EC 2.1.2.5) - (EC 43.1.4) a été résotue en
présence d'un analogue de produit, l'acide folinique, ainsi qu'en présence d'un
analogue de substrat, le C5,ClOdideazatetrahydrofolate. La protéine existe sous la
forme d'un homodimère dont chaque sous-unite est composée de deux sousdomaine
de type a@ qui adoptent un repliement protéique unique à ce jour. On observe, a
t'intérieur de chaque sous-unite, un tunel éléctrostatique qui traverse la molecuIe en
largeur en incluant les sites de liaison des ligands. La distriiution des résidus chargés
le long du le tunel, nous permet de proposer un mode de liaison pour le substrat
naturai: le tetrahydropteroylpolyglutamate. Des études de modélisation on indique
que la forme triglutamate (liaison gamma) du substrat, le tetrahydrofolate, peut être
accomodée dans le sens de la longueur du tunel. La densité éléctronique indique aussi
qu'une moiecule de glycerol est liée à chacun des protomères; elle se retrouve aussi i
la base d'un second tunel qui communique avec le tunel éléctrostatique principal. Ce
second tunel est une entrée vers le site actif probable pour le fomiiminog[utamate.
Cette sûucnire nous permet de propose le résidue His82 comme étant un élément
important dans le rnéchanisme catalitique de la réaction de tramferase.
Acknowiedgments
I'd like to express my heartfelt gratitude to my thesis supervisor, Dr. Mice
Vrielink, for her support, encouragement and guidance in this endeavour. She has
provided me with many opporhmities to expand my scientific abilities. 1 am also
appreciative of her prompt and careh1 reading of this thesis.
1 would also like to thank my colIaborators Dr. Robert E. MacKenzie, Dr.
Enrico Purisima and Traian S u l a for their advice and assistance in this project.
I am indebted to put and present members of tfie VrieIink laboratory for
creating an enjoyable envimament in which to work. Thanks to Jaime Cheah,
Nathdie Cmteau, Paula Lario, Kimberiey Yue, Rakesh Khanna and Enrico Schlieff
who had first wekomed me to the !ab, provided me with technical assistance and
animated scientific discussions. 1 am gratefiil to Dr. Bijan SobeI-Ahvazi for his
Enendship and advice. I woutd aiso hie to acknowledge the camaraderie of Alpesh
Patel, Rupert Abdalian and Meak Chhuom.
1 would dso Iike to thank René Coulombe for translating my abstract into
French as well as pmviding assistance in the laboratory. AIso thanks to Dr. M. Cygler
and members of his group for their assistance in structure soIving.
i am gratefd to Hervé Hogues and Stephane Raymond for providing excellent
technicd support with the cornputers. I wouid ais0 Iike to acknowledge ai i members
of the laboratory of Dr. K. Gehring, in particular, N.S- and A S . for making the
arduous trek to BR1 easier to bear.
I would Iike to thank my famiIy for their inspiration, encouragement and
support all of which is appreciated. They have provideci me with so much
understanding and patience. Thanks to al1 my &ends in Montreal and Vancouver
who have also provided support.
1 would like to express the sùicerest gratitude to my fiancée, Jaime Cheah,
whose love and passion bolsters rny spirit and who provided Iimitless encouragement
and support.
Thanks to the Medical Research Council, Canada (research grant to Dr. A.
Vrielink) for financial support.
Contributions of authors
Portions of the work presented in this thesis have been published in the
following journais:
Crystallization and preliminary X-ray andysis of the formiminotransferase
dornain h m the bifictional enzyme fomiminotransferase£yclodeaminase. Kobls,
D., Croteau, N., Mejia, N., MacKenzie, R.E. & Vnelink, A. (1999). Acta Crysr. D 55,
1206-1208.
The crystal structure of formiminotransferase domain of
fonniminotransferase-cyclodeaminase: implications for substrate channeling in a
synthetase and more recently, E i ,2-oxoisovaIerate dehydrogenase. With tryptophan
synthase the intermediate, indole, is transferred h m the a-site to the P-site through a
25A long tunnel (Hyde et al. 1988). This m effect sequesters the non-polar
intermediate h m the aqueuus environment and increases the efficiency of overall
catalysis. Further cqstallographic studies have revealed confornational changes to
the structure as a result of monovalent cation binding, which affects the interactions
between the a- and P-subunits (Rhee et al. 1996). in addition, studies have s h o w
that the channeling and the coupling of'activities of the two active sites are contmlled
by allosteric signals that cause the two catalytic cycles to occur in phase (Pan et al.
1997).
The structure of carbarnoyl phosphate synthetase h m E. coli reveaIs a tunnel
of 96A through which the enzymatic intemediates are passed between three active
sites (Thoden er al. 1997). This effectively resuhs in intermediate channeling with
100% efficiency as well as protection of the IabiIe intermediates, carboxytphosphate
and carbarnate, from decomposition. Recent structwal studies have shown that the
active sites communkate with each other via domain movements as a result of the
binding of a nucleotide triphosphate (Thoden et al, 1999). EL.2-oxoisovaferate
dehydrogenase aiso shows the presence of a long hydrophobie tunnel where the E2
IipoyI-lysine arm teads to the active site and ailows channeling of the enamîne
intermediate (Ævarsson et al. 1999).
In contrast to the tunnels observeci in the structures of tryptophan synthase,
carbamoyl phosphate synthetase and E l,2-oxoisovaierate dehydrogenase, the structure
of the bihctional enzyme thymidylate synthase-dihydrofolate reductase reveaIs that
the transfer of dihydrofoIate between the active sites occurs by movement of the
Ligand across the surface of the protein (Knighton et al. 1994). The u n d surface
charge distniution is believed to account for the channeiing of the intermediate
beîween active sites. This charged surface Iinking the thymidylate synthase active site
and the dihydrofolate reductase site, 400A away, has been temed the electrostactic
highway (Stroud, 1994).
The crystal structure of FTCD will provide insight into the mechanism by
which this enzyme channels and will provide evidence for the location and number of
Iigand binding sites. information for the cataiytic mechanism of both the transferase
and deaminase reaction cm also be gleaned fiom an analysis of the three dimensiond
structure of FTCD. The anaiysis of the three dimensionaI structure of FT domain
provides an initiai view of the cataiytic mechanism for the tramferase reaction and
provides a basis for enhancing our understanding of the channeling mechanism.
Cbapter two
Materials and Experimentai Methods
2.1. Protein m d h t h o n and concentration
Hexahistidine-tagged formiminotransferase domain was overexpressed using
the pBKe-Cm1 expression vector in Escherichia coli strain BL21tDE3. Purification
was performed as describeci previously (Murley & MacKenzie, 1995) omitting the last
DEAE Sepharose column. The pooled fractions containhg activity for the Fï domain
were then dialyzed into 25mM MOPS pH 8.2, lûrnM K2S04 pH 7.3,35rnM BME and
10% (vlv) glycerol with the addition of ZûmM EDTA to remove any N?+ which may
have leached h m the colurnn. The protein was dialyzed again in the same buffer
excluding EDTA to remove any chelated Pli'' and EDTA. The protein solution was
then concentrated to 8mg m~-' using a Centriprep 10 and Centricon 3 (Amicon, Inc.).
Protein concentrations were detemiied using a Bradford assay with BSA as the
standard.
2.2. Cwstallization with folinic ocid
CrystaIIization conditions were screened by the hanghg-&op vapour diffusion
technique (McPherson, 1990). Protein for these trials was pudied as above. Prior to
these trials, the protein soIution was transferred to a 0.22 pm Eppendorf mter
e (Mïilipore Corporation, Bedford, MA) and filtered by spinning the sampie at hi@
speed (14,000 rpm) in an Eppendorf centrifuge 54132 for 2 minutes at 4OC to remove
any particulate matter which may interfere with crystallization. initial trials using a
sparse-matrix screen, descnbed by Jancarik & Kim (1991), showed a promising
p u l a r precipitate with some precipitants (ammonium suifate, sodium formate,
sodium citrate, polyethylene glycol 2000 and 8000). Further crystallization
experiments around these conditions did not resuit in any significant improvement. A
source of structural heterogeneity is that many proteins have considerable
conformation flexibilty. Such flexibility may act to inhibit crystallbtion. This
flexibility is especially prevalent in muitidomain proteins where the interdomain
contacts may be flexible (Sousa, 1997). It was thought that the introduction of a
ligand to the crystallization trials of FT domain may act to enhance crystailization,
particularly if such conformational flexibility exists. The ligand may 'lock' the
protein in a single conformational state, thus rendering the protein more amenable to
crystalIization. For this reason, it was decided to screen crystailization conditions of
Fï domain in the presence of various substrate and product analogues. in particdar,
the product analogue analogues 2mM folinic acid and 2mM glutamate were added,
either together or independently, to the conditions h m the randorn screen which gave
the most granular precipitates. X-ray diffraction quaIity crystals were obtaïned using
1M Na3citrate, IOOmM Tris pH 8.0 and 15% giycerol as mother Iiquor with the
addition of 2mM foluiic acid to the protein buffer describeci above. Tt should be noted
that some variation in the conditions would still yield crystais. Specifically, the
concentration of Na3citrate couid Vary between 0.95M to 1. tM. The glycerol
concentration could Vary h m 10% to 20%. It was Iater noted that the pH of the
@ mother liquor was approximately 9, despite the pH of the Tris buffer added. It is
thought that the relatively hi& concentration of Na3citrate would saturate any effect
that the Tris buffer would have. Equal volumes of mother liquor and protein were
mixed in a cirop and the ûays were incubated at 290K for two weeks, after which
crystals with a typical size of 0.7 x 0.25 x 0.25 mm and in a fin-like morphology
appeared.
Crystallization conditions of FT domain with substrate andogues were
pursueci in order to obtain a data set that would yield a structure that would give more
information on the binding of a substrate. Crystailization experiments were attempted
with folic acid and C5,ClO-dideazatetrahydrofolate as well as CO-crystdlization
experiments with folinic acid and glutamate together and C5,Clû-
dideazatetraùydrofolate together with formirninogiutamate. Figure 2.1 shows the
chemicaI structures of the compounds used in this study. Since the stnrcture h m the
crystals with folinic acid revealed the binding of glycerol (discussed below),
crystallization experiments were also attempted in the presence and absence of
glycerol. Al1 crystallization experiments were perfomed as describeci in Sections 2.1
and 2.2 with regards to pmtein solution, incubation temperature and the ratio of
mother liquor to protein solution used. Crystais coufd be grown in the absence of
gIycemI and either in the presence of 3mM folinic acid or in the presence of 2mM
C5,C 1 O-dideazatetrahydrofolate. These crystais appeared within two weeks and were
both of smaii size and poor quaiity.
Figure 2.1
Chemicai structures of a) tetrahydrobIate, b) folinic acid and c) C5,CIO-
dideazatetrahydrofolate.
Crystals were also produced in the presence of glyceroi with the addition of
2mM folinic acid and IOOmM giutamate, or, with the addition of 2mM C5,CIO-
dideazatetrahydrofolate. These crystaIs were of a bipyramidal morphology with
largest dimensions obsewed being 0.25 x 0.20 x O.OSmm and generally appear in four
to eight weeks. Crystals with a morphoIogy identicai to crystals obtained in Section
2-2 were observed to grow in the same drops as the bipyramidai crystals. These
crystals have approxirnate dimensions of 020 x 0.15 x 0.15rnrn and appeared in four
to eight weeks.
2.4. Heavv atom derivative screenin~ and prewratikn
Heavy atom derivatives were screened by soaking crystals in a solution of
mother liquor and the heavy atom to be screened. The concentration of the heavy
atom and the duration of the soak were var id in an attempt to optimize conditions to
produce a successfu1 derivative. To test whether a crystai soaked in a particular heavy
atom solution was a successful derivative five to ten X-ray diffraction images were
collected h m the soaked crystai. These images were then processed and scaied to a
native data set using the program SCALEIT h m the CCP4 suite of programs
software (Collaborative Compuwtional Pmject, 1994). Possiile derivatives would be
indicated by having an above 12%. Data collection was aiiowed to continue on
derivatives that showed promisîng RdarvS.
Isomorphous difference Patterson maps and, once an initiai set of phases had
been determineci, difference Fourier maps, were caicuiated h m these heavy atom
derivative data sets using programs h m the CCP4 software (Collaborative
Computational Project, 1994). The coordinates for the position of these heavy atoms
were detemineci h m these maps and relined with MLPHARE, h m the CCP4 suite
of programs. A total of three heavy atom derivatives were found. By çoaking crystaIs
in pchlorornercuriobenzoate suIfonic acid (PCMBS) (FI& Chemie, AG CH-9470
Buchs) at a concentration of ImM for approximately 4 hours, a mercurial derivative
was successhlly obtained. A platinum derivative was obtained by soaking crystais in
mother Iiquor containing LmM K2m(CN)4] (Fluka Chemie, AG CH-9470 Buchs) for
17 to 24 hours. Finally, a goid derivative was obtained by soaking crystals in mother
liquor containing 6mM K[Au(CN)2] (Aldrich Chem Co.) for 40 to 44 hours.
2.5. Duta collection and m a r e determinmion
2.5.1. Crysral in complerr wath/olinic acid
Data were collected at 83K on a MAR image plate detector mounted on a
Rigala RU-200 rotating anode X-ray generator (CuKa radiation). Synchrotron data
were collected at 1.04A for the goid derivative and 1.07A for the high resolution
native data on bearnline X8-C (NSLS, Brookhaven NationaI Laboratory, New York).
The X-ray images were processeci using DEEiZO m the HKL suite of software
(ûtwinowski, 1993; Minor, 1993). ScaIing and merging of the data was continued
with SCALEPACK h m the EML suite of software (Ohrinowski, 1993; Minor,
1993). The intensity data were sorted and structure bcîor amplitudes caldated
using the progams SORTMTZ and TRUNCATE h m the CCP4 suite of software
(ColIaborative Computationd Projeci, 1994). The hi@ resolution native data set,
collected at the synchrotron radiation faciiity was ody 73% comp1ete in the lowest
resolution bin (50.0 - 3.66A) due to spot intensity ovdow. In ordw to complete the
data, the high tesoiution (1 .74 and low resolution (2.8A) data sets were merged
using the program SCALEPACK h m the HKL suite olsoftware.
The structure was soivcd by the rnethod of multiple isomorphous replacement
using 3 heavy atom derivatives. Difference Patterson syntheses were usai to identiîy
the heavy atom positions for the mercuriai derivative and an initid set of phases was
calculated using the program MLPHARE (Collaborative Computationd Project,
1994). Positions of the other heavy atom derivatives (gold and platinun) were
determined h m difference Fourier maps ushg the initiai set of phases h m the
mercuriai derivative. The anornaiou signal h m the gold derivative was obtained
h m data colkcted at the synchrotron facility and used together with the isomorphous
signa1 h m the three derivatives in order to obtain the best set of MIR phases. The
MIR phases were fwther optimized by sdvent flattening and histogram matcbg
using the program DM (Collaborative Computationd Ehject, 1994), with a solvent
content of 50% (assuming two mofecuies per asymmetcic)- The electron density map
calculated h m the improved phases clearly deiineated ttie two protomers in the
asymmetric unit and showed elements of secondary structnre which were related by
ttonnystallographic symmetry (NCS). A prelnnniary mode1 was constructed for a B-
strand and an a-helix in both protomers and the atoms in the mode1 as welI as the
heavy atom positions used to obtain the non-crystallographic symmetry matrices. A
mask was built around one of the protomers and, using the non-crystallographic
symmetry matrices, two-fold averaging was perfomed using the RAVE software
(Jones, 1992; Kleywegt & Read, 1997). The initial mode1 for a single protomer was
built h m the resulting electron density map ushg the program O (Jones et al. 1991).
This Lirst model consisted of 286 of the 328 residues with 77% of the full amino acid
sequence. However, only an alanine side chah was included in the model when the
electron density was unclear.
Crystallographic refinement was initially perfomed with the program XPLOR
(Brünger et al. 1987) and, in later stages, the program CNS was used (Briinger et ai.
1998)- Initially constrained refinernent was carried out to 2.8A resolution. Once the
resolution was extended to 2.2A the constraints were removed and the protomers were
refined as separate molecules with no non-crystallographic symmetry imposed. Each
cycle of refinement was followed by a manual rebuild using the program O (Jones et
ot 1991). SIGMAA weighted maps calculated with coefficients 3Fo-2Fc and Fo-Fc
were used for the model rebuilds. In the h a 1 stages of refinement 2 b F c maps were
used. The difference electron density for the f o h c acid ligand appeared clearer for
one of the two pmtomew however both were included in the model. Figure 2.2
displays the electron density map for the two separate ligands. Water moIecuies were
built where difference electron density above 3a was observed and where hydrogen
bond contacts were made to other polar atoms. In the final stages of refinement,
mulriple conformations for the side chains of 24 residues were modeleci and refined
with CNS. The h a 1 mode1 consists of residues 2 - 326 for one protomer and 2 - 207
0 and 214 - 326 for the second protomer, two molecules of folinic acid and 771 water
Figure 2.2
The difference in the quality of the 2Fo-Fc electron deasity map at 1.7A resolution,
contoured at I.3a, for the folinic acid ligand and giycerol molecule modeled in a)
protomer "A" and b) protomer "B" (produced with the program, SETOR, Evans,
1993).
molecules. Mer the final round of refinement, the program PROCHECK (Laskowski
et al. 1993) was used to calculate a Ramanchadran plot which indicated that al1 of the
residues are in favorable regions of cp/W space (Figure 2.3). Coordinates have been
deposited in the Bmkhaven Protein Data Bank (Bernstein et al. 1977) (accession
number I QD 1).
2.5.2. Crystal in cornplex with CS, CI O-dideazatetrahydrofolate
Data were also collected at 83K on a MAR image plate detector mounted on a
Rigaku RU-200 rotating anode X-ray generator (CuKa radiation) with double
focussing mirrors (Supper Ltd.). The X-ray images were also processeci, scaled and
merged using DENZO and SCALEPACK h m the HKL suite of software
(Otwinowski, 1993; Minor, 1993). The data were sorted and structure factor
amplitudes calculated using the pmgrams SORTMTZ and TRUNCATE h m the
CCP4 suite of software (CoIlaborative Computational Project, 1994). Data were then
convmed to a format usable by CNS (Briinger, 1998). Since the space group is the
same and the ce11 dimensions of this crystal are identical to the crystai complexed
with folinic acid, difference Fourrier techniques was used to solve the structure of the
C5,ClMdeazatetrahydrofolate complexed crystal. Prior to rehement with this
model, the folinic acid, the glyceroi and the water molecules were removed h m both
protomers. in order to prevent model bi s , Ioop regions that either made contact with
the folinic acid, or were near the binding site of folinic acid, were aIso removed h m
Figure 2 3
The Ramanchandran plot (Ramakrishman & Ramanchandran, 1965) of the main-
chain dihedral angles for 558 non-glycine and non-proiine residues modeleci in the
structure of FT domain in cornplex with folinic acid, as calculated by PROCHECK
(Laskowski et al. 1993). GIycine residues are represented as triangles. A total of 5 13
residues (91.9%) are in the most energeacally favoumble regions (A,B,L). A total of
45 residues (8.1%) are in the energeticaity les favourable regions (a,b,l). No residues
fa11 in either of the generously albwed regions or disaIlowd regions.
both protomers. Specificaily, the loop regions removed consisted of residues 37-
48,78-85,137-142, l75-183,224-23 1 and 267-272. Also, the loop region consisting
of residues 208-213 was also removed due to the fact that density could ody be seen
in one of the protomers h m the original model. Refinement was carried out using
the data collected h m the C5,ClO-dideazatetrahydrofolate complexed crystal against
the modified model. Restrained rigid-body refinernent, sirnulateci annealing, least-
squares minimization and group temperature factor refinement were initidly
performed with this new data. SIGMAA weighted electron density maps, calculated
with coefficients 3Fo-2Fc and Fo-Fc, were used for rnodel rebuilding. The loop
regions that were previously removed were rebuilt Uito the new electron density. The
density conesponding to both the C5,ClMideazatetrahydmfoIate molecuIe and
glycerol molecule was present but of fairly poor quaiity, therefore another round of
restrained refinement was performed as descri'bed above. The electron density maps
caIculated h m this refinement resulted in significant hprovements for the ligand
regions of the sûuchue. ïhe mode1 was manually rebuilt again and both the gIyceroI
molecule and C5,ClO-dideazatetrahydroflate molecule were included and a final
mund of refinement was performed. Figure 2.4 shows a Ramachmdran plot which
indicates that most residues lie in favourable regions of q/y space.
2.6 Substrate docking
The (6S)-tetrahydropteroyI-trigiutamte-Nme andogue of the naturai substrate was
@' docked into the binding site of the FT-domain using Sybyl6.5 molecular modeiing
Figure 2.4
The Ramanchandran plot (Rarnakrishman & Ramanchandran, 1965) of the main-
chah dihedral angles for 562 non-glycine and non-proIine residues modeled in the
structxe of FT domain in cornplex with C5,CIO-dideazatehahydrofolate, as
calculated by PROCHECK (Laskowski et al. 1993). Glycine residues are represented
as triangles. A total of 496 residues (88.3%) are in the most energetically favourable
regions (A,B,L). A total of 64 residues (11.4%) are in the energeticaily l e s
favourable regions (a,b,l) and 2 residues (0.4%) fdl are in generoudy ailowed regions.
Glul45, which is one of the residues in the genemusly ailowed region, is slightIy
strained as it is part of a loop structure bridging two short a-helices. The electron
density allows for unambiguous modeling of tbis strained conformation. No residues
are in disalIowed regions.
-1b - 3 5 -40 4s 4'5 Phi (degrees)
software (Tripos, Inc. St, Louis, MO). Structural refinement was performed in Sybyl
6.5 by energy minimization using AMBER 4.1 all-atom force-field with a Powel
minimizer, distance dependent (4r) dielectric constant and an 8A non-bonded cutoff.
The energy tninimization was carried out until the mot-rnean-square of the gradient
was smaller than 0.05 kcal/molk
The coordinates of the protomw with better defined electron density were used as a
starting point for the molecular docking. The folinic acid, glycerol and al1 water
molecules were removed. The hydrogen atoms ad AMBER 4.1 point charges for the
protein part were added with the Biopolymer module in Sybyl 6.5. For consistency
with the force-fieId employed in energy minimization the atomic partial charges of the
substrate molecule were determined on the "fragment-additivity" bais using (6s)-
teîrahydropteroyl and y-linkable glutamate as hgments. Charge calculations were
performed on the neutrai (6s)-tetrahydropteroy1-Nme and negatively charged Ace-
yGlu-Nme rnolecuIes in an extended conformation at the 6-3 IG* ab initio level using
Gaussian 94 (Gaussian, inc., Pittsburgh, PA) without geometry optirnization and with
subsequent fitting to the electrostatic potentiai. Missing atom types as welI as
m d e h e d equilibrium vaIues and force constants for the ligand moiecule were
assigneri by analogy with those parametenzed in the AMBER 4.1 force-field.
Dockhg of the substrate rnolecuie was carried within a "ligand-C~s"' stepwise
protocol. The (6s)-tetrahydroptmyl-Nme rnolecuie was positioned in the binding
site in the sirnilar fashion with corresponding fiagrnent of the crystailized (6R)-folinic
acid and relaxeci in the h e d protein environment. Each of the foliowing y-linkabIe
giutamate units was then joined-up in two steps, 6rst as an aminobutyrate and
subsequently as a complete y-Glu-Nme. The conformation of the added fragment was
selected manuaIIy by considering several stmctural features of the enzyme binding
site such as steric allowance, polarity, H-bonding capabiIities and position of water
moïecules in the original crystal structure as well as conformation strain in the ligand
molecule. Following energy minimization with the protein atoms constrained to their
crystallographic positions the next hgment was added to this docked partid substrate
molecule. M e r accommodation of the complete substrate analogue, four
crystallographic water molecules that allow H-bonding with the ligand molecule were
added and relaxed in the fixed complex environment. Finally, the ligand and water
molecule dong with the protein residues 8A fiom the ligand were dlowed to move
during energy minimization.
Chapter three
Results and Discussion
The formiminotransferase domain with the product analogue, folhic acid,
bound was crystallized in the orthorhombic space gruup P2i212r, with unit ce11
dimensions a+.4A b=103.7A c=1223A (Figure 3.1). Using a molecular mass of
74kDa (for the homodimer) and assuming the presence of one dimer per asymmetric
unit, a Vm value of 2.95A3~a-' was obtained. m-s vaiue is witùin the range observed
by Matthews (1968) and corresponds to a solvent content of 50%.
Crystals with foünic acid w m aiso obtained with identical conditions reported
above with the omission of glycerol. These crystais grew in tight sphencal clusters
and had jagged edges, characteristic of crystaIs growing too fast, where individuai
motecules are unable to orient ttiemselves quickly enough to form an ordered crystaI
Iattice. These crystals were not of X-ray diffraction quaiïty. No fitrttier improvernent
of these crystals was attempted. Perhaps the addition of additives 0th than giycerol
such as plutamate or FIGLU andlot decreasing the CryStaIIization temperature may act
to sIow the qsîahation process, thus improving the quality of crysîais obtained,
S m d bipyramidal crystais were obtained m Simrlar experimentd conditions
as above, except the absence of giyceml and in the presence of 2mM CS,C10-
Figure 3.1
A crystal of FT domain grown in the presence of 2mM (6RIS)-folinic acid and 10%
(vlv) glycerol by vapour d i m o n using the hanging drop method. The crystal size is
approximately 0.7 x 0.25 x 0.25mm.
dideazatetrahydrOpten,yhono~utamate. These crystals appeared within two weeks.
Attempts to mamseed these crystais in order to obtain larger crystals were not
successfiii and streak seeding crystais onIy yieIded more crystals of a similar size.
Two crystal morphologies were obtained using conditions similar to those
with folinic acid and giycerol. In the presence of either 2mM C5,ClO-
dideazatetrahydrofolate or 2mm folinic acid and 80mM glutamate, bipyrarnidal
crystals were grown. A satisfactory data set has yet to be obtained h m crystals with
this morphology. The highest resolution data collected to date for this crystai form is
approximately 3.1& aithough crystals have shown potentiai to diffiact X-rays near
Z ~ L resolution using a synchrotron Iight source.
A complete data set has been collected h m a crystal grown in the presence of
2mM folinic acid and IOOmM glutamate. This data set yielded an ambiguity in the
space group as the data could be successfuIIy processecl in either the orthorhombic
crystal system, with ceIl dimensions of a=10 l.4A b=101 .SA c=l36.1& or the
tetragonal crystal systern, with ceIl dimensions of a=b=10 1 .SA c=136.l A, using the
HKL software (Otwinowski, 1993; Minor, 1993). Upon scaiing the data, the
orthorhombic space group seems more Iikely to be correct as the R-mage obtained is
17.2% as opposed to 26.9% for the tetragond space group. WhiIe these R-merge
values are very large indicating that these data sets will not be usehl for caicuiating
an electron density map, the difference between the two values is significant and may
be used as an indication for the correct space group. Further attempts to work on the
structure will solve this ambiguity as weU as provide more information on the binding
of the substrate, FTGLU. Further work may benefit fiorn screening a different
cryoprotectant as the data collected suffered h m excessive mosaicity (>1.5),
Rod-like crystals were also obtained with FT domain in complex with 2mM
C5,Cl O-dideazatetrahydrofolate. These crystals are of the same space group and ce11
dimensions as those grown with foiinic acid. The resolution bits to which it
d i f i c t s are also similar to that of the crystais grown with f o l k acid. A complete
data set was coIIected to 2.8A resolution. (Further discussed in Section 3.1 1)
3.3. Phase deîermination
ï h e î h t heavy atom to be identified was a mercuriai derivative using the
heavy atom reagent pchloromerc~benzenesulfonic acid (PCMBS). Two Hg-atom
positions were identified h m the Patterson vector superposition method in SHELXS
(Sheldnck et al. 1993). The heavy atorn parameters (coordinates, occupancy and
isotropic- temperature factor) for these two sites were refmed using the program
WHARE h m the CCP4 suite of software. These initiai phase estimates revealed
that the rnercuriai derivative is weakly occupied, as seen by the phasing power 0.91
and the Rd& of 0.79 (see Table 1 and 2 for a summary of statistics). This initiai set
of phases allowed for the determination of the other heavy atom derivative sites usiug
the difference Fourier rnethod.
A second derivative was identifiai utiliang ImM &(Pt(CN)4] soaked for 17
hours (See Table 1). Using the singie isomorphous repIacement (SIR) set of phases
fiom the mercurial derivative, two platinyl sites w m determined h m difference
Table 1
Data Collection Slatistics for Fonniminotransferase Domain complexed witli foliiiic acid.
7 MAR, MAR Rcsearch X-ray plato detector with a double mirror focussing system, mounted on a Rigaku RU200 rotaiing anode gencrator using CuKa radiation. $ SRS, Synchrotron radiation light source al wavelength 1.0397A, beamline X8C Brookhaven National Light Source, Upton, New York. 1 R-nierge = C C - 11 1 C C Ih,, (summed over al1 intensities) 8 R-deriv - C lFderivh - Fnathl/C F nath (resolution range 40A Io 2.8 A) Y In the lowest resolution bin (50.0 A Io 3.66A) data was only 73.1% complete thus the data was scaled and merged witli a low resolution native daln set. *In the lowest resoluiion bin (50.0 A to 3.66A) data is now 94.8% complete.
Table 2
Heavy Alom Refinement Statistics for Formiminoiransferase Domaiii complexed with folinic acid.
1 Resolution Range (A) ( R-cullist 1 Phasing Powerf 1 No. of sites 1 Occupancy
PCMBS 1 1 0.79 0.9 1 2 0.407
1 1 1 1 1 0.230 t R-cullis = Et, (IFPI, f Fpl - FII(CIIIC)I 1 Zt, lFp,l *Fp( $ Phasing power = r.m.s, heavy-atam structure factor / r.m.s. lack oîclosure. Overall Figure of Mcrit is 0.57 using al1 reflections frorn 15.0A lo 2,8A
Fourier qntheses. With the hown positions of the platinum atorns a set of phases
could be calculated with MLPHARE (CCP4 suite, 1994). Two platinyl derivative
data sets were collected with different soak times to obtain differently occupied heavy
atom sites, and slightly different phase information (See Tables 1 and 2).
A third derivative was identified by soaking native crystals in 6mM
K[Au(CN)2] for 40-45 hours. Two gold sites were Iocated by difference Fourier
techniques with phases obtained h m the mercurial derivative. ùiterestingiy, the
coordinates for the gold sites were equivaient to those of the platinum derivative.
These new gold sites were aiso refined as above using MLPHARE (CCP4 suite, 1994)
to give an initial estimate of the phases. The occupancies and B-factors of the gotd
and pIatinum derivatives were aiso reflned sirnultaneously. Possi'bly due to the
difference in occupancies between the gold and platinum derivative, the concurrent
refmernent of the two derivatives yielded satisfactory statistics.
Finally, the heavy atom parameters (coordinates, occupancy and isotropie
temperature factors) for each of the fou. derivatives, were rehed using MLPEMRE
(CCP4 suite, 1994) and an electron density map was calculated. The finai Figure of
Ment (FOM) after heavy atom refinement was 0.50. Upon visualization of this rnap
in O (Jones et al. 1991), as well as the bones atorns comûucted fivm this map, some
areas of secondary structure were visible.
A second gold denvative data set was coiiected, mcluding the anomaious
signai, using a synchrotron light source using a wavelength corresponding to the
absorption edge of gold (1.04A)- This data set was included in the heavy atom
cefiement to give a FOM of 0.57. map was slightiy improved when compareâ
with the electron density rnap caiculated without the anornaIous signal and reveals
larger regions of interpretable stmcture. Further improvement of this map is
discussed in Section 3.4. Figure 3 2 displays the impmvemwts observed in the
Although the electron density map had regions that were clearly interpretable,
density modification methods was undertaken in order to furthe; improve the phases.
To this end, solvent Battening and histogram matching (using a solvent content of
SPA) were perfonned As seen in Figure 3.2 this had significantiy improved the
electron density map and allowed some regions of the mode1 to be built. Since the
asymmetric unit contains two molecules, it was thougbt that two-fold averaging
would dso improve the electron density map. Figure 3.3 shows the mask that was
built around one of the protomers for two-fold averaging- Upon two-fold averaging,
using the RAVE software (Jones, 1992; KIeywegt & Read, I997), a M e r improved
etectron density map was observed, with regions of we11-dehed etectron density for
both the polypeptide main chah and side challis (See Figure 32).
3.5. Model Buildin~ and Retlnement
M e r soIvent flattering, histogram-matcbg and two-fold averaging, the
eIectron density map was of sufncient quality to allow the initial mode1 to be bu&
Figure 3.2
A view at 2.8A resolution of the 3Fo-2Fc electron density map, contoured at 1.3 cr, in
a P-sheet region of the FT domain in complex with folinic acid. The electron denisty
map shows improvernent as denisity modification and averaging techniques are
appIied. The electron density map was calcuiated sequentiaIly h m a) the phases
determined h m the heavy atom refiement with out contribution h m the anmaIous
signai, b) with contriiution of the anomaious signai, c) after solvent flattening and
histograrn matching and d) f ier two-fold averaging.
Figure 3 3
A mask built around se!ected bones atoms for two-fold averaging The mask was
calculated and improved using the program MAMA (Kleywegt & Jones, 1994). The
matrices used in two-fold averaging were improved with the program iMP (Kleywegt
& Jones, 1994)
One protomer was built by fitting the amino acid sequence, as detemined by Murley
& MacKenzie (I995), into the electron density rnap. The other protomer was then
generated by using the NCS matrices. This initial model was subjected to a strict
NCS rehement using XPLOR (Briinger, 1987). SimuIated aanealing, least-squares
minimization and group temperature factor refinement were included. The first
rounds of refinement were canied out using data to 2.8A resolution. SIGMAA
weighted 3Fo-2Fc and Fo-Fc electron density maps were calcutated with the phases
h m the refined model and used for the manual rebuilds.
Once the high resolution (1.74 native data set was coIlected, the refinement
was extended to 2.2A resolution. During these stages, the NCS constraints were
loosened such that the two protomers wouId be refined independently of the other and
the ligands, folinic acid and glycerol, were included in the model. in the Iater stages
of refinement to 1.7A resolution, the NCS restraints were removed, and 2Fo-Fc
electron density maps were used for manual rebuilds. 771 water molecules were
included into the mode1 using differwce Fourier eIectron density maps. In addition,
alternate conformations for 24 side chains were modeled with quai occupamies
assigned. The folinic acid Iigand was assigned hl1 occupancy in both protomem. The
h a 1 refinernent statistics are shown in Table 3. Figures 3.4 and 3.5 give examples of
the electron density d e r the Iast round of refinement.
Table 3
Modei refinement statistics for mode1 with folinic acid.
Resolution Range (A) 50.0 - 1.7 R-factor 19.1 R-free * 213 R.m.s.d. bond Ieagtbs (A) 0.005 R.m.s.d. bond angles (3 1.25 Number of non-hydrogen atoms 5035 Number of water moIecules 77 1 Average B-factors (2) - Overall 2221
- Protein atoms 20.19 - Water rnolecuies 34.69
* 10% of the reflections was used to calculate R-fiee.
Figure 3.4
A s t e m view of the 2Fo-Fc electron density rnap depicting the folinic acid ligand and
glycerol molecule for a single protomer, Some side c h a h and water molecules are
also displayed. The rnap is contoured at 1 . 3 ~ (produced with the program, SETOR
Evans, 1993).
Figure 3.5
A stereo view at 1.7A resolution of the SFo-Fc electron density map depicting the two
conformations of Val303. The map is c o n t o d at 1 . 3 ~ @roduced with the program,
SETOR, Evans, 1993).
The structure of the FT domain forms a homodimer; the two protomers are arranged
such that the dimeric unit adopts a 'W7-shaped morphology (Figure 3.6). The two
protomers withh the dimer are related to each other by a non-crystalIographic two
fold rotation axis. The overall dimensions of each protomer are 50A x 43A x 35A.
The N- and C-termini of each protomer are Iocated in close proximity to each other
but due to the non-crystailographic two fold rotation axis, the termini on one protomer
are located on the opposite face of the molecule than the termini of the second
protomer wiihin the dimeric unit. The coordinates for a singIe protomer were
submitted to the Daii server (Hoim & Sander, 1993) in order to identiQ any
topological similarities with previously identified protein motifs. No significant
stnicturaI sirnilarity was observeci indicating that the FT domain adopts a novel
protein hld.
3. Z The structure of the protomer
The protomer is made up of two a@-units comprishg an N-terminal and a C-
tenninal domain. A topology diagram showing the secondary structure elements in
each domain is shown in Figure 3.7. Figure 3.8 shows an Ca trace of the pmtomer
with numberhg for every 25 residues- Each domain consists of a P-sheet with a-
helices Iocated on the extemal sinface (Figure 3.9). The $-sheet of the N-terminal
domain fxes that of the C-terminai domain to form a double P-sheet layer between
Figure 3.6
Ribbon diagram of the dimer of formirninotransferase domain. The different
protomers are a light and dark grey. The product analogue, folinic acid, is depicted in
a bail-and-stick representation. The dashed lines correspond to residues 208-214
which were not rnodeled due to the poor quality of eIectron density @roduced with
the program MOLSCRIPT, KrauIis, 199 1).
Figure 3.7
Topology diagram for FT domain with p-strands and a-helices numbered in the order
they appear in the primary sequence. The N-terminal domain is depicted in light grey
and the C-terminal is depicted in dark grey. The arrows represent the ~-strands with
their directionality, whiIe the a-heIices are represented as cylinders.
Figure 3.8
A stereo view of the Ca trace for a single protomer of FT domain with every 2 5 ~
residue labelleci. The dark grey trace repments the N-terminal domain and the C-
terminal domain is in light grey. Folinic acid and giycerol are show in a black stick
representation (produced using the program MOLSCRIPT, Krauiis, 1991).
Figure 3.9
Riibon diagram of single protomer of FT domain displayed in stem representation.
The domains are shaded such that light grey represents the N-terminal domain and
dark grey depicts the C-texminal domain (produceci with the program MOLSCRIPT,
KrauIis, 199 1).
the a-helices. The a-helices in the C-terminal domain forrn the bottorn d a c e of the
"V-shaped dimer while those in the N-terminal domain make up the top sides of the
dimer (Figure 3.6). A cIeft making up the binding site for the ligand, folinic acid, is
Iocated between the P-sheets of each domain.
Due to the high remlution of the native data (1.7A) used for crystai1ographic
rehemew the two protomers were refined without irnposing any non-
crystaliographic symmetry (NCS) restraints. Superposition of the alpha carbon trace
for the two protomers, yields a mot-mean-square (r.m.s.) difference of 0.45A between
the 3 18 structuraily homologous Ca atoms indicating no signifiant difference in the
overall fold of the two protomers. Upon superposition of the two protomers, it was
noted chat the best fit was observed in the B-sheet regions of the structure. The
largest differences were found in a number of loop regions of the structure (residues
223 - 232,310 - 314, 318 - 326 and 204 - 214) and at the a4 helix (tesidues 131 -
146). The Ioop region between residues 310 and 314 is involved in dimer contacts.
Movements in this region, away h m exact two fold symmetry, may act to optimize
the interactions between the two protomers. Further differences in the twp region
between residues 223 and 232 may be correiated with the ptedicted
fomiminogIutamate bmdmg site (to be firrther discussed beIow).
The N-terminal domain of the protomer consists of resihes 1-178. It is made
up of a six stranded mixed fi-pleated sheet (pl - 86) and five a-heIices (a l - a5)
(Figure 3.7). Strands BI - B3 are arranged in an anti-pdleI fashion whereas 84 - 86
are parailel. The five a-helices are arranged on the extema1 surface of the p-sheet.
An extended hop between helix a2 and strmd a4 is observe& This loop folk back
over the structure, from the extemal surface of the dimer, m s s the f!-sheet in the N
terminal domain and lies near the giutamate portion of folinic acid. A second
extended loop is seen between residues 128 and 138, on the surface of the molecule
where the cyclodeaminase domain is expected to lie. A glycine cesidue at position
127 and a proline at position 139 enabte the loop to fold back h m the glutamate
portion of the folinic acid ligand, This region of the structure may interact with the
cycIodeaminase domain.
The C-terminal domain consists of residues 182 - 326 and also folds into a
mïxed a@ structure similar to the N-temiinal domain but with a four stranded anti-
parailel P-sheet (B7 - p 10) ( s e Figure 3.7). The topology of this four stranded B-
sheet and two a-helices (a6 and a7) is similar to strands PI - P4 and helices ai and
a2 of the N-terminal domain. A superposition of 67 structurally homologous alpha
carbon atoms comprising the secondary structure elements of this region of the C-
terminal domain with the equivalent region in the N-terminal domain resulted in a
rms. difference of 2.9A. A major difference is seen m the orientations of the h t
heIix in each domain (al and a6) dative to the position of the P-sheet and the
second a-helix. Also, the relative orientations of the Ioop regions between strands B2
and B3 and strands $8 and P9 is significantIy different in the superposition of the two
domains. Finally, the loop region h m residues 260 - 266 is much shorter than the
equivalent bop (residues 73 - 89) in the N-terminal domain; the latter loop extends
across the p-sheet to lie over the foIinic acid ligand Sequence c o ~ s o n s between
residues 2 - 95 of the N-temiinal domaiu and residues 182 - 270 of the C terminal
domain did not show any significaat sequence homology. The h a i two heIices in the
C-terminal domain, a8 and a9, are located near the dimer interface and have residues
involved in intersubunit interaction.
Residues 208 - 214, which are in a Ioop region between a6 and a8, are poorly
defined in the electron density map and codd ody be modeled for one of the two
protomers. The temperature factors in this region of the structure are significantly
higher than observed in the rest of the structure suggesting some conformational
ffexibility in this region of the structure. The C-terminus of the molecule adopts a
short 310-heIix. This region of the structure is the expected entry-point into the
cyclodeaminase domain.
The two domains are separated by a short linker region (residues 179 - 18 1).
The side chain of Ars1 79 makes hydrogen bonding contact to the y-carboxylate group
of folinic acid. The temperature factors in this Iinker region are not significantly
higher than in other regions of the protein chah indicating that the linker is not more
flexible than the remainder of the molecule.
The dirner interface has been implicated as important for the hmction of the
FI' domain, since dissociation of the domain into protomers results in a Ioss of
catdytic activity (Murley & MacKenzie, 1997). Using the program GR4SP
(Nicholls et al. 1991), the buried surface area between the two protomers was
calcuiated to be approximately 1901 A'. The mterface is made up solely of residues
in the C- terminai domain. Three loop regions, between 07 and P6 (residues 189 -
192), P8 and P9 (residues 229 - 230) and a 7 and $10 (residues 260 - 266) and
residues 288 - 316 in the C-terminal a-helix make hydrogen bond contacts as welI as
hydrophobie interactions across the dirner interface. This C-terminal helix is
comprised of a poIar face made up of residues Gln295, G1397, His298, Arg301,
As11305 and Arg306. The side chains of al1 of these residues with the exception of
Arg301 make hydrogen bonding contacts with residues in the two loop regions
between p8 and P9 and behveen a7 and Plo. At the central region of the dimer
where the NCS two-fold symmetry axis is located a pocket of water molecules
rnaking contact to both protomers. Interestingly, a water molecule is present exactly
where the NCS two-fold symmeay axis lies. This waler molecule makes hyd&en
bonding contact with the main chah oxygen atom of Asn3O5 of each protomer as well
as well as two other NCS related water molecules. (Figure 3-10)
3.9. Ligand bindinp sites
The folinic acid binding site lies between the two domains of a protomer and
makes extensive contacts with residues in both domains. Significant diffmces were
observed in the positions of the foiinic acid ligand in the two protomers, particuiarly
the p-aminobenzoyl portion of the iigand. The electron density for the ligand in one
protomer was considerably weaker than obsened in the second protomer. Figure 3.4
shows the electron density for foIinic acid as well as select residues and water
molecules in the vicinity of the ligand in one monomer. IntereStingiy, the co-
q crystallization was carrieci out using a racemic murture of foiinic acid- Since the
Figure 3.10
A stereo view of the dimeric interface with the water moIecule (1abeId wat) at the
NCS two-fold symmetry axis. Some of the residues present at the dimeric interface
are also displayed. (producd using the program MOLSCRIPT, Krauf s, 1991).
physiological substrate for the enzyme is (65')-teûahydrofolate, it was expected that
the 6S enantiomer of folinic acid would bind preferentially. However, to our surprise,
it is clear fiom the electron density maps that the 6R enantiomer of folinic acid
preferentially binds to the enzyme. Attempts to mode1 and refine 6S enantiomer
clearly revealed difference electron density that confinned the presence of (6R)-folinic
acid, Co-crystdlizations were carried out with enantiomencdly pure (6R)-folinic acid
and (6s)-folinic acid (Schircks Laboratorks) using the same conditions as for the
racernic mixture. Ctystals ody appeared with (6R)-folinic acid confirming that the
enzyme that was crystallized has preferentiaI1y selected the 6R isomer of the ligand.
When the protomers are superimposed on each other it is observed that most
of the side chains which interact with the ligand adopt similar conformation in the two
protomers. The contacts between the ligand and the protein side chains differ in a
number cases (see Table 4) due largely to the differences in the position of the ligand
in the active sites of the two protomers.
in the protomer where the electron density for folinic acid is better defined, the
ligand makes 25 hydrogen bonding contacts with the protein and a M e r 7 hydrogen
bonding contacts with water motecdes. In the second protomer, the ligand makes
hydrogen bonding contacts with 24 pmtein atoms and a f i e r 5 hydrogen bonds with
water molecules. The hydrogen bond contacts made between f o h c acid and the
protein are given in Table 4. The teuahydropteridin ring system for (6R)-folinic acid
makes hydrogen bond contacts with the side ch in of Asp39, Ser40, Thr44 and
Glu228 (see Table 4). The carbanyi oxygen of the paminobemoyl moiety makes
e more extensive hydrogen bonds with the protein in the protorner (B) exhibiting the
Table 4 Hydrogen bonding contacts between folinic acid and FT domain.
*Col - glycerol Y - see Fiam 1.3 for atom name reference
Pmtomer "A" Ser4O-O
Gé1i'8-OEZ Wat588-0
Gln268-NE2 -39-ODI
Ser4O-O Thr44-W 1
Wad88-0
Asn 1 CLOD 1
hsnU7-ODI
GoI780*92 Wai829-0
HisBî-NE2 kg46-NH2
W 7 9 D I Go1780-03
AsnlO-ODI His8Z-NE-
Argl42-NH I
Asn 186-ND2 Gln233QEl Gln2684E1 l'y1 26-OH Arg t 79-NE Wat667-0
M 7 - 4 3 D I Asn237-ND2 His82-NE2 Go1780-02 - Wat909-O -
- Ars 142-NH2 Am237-OD I Go178 092 Go178 0.03 AsniO-ODI
*142-NH I Arg l 42-NH2 Asn 1 86-OD I
Tyr 1 Z-ûH Arg 179-NE W&28-û
Tyr 12WH Arg 179-NE Ar1 79-NH2 ws 1 I Z ~
better electron density for the ligand (Table 4). Furthemore, the side chahs of Val48,
His82 and Vd270, and the aliphatic portion of kg46 make favorable van der WaaIs
contacts with the ring of the p-amuiobenzoy1 moiety. Figure 3.11 shows the
interactions made by the folinic acid ligand and the protein molecule for a single
protomer.
During the course of the crystallographic rehement some density of unknown
origin was observed near the p-aminobenzoyl portion of folinic acid. inspection of
both the density and the crystallization conditions suggested that a single glycerol
molecule (10% in the crystallization mixture) was bound to each protomer (see Figure
3.8). As is the case with the folinic acid ligand, the quality of the electron density for
the giycerol molecules differ in the two protomers. Glycerol makes a total of four
hydrogen bond contacts with protein cesidues around the folinic acid binding pocket
(NE2-His82, NH2-Arg142, N-iie222, GSer235). In addition, the glycerol molecuIe
contacts the glutamate carboxyfate group of folinic acid (Figure 3.8).
An inspection of the molecular surface was carried out using the program
GRASP (Nicholls et al. 1 991) (see Figure 3.12 and 3.1 3). From this anaiysis we are
able to visuaiize the folinic acid Ligand buried between the two domains of the
monomer, in a tunnel which spans the width of the protein (see Figure 3.10). The
tunnel is approximately 38A long and 8A wide. The electrostatic surface of the
tunue1 reveals a concentration of negatively charged residues (e-g. Asp39. Glu228) at
the tetrahydropteroyl binding region of the protein and a trail of positively charged
residues (Arg142, Argl72, Lys180 and Lys218) where the y-Linked polyglutamate
moiety of the natural substrate is expected to bind. The folinic acid Iigand contains
Figure 3.1 1
Stereo view of the folinic acid binding site of FT domain. The main chah of the
protein is depicted in a nibon representation. The folinic acid and gLycerol moIecuIes
are disptayed in a dark baI1-and-stick styIe whiIe amino acid residues that make
hydrogen bonding contacts are displayed in Iight ball-and-stick bonds. Water
molecules are dispiayed as spheres, while the hydrogen bond contacts are displayed as
dashed bIack lines (produced using the program MOLSCRIPT, ffidis, 1991).
Figure 3.12
Mdecular surface representatian of the FT domain dimw. The electrostatic potential
of the protomer moIecules are mapped ont0 the surface between -1SkT ( r d ) and
+15kT (blue). Surface accesibk atorns of folinic acid are depicted in yelIow as a
space-fiIling mode1 (produced with the program GRASP, Nicholls, 199 1).
Figure 3.13
The cross-section through the surface representation of one the protomers tevealing
the electrostatic charges (-15 kT is represented in red and +15 kT is represented in
blue) king the tunnel. A backbone trace of the pmtein is represented as tubes, while
the ligands. (6R)-CoIinic acid and giyceroI are sbown in grey stick representation
(produced wi th the program GRASP, Nicholls, 199 1 ).
only a single glutamate group thus the remaining part of the tunnel which would
constitute the expected polyglutamate binding regions is occupied by water
molecules. Inspection of the two sequences of FTCD show that most of the residues
that make up the d a c e of the electrostatic tunnel, ùz particultir, the residues that are
thought to be important in bùiding the polyglutamate tail (specifically AsnlO, Glu128,
Arg179, Glu220, Asn237) are conserveci- This demonstrates the importance of these
residues. Studies by MurIey and MacKenzie (Murley & MacKenzie, 1995) have
s h o w that the predorninant glutamate binding site resides in the CD domain. The
base end of the tunnel, containhg the polyglutamate binding sites, lies near the same
surface as the C-terminus of the FT domain. Entry into the CD domain commences at
residue 334, with a Iinker regions of 8 residues behveen the two domains. Thus, the
location of the polyglutamate binding region should lie near the approximate position
of the CD domain. By docking a substrate andogue we were able to position a total
of three glutamate binding sites in the Rdomain.
Further inspection of the molecular surface revealed the presence of a second
tunnel approximately 9A long that intersects with major tunnel near the p-
aminobenzoyl portion of fohic acid A glycerol molede is located at the base of
tbis shorter second tunnel, where it intersects with the folinic acid tunnel. The
electrostatic surface of the tunnel is only slightIy positiveIy charged in contrast with
the surface of the main tunnel (see Figure 3.13).
3.10. Producc analome versus snbsfrate binding
Using the tetrahydrupteridin ring in the crystal structure as an anchor, we
modeled the substrate analogue, (6S)-tetrahydroptemyItriglutamate-Nme in the
binding site of FT-domain. This moIecuIe has the same chirality at the C6 position as
the natural substrate. As descnied below, this modeled complex appears to have
binding interactions with the protein that are equalIy favourable as those of (6R)-
folinic acid. Why then is the (6R) isorner of folinic acid preferred by the protein?
The answer seems to be due to the fact that folinic acts more like a product analogue
as a result of the presence of the formyl group at NS of the tetrahydropteridin ring.
This presence of the formyl group results in a steric repulsion between C9 and the
formyl oxygen thus destabtizing the bund conformation of the (65') isomer and
decreasing its binding affinity relative to the (6R) isomer which, as the c y t a i
structure shows, does not exhibit any steric repulsion. The substrate analogue is
unsubstituted at NS and is thus more readily accommodateci in the active site without
unfavorable steric interactions. One might raise the objection that the naturai
biosynthetic product of the substrate is in fact the N5-formimino derivative which is
isosteric with the fomyl group in folinic acid. We argue however, that this is not
inconsistent with the nature of the enzyme. ï h e formimino p u p is only present in
the product of the formiminomsferase reactian, Thus, upon product formation, the
unfavorable steric repulsion exhibiteci in the proteidproduct complex acts as a dnWig
force to release the ligand h m the FT binding site. Furthermore, these sterk clashes
wodd prevent the product h m rebinding to the transferase active site and would
drive the channehg of the product to the deaminase active site.
We now describe the main features of the substrate analogue mode1 (69-
tetrahydropteroyl-triglutamate-Nme. The optimum number of glutamates for
channehg of the product to the CDdomain is five, however, ody three can be
accomodated in the main fl tunnel. This suggests that the remaining two
glutamate binding sites residue in the CD-domain.
The overall charge distriiution of the docked substrate is compIemenmy to
that at the surface of the main tunnel of the domain. Positively charged pockets in the
tunnel accommodate negatively charged a-carboxylate groups of the three y-Iinked
glutamyl residues. in contrast, the region of the tetrahycîropteroyl rnoiety which
carries partid positive charges is sequestered within the opposite end of the tunneI
which is negatively charged (Figure 3.13). Molecula. eIectrostatic dipole
calculations, performed with the program GRASP (Nicholls, 1991) on the
uncomplexeri single protomer molecule are striking in that it shows a signifiant
dipole moment (237 Debye) positioned in the tunnel and directed towards the
negatively charged surface near the binding site of the tetrahydropteroyl rnoiety- This
dipole moment is anti-paralle1 to that of the isolateci substrate andogue molecule
calculated for its bound conformation, ïhus, the electrostatics in the tunnel is
expected to aid in the channehg mechanism by guiding the highly charged and
poIarïzed subsûate h m t3e primary glutamate binding site in the CD domain to the
FTdomain where the reaction is to occur.
The docked substrate analogue is predicted to establish a number of favorable
interaction contacts within the tunnel of the FTdomain (Figure 3.14). The
amphiphilic substrate molecule packs in a complernentary fasbion against the
intercalating polar and hydrophobie patches at the tunnel surface. in the mode1 of the
6S substrate analogue the methylene substituent at C6 lies in the equatorial rather than
the axial position as observed in the (6R)-foiinic acid complex. Nevertheles, the
(69-tetrahydropteroyl rnoiety of the docked ligand interacts with the protein residues
in a fashion which is sirnilar to that of the (6R)-tetrahydropteroyl part of the
crystailized folinic acid. The hydrogen bond interactions of the tetrahydroptendin
rnoiety with Asp39 and Glu228 as well as with a burieci moIecule are preserved (see
Figures 3.8 and 3.1 1). Energy minimization ailowed formation of a novel hydrogen-
bond between NI of the tetrahydropteridin ring and Gin268 residue.
The p-aminobenzoy1 fragment of the substrate anaiogue undergoes a
translation of approximateiy 1.8A W e r into the tunnel relative to its position as
seen in the compIexed folinic acid. This translation is a direct result of the more
extended structure of the equatorial versus axial conformation at the tetrahydropteroyl
C6 atom. in fact, we observed some W o m in the accommodation of the p-
aminobenzoyl p u p of the folinic acid in the two protomers.
The y-linked triglutamate part of the substrate is predicted to bind in an
extended conformation which follows the nanow channel of the enzyme domain
Translation of the paminobenzoyI fiagrnent alters the binding mode of the first
giutamate in the substrate relative to that in the folinic acid. in the substrate analogue,
the a-carboxylate of the first glutamate occupies the mean position of the y-
Figure 3.14
Stereo view of the enzyme in the region of the docked substrate, (6s)-
tetrahydropteroyltrigiutamate-Nme, ï he main chain of the protein is depicted in a
nibon representation. The substrate analogue is displayed in a dark ball-and-stick
style while arnino acid residues that make hydrogen bonding contacts are displayed in
tight ball-and-stick bonds. Water molecules are displayed as sphetes, while the
hydrogen bond contacts are displayed as dashed black lines (produced using the
program MOLSCRIPT, Ktaulis, 199 1 ).
carboxylate and a buried water molecule in the folinic a c i m domain complex. This
change in binding mode is quite reasonable, given that the y-carboxylate in folinic
acid becomes a y-linked amide in the substrate analogue. The a-carboxylate p u p
rnakes hydrogen-bond interactions with the side chains of AsnlO and kg179 as well
as with a buried water molecule. This water molecule is also hydrogen-bonded ta the
arnide NH p u p of the b t glutamate. The y-amide carbonyl interacts with the side
c h a h of Tyr126 and Arg179. in addition to the polar interactions, there is also a very
good hydrophobic packing between the aliphatic portion of the fïrst glutamate and the
side chains of Arg142 and k g 4 6 are no longer involved in salt-bridge interactions
with a-carboxylate group of the hrçt glutamate of the substrate. These positively
charged residues are now free to interact with the second substrate,
forniminoglutamate, and to stabilize a tetrahedral intermediate that woufd be fonned
during the transier of the formirnino group.
The second glutamate residue of the docked substrate analogue is also mostly
bwied in the putative binding tunnel. Its a-carboxylate makes hydrogen-bond
contacts with the side chab of Gln220 and the main chah MI group of Leu238. The
arnide NH groq interacts with a burieci water molecule. The y-amide carbnyl is
partially exposed to the solvent. The aliphatic portion of the second glutamate makes
hydrophobic interactions with the side chahs of Leu239, Leu182 and the aliphatic
portion of the k g 1 79 side chain.
The third giutamate residue is the most solvent exposed part of tûe modeIed
substrate complex. Its a-carboxylate group is hydrogen-bonded to the backbone
amides of Lys180 and Glu128, and is positioned in close proximity to the ammonium
group of Lysl80. The amide NH group interacts with a buried water molecule
whereas the y-amide carbonyf makes hydrogen-bond interaction with the side chah of
Lys 180.
3.1 1. Catafvtic mechanism
The reaction catalyzed by FTCD (Figure 1.2) transfers the formimino group
from formiminoglutamate to tetrahydrofalate and subsequently carries out a
cyclodeamination to give NS,N10-methenyl-THF- The precise mechanism for the
reactions and the residues important for catalysis and substrate binding are not yet
known. The structure of FTdomain provides us with a first view of the enzyme
active site and enable us to identify potential residues which may be implicated in the
catalytic mechanism. The N5 atom of the substrate, the nucleophile expected to
attack the irnino carbon of formiminoglutamate, is completely buied h m the
external surface of the protein. Such a buried environment will protect the labile NS-
formimino-tetrahydrofolate product of the FT reaction h m hydrolpis. The presence
of the bound glycerol molecule (a close mimetic of the product glutamate) at the base
of the second tunnel suggests that this short tunnel may be a route through which the
formiminoglutamate substrate enters and the glutamate product Ieaves. However,
examination of this route shows that access to N5 of the tetrahydropteridin ring is
bIocked by the side chain of His82- The presence of a histidine (His82) near the
submte is tantalizing, since previous studies have indicated that a histidine residue
may be important in the formimùiotransferase reaction. A posaile role for His82 is
that of a base abstracting the proton fiom N5 of tetrahydrofolate, thus increasing iis
nucleophilicity for attack at the imino carbon atom of formiminoglutarnate (see Figure
3-15), The protonated fiis82 could subsequently facilitate the breakdowu of the
intermediate by protonating the amino group of the giutamate yielding the products.
in the cunent crystal structure, His82 is positioned approximately 7 A h m NS of the
substrate which is too far to fulfill its proposed role as a base. In order for both the
formiminoglutamate and the side chain of His82 to lie in close proximity to the N5 of
tetrahydrofolate, the protein must undergo a change in conformation, including a
change in the histidine side ch in position in the substrate bound structure in order to
provide access to N5 of tetrahydrofolate. His82 c m be brought within 3.5A of N5
when Chi 1 is rotated. Other movernents in the structure would need to be made in
order to accommodate formiminoglutamate near the nucleophilic center of
tetrahydrofolate. ïhese changes in the protein may occur not just as srnail torsional
changes of side chahs but may involve large loop movements. in order to address
these questions it will be necessary to determine the structure of a subsmte analogue
complex of the Mdomain.
3.12. Data colIection and mode1 mfinement of FT domain comdaed with C5,CIû- dideazatetrah ydrofolate
Data were collected to 2.8A resolution using CuKa radiation and a MAR
image plate detector. Statistics h m the data collection are shown in Table 5. Since
the space group and the unit ce11 dimensions are identicai to those of the folinic acid
complex model (FT-fol), this model was used to soive the C5,ClO-
dideazatetrahydrofoIate complex structure (FTddf), The water molecules, Ligands
Figure 3.15
Schernatic of a possible catalytic mechanism far the fomimino-tramferase
reaction using His82 as a base catalyst.
Table 5
Data collection statistics for FT domain in complex with C5,ClO- dideazatetrahydrofolate.
Resolution (A) 50.0 - 2.8 Number of unique reflections 207 15 Total number of reflections 82 106 % complete 98.9 l/c 11.0 R-merge (%) 6.1
and severai loop regions were rernoved (as discussed in Section 2.5.2) and this edited
model was used to calculate the initial phases for the C5,CIO-dideazateûahydrofolate
complex structure. This was accomplished by refining the data obtained against the
modified folinic acid model. SIGh4AA weighted 3Fo-2Fc and Fo-Fc electron density
maps were calculated fiom the refined modified folinic acid model and used to
rebuild the rnissing loop regions as weU as the mode1 for the C5,ClO-
dideazatetmhyrofolate substrate analogue. The refinernent statistics for the FT-ddf
model are shown in Table 6.
3.13. Structure of FT domoin in com~fex with C5.C1Oilideazatetrahvdrofolate
The overall structure and fold the FTddf model is very similar to that of the
FT-fol structure. The major differences that occur between the two structures are in
and around the ligand binding site and in the actual conformation of the ligand itself.
The model consists of two protomers, each a single polypeptide chah
encompassing residues 2-327. While, in contrast to the originai model where
protomer "A" consisted of two polypeptide chahs (residues 2-207 and 214-326) and
protomer 'B" consisted of a single polypeptide chab (residues 2-326), the two
protomers in the FTddf mode1 consists of a single polypeptide chah each (nisidues
2-327). ïhis is likely due to the fact that NCS-restrahts were applied to the
refînement of the latter model, to 2.8A cesolution, while the former, to 1.7A
reoslution, was refined without imposing any NCS symmetry. The new model aiso
contains a moIecule of giycerol and a molecuie of (6R)CS,C10-
Table 6
Model refinement statistics for mode1 C5,ClO-dideazatetrahydrofolate.
Resolution range (A) 6.0-2.8 R-factor 22.7 R-fiee * 28.1 R.m.s.d. bond lengths 0.008 Rm.s.d. bond angles 1.30 Number o f non-hydrogen atorns 5057 Average B-factors (A2) - Overail 24.34 - . - -- --
* 10% of data was used to cdcuiate R-ke.
dideazatetrahydrofoIate per protomer. The eIectron density for the glycerol is not very
well defineci, due to the reiatively low resolution, making the exact torsional
conformations difficult to determine. No water molecdes were built into the model,
alsa due to the resolution of the electron density map. Figure 3.16 shows a regioa of
the elecmn density map.
Superposition of the two protomers h m FTddf mode1 resuIted in an r.ms.
difference of 0.3 1A between 326 structurally homolugous Ca-atoms- Superposition
of the first and second protomer of ihe FT-fol model ont0 the respective protomer of
the FTddf mode1 yields a r.m.s. difference of 0.31A and 030A between 318
snucturaIly homologous Ca-atorns. Thus the structures of the two models are vwy
similac Rowever, local differences are observed, for example, in protomer "B" of the
FTddf model, the loop region between residues 224-231 shi% as much as 2.5A
(Figure 3.17). The ligand, C5,CIO-dideazatetrahydrofolate, appears to be in a more
extended confoxmation that places the pterh ring closer to the extemal surface of the
protein when compareci to the bindhg of the folinic acid ligand. Such a shifl in the
Ioop region avoids unfavourable contacts that wodd occur between the pterin ring of
the ligand with the side chahs of Leu226 and Lys 229 h m this loop region (Further
discussed beIow).
InterestingIy, upoo initial inspection and cornparison of the FTddf model with
the FT-fol mode1 it would appear &at the better dehed densrensrty for the ligand resides
in the protomer "A" as opposed to protomer "B" as it did in the FT-fol model.
Attempts to model (6R)CS,Clûdideazatetrahydrofolate m the electron deLlSity of both * protomers was s u c c d - The (6R)CS,C 1 Mdeazatetrahydrofolate is analogous to
Figure 3.16
A stem view of the 3Fo-2Fc electron density map depicting the substrate analogue,
C5,ClO-dideazatetrahydrofolate, and giycerol molecule for a singie protomer. Some
side chains and water moiecules are also displayed. The map is contoured at l a
(produced with the program, SETOR, Evans, 1993).
Figure 3.1 7
Stereo view of the C5,ClMideazatetrahydrofoIate binding site of FT domain. The
main chah of the protein is depicted in a nibon representation. The substrate
analogue, 3,1 O-dideazatetrahydrofolate, and glycerol molecules are displayed in dark
grey bonds. Amino acid residues that make hydrogen bonding contacts are dispIayed
in light grey bonds, the hydrogen bond contacts are displayed as dashed black lines
(produced using the program MOLSCRIPT, Kraulis, 1991).
toi
(69 tetdydrofolate in t a s of stereochemistry (Figure 3.18). While the electron
density is not very well defined due to the relatively poor resolution ( U A resolution
for the F ï d d f modei as opposed to 1.7A for the FT-fol model), the physiologically
relevant stereoisomer can be satisfactorily modeled into the density. However, due to
the poor electron density around the C6 atom of C5,C1O-dideazatetrahydrofolate,
higher resolution data will be needed to unambiguously confirm whether the correct
stereoisorner has been modeled (Figure 3.16).
in comparing the conformation of the ligands between the two structures,
changes in the relative position of the pterin ring system are observed. There is an
inversion in the stereochernistry at Cd and a change in the pucker of the pterin ring
systern (Figures 3.17 and 3.19). This difference in ring pucker results in an equatorial
substituent at C6 rather than the axial as was observed in the FT-fol model and results
in an extended conformation that pushes the pterin ring towards the extemal M a c e
of the protein. With out a shifi in the loop region consisting of residues 224-229, this
conformation of the ligand would resuit in unfavourable contacts between the
substrate analogue and some of the side chains of residues in this Ioop region. This is
in agreement with what was predicted in the substrate docking study for pterin ring of
(65')-tetrahydrofolate ( s e section 3.10).
Similar contacts are made to the C5,ClO-dideazatetrahydrofolate ligand as to
the fotinic acid Ligand in their respective models. Interestingly, the contacts made in
protomer "B" of the FT-fol model a~ more similar to those observed in protomer "'An
of the FTddf model. For example, identical hydrogen bond contacts between OEl
and OE2 of G l m 8 and NA2 and N3 of the pterin ring are observeci for protomer "B"
Figure 3.18
Chemical structures of the naturai substrate (65')-tetrahydro fo Iate and the substrate
analogue, (6R)-5, IO-dideazatetrahydrofolate depicting the chirality at C6.
Figure 3.19
Stem view of the superimposition of FT-fol and the FTddf ligand binding sites. The
main-chah of the protein is depicted in a ribbon representation and the side chahs,
foliuic a d , C5,Cl Oaideazatetcahydro folate and glyceroI are shown in a ball-and-
stick representation. Dark grey represents the FT-fol mode1 and light grey corresponds
to the FTddf mode1 (produced using the program MOLSCRIPT, Kraulis, 199 1).
of the ET-fol model and protomer ""A" of the FT-ddf model. This suggest that
dlosteric interactions exist as the binding of a ligand to one protorner of Ff domain
appears to interfere with ligand binding in the other protomer. Further structural and
kinetic analysis is necessary to test this postulation. Table 7 shows a list of hydrogen
bond contacts made between the C5,ClOdideazatetrahydrofolate ligand and the
protein, for each protomer.
Table 7
Hydrogen bonding contacts between C5,ClO-dideazatetrahydrofolate and FT domain.