Sie sind auf Seite 1von 29

.I. Mol. 13iol.

(1984) 174, 497-525

Mechanism of Action of Aspartate Aminotransferase


Proposed on the Basis of its Spatial Structure?

.JACK F. KIRSCH~, GREGOR EICHEL@, GEOFFREY (2. FORD/~


MICHAICI, G. VINCRNT, JOHAN plj. JANSONIUS~

.-I btrilung Strukturbiologie, Biozentrum der Uninersitiit Base1


CH-4056 Basel, Switzerland

HEINZ GEHRING AND PHIMPP CHRISTEX

Biochemisches Znstitut der liniversitiit Ziirich


CH-8028 Ziirich. Switzerland

(KPceiced 16 ilugust 1983, and in revised form 19 DerrmbPr 1983)

Aapartate aminotransferase is a pyridoxal phosphate-dependent enzyme that


c&alyses the transamination reaction: r,-aspartate + P-oxoglutarate 2
oxaloacetat’r + L-glutamate. The enzyme shuttles between its pyridoxal and
pyridoxamine forms in a double-displacement process. This paper proposes a
mechanism of action that delineates the dynamic role of the protein moiety of this
rnzymcb. It is based on crystallographically determined spatial structures (at 2.8 w
resolution) of the mitochondrial isoenzyme in its unliganded forms and in
c,omplexes with substrate analogues, as well as on model building studies.
The enzyme is composed of t,wo identical subunit,s, which consist, of two
domains. The coenzyme is bound to the larger domain and is situated in a pocket
near the subunit interface. The proximal and distal carboxylate group of
dicarboxylic: substrates are bound to Arg386 and Arg292, respectively, the latter
residue belonging to the adjacent subunit. These interactions largely determine
t hr substrate specificity of the enzyme. They not only position the substrate for
rfK%nt catalysis but also bring about a bulk movement of the small domain that
closes the active site crevice and moves Arg386 about 3 A closer to the coenzyme.
The replacement of the E-amino group of Lys258 by the a-amino group of the
substrate in the aldimine bond to pyridoxal phosphate is accompanied by a tilting
of t,he cornzymr by - 30”. The released c-amino group of Lvs2.58 serves as a
proton acceptor/donor in the l$prototropic shift producing the ketimine
intermediate. At this stage, or after hydrolysis of the ketimine bond, the
cornz.vmr rotates back to an orientation between that in the “external” aldimine
intermediate and that in the pyridoxal form. Throughout this process. the

‘r This paper is dedicated to Professor Alexander E. Braunstein on thr occasion of his 80th birthday.
$ l’rwnantwt~ address: [Tniversity of C!alifornia, Department of Biochemistry. Berkeley. (‘A 94720.
I’.S.A
8 l’rrsrnt address: Vnivrrsity of (Aifornia, School of Medicine. Department of Biochemistry and
Nophysics. San Francisco. (:A 94143, L’24.A.
11 Present address: LTniversity of Shefield. Department of Biochemistry. Shefield Sit) 2’I’S
lhglanti.
’ Author to whom reprint requests should be addressed.
197
498 J. F. KIRSCH ET AL.
protonated pyridine nitrogen atom maintains a hydrogen bond to the
p-carboxylate group of Asp222. Upon formation of the pyridoxamine form, the
small domain moves back to its original position. The proposed mechanism is
compatible with the known kinetic and stereochemicalfeatures of enzymic
transamination.

1. Introduction
Many important reactions in amino acid metabolism, including transamination,
decarboxylation, racemization, and p- and y-replacement reactions are catalysed
by pyridoxal phosphate-dependent enzymes. In these enzymes, the cofactor is
bound covalently through an azomethine linkage between its C-4’ carbon atom
and the a-amino group of an active site lysine residue. The first reaction step is
invariably a transaldimination converting this “internal” aldimine into an
“external” aldimine intermediate in which the a-amino group of the lysine residue
has been displaced by the a-amino group of the amino acid substrate. The
reaction pathways of the various pyridoxal phosphate-dependent enzymes diverge
in the subsequent steps to give rise to different products (Snell & di Mari, 1970;
Dunathan, 1971; Braunstein, 1973).
Aspartate aminotransferase is the most extensively studied PLPt-dependent
enzyme functioning in amino acid metabolism. It catalyses the reversible transfer
of the amino group of glutamate or aspartate to oxaloacetate or 2-oxoglutarate.
The reaction follows a double-displacement mechanism with the enzyme shuttling
between its pyridoxal and pyridoxamine form. The enzyme is a dimeric protein,
consisting of two identical subunits of M, N 45,000 (Braunstein, 1973). Following
the fundamental studies reported by Braunstein (Braunstein & Shemyakin, 1953)
and Snell (Metzler et aE., 1954), a vast amount of knowledge on the interaction of
coenzyme and substrate has been accumulated. In contrast, an understanding of
how the protein moiety participates in catalysis has remained elusive. Apart from
Lys258 (Ovchinnikov et al., 1973; Doonan et al., 1974), which provides the
covalent linkage to PLP (Hughes et al., 1962), no active site residue had been
identified prior to crystallographic analysis of the enzyme. On the basis of
spectroscopic data and stereochemical considerations, a dynamic model of
transamination has been proposed that postulates a reorientation of the coenzyme
relative to the protein during the catalytic process (Ivanov & Karpeisky, 1969).
Both differential chemical modifications (Birchmeier et al., 1973; Gehring &
Christen, 1978) and measurement of peptide hydrogen exchange (Pfister et al.,
1978) have indicated that the interaction of the enzyme with a transaminating
substrate pair induces conformational changes in extended parts of the protein
matrix.
Recently, crystallographic structure determinations have been carried out on
both the cytosolic and the mitochondrial isoenzyme of AATase (Borisov et al.,
1980; Arnone et al., 1982; Harutyunyan et al., 1982; Ford et al., 1980). In this

t Abbreviations used: PLP, pyridoxal 5’-phosphate; AATase, aspartate aminotransferase; PMP,


pyridoxamine 5’.phosphate; PEG 4000, polyethylene glycol 4.000; PPL-amino acid,
N-(5’.phosphopyridoxyl)-L-amino acid.
ME(‘HASIRM OF ASPARTATE AMINOTRASSFERASE 109

paper. a mechanism of action is proposed that is based on the active site geometry
of mitochondrial AATase. Insight into the dynamic aspects of the catalytic
process has been gained through crystallographic studies of analogues that model
reaction intermediates. Structural details of the native enzyme and the various
derivatives are reported only to the extent to which they are relevant to this
article. A more complete account of the structural data will be given elsewhere.
The proposed mechanism is based on unrefined structural data, to which, in
general. an error level of up to 0.5 A in the atomic positions is attributed (Rlundell
& Johnson. 1976). This estimate seems somewhat high for our model, which is
derived from an electron density map of exceptionally good quality (Eichele.
1980). However. even an error level of O-5 a would not alter the assignment of the
groups believed to take part in the mechanism of catalysis.

2. Materials and Methods


(a) Enzyme forms
Mitochondrial AATasr from chicken heart (PLP form) was purified and crystallized
ceveral derivatives
(space group Pl; Gehring et al., 1977). S of the native enzyme were
prepared in crystalline form.

The cofactor was removed by dialysis (Schlegel & Christen, 1974). Isomorphous crystals
of the apoenzyme were obtained by the same procedure as used for the PLP-enzyme.
(ii) PNP-enzyme
PLP-enzyme crystals were soaked in 5 mM-L-cysteine sulphinate. 10 mM-sodium phosphate
(pH 7.5). 20”; [w/v) PEG 4000 at room temperature for several hours. Alternatively.
c,rystals were grown from enzyme that had been converted into the PMP form prior to
crystallization. These crystals were also isomorphous to those of the PLP-enzyme.
(iii) tlrduced etlzyme
The azomethine bond was reduced by immersing PLP-enzyme crystals in 5 mM-sodium
cyanoborohydride, 10 mM-sodium phosphate (pH 5.7), 2Oc/, (w/v) PEG 4006 at room
temperature until the yellow colour had disappeared. Isomorphous crystals could also be
grown from PLP-enzyme previously reduced in solution.
(iv) P:yyr%doxolphosphatr enzyme
This form was prepared by diffusing the coenzyme derivative into apoenzyme crystals
(1.5 rnbt-pyridoxol 5’.phosphate, 10 mm-sodium phosphate (pH 7.5). Zoo/c (w/v) PEG 4000,
one day at room temperature).
(V ) Phosphopyridoxyl amino acid derivatives
Three s-(5’-phosphopyridoxyl)-L-amino acid derivatives were prepared either by
cocrystallization or by soaking apoenzyme crystals in 20% (w/v) PEG 4000, 10 m&r-sodium
phosphate (pH 7.5) containing the appropriate PPL-amino acid. The concentrations of the
PPL-amino acid. soaking times and temperatures were: PPL-Asp, 0.2 mM, 12 to 15 h. 4°C:
PPL3-iodo-Tyr. 1 mM, 4 days, room temperature; PPL-Ala, 0.6 mM. 1 dav, room
temperature. All 3 compounds were synthesized as reported earlier for PPL-3-iodo-Tyr
(Nchelr rl nl.. 1979). PPL-Asp and PPL-Ala were further purified on an anion-exchange
c.olumn (A(: l-X4 from BioRad, eluant 0.1 M-formic acid) and subsequently lyophilized.
On soaking in PPL-Asp, many crystals shattered and had to be discarded.
500 .J. F. KIRSCH ET AL.

(‘oc,,:ystallization with PPL-Ala (7 mM) was performed under essentially the conditions
drsrribed for PPL-J-iodo-Tyr (Eirhele et al.. 1979).

Difl’ractometer data collection (PPL-Asp derivative to 34 A. all others to 2.8 A) and


processing were performed as described (Eichele et al., 1979; Ford rt al.. 1980). Difference
electron density maps were computed with coefficients ~rn(Fo-B’~) exp icr,. in which FD and
Fp are the derivative and native (holo- or apoenzyme) structure factor amplitudes,
respectively, CQ, is t’he native protein phase and no its figure of merit. Watson-Kendrew
models of one of the active sites (with residues from both subunits contributing) and of the
small domain of one subunit were built in a Richards optical comparator (Richards, 1968)
with the mirror parallel to the electron density sections (Colman et ul.. 1972). The amino
acid sequencr used is that of mitochondrial AATase from chicken (Graf-Hausner et al..
I!t83). Atomic c*o-ordinates based on these models were further improved by model building
with c~omputer graphics. A careful comparison of the electron density in sections normal to
thr molecular d-fold axis suggested no deviations from S-fold symmet,ry other than perhaps
ill a few sidtt-chains in molecular contact, areas. The electron density was t,hen averaged
about this axis prior to model building of a complete subunit, on the Vector General 3 D3
graphic*s display system at the Base1 IJniversit’y romputer (‘entre. The software used was
originally written by Dr (1. N. Morimoto at College Station, Texas (Morimoto & Meyer.
1!476) and updated by Dr E. L. Mehler at the Biozentrum, Base1 (unpublished work). A
largcl-scsalr st.ereo viewer, developed and built by E. Hratschi, was used throughout. A
tentative cr-(barbon model of the “closed” st’ructure (see Results) was obtained by applying
a transformation matrix to the cc-carbon atoms of residues 15 to 47 and 326 to 410. This
transformation matrix corresponds to a 13” rotation towards the active site around an axis
through Gly325 CA and parallel to the molecular S-fold axis (Z). Difference Fourier maps
\vert’ inspect,ed with both the graphics system and the Ricahards box. (‘rystallographic
rt+inement of the model is underway ((i. C’. Ford, 11. (:. Yincent 8r *J. N. .Jansonius.
unpublished results).

3. Results
The structure of the active site of mitochondrial AATase is shown in Figure 1.
The model includes the protein side-chains that interact with the coenzyme or the
substrate. A description of the course of the polypeptide chain in the act,ive sit,e
area has been reported (Ford et al., 1980). Distances between the relevant atoms
are listed in Table 1.

(a) Ligands to the coenzymr

The positive charge of the protonated pyridine nitrogen atom (Jenkins & Sizer.
1957) is balanced by hydrogen-bonding to the B-carboxylate group of Asp222,
which forms a second hydrogen bond with His143 NE2. The 2-methyl group of
PLP is located in a pocket formed by residues Trpl40, His143, His189, His193.
Asn194, Asp222, Ala224 and Tyr225. The methyl group of Ala224 is at van der
Waals’ distance behind the top part of the coenzyme ring. and prevents the
coenzyme from moving backwards. The 3’-oxygen atom of the cofactor is at short
hydrogen-bonding distance from the phenolic OH of Tyr225. The best fit to the
continuous electron density between Lys258 and the cofactor ring is obtained
when the imino nitrogen is located on the 3’-oxygen side of the cofactor ring
M EC’HANISM OF ASPARTATE AMINOTRASSFERASE

FIG. 1. Stereo drawing of the active site of mitochondrial AATase in the unliganded PLP form (open
~~cnlformation) as seen from the solvent. The N-terminal segment, which runs in front of the active site
t)t&)w Argd92 and ArgSX6, is not shown (cf. Fig. 2). a-Carbon atoms of sequential residues are
~~onnectcd hy thin lines (exceptions: residues 38 to 40 and 107 to 109 whose non-hydrogen backbone
atoms are included). The directions of the axes in the Cartesian molecular co-ordinate spstem arr
indicated. The X-axis corresponds roughly to the longest dimension of the dimer and the Z-axis
~Gnc*idrs with thr molrcular 2-fold axis. Residues Tyr70, Arg292, Ser296 and Asn297 belong to t ht*
~tdjac~t subunit.

TABLE 1
Srlrrted interatomic distances in the active site for mitochondrial AATase
and its derivatives

Distance (A)
PLP- PMP- PLP-enz.- PPL-Asp PPL-Ala PPL-3-iodo-Tyr
Atoms enzyme enzyme reduced derivative derivative derivative

(‘oenzyme N-I Asp222 ODI 3.0 3.1 34 2.9 3.5 3.1


(‘ornzpme N-l Asp222 OU2 3.6 3.0 3.1 2.7 3.5 3.4
( ‘ornz,vme (b-3’ Tyr22.5 OEH 2.3’ 2.8 2.4 3-H 2.7 2 3
(‘cwnzyme O-3’ Lys258 NZ 2.6 5.2 2.9 (5.R)h (54) (5.2)
(‘ornz,vme O-3’ Arg386 CZ 9.8 9.7 10.0 6.3 !b I 9.4
(‘ctrnz,vmr (‘-4’ Tyr70 OEH 4.3 4.6 4.3 4.6 3.X 4.I
( ‘omzymr (‘-4’ Lys2.58 NZ 3.5 (3.6) (3.1) (33)
( ‘ornz,vme N-4’ Tyr225 OEH 3.5 3.1 3.4 3.7 4.0 3.2
( ‘oenzyme S-4’ Lys258 NZ 2.5 (4.0) (3.5) (3.7)
Tyr70 OEH Lys258 NZ 4.1 3.2 4.8 (2.9) (34) (3.0)
Arp292 (“Z Arg386 CZ IO.3 10.3 10.3 (9.5) (11.2) (I 1.2)
Suhstratr (‘A Tyr70 OEH 4.3 34 4.2
Substrate (‘.I Lys258 NZ (3.7) (4.3) (4.5)
Suhstrxtr (‘ Arg292 cz (6.9) (-5.1) (.5.6)
Yubstl~ate (‘ Arg386 CZ 4.1 8.X 9.3
t~uhstrate
. (Y’ Arg292 ( :z (4.5)
Substrate O-; Trp140 NE1 3.9 1.5 34
Substrate ()I)1 Trpl40 NE1 2 -5

l)ist ancrs (8) of phosphate oxygen atoms to phosphate ligands in PLP-enzyme. OP2 to Tyr70 OK H,
2.9: ()I’% to Arg266 NH2, 3.1; OP3 to SerlO7 OG, 2.6; OP3 to Gly108 N, 2.7: 01’3 to Ser255 OG. 2.9:
01’4 tr) ThrlO9 N. 2.7: OP4 to ThrlO9 OGl. 3.1; OP4 to Arg266 SHl. 2.X (essentially identical valws
ill all tlrrivatives).
A This and a few other too short distances between hydrogen-honded atoms must, be attributed to
t hr errors in the, co-ordinates of the unrefined models.
b Distances involving the flexible side-chains of Lys258 and Arg292 are given in parentheses; the
positions of these side-chains were determined on the basis of stereochemical considerations, rather
t ban direct. S-rag widenw.
502 J. F. KIRYCH ET AL.

(cisoid conformer). The alternative transoid conformer can be fitted too, but is less
likely, because no hydrogen bond can be formed between N-I and Asp222 and
because of the larger distance between O-3’ of the cofactor and Tyr22.5 OEH. The
elect~rondensity map suggeststhat the azomethine double bond is 5’ to 16” out of
the plane of the cofactor pyridine ring. Releasing Lys258 from the covalent bond
with the cofactor, e.g. in the PMP-enzyme, causesthe distal portion of the lysine
side-chain to move forward forming a, hydrogen bond between its a-amino group
and Tyr70 OEHP. Although not, obvious from electron density difference maps. in
the PPL-amino acid derivatives (see below) the position of t,he a-amino group of
Lys258 must differ from that in the PMP-enzyme, due to steric hindrance by the
amino acid moiety. Atomic distances in the active site as measured for the
various derivatives are given in Table 1.
The spatial model (Fig. 1) suggests that the phosphate group of the coenzyme
rec*eivesseven or eight hydrogen bonds to the three non-bridging oxygen atoms.
Involved are t.he guanidinium group of Arg266 with probably two hydrogen
bonds, Tyr7O OEH, Serl07 OG, ThrlO9 OGl, Ser255 OG and t,he peptide NH
groups of Gly108 and ThrlO9. It has been shown by 31P nuclear magnetir
resoname that the phosphate group carries a double-negat’ive charge in the
cytosolic PLP-enzyme (see Discussion). The double-negative charge is largely
~(~~~lpensat~d by Arg266 and the positive electrostatic field at the ?i terminus of
an E-helix (HOI et al., 3978), consisting of residues 108 to 122. The covalent, bonds
between the phosphate group and the pyridine ring are strained. Although the
dihedral angle around the C-5--C-5’ bond (defined by the atoms C-4. (‘-5, C-5’ and
O-5’: seeFig. 4 for the numbering convention) cannot be determined accurately at
t,he present resolution. t,he electron density is incompatible with a vaIue of more
than 20” to 25”. in cont,rast to the P~IP-enz~~n~ and t~he PPL-amino acid
derivalives, where this angle varies in the range from 40” to 70”. corresponding to
one of t’he two values (+60”, -60”) predicted for an unstrained internal aldimine
structure (Tumanyan t-t al.. 1974).
The coenzyme is reoriented in the pyridoxamine form of the enzyme (see
below). The torsion angle around C-5-C-5’ is about 55” in t’he model. The PMP
amino group is somewhat behind the ring plane, where it simultaneously donates
hydrogen bonds to O-3’ and to Tyr225 OEH. It accepts a hydrogen bond from the
protjonated Lys258. which is also properly positioned Lo donate a hydrogen bond
to Tyr70 OEH. This arrangement lowers t’he pK, value of the PMP amino group.
The structure of the pyridoxol phosphate derivat,ive is very similar to that of the
P~~I’-enz~n~e.

(b) Cofactor tilt


The cofactor ring in the PLP-enzyme is tilted about 20” relat’ive to a plane
through the molecular 2-fold axis and the long axis of t,he dimer (X2 plane,
Fig. 1). Good estimates for the tilts of t,he ring in t,he various derivat,ives were

t The active site is composed of amino acid residues from both subunits. Most of these belong to the
same subunit as the Schiff base forming Lys258. A minority, including Tyr70, Arg292, Ser296 and
Asn297, belong to the second subunit, a fact not indicated by any special notation.
~r~~‘~A~~S~l OF ASPARTATE A~I~~T~~~~S~E~AS~ lifi:i

obtain& by mctdet building into difference Fourier maps computed with


~~rnpfit~~des - Fapof.The resulting tilt angles relative to the XZ plane are
(~~=rivative
about 4W for the PMP-enzyme and the reduced enzyme, 45” to 50” for t,he
pyridoxol-phosphate enzyme, the PPL-Ala and PPL-3-iodo-Tyr derivatives and
~0‘ to 59 for the PPLAsp derivative. TXfference Fourier maps exclude an)
+nificant translation of the phosphorus atom, but not slight rotat’ions of the
l,hosphate moit$y. The t,ilt movement can be modelled by rotations around that
jingles bonds of the 5’ side-chain. The main component is a rotat,ion around thcs
f ‘-.5-- (‘6 bond. The sockets for the ring rotation are the phosphat’e binding
poekrt on one side and t*he ~-~arboxylate group of Asp222 int,eracting with the
pyridinr N-f t,ogetSherwith the pocket for the 2-methyl group on t.he other. The
tfista~~~ bet.wrrn Asp222 and the N-t atom of the cofactor is largest. in t~heYLP-
ranzytne buf, does not vary si~ni6~antl~ in the derivative ftlrms of r?lATase. Asp282
:atipart, of t,he rigid a/B fold of the PLP binding domain (Ford et al.. 198t)) prevents
t hta iowt~ part of the cofact,or ring from moving backwards by more than 0.5 to
I iy llnit’. The indole group of Trpl4U located in front of the lower part of the
c~nzyrnc~ ring limits the tilt to approximately 70”.

(c) Confomaa~ionaE chm2ge.s of the protein

The snlphydryl group of Cys166 has been shown to be an indi&~t~)r of


~~~~nformationalchanges of the protein matrix in the m~to~hondrial enzyme
(Gehring & Christen, I%%). Its reactivity towards sufphydryl reagents increases
upon in~era~?tionof the enzyme with maleate or the t.ransaminatin~ substrate pair
(Table 2). A similar increase in reactivity is observed when PPL-Asp or PPL-Glu.
:\nalt)gues of rkovalent enzyme substrate intermediates, are bound to thp

f?iridoxat enzyme 1”
pfw mttieate~ 4
H&enzyme plus aspart&*
and oxitltracutate 6 to Sd
Complex of apoenzpme:
with I’PL-aspartate ti
with PPL-glutamatee 6
with PPL-&nine’ 1.2
504 ,J. F. KIRSPHi ET AL.
apoenzyme. PPL-Ala, in contrast, does not change the reactivit’y of the
sulphydryl group; only analagnes of specific substrates seem to be effective.
The globat conformational change of the enzyme mat,rix resulting from the
binding and turnover of specific sub&&es was first seen in the difference Fourier
map between PPL-Asp derivative and apoenzyme (Eichele et al., 1979). The
subunit of the crystalline enzyme dimer Lhat is least constrained by the crystal
lattice (S$, hop subunit, in Fig. 2) shows major conformational changes
(corresponding to a bulk movement of the entire small domain (Eichele at al., 1919;
Eichefe, 1980) conco~nit~nt with full occupancy of the active site by PPL-Asp.
The second subunit (8, bottom subunit, in Fig. 2) is occupied only t,o 30 to 40%
and fails to undergo t,he conformational change. Cocrystallization of apoenzyme
with PPL-Asp resulted in two new crystal forms, space groups P2, (Thsller et nl.,
1981) and 6222,, in which both subunits have undergone t,he large conformational
change and molecular symmetry is maintained (31. G. Vincent, I>. Picot,
C. Thailer, G. Eichek, G. C. Ford & J. X. Jsnsonius, unpublished results). This
finding is consistent with the functional equivalence of the t,wo subunits of the
enzyme in solution (see Schlegel & Christen, 1978. and references cited therein).
Apparentjly, movement of the small domain of subunit, R is incompatible with the
cryst~al lat,tica in the triclinic crystals, in which the environments of the t-70
subunits differ. Inspection of the Fourier map and model building indeed showed
that. this movement8is blocked by a r~ei~hbouring molecule (&I. G. ‘C’incent & J. X.
Jansonius, unpublished resu1t.s).The interference of t’he crystal lattice with the
binding of PPL-Asp is demonstrated also by the shattering of many cryst,als,
which becomes more frequent at higher tempe&ure or with increasing
concentration of PPL-Asp.
In contrast to PPL-Asp, Ihe analogues of unspecific sub&rates PPC-3-iodo-Tyr
and PPL-Ala fully occupy both active sites. No conformational changes were
observed either upon soaking (with PPT,-3-iodo-Tyr or PPL-Ala) or upon
cocrystallization (with PPL-Ala). This observation explains the unchanged
reactivity of Cysl66 in the PPLAla derivative in solution relative to the PLP-
enzyme (Table 2).
The small domain has been defined to include residues 15 to 47 and 359 to 41%
In view of recent studies, however, the second part is redefined to extend over
residues 326 to 410 (M. C. Vincent, 1). Picot, C. Thaller, G. Eichele, G. Cr. Ford &
,J. N. Jansonius, unpublished results). ‘Residue 325 is s glycine in the 50 A long
helix (residues 313 to 344) and is positioned at. a kink, which could serve as one of
t,he pi~oth~ of t‘he rotation of the smal1 domain. A small change of the dihedral
angles at GIy32ft (A$= - IO”, A$ = IO”) is indeed sufficient to bring about the
observed rearrangement,. The motion can he described as a 12” to IS” rotation of
the small domain t,owrtrd t)he coenzyme around a,n axis through Gly325,
approximat,ely parallel to the molecular ?&fold axis (2). At,oms far from the hinge
shift by up to 7 A. Structurat ~arrangements in the regions of residues f 1 to 14
and 3X to 48 ac~~on~~~any the movement of the S-terminal part of the smaif
domain, The net effect of the process is a tightening of the protein structure
around the substrate, which thus becomes locked into the active site pocket.
Figure 2 in a stereo drawing of the a-carbon backbone of mitochondrial AATase,
I+:. 2. Thr a-carbon haekbone of the rni~~chot~drial AATase dimer, stereo riew along 6hr moiet~lar
%-axis (“-fold axis). Every 10th residue in the sequenre is marked with an enlarged open circle. Thta
srrtxll domains of b&h subunits are indicated hy double line connections between the z-carbon atoms.
The bottom subunit is shown in the open conformation. The top subunit is drawn in thr putativr
~~I(~srd (*onformation. as it results from a r&at& of the small domain around an axis through C:1~325
and parallel to %, over 13” towards the coenzyme (shown in heavy lines). This conformatIonal
c&nge VIOSPS the entrance to the active site. Tt also unmasks Cysl66. The cl-carbon atom of this
rc,sidue its well as those of Arg292 and Arg386 are marked by enlarged filled (lircles. The asymmrtrir
open/closed hyhrid structure shown here act,ually exists in the derivativr obtainrd hy diffusion 01
I’t’L-Asp int,o itn aporneymr crystal.

ill which one subunit (bottom) is in the open conformati~)n and the other in thrl
c40sedform, as mode~ledaccording to the above des~ripti(~n of the small domain
movement. The unmasking of Cysl66 in the closed structure and the ti~hteni~~~ of
t tttb protein around the active site pocket are clearly seen. Rigid body rotations
similar to that observed in AATase have been described for other enzymes, e.g.
hc,sokinase (Bennrtt & Steitz, 1978) and citrate synthase (Remington ~4rxl.. 1982).

(d) Substrate binding de


‘l’hr difference Fourier map comparing the PPLAsp derir-ativr and the
apoenzyme shows interpretable features in the active site pocket of subunit Ss.
allowing an analysis of the position of the guanidinium group of Arg386, which
prot r&es into t,he crevice. Figure 3 shows the atomic model su~~erin~~~osed
on the
tlif&n*n~ map in the active site area. The side-chain of Arg386 before and aftt~
506 et. F. KTRSCH ET AI,.

PIG. 3. Stereo drawing of difference electron density map with coefficients (i(‘t,r.i,~,J~~, - F’,,,) exp iala in
thr region of the active site of subunit, St. Positive contours have Bern drawn at 1.5 and 2.5 Cmes the
standard deviation of the electron density; negative contour lines have been omitted for &ritg.
Positive den&y is visible for the PPL-Asp molecule with the exception of it,s phosphatr moiety. whose
hinding site is occupied by inorganic phosphate in the apoenzyme. Superimposed on the density are
the PPL-Asp molecule, Tyr70 (which does not move), A@86 in the closed (continuous lines) and in
the open conformation (broken lines, labelled X386), Lys2.58 and Arg292. A shift of 4rg386 by -3 .A is
clearly indicated, The positions of the side-chains of LysYAH and Arg292 are not defined by thra
difference map, hut were oriented into stereochemically reasonable positions. The relevant interatomic,
distances are given in Table 1, The directions of thr X. Y and Z-axps are given in the lpft lower VW~W.
For further information, see the text.

the conformational change hns been included. The movement of the small domain
brings the guanidinium group of Arg386 about, 3 A closer to the coenzyme, to a
position where it can make two hydrogen bonds to the a-carboxylate group of t,he
aspartate moiety, The left side of the active site, which includes Tyr’70, exhibits
no main-chain rearrangement. The substrate moiety of PPL-Asp spans Arg386
and Arg292 with its j?-carboxylate group hydrogen-bonded to Arg292 and the
indole nitrogen of Trpl40. The two carboxylate groups are in cis orientation
(Xl z 60”). The binding of the substrate portions of PPL-3-iodo-Tyr and PPL-Ala,
which in contrast to PPL-Asp induce no conformational changes in the small
domain, is different. Their a-carboxylate groups are hydrogen-bonded to the
indole nitrogen of Trpl40 in an energetically favourable posit,ion between Arg386
and Arg292 (Table 1).

(e) Model building


Models of the Michaelis adsorption complexes and of five covalent
intermediates, namely the tetrahedral intermediate in transaldimination, the
external aldimine, quinonoid and ketimine intermediate and the tetrahedral
intermediate in ketimine hydrolysis were built by fitting the respective forms of
coenzyme and substrate into the active site structure (open conformation) with
the use of computer graphics. Despite limited knowledge of the closed structure
(from the PPL-Asp derivative), several important conclusions could be drawn
that ~ontribu~ to the delineation of a hypothetical mechanism.
Models of the intermed~a~s are sketched in Figure 5 in the Discussion.
The PLP-enzyme in its open conformation can bind aspartate by making
double hydrogen bonds and charge interactions between Arg386 and the
r-c*arboxylate group and between Arg292 and the distal carboxylate group of the
suhst,rate. The latter group is involved in an additional hydrogen bond to Trpl40
NE1 The ctarboxylate groups of the substrate are in cis orientation (xl =: 60”).
The arnino group of the substrate points towards the coenzyme. but is not in
direct van der Waals’ contact, with it. In this and the subsequent models, the side-
chain of Arg292 that protrudes into the solvent had to be reoriented in order to
allow l~ydrt)~en bonding to the distal carboxylat,e group. In cont,rast, the Arg386
side-chain is held rigidly in the small domain structure and can move closer to the
substrate only together with this domain. The position of Arg386 in the closed
slructure is known from the difference map of the PPL-Asp derivative and has
l#Jen used in building the model of the Michaelis complex with aspartate in the
closed form (Fig. 5, model I). With the carboxylate groups of t’he substrate still in
ci.u orientation, essentially the same interactions can be made as in t)he open
conformat,ion. but t’he substrate is now in close van der Waals’ contact with thr
cornz>-mr t,hrough its amino group (perpendicular to the aldimine double bond at
3-4 x from (‘-4’) and its a-carboxylate group, and with the ring of Tyr70 t’hrough
(‘A and CK. in the closed form of the enzyme-substrate complex, the distancxe
l)t%wetLnthe a-carboxylate group and the guanidinium gI;oup of Arg386 is somewhat
too large for hydrogen bonding. Left of the subst~rate(in the view of Fig. 1) t,here
is room for Lwo water molecules. One water molecule, H,U(l), could make
hydrogen bonds to Asn297 ND2, Ser296 OG, PLP OP2, Tyr70 OEH and to a
scwnd water molecule, H,0(2), posit,ioned at’ hydrogen bonding distance from
ThrlO9 O(:l along with Ser296 OG and PLP OP2. H,O(l) would be at or near
hydrogen bonding distance from the a-amino group of aspartate. and might be of
importance in orient,ing the substrate.
Model building experiments with glutamate showed that it can bind in a similar
mannrr to aspartate. It makes the same interactions. Arg292 can be reoriented
wsily to make orw or perhaps two hydrogen bonds wit.h the y-narhoxylate group.

4 rotation of 30” t,o 35” around C-5-C-5’, coupled with a small rotation around
(‘%-O-5’. brings the coenzyme into the right position and orientation for
forming the external aldimine intermediate with aspartate (Fig. 5, model III). The
(ooenzyme t)ilt in this model is 50” to 55”. Most of the atomic positions of the
aspartate moiety are within 1 A to 1.5 f! from those found in the PPL-izsp
dc~rivat~iveand t’he same hydrogen bonds are formed (see Fig. 3). The CA atom is
situated 3 A to 3.5 b underneath the c-amino group of Lys258 (positioned as in
Fig. 3). suggest,ing that this group serves as acceptor of the CA proton. Rotation
of t,hr coenzyme ring around C-5-C-5’ and C&-O-5’ brings the aspartate moiety
either t,no close to Tyr7O and Gly38 or too close to TrpllO. In view of t#hese
narrow ~~~)nstra,ir~tswe believe that our model must be close to reality. A model
508 J. P. KIRSCH ET AL.

was also built of the tetrahedral intermediate in transaldimination (Fig. 5,


model II) in which the coenzyme ring tilt is 35” and single bonds are formed from
the C-4’ atom to Lys258 NZ behind and t,o the substrate nit,rogen atom in front of
the coenzyme pyridine ring. The a-carboxylate group is at slightly larger than
hydrogen bonding distance from Arg386.

(iii) Quinonoid intermediate


In this intermediate (Fig. 5, model TV), the CA-CB and CA-C bonds of
aspartate lie in the plane of the pyridine ring. Further rotation around C-5--C-5’,
resulting in a ring tilt of 66”, is necessary to avoid too close contacts to Tyr70 and
Gly38 and to keep t,he interactions with Arg386 and drg292 int,act. However, the
P-carboxylate can no longer receive a hydrogen bond from Trp140. Instead.
hydrogen bonds can be made with Ser296 OG and Asn297 ND2. Tn this model, the
pyridine ring is exactly parallel to the indole ring of Trpl40 at a distance of 3.4 A.
The e-amino group of Lys258 is situated 3.5 A from C-4’. Its distance from C4 is
4.2 A, the increase relative to the external aldimine structure being due to the
increased tilt of the coenzyme.
The stability of the quinonoid intermediate formed with the substrate analogue
erythro-3-hydroxyaspartate (Jenkins. 1964) can be explained by two hydrogen
bonds of the 3-OH group with phosphate OP2 (as donor) and with Tyr7O OEH (as
acceptor). The tyrosine side-chain has been released from the hydrogen bond to
phosphate OP2 and has moved somewhat, up and to the right (in the view of
Fig. 1) by rotation around CA--CB.

(iv) Ketimine intermediate


In the ketimine model (Fig. 5, model V) the coenzyme tilt is essentially
identical to that in the PMP-enzyme. The ,&carboxylate group of the C,
substrate is hydrogen-bonded to Trpl49 and t,o Arg292, while both the
a-carboxylate group and the ketimine double bond lie essentially in a plane
parallel to the X Y plane. Arg386 in the closed conformation can form one or two
hydrogen bonds t.o the a-carboxylate group as has been shown for the PPL-Asp
derivative (Fig. 3). The CR atom of the substrat~e is in van der Waals’ contact with
Tyr70. The corresponding model with the C, substrat,e has CB and CG at van der
Waals’ distance from Tyr70. The y-carboxylate oxygen OEl is hydrogen bonded to
Ser296 OG, while OE2 is engaged in a hydrogen bond with Arg292. So hydrogen
bond is made with Trpl40 NEI. The s-amino group of Lys258 is presumed to be
unprotonated at’ t’his st’age. It can make hydrogen bonds t’o Tyr70 and to the
phosphate oxygen OP2. The next’ step, hydrolysis of the ket’imine int)ermediate,
requires that a water molecule has access to CA. The only approach is from above.
There is room for a water molecule, H,0(3), in front of Lys258 and 305 A above
the a-carbon atom of the substrate. It can simultaneously be hydrogen-bonded to
Tyr70 OEH, Gly38 0 and Lys258 ?uZ. Th us, the stage is set for the first step in
the hydrolytic release of the 0x0 acid product.
~11~:~~~~4~IS~l OF ASP.4RTATE AMISOTR~~SSFEK~4sE 5(1!4

(S ) ~‘etr~he(lr(~l ~~~t~~rn.~~~ute in the ~.~~,~o~~,~i,s of the ~e~i~~,~~,~

This model (Aonfirms that hydrolysis of the ketimine intermediate can be


atacomplished by a wa,ter molecule from above. The hgdroxyl group point,s in the
dirpct,ion between Tyr70 and Gly38.

Both oxo-substrates can be fitted in the a&re site of the PMPenzyme in a


similar way. Their carbony and a-carboxylate groups are presumed to be (nearly)
coplanar and oriented essen&lly parallel t,o the cboenzymering (Fig. 5, model VT).
The tarbony group points approximately in the + Z-direction. Its oxygen atom is
posit,ioned between Tyr70 and clly38. This ori~i~t,at~io~~~~orresI)ondsto that
observed in a difference elect,ron dens&y map of the PLP-enzyme with and
wit bout, %-oxogluCarate (Eichele, 1980; and unpublished results). The oc-carboxylate
group hydrogen bonds to Arg386. The amino group of the coenzyme in the model is
somewhat, in front) of t’he coenzyme ring (torsion angle X-4’- X-4’ (‘-4 -C-3 20” to
25’ ) ant1at N 3.5 A distance from t,he carhonyl carbon on a line perpt~ndictulwrto the
(~arbonyl group. (:K of oxaloacetate aswell as(1n and CX of 2-oxoglutarate are in van
dt~. \Vaals contact with Tyr70. Their distal carboxylate groups are directed awa!
frt)m thr c~oenzyrnc~ and are involved in one or two hydrogen bonds to ArgZW.

4. Discussion
Sinctt the elucidat’ion of the role of PLP in enzyme-cata.lysed transamination
(Hraunstein & Shemyakin. 1953; Metzler et al., 1954). a vast amount of
experiment~al data has been accumulated and numerous mechanistic proposals
bilv(h hchenmade to account for the combined role of the protein and the coenzymr
(cb.g.SW Velick CLVavra, 1962; Fasetla et nl., 1966; Ivanov & Karpeisky, 1969:
Sl~tzier. 1979). The erystjallographic and model building experiments described
above c*onsidera.blyc>xpand the experimental basis for delineating a hypothetical
mt*t*hanismof action. An out,line of the course of events during enzyme-catalysed
t r~~r~sal~lirla~,ion
that, is ~~OrnI~atiblewith the present and previous experim~~ltal
rlat a is given in the first part of t,his section. Sut)sequently. particuIar aspects of
tht, proposed mc&anism are discussed in more det,ail.

(a) Outline of hypothetical mechanism1 qf A A Taw


In this section, a half-reaction of transamination is desrribed starting from the
l’I,P form of the enzyme plus an amino acid substrate and ending with the P&W
fi)rm plus t,hr 0x0 acid product. Figure 4 illustrates the sequencte of chemical
c~vc~nts.while Figure 5 gives an impression of the dynamics of the process t,hrough
spatial models of the ~~~rres~~on(Iin~ intermediat,es, as built on a graphics display
s?,strm (SW Results).
510 .J. F. KIRSCH ET A I,.
Arg 292 Arg 366
f +

Lyr 256

co;
l I
H3N-C-H

:H
I 2
co;

Arg 292 Arg 366


+
al l

III H..O-
I.30 “Ill
‘C-Asp 222
0”

Arg 1 292 Arg 366


f Arg 292 Arg 266
1 f

-0opc -Cl+2

FIG. 4. Sequence of chemical events in t,he half-reaction of ensymic transamination: PLP-enzyme +


aspartate # PMP-enzyme + oxnloacetate. The most important hydrogen honding and charge
interactions of coenzyme and substrate with active site groups are shown. The intermediates depicted
are: I and Ia, PLP-enzyme (“internal” aldimine) and amino acid substrate at high and low pH.
respectively; II, tetrahedral intermediate in transaldimination; III. “external” aldimine intermediate;
IV, quinonoid intermediate; V, ketimine intermediate; VI, PMP-enzyme and 0x0 acid product. The
approximate wavelengths of the absorption maxima characterizing the various species are indicated.
The 90” rotation of the C-4 C-4’- -N plane occurring during transaldimination (between species II
and III) is shown. The protons involved in the l$prototropie shift III to \’ are encircled. The
numbering of the atoms of the (noenzyme ring starts at the ppridine nitrogen atom and runs
anticlockwise.

Tn the PLP-enzyme, the coenzyma is anchored to the protein with its pyridoxal
moiety in the reactive bipolar ionic form (French e:t al., 1965; Ivanov &
Karpeisky, 1969). The protonated N-l is stabilized by a hydrogen bond and a salt
bridge to Asp222, below and behind it, (Fig. 1). The deprotonated Shydroxyl
group is involved in a strong hydrogen bond with Tyr225. The tilt of the
coenzyme of -20” (relative to the X2 plane, Fig. 1) corresponds to a strained
situation as evidenced by a nearly eclipsed conformation around C-5-C-5’ and
non-planarity of the aldimine bond with the pyridinr ring, the torsion angle
around C-4--C-4’ being 5” to 10”.
Roth crystallographic evidence and experiments in solution indicate that the
unliganded enzyme exists predominantly in the open conformation. The Michaelis
adsorption complex must be formed between t)he open conformation of the PLP-
enzyme and the amino acid substrate, since there is no access for the substrate to
ME(‘HANISM OF ASPARTATE AMINOTRASSFERASE 61 I

the active site in the closed structure (at least not in its time-averaged form). The
suMrate is guided into the active site by the negative charge on O-3’ of the
~o~‘r~zyrl~t~ and the positive charges on the guanidin~um groups of Arg386 and
Arg292. The arrangement of these charges and the shape of the active site pocket
rc.stric+t the possibilit8y of non-productive modes of binding and binding of
n;itnino acids. The substrate is bound with its two carboxylate groups in cis
orirbtrtation. as had been indicated by binding studies with fixed conformational
isomers (,Jenkins et ctt., 1959: Angielski oi Rogulski, 1962: Khomutov et al., 1968:
~Ji~.h~ldi~ B ~~arti~lez-(‘arrion, 1970: ~raunstein, 1973). The comI)ensation ttf

c+;irges by the interaction of the substrate a-carboxylat,e group w&h Arg386 and
of’ its distal c4~oxylate group with Arg292 has two effects. Firstly. it changes t,ht,
p/i values of’ the x-amino group and the aldimine group in a way favouring
prot.onation of the latter and deprotonation of the former. Further. it shifts the
c*orlfot~nlatiortal ~~~~UilibriLlrn more towards the “closed” form, thus ~)ringing
ArgMti, which is rigidly built into the small domain structure, - 3 ,% closer to the
c~oc~nzytnr. In this c*orrformation (Fig. 5, model I), t)he substrate van bind with
hot h its a-carboxylate group and its amino group wedged between the coenzyme
HIMI thtk indole ring of Trp140. The atnino group is placed - 3.5 x from C-4’ on t,ht
normal to the aldimine double bond. This position is presumably stabilized by a
hy(lrogirt*rt ttond between the atnino group and one of two water moiecules that can
tit int,o the a&v? sit,<\ por:ket below Tyr70. The distal carboxylat,e group is
h~rlrogrtI-l)ondrd to Arg292, but the distance between the x-carboxylate group
air<1 At&M is somewhat too large for hydrogen bonding. Transaldimination is
initiated 1,~ rotat,ion of the coenzyme, mainly around an axis through C-2 and
f ’ 5, its qggested t~,v lvanov CL Karpeisky (1969), but with additional cxomponents
arount-1 1’.5’ 0~5’ inid O-5PP, keeping the pyridine S-1 at. hydrogen bonding
ciistamta from Asp222. The nucleophilic at,tack of the deprotonat.4 substrate
amino group on (‘-1’ is facilitated by releasing the strain upon rot&ion of t,he
c~o~~tizytnc. Tfie two-step process 2%~a tetrahedral intermediate (TT) involves a 90’
rotation of the two (l----X bonds around C:SmPC1’ (Ivanov dt Karpeisky. 1969:
Furbish csfnl., 1969: Federiuk & Shafer? 1983) as is evident from Figure 3. stages
II anti I II. Thr a-amino group at.tacks C-4’ frotn the front side in a dire&ion
lit~rpcndi~ular to the plane of the pyridine ring. The 90* rotation that brings the
bond to the Lys%TiX nit8rogen atom behind C-4’ and the plane of the pyridine ring
is t’acilitatjed t,y attrac+ion of the opposite charges on the substrate nitrogen atom
attct 0.3 of’ the t~otnzytne. The a-amino group can then be released. The
tr~ltlsi~ltlirrriiratiori process is driven in part by the ~oenzynl~~‘s preference for a tilt
anglr~ of - 40”. In the external aidimine (TII), hydrogen bonds exist again between
the, a-c.;nl)oxylate group and Arg386 in addition to t,hose between the distal
(‘ii rl)oxylat,t~ group and Arg292. The released uncharged a-amino group of Lys25X
is plac~c4 between Tyr70 OKH and Gly38 0, forming a hydrogen bond to the
former. It is ~}ositi~)n~~l dire&y above the substrate CA at van der Waals’
tlistamca. The <‘.A H bond points towards t,he a-a.ntino nitrogen atom.
lJt‘t’rJ~“(lic‘t~lar to t hth plane of the pyridine ring. The ~-amino group can thus
atsc.eJrt thth proton (encircled in Fig. 4, stage TTI) from (!A. Snrll (1962) w-as the
tirst to propose this possibility.
512 .J F. KIRRt'H ET AL.

I Rl id~aetis adsorpt~ion complex of


I’I,I’-enzyme with asp&ate (c4osed
c‘otlforlllRtitltl1

II l’rtrahrdral intermediate in
t,rilrlsaldiminatiot,

Frc:. 5. Spatial tnodets of the active site in selrtrted intermediates of the enzynic transamination
reaction (for details, ?iee the trxt).
ME(‘HANISM OF ASPARTATE AMINOTRAXSFERASE

IV Quinonoid intrrmrdiatr

V Krtimine intermediatr

\-I Michaelis adsorption complrx of


PMP-enzyme with oxaloacrt,etr
(op=n conformation)

FTC:. 5. IV to \
514 J. P. KIRSCH ET AL.

In the external aldimine structure, the coenzyme tilt is increased to -50” in


!)rder to accommodate the substrate moiety. Consequently, the hydrogen bond of
Tyr225 to O-3’ is weakened or broken. The resulting increase in the pK value of
the aldimine N (see below) further facilitates deprotonation of CA giving rise to
the quinonoid intermediate (TV) (Fig. 5. model IV), which is stabilized by the
elect’ron sink of the protonated pyridine ring. In the quinonoid intermediate, CB
and the m-carboxylate group are coplanar with the pyridine ring, necessitating a
further increase in the tilt angle to 60” in order to avoid too close contacts with
Tyr70 and Gly38. The increased tilt orients the pyridine plane parallel to Trp140
at a distance of -3.5 8. The distal carboxylate group, which up to this point was
hydrogen-bonded to both Arg292 and Trpl40 NEI, is reoriented to a more
extended conformation and now makes, in addition to the hydrogen bond/salt
bridge to Arg292, hydrogen bonds to Ser296 OG and Asn297 ND2. This position
of t,he distal carboxylate group is incompatible with binding of a water molecule,
H,O(l), in the active sit,e pocket. Presumably, the water molecule is expelled
concomitantly with formation of the quinonoid intermediate. Lys258 is now
positively charged. Its XH: group is attracted by the negatively charged
phosphate group and forms a hydrogen bond to phosphate oxygen OP2. Its
hydrogen bond to Tyr70 is maintained. The third proton is now directly above
C-4’ short van der Waals’ distance. The backward movement of the lysine side-
chain provides room for a water molecule, H,0(3), to position itself between the
amino group of Lys258, Tyr70 OEH and Gly38 0.
Lys258 protonates C-4’ from the si side (see encircled hydrogen atom in Fig. 4,
stage V) giving rise to the ketimine intermediate (V). The resulting tetrahedral
geometry around C-4’ allows the pyridine ring to relax back into its preferred tilt
of w 40”. The fi-carboxylate group of the C, substrate rotates back into
hydrogen-bonding contact with Trp140 NEl, allowing the H,O(l) position to be
occupied again, while the y-carboxylat’e groups of the C5 substrate remains
hydrogen-bonded to Ser296 OG. In bobh cases the hydrogen bond/charge
interaction with Arg292 remains intact, due to the flexibility of its side-chain. The
hydrogen bond between the phenolic hydroxyl group of Tyr225 and O-3’ of the
coenzyme is reformed or strengthened. Subsequently, H,0(3) carries out a
nucleophilic attack on CA, losing a proton to Lys258, which remains hydrogen
bonded to Tyr70 and perhaps to a phosphate oxygen atom. The coenzyme tilt in
the tetrahedral intermediat’e is probably increased again to 45” or 50”, and the
distal carboxylate group is only hydrogen bonded to Arg292. The tetrahedral
intermediate dissociates into PMP and the 0x0 acid product. Rotation of the
coenzyme to the -40” tilt angle of PMP allows breaking of the N-CA bond
without substantial movement of the 0x0 acid product. The unprotonated amino
group of PMP reorients itself t’o make hydrogen bonds with O-3’, Tyr225 OEH
and with the E-amino group of Lys258. The product is released from its complex
with t,he PMP form (VI). For steric reasons (see above), the product can be
released from the enzyme molecule only in its open conformation (Fig. 5,
model VI). However, it is not known whether the equilibrium in the complex VI
lies more towards the open or towards the closed conformation. l
MEC‘HANISM OF ASPARTATE AMINOTRANSFERASE 515

This mechanism excludes the enzyme monomer as an efficient catalyst. Four


residues, Tyr70, Arg292, Ser296 and Asn297, that play a role in the catalyt’ic
process, belong to the adjacent subunit.

(b) Movements of the coenzyme

The hypothesis of rotational movement of the coenzyme during


transaldimination was first advanced by Ivanov & Karpeisky (1969). Their
detailed mechanistic scheme shows remarkable insight and intuition. The different
coenzyme tilt angles found experimentally by us for the PLP-enzyme ( - 20”), the
PMP-rnzyme (-40) and the PPL-Asp derivative (50” to 55”) of mitochondrial
AATase as well as similar results obtained by Arnone et al. (1982) for cytosolic
AATa,se from pig establish the occurrence of such a rotation. Our model building
experiments confirm the necessity, for steric reasons. of a coenzyme ring
reorientation. This is achieved by rotations around the bonds between the
pyridine ring and the phosphorus atom. This movement, of which the main
component is a rotation of the pyridine ring around C-5-C-5 (t’he axis suggested
by lvanov & Karpeisky, 1969) allows the formation of a covalent single and
subsequently double bond between C-4’ of t’he coenzyme and the substrate amino
nit’rogen atom which displaces the E-amino group of Lys258. In this process, the
hydrogen bond/salt bridge between Asp222 and the protonated pyridine E’- 1, that
stabilizes the latter and keeps the “electron sink” intact, is reinforced. The
st rurt UIW of PMP-enzyme, reduced PLP-enzyme and the pyridoxol phosphate
derivative suggest that a tilt angle of - 40” corresponds to the relaxed orientation
of t)he coenzyme. This tilt angle lies halfway between the minimum value of - 20”
in the PLP-enzyme and the limiting value of 60” to 70” as set by the indole ring of
Trp140; the dihedral angle around the C-5-C-5’ bond lies in the range 40” to 70”.
ltr contrast,, in the PLP-enzyme this dihedral angle is 25” or less, which is
sterically unfavourable. A strained internal aldimine structure is also signalled bj
non-planarity of the bonds on either side of the aldimine double bond and by a
torsion angle around C:-4-C-4’ of 5” to 10”. This strained conformation; which
appears t,o be stabilized by the strong hydrogen bond between Tyr225 OEH and
t)hr unprotonated O-3’ of the coenzyme, presumably facilitates transaldiminatiorr.
Such a possibility has been considered by Tumanyan et al. (1974). The coenzyme
tilt atlgle in the covalent intermediates is, within the range given above,
determined by the geometric constraints of the coenzym+substrate compounds
(I’.#. planarity around double bonds), by the tendency of the t,wo carboxylat’e
groups to make favourable hydrogen bonding and charge interactions, and by
steric constraints on possible positions of the subst~rate moiety, imposed mainly
t)y Tyr70 and Gly38 above the active site pocket and by Trp140 below.
Thr analogue T’PLAsp, although lacking the aldimine double bond, has a
tlclarly planar conformation of ring, C-4’, K and CA (Fig. 3). It is thus a good
model for the external aldimine intermediate with aspartate. The interactions
with Arg292. Arg386 and Trp140 orient the aspartate moiet,y in such a way that
thy (‘A- H bond is nearly normal to the coenzyme ring plane. wit)h t>hta

IX
516 J. F. KIRSCH ET AL.
a-hydrogen positioned close to the &-amino group of Lys258. The corresponding
situation in the external aldimine intermediate promotes labiiization of the CA
prot#on (Dunathan? 1971). The coplanarity of the nitrogen and CA atoms with the
pyridine ring in PPL-Asp is a clear indication that the active site structure in its
closed conformation stabilizes the external aldimine intermediate. The same
conclusion can be drawn from the interaction of 2-methylaspartate with AATase.
Both soaking experiments with cytosolie AATase (Arnone et al., 1982) and
cocrystallization experiments with mitochondrial AATase (%I. G. Vincent,
D. Picot, C. Thaller, G. Eichele, G. C. Ford & J. N. Jansonius, unpublished
results) have shown that this substrate anaiogue predominantly forms t,he
external aldimine rather than the Michaelis complex.
Evidence for a reorientation of the coenzyme ring during the transamination
reaction has been provided also by linear dichroism measurements on single
crystals of both the cytosolic (Metzler et al., 1978; Makarov et al., 1981) and the
mitochondrial isoenzyme (Vincent, et al., 1984).

(c) State of ionization of active site groups


The shifts in the pK values of coenzyme and substrate groups that accompany
and facilitate the individual elementary steps of enzymie transamination have
been described and reviewed extensively (Ivanov & Karpeisky, 1969; Braunstein,
1973). The present structural data are consistent with most of the previous results
and interpretations.
The positive charges of Arg292 and Arg386, together with the protonation of
the pyridine N-l and the hydrogen bond of O-3’ of PLP with the hydroxyl group
of Tyr225, lowers the pK, value of the Schiff base from an expected value of
roughly 11 (Metzler et al., 1980) to 6.2 (Eichele et al., 1978). The effect of the
arginine residues is demonstrated convincingly by the shift in pK, value up to
about, 8.0 on binding dicarboxylate inhibitors (Bonsib et al.. 1975). Neutralization
of the a-carboxylate of the amino acid substrate lowers the pK value of the
a-amino group from -9.7 to -7.4 (Perrin, 1965). In the covalent intermediates,
the breaking of the hydrogen bond between Tyr225 OEH and coenzyme O-3’,
which exists in the unliganded PLP-enzyme, would favour the formation of a
hydrogen bond between O-3’ and the protonated imine nitrogen atom. That
would, in turn, increase the pK, value of the latt)er b.y about two units
(Braunstein, 1973). This factor, combined with the effective arginine
neutralization by substrate, may explain the inability t,o titrate the external
aldimine below pH 11 (Jenkins & d’,Ari, 1966; Hammes & Tancredi, 1967; Harruff
& -Jenkins, 1978). The formal positive charge residing on the imine nitrogen atom
increases the acidity of t,he adjacent C:-H bonds over that normally provided by
the electron sink of the pyridine ring, and thus facilitates the formation of the
carbanion required in the next, step of the reaction.
The V,,,/K, {aspartate) versus pH profile shows inflection points at pH 6.3 and
pH 94 (Velick & Vavra, 1962). A pK, of 9.2 in V,,,jKm is also seen for
2-oxoglutarate (Kiick & Cook. 1983) and must therefore reflect a prototropic
MEC’HANISM OF ASPARTATE AMINOTRANSFERASE 517

group of the enzyme. A possible candidate for the high pK, acidic group is the
pyridinium nitrogen atom. Due to hydrogen bonding and charge interaction with
Asp22S, its pK, value is expected to be raised drastically above the value of 5.9
(protonated imine nitrogen hydrogen-bonded to O-3’; Metzler et al.: 1980) or 8-4
(deprotonated imine nitrogen; Tvanov & Karpeisky, 1969) proposed for the non-
enzyme-bound coenzyme. The corresponding moiety of the 5’-deoxy-pyridoxal
complex with pyridoxamine pyruvate transaminase has a pK, value of 8.35
((iilmrr bz Kirsch, 1977). Tt is not known if similar interact,ions between the
pyridoxal derivat,ives and the protein exist in this case.
However, the pyridine N-l does not contribute to the binding of the substrates,
thus its pK value would not only be reflected in the V,,,,,/Km vwsus pH profile but
also in the V,,, uersu~s pH plot, which, however, is not the case. Binding of
substrate may even further increase its pK value due to neutralization of positive
charges and thus keep t,he electron sink intact over t,he whole invest,igated pH
range. This mechanism could explain the observed independence of V,,, from pH.
A possible explanation for the pK, value of - 9 observed from the 0x0 acid side
of’ the reaction (Kiick & Cook, 1983) could be the first protonation of the pair of
amino groups of PMP and Lys258, which are at hydrogen-bonding dist’ance in thr
free pyridoxamine form. The two amino groups thus may share the additional
proton, a suggesti~)nmade by Fasella et al. (1966). Lys258 has a pK, value of X in
the apo form (Slebe & Martinez-Carrion, 1976) and of 8.2 in the complex of the
rnzyme wit,h the nuclear magnetic resonance probe phosphopyridoxyl
trifluoroethylamine (Plebe & Martinez-Carrion, 1978). Hydrogen bonding will
increase this pK, value and could therefore explain the inflection point at pH -9.
A positive partial charge on t~heamino group of PMP would explain also the low
af?inity of amino acids toward the PMP form of the enzyme. Binding of an oxo
atsid increases the p& value due to the neutralization of the positive charges of
thca two arginim residues. Nucleophilic attack by the amino group of thr
coenzyme requires the hydrogen bond t,o be disrupt’ed and the proton t)o tw
localized on Lys2.58. On elimination of water, the charge is transferred to thtb
ketimine N. The absence of any inflection point in the lT,,% 2lersuspH plot that
u-ould indicate a pk’, value of a group in the enzyme-substrat,e ctomplex is in
at.cordancbewith this proposal.
The lower inflection point at pH 6-3 observed with aspartate almost certainI!
can be ascribed to Schiff base protonation (Eichele et al.? 1978). The pK, value of
5.X detected with Soxoglutarate, reflecting an acid/base group in the PMP form.
may result from the second protonation of t,he PMP-Lys258 diamine complex
(Fasella et al., 1966: Kiick & Cook, 1983).

(d) Kinetic considerations


In Figure 4, the most, abundant protonic forms of the amino acid substrate and
enzyme bound PLP between pH 6.2 and 9.9 are shown to produce tahetetrahedral
intermediate (stage II). The kinetically indistinguishable pathway involving the
deprotonated amino acid and protonated Schiff base would, in view of the high
518 ,J. F. KIRRCH ET AL.

rate constant of the transaldimination step (10’ to lo8 M-~ s-l) as determined by
temperature jump measurements (Fasella & Hammes, 1967), require association
rate constants exceeding the diffusion controlled limit’.
Carbonyl substitution reactions such as transaldimination and ketimine
formation are also extremely rapid for non-enzymic intramolecular
transimination reactions catalysed by PLP, and it has been suggested that in
these stages of the reaction sequence (Fig. 4, stage I to III and V to VI) the
enzyme provides only the entropic assistance of concentration and orientation
(Tobias & Kallen, 1975). Our hypothetical mechanism includes participation of
the protein, namely through strain imposed on the internal aldimine and through
the assistance of Lys258 in proton transfer during ketimine hydrolysis. In the
closed active site, which has room for only a few water molecules. we believe the
lysine e-amino group to be the most likely acceptor of the water proton.
Intramolecular proton transfers between heteroatoms as occurring between the
two nitrogen atoms during transaldimination are very rapid (e.g. see Eigen, 1964).

(e) Candidates for the proton donor and acceptor function


A comparison of the rate of enzymic transamination with that of the
corresponding non-enzymic reaction indicates that in the enzyme the
13prototropic shift must be catalysed by an acid/base group (Auld & Bruice,
1967). In the enzymic reaction, tautomerization is at least partly rate-
determining, as shown by the decrease in V,,, values by a factor of 1.5 to 2 when
a-deuteroamino acids are used as substrates (Banks et al., 1968; Fang et al., 1970;
Kirsch & Julin, 1982). On the basis of indirect experimental evidence, the lysine
residue forming the internal aldimine, i.e. Lys258, as well as a histidine residue
have been longstanding candidates for the role of acid/base catalyst (Snell, 1962;
Braunstein, 1973; Fasella et al., 1978). The active site geometry rules out a
histidine residue, because none is situated close enough to CA and C-4’. The
inactivation accompanying chemical modification of a histidine residue (Martinez-
Carrion et al., 1967; Yamasaki et al., 1975; Polidoro et al., 1976; Azaryan et al.,
1976) might be due to destruction of His143 involved in a hydrogen bond with
Asp222.
In the PPL-Asp derivative, and thus presumably also in the external aldimine
intermediate, the CA-H bond points “up” (positive Z direction, Fig. 3). The
only potential proton acceptor groups close enough to act as the base in the
deprotonation step are the amino group of Lys258 and the phenolic oxygen of
Tyr70. In all derivatives of mitochondrial AATase studied crystallographically in
which Lys258 is not covalently connected to the coenzyme, its a-amino group
forms a hydrogen bond with the phenolic OH of Tyr70. The s-amino group of
Lys258 is anchored above CA in an excellent position for proton transfer (distance
in the PPL-Asp derivative -3.7 A, see Table 1). Tyr70 OEH appears the less
likely alternative because of its less favourable distance from CA (4.3 A in the
PPL-Asp derivative) and to C-4’ (4.6 A in the PPL-Asp derivative). Its proton is
engaged in a hydrogen bond with an oxygen atom of the phosphate group of PLP.
Reports on nuclear magnetic resonance studies on the state of ionization of the
MECHANISM OF ASPARTATE AMINOTRANSFERASE ,519

phosphate group in the mitochondrial (Mattingly et al., 1982) and the cytosolir
isoenzyme (Martinez-Carrion, 1975; Martinez-Carrion et al., 1984; Schnackerz.
1984) agree that the phosphate group is dianionic in the PLP form of the enzyme
at neutral pH but disagree with respect to its pK, value and the state of
ionization in other functional states of the enzyme. Furthermore, the
interpretation of these data may be inconclusive because the observed chemical
shifts might not (or not only) reflect a change in the state of ionization, but rather
a distortion of the tetrahedral symmetry of the phosphate group (Gorenstein,
1975; Mattingly et aE., 1982). Thus, it remains unclear whether the phosphate
group could abstract a proton from Tyr70, rendering it a possible acceptor for the
a-proton. It seems more likely that the importance of Tyr70 is the proper
positioning of the amino group of Lys258. A schematic representation of the
proton shift process is given in Figure 6. The Lys-Tyr-phosphate network in its
active state contains three protons when the phosphate moiety is dianionic. The
pH dependence of V,,, (Velick & Vavra, 1962; Kiick & Cook, 1983) suggests that
the pK, value of the enzyme-substrate complex that might correspond to
proton&ion of the lysine residue is <6 in the intermediate stages III to V of
Figure 4. We propose that Lys258 is also the donor in the protonation of C-4’. In
the PPL-Asp derivative, the distance from C-4’ to NZ is 3.6 8, while Tyr70 OEH
is 1.6 a away. The increase in tilt of the coenzyme in the quinonoid intermediate
would move CA away from the E-NH, of Lys258 and improve the position of
(‘-4’ for protonation by the latter. The observed transfer of 3H from CA of the
substrate glutamate to C-4’ of PMP during transamination (Gehring et aE., 1983)
supports the hypothesis of a single proton acceptor/donor group effecting fhe
1.%prototropic shift. A similar intramolecular proton transfer excluding concerted
tleprotonation/protonation has been found for pyridoxamine pyruvatr

Tyr70 Lys258

/ .,.
H. ‘H

FIG. 6. The hydrogen bonds between Lys258, Tyr70 and the phosphate group of the coenzyme in the
external aldimine intermediate. The 1,3-prototropic shift is indicated, but the drawing should not be
interpreted as suggesting concerted deprotonation/protonation.
520 J. F. KIRSCH ET AL.
aminotransferase (Dun&than, 1971) and has repeatedly been proposed for AATase
(seeBraunstein, 1973).

(f) Conformational changesof the protein


The movement of the small domain during the catalytic process appears to be
the cause of the enhanced reactivity of Cys166 in the mitochondrial (Gehring &
Christen, 1978) and of Cys390 in the cytosolie enzyme (Birchmeier et al., 1973) in
the presence of an amino acid/ox0 acid substrate pair. Cysl66 is situated at the
“tip” of the PLP-binding domain most distant from the 2-fold axis at the
interface with the small domain, the movement of which changes the environment
of the cysteine residue (see Fig. 2). The retardation of peptide hydrogen exchange
observed in the (cytosolic) enzyme undergoing the transamination reaction
(Pfister et al., 1978) presumably relates to the tightening of the protein structure
around the substrate.
In the crystallographic studies of mitochondrial and cytosolic AATase, the
movement of the small domain was observed in complexes with the dicarboxylic
inhibitors maleate and succinate and with analogues of covalent intermediates of
transamination. The external aldimine with 2-methylaspartate (Fasella et aE.?
1966) and the quinonoid intermedia~ with er~~hro-3-hydroxyaspart,ate (Jenkins,
1964; Hammes & Haslam, 1969) are also in the closed conformation as expected
(Table 3). Because of its structural resemblance to the PPL-Asp derivative, we
expect the ketimine intermediate to be in the closed form as well. The same
inhibitors and substrate analogues that bring about the closed conformation also
induce an enhanced reactivity of Cysl66 (Table 1, and unpublished results). What
triggers the conformational change? The fact that the apoenzyme in the presence
of maleate crystallizes in the closed conformation (Table 3) suggests that the
compensation of the charges of Arg386 and Arg292 plays a major role. This
charge compensation probably also explains the closed conformation in crystals of
eytosolic AATase from chicken (Table 3) in the presence of 60% saturated
ammonium sulphate and O-15 M-potassium phosphate (Borisov et al., 1978). The
present crystallographic finding that the closed structure is induced by the
formation of the adsorption complex and the increased react,ivity of Cysl66 in the
presence of maleate (Table 2), are in apparent conflict with previous studies with
the enzyme in solution (Gehring & Christen, 1978). Apparently, a major shift in
the open/closed equilibrium of the PLP form of the enzyme is not “syncatalytit”,
i.e. it is not concomitant with the changes in covalency but rather is a ligand-
induced conformational change, which takes place on formation of the enzyme-
substrate adsorption complex. The corresponding data for the PMP form are thus
far inconclusive. Previous differential chemical modificat,ion studies of Cys166 had
indicated that the formation of the non-productive enzyme-substrate complexes
(PLP-form with 2-oxoglutarate or oxaloaeet,ate and PMP-form with glutamate or
aspartate) gave no or only a slight increase in reactivity of Cysl66, while an
increase by about one order of magnitude was observed in the presence of a
transaminating substrate pair. A likely interpretation of the previous and present
data is that the observed shifts in the equilibrium between the open and closed
ME(IHANISM OF ASPARTATE AMINOTRANSFERASE ,521

TABLE 3
Occurrence of the open or closedconformation in unliganded and liganded forms of
AATase in the crystalline state

Open conformation Closed conformation

.A. Mitochondrial enzyme from chicken


PLP-enzyme PPL-Asp derivative”, b
PMP~enzymr PPL-Glu derivati&
Apoenzyme PLP-enzyme + maleate”~b
Rrdwed PLP-enzyme PLP-enzyme + 2-methylaspartateb
Pyridoxol-P derivative Apoenzyme + maleate’
I’l’LAla derivative
I’I’L-:I-iodo-Tyr derivative
PLI’-cwzyme + 2-oxoglutaratx!

1% (‘ytosdir rnzyme from chicken’


I’LP-enzyme Apoenzyme
PMP-enzyme PLP-enzyme + maleate
I’l,P~isonic,otinyl hydrazide derivative PLP-enzyme fsuccinate
PLP-enzyme + 2-methylaspartate
PLP-enzyme + erythro-3-hydroxyaspartate

C’. ( ‘ytosolic rnzymr from pig’


PLP-enzyme PLP-enzyme + 2-methylaspartate

* If produced by soaking, only one subunit (81, top subunit in Fig. 2) changes into the closed
conformation, the other (S, bottom subunit in Fig. 2) is constrained in the open conformation.
’ Cocrystallization gives a different crystal form, in which both subunits are in the closed
conformation (M. G. Vincent, D. Picot, C. Thaller, G. Eichele, G. C. Ford & J. S. Jansonius.
unpublished results).
’ 11. Picot (unpublished data).
” Eichele (1980).
’ Borisov rt al. (1978,198(j).
f Arnone rt al. (1982).

structures correspond to rather small free energy increments. Because of the small
free energy changes, lattice forces might interfere with the transconformational
equilibrium. High-resolution studies of the enzyme structure in the closed
conformation should clarify whether small local conformational changes of the
protein correlated with the coenzyme movement occur. Similarly, subtle
differences in the geometry of the enzyme substrate intermediates of C4 and C,
substrates, as suggested by model building, must be confirmed by future high-
resolut,ion studies.

(g) Substrate specificity


The crystallographic studies with PPL-amino acid derivatives indicate that the
substrate specificity of the enzyme is determined largely by the interactions of the
two substrate carboxylate groups with Arg386 and Arg292. The importance of
these interactions for both the narrowing of the active site cleft and efficient
cat,alysis is demonstrated by the failure of PPL-Ala and PPL-3-iodo-Tyr to
induce the conformational change and by the lo5 and IO3 times slower turnover
522 J. F. KIRSCH PT AL.

of alanine and aromatic amino acids, respectively, as compared to that of


dicarboxylic substrates (Mavrides & Christen, 19’78). The different orientation of
the a-carboxylate group of the amino acid moieties in these two derivatives with
respect to that of the PPL-Asp derivative implies a different position of the
u-proton, which would make deprotonation by Lys258 in the corresponding
external aldimine intermediate more difficult. Thus, the PPL-amino acid
derivatives explain at least qualitatively the substrate specificity of AATase. The
difference in turnover of alanine and aromatic amino acids is more difficult to
explain. Perhaps the much smaller alanine, which does not fill up the active site
pocket, has more possibilities for unproductive binding.
The role assigned above to Arg292 is fully compatible with the complete
protection of this residue by a transaminating substrat#e pair against chemical
modification by phenylglyoxal and with the functional properties of the modified
enzyme (Sandmeier & Christen, 1982).

(h) Comparison with stereochemical data


In the internal aldimine, the B face of the coenzyme (Ford et al., 1980) is
exposed to solvent, and the aldimine double bond is &s&d, i.e. the aldimine-N is
on the 3’-OH side (Fig. 1). This situation corresponds to solvent exposure of the re
face with respect to C-4’. Indeed, reduction of the double bond in the “internal”
aldimine takes place on this face (Austerm~hle-Bertola, 1973). Dunathan (1971)
concludes, on the basis of stereochemical considerations, that the external
aldimine intermediate (Fig. 4, stage III) exists also in the cisoid conformation and
that the a-carboxylate group lies on the same side as the 3’-OH group. This again
is in agreement with our findings. From the discussion above, it follows that the
13prototropic shift must be a cis process. The acid/base catalyst, Lys2.58 lies above
the A face of the ring plane of the forward tilted coenzyme. The CA-H bond in
the PPL-Asp derivative is oriented approximately perpendicular to the coenzyme
ring plane, in accord with Dunathan’s postulate for the most favourable
conformation for proton abstraction in the external aldimine (Dunathan, 1971).
Consistent with the change in tilt of the ~oenzyme upon tr~saldimination is the
ending in Arigoni’s laboratory that the internal aldimine is reduced by
borohydride from the re side, whereas the external aldimine is reduced from the si
side (Austermiihle-Bertola, 1973). In contrast, in AATase carbamylated at
Lys258, the reduction of the external aldimine with borohydride occurs from the
re side (Zito & Martinez-Carrion, 1980). Apparently, in this enzyme derivative the
si side is even less accessible than the re side due to the carbamyl group on
Lys258. The stereochemistry of the C-4’ protonation in ‘Hz0 during a single
turnover reaction of the carbamylated enzyme nevertheless was found to agree
with that of the native enzyme. This finding has been interpreted as indicating
that Lys258 is not the proton donor/acceptor (Martinez-Carrion et al., 1979). The
extremely slow rate of the reaction of the earbamylated enzyme, however, does
not seem to justify such a conclusion.
M E(‘H.4NISM OF ASPARTATE AMINOTRANSFERAGE 523

(i) ConclwEing remarks


The proposed mechanism appears to be compatible with the spatial structures
reportSed for the cytosolic isoenzyme of AATase from pig (Arnone et al., 1982) and
(+icken (Korisov et al., 1980; Harutyunyan et al., 1982). In particular, the
movements of the small domain and of the coenzyme have also been observed in
the cytosolic isoenzyme. A detailed comparison of the two homologous isoenzymes
will be necessary in order to identify the structural features underlying their slight
but definite functional di~erences (see Brauustein, 1973). A more accurate and
c*ttmplete description of the mechanism of action will have to await the
determination of the structure of productive enzyme substrate intermediates (by
(Lryocrystallography) and a closer delineation of the conformational changes
acncompanying the catalytic process. In particular, the interplay between the
reorientations of the coenzyme and the conformational changes of the protein
matrix will have to be investiga~d in detail. Another important question is t,hat
of thr triggering mechanism for the small domain movement.
The present hypothetical mechanism should provide a framework within which
the problem of protein-assisted pyridoxal catalysis can be pursued further. The
elucidation of the structures of further pyridoxal enzymes will eventually allow an
approach at the atomic level to the challenging problem of how the apoprotein
moietiex of the various PLP-enzymes effect reaction- and substrate-spe~~ti(
c~tl alysis.

We thank P. Battig, E. Bratschi, U. Fiirstenberger, A. Uasser. .4.-S. Niklaus and


E. Wilson for expert technical assistance; T. Teshiba for skilfully preparing the artistic,
drawings in Fig. 5; D. Picot for discussions and the permission to use some of his
linpublished results; and T. Balsiger for her invaluably help in the preparation of the
manuscript. We are indebted to the director and staff of the Base1 University Computer
f ‘entre for the useof computing and graphic display facilities and to Dr E. I,. Mehler for
c*omputing assistance.
We gratefully acknowledge financial assistance from the Swiss National Science
Foundation (grants 3.217-0.77, 3.569-0.79 and 3.142-0.81 to P.C.; grants 3.415-0.78,
3.415 I .78 and 3.224-0.82 to J.N.J.) as well as from the Fritz Hoffmann-La Roehe Stiftung
(to P.f ‘. and ,J.N..J.. Arbeitsgemeins~haft no. 159).

REFERENCES
.ingielski, S. & Rogulski, J. (1962). Acta Biochim. Pal. 9, 357-362.
At-none, A., Briley, P. D., Rogers, P. H., Hyde, C. C., Metzler, C. M. & Metzler, D. E.
(1982). In ~o~~~~ZarStructure and BioZogicd Activity (Griffen, .J. F. & Duax. ‘17’. L..
eds), pp. 57-74, Elsevier North-Holland Inc., New York.
Auld, If. S. & Bruice, T. C. (1967). J. AmeT. Chem. Sot. 89, 2090-2097.
Austermiihle-Bertola, E. (1973). Thesis no. 5009, Eidgenijssische Technische Hochschule.
Zurich.
Azaryan, A. V., Mekhanik, M. 1,. & Torchinskii, Yu. M. (1976). Hiokhimiya, 41, 2075-2077.
Banks. 12. E. C., Bell, M. P., Lawrence. A. J. $ Vernon, C!. A. (1968). In f>yridoraZ
C&alyais: ~nzyrn~s and Model Systema (E. E. Snell et al., eds). pp. 191-202. Wiley,
New York.
Bennett, W. S. & Steitz, T. A. (1978). Proc. Nut. Acad. AS&, U.S.A. 75, 4848-4852.
Birc*hmeier, W., Wilson, K. J. & Christen, P. (1973). J. RioE. C&m. 248, 1751-1759.
Blundell, T. I,. & Johnson, L. N. (1976). Protein CrystaZlography, pp. 420-421, Academic
Press. New York.
524 tJ. F. KIRSCH ET AL.

Bonsib, S. M., Harruff, R. C. & Jenkins, W. T. (1975). J. Biol. Chem. 250, 8635-8641.
Borisov, V. V., Borisova, S. N., Kachalova, G. S., Sosfenov, N. I., Vainshtein, B. K.,
Torchinsky, Yu. M. & Braunstein, A. E. (1978). J. Mol. Biol. 125, 275-292.
Borisov. V. V., Borisova, S. N., Sosfenov, N. I. & Vainshtein. B. K. (1980). Nature
(London), 284, 189-190.
Braunstein, A. E. (1973). In The Enzymes (Boyer. P. I)., ed.). 3rd edit.. vol. 9. pp. 379481,
Academic Press, New York.
Braunstein, A. E. & Shemyakin, M. M. (1953). Biokhimiya, 18, 393411.
Colman. P. M., Jansonius, ,J. N. & Matthews, B. W. (1972). J. Mol. Biol. 70, 701-724.
Doonan, S., Doonan, H. ,J., Hanford, R., Vernon, C. A., Walker. J. M.. Bossa, F., Barra,
D., Carloni. M.. Fasella. P., Riva. F. & Walton, P. L. (1974). FEBS Letters, 38. 22%
233.
Dunathan, H. C. (1971). Advan. Enzymol. 35. 79-134.
Eichele, G. (1980). Ph.D. thesis, University of Basel.
Eichele, G., Karabelnik, D., Halonbrenner, R.. Jansonius, J. N. & Christen, P. (1978). J.
Biol. Chem. 253, 5239-5242.
Eichele. G., Ford, G. C.. Glor, M.. Jansonius. J. S., Mavrides, C. & Christen, P. (1979). J.
Mol. Biol. 133. 161-180.
Eigen, M. (1964). Angew. Chemie Znt. Ed. 3, l-19.
Fang, S.-M., Rhodes, H. J. & Blake, M. I. (1970). Biochim. Biophys. Actu, 212, 281-287.
Fasella. P. & Hammes, G. G. (1967). Biochemistry, 6, 1798-1804.
Fasella, P., Giartosio. A. & Hammes, G. G. (1966). Biochemistry, 5, 197-202.
Fasella, P., Carotti, D.. Giartosio, A., Riva, FE‘.& Turano, C. (1978). In Enzyme-activated
Irreversible Inhibitors (Seiler, N.. Jung, M. J. & Koch-Weser, ,J., eds), pp. 87-107,
Elsevier North-Holland Inc., New York.
Federiuk. C. S. & Shafer, cJ. A. (1983). J. Biol. Chem. 258, 5372-5378.
Ford, G. C., Eichele. G. & ,Jansonius, J. K-. (1980). Proc. Xzt. Acud. Sci., 1I.N.A. 77, 255%
2563.
French, T. C., Auld. D. S. & Bruice, T. C. (1965). Biochemistry, 4. 77-84.
Furbish, F. S., Fonda, M. L. & Metzler, D. E. (1969). Biochemistry, 8. 516%5180.
Gehring. H. & Christen, I’. (1978). J. Biol. Chem. 253, 3158-3163.
Gehring, H.. Christen, P., Eichele, G., Glor, M., Jansonius? J. N., Reimer. A.-S.. Smit.
.J. I). G. & Thaller. C. (1977). ,I. Mol. Biol. 115. 977101.
Gehring, H., Sandmeier. E., Kirsten, H., Carrassi. R., Christen, P. & Jansonius, tJ. N.
(1984). In Chemical and Biological Aspects of Vitamin B6 Catalysis (Evangelopoulos.
,4. E., ed.), Liss, New York, in the press.
Gilmer. 1’. .J. & Kirsch, .J. F. (1977). Biochemistry, 16, 524&5253.
Gorenstein. 1). G. (1975). J. Amer. Chem. Sot. 97, 898-900.
Graf-Hausner. IT.. Wilson. K. J. & Christen, P. (1983). J. Biol. (‘hem. 258, 8813-8826.
Hammes. G. G. & Haslam, ,J. L. (1969). Biochemistry, 8, 1591~1598.
Hammes, G. G. & Tancredi, J. F. (1967). Biochim. Biophys. Actu, 146. 312-313.
Harruff, R. C. & <Jenkins, W. T. (1978). Arch. Biochem. Biophys. 188, 3746.
Harutyunyan, E. G.. Malashkevich, V. N., Tersyan. S. S., Kochkina, V. ,M.. Torchinsky.
Yu. M. & Braunstein, .4. E. (1982). FEBS Letters, 138, 113-116.
HOI. W. G. ,J., van Duijnen, P. T. & Berendsen. H. .J. C. (1978). ,%ture (London), 273,
443-446.
Hughes, R. C., Jenkins, W. T. & Fischer, E. H. (1962). Proc. iVat. Acud. Sci., U.S.A. 48,
1615-1618.
Ivanov. V. I. & Karpeisky, M. Ya. (1969). Advun. Enzymol. 32. 21-53.
Jenkins, W. T. (1964). J. Biol. Chem. 239, 1742-1747.
Jenkins, W. T. & d’Ari, L. (1966). J. Biol. Chem. 241. 2845-2854.
Jenkins, W. T. & Sizer. I. (1957). J. Amer. Chem. Sot. 79, 2655-2656.
Jenkins. W. T., Yphantis, D. A. & Sizer, I. W. (1959). J. Biol. Chem. 234, 51-57.
Khomutov, R. M., Severin, E. S.. Govaleva, H. K.. Gulyaev, N. N., Gnuchev, N. V. BE
Sastchenko, L. N. (1968). In Pyridoxul Catalysis: Enzymes and Model Systems (E. E.
Snell et al., eds), pp. 631-650, Wiley, New York.
M~(~~A~IS~ OF ASPARTATE A~I~~TR.~~SF~RAS~ .x5

Kiick, D. M. & Cook, P. F. (1983). ~~~~rn~~~r~, 22, 375-382.


Kirsch, ,J. F. & ,Julin, D. A. (1982). Fed. Proc. Fed. Amer. &c. Exp. Bioi. 41. 628.
Makarov. V. L., Kochkina, V. M. & Torchinsky, Yu. M. (1981). Biochim. Biophys. .4&r.
659 219-228.
Martinez-Carrion. M. (1975). Eur. J. Biochem. 54, 39-43.
Martinez-Carrion. M.. Turano, C., Riva? F. & Fasella, P. (1967). J. Biol. Chem. 242, 1321i-
1430.
Martinez-Carrion, M., Plebe, J. C. & Gonzalez, M. (1979). J. Biol. Chem. 254, 3160-3162.
\lartinez-Carrion, M., Mattingly, J. & Iriarte, A. (1984). In Chemical and Biological Aspects
oj liitnmin R6 Cata1ysi.s (Evangelopoulos, A. E.? ed.), Liss, New York, in the press.
Xlxttingly. M. E.. Mattingly, J. R. & Martinez-Carrion, M. (1982). J. Biol. Chem. 257. 88%
8878.
Mavrides. C. Cy:Christen. P. (1978). ~,~oc~ern. Bio~~ys. Res. Commun. 85. X9-773.
JJetzler. (1. 31.. Metzler, D. E., Martin D. S., Newman, R., Amone. A. & Rogers. P. (1978).
.I. Biol. C&m. 253. 5251-5254.
Metzler. (1. M.. Cahill. A. & Metzler, D. E. (1980). J. ilmer. L’hem. L&X. 102, 6075-6082.
Metzler. 1). E. (1979). Advan. Enzymol. 50. l-40.
1letzler. D. E.. Ikawa, M. & Snell, E. E. (1954). J. Amer. Chem. Sot. 76, 648-652.
Jlirhuda, C. M. & Martinez-Carrion, M. (1970). J. Biol. Chem. 245. 262-269.
\lorimoto. (‘. S. Nr. Meyer, E. F. Jr (1976). In Crystallographic Computing Techniques
(Ahmed. F. R.. ed.), pp. 4888496, Munksgaard. Copenhagen.
()vchinnikov. Yu. A.. Egorov, C. A., Aldanova, N. A., Feigina. M. Yu., Lipkin, V. M..
Abdulaev. S. G., Grishin, E. V., Kiselev, A. P., Modyanov. ,hu‘.R;., Braunstein. A. E..
Polyanovskv. 0. L. & Kosikov, V. V. (1973). FERS Letters, 29, 31-34.
l’errin. C. (1965). ~~~~~~~~ci~~~on Constants oj Organic Bnsrs in .3queous Solution,
But~,erw(~rt,hs, London.
t’fister, K., KS@, J. H. R. & Christen, P. (1978). Proc. Nat. Acad. Sci.. C’.S.B. 75. 145 148.
f’olidoro. G.. di Cola. D., di Ilio. C.. Politi. L. & Scandurra. R. (1976). .Wol. cell. ~~oc~e~~.
11. 1.55--1.59.
Remington, S.. Wiegand. G. & Huber, R. (1982). J. &fol. Biol. 158, 111-152.
Richards. F. M. (1968). J. rWoE.Biol. 37, 225-230.
Sandmeier. E. & Christen, P. (1980). J. Biol. Chem. 255, 10284-10289.
Sandmeier. E. 6%Christen, P. (1982). J. BioE. Chem. 257, 674Ft6750.
Schlepel, H. & Christen, I’. (1974). Biochem. Biophy,s. Res. Commun. 61, 117-123.
Schlepel. H. $ Christen, I’. (1978). Biochim. Biophys. Acta, 532, 6-16.
Schnackerz. K. 11. (1984). In Chemical and Biological Aspects of Vitamin B6 Cata.lysis
(Evangelopoulos. A, E.. ed.), Liss, Xew York, in bhe press.
Slebr, .J. C. Sr Martinez-Carrion, M. (1976). J. BioE. Chem. 251. 5663-5669.
Stebe. ,J. (‘. & Martinez-Carrion, M. (1978). J. Biol. Chem. 253, 2093-2097.
Snell, It:. E. (1962). Broo~h~ven Symp. BioE. 15, 32-51.
Snt~Il. F:. E. & di h1ari.S.J. (1970). InTheEnzym.vs(f3over. 1’. I)..ed.).Yrderiit.. vol. 2, pp, 33h
370. Academic Press, New York.
‘I’hallrr, (‘.. Weaver. L. H., Eiehele, G., Wilson, E., Karlsson, R. & ,Jansonius, f. S. (1981).
.I. Moi. Niol. 147, 46fA69.
Tobias. I’. S. I(r Kallen. R. G. (1975). J. Amer. Chem. Sot. 97. 6530-6539.
Tumanyan, V. (:.. Mamaeva, 0. K.. Bocharov, A. I,., Ivanov. V. T., Karpeisky. M. Y. &
Yakovlrv. (:. I. (1974). Eur. J. Biochem. 50, 11%127.
\.rlick. S. F. &, Vavra. ?J. (1962). J. BioE. Chem. 237, 2109-2122.
Vincerlt. M. r:.. Picot. D., Eichele, G., ,Jansonius, J. N., Kirsten, H. & Christen, I’. (1984).
III (‘hwaical I& Biological Aspects of Vitamin B6 Catalysis (Evangelopoulos, A. E..
WI.), Lisx, Krw York, in the press.
t.arnasaki. M., Tanasr, S. & Merino, Y. (1975). Biocham. Biophya. Rex f ‘ommun. 65. 652-
657.
%ito. S. VV. & Martinez-(~~arrion, M. (1980). J. Biol. Chem. 255, 8~.~8649.

Educedby R. Huber

Das könnte Ihnen auch gefallen