Download BBSRC 24/B11662 "Protein processing and electron transfer in

FULL REPORT
Objectives 1&2 : to investigate the chemical mechanism of formation of the thioether bond at the
active site through model compound and protein studies.
to investigate the chemistry underlying pro-sequence processing events through
mutagenesis and characterisation of resultant enzyme species.
Galactose oxidase (GO) (EC 1.1.3.9) catalyses the oxidation of primary alcohols including D-galactose, to
aldehydes. It is a monomeric, principally beta structure protein of three domains with a single copper atom and a
protein-derived tyrosine radical as cofactors. It belongs to a growing class of proteins that self-catalyse assembly of
their redox-active cofactors derived from active site amino acids. The enzyme exists in three redox states: reduced
(Cu(I)-Tyr); semi-reduced (Cu(II)-Tyr) and oxidised (Cu(II)-Tyry) that are distinguishable by UV-vis
spectroscopy.
41aa
Precursor GO
GO requires several post-translational modifications to generate the
active enzyme;
1. cleavage of a secretion signal sequence,
2. cleavage of an N-terminal pro-sequence,
3. formation of a thioether bond between Cys228 and Tyr 272, and
4. oxidation to the radical state.
An important tool in the analysis of processing is the resolution of
distinct forms of GO by SDS-PAGE. This allows ProGO, Premat GO
and mature GO to be distinguished by their distinct migration. Growth
under copper limited conditions led to three species that were
identified by N-terminal amino acid sequencing. Subsequently we
perfected the process of preparing copper-free GO from Aspergillus.
Processing step (1) will occur in the ER through signal peptidase
action. To date, we have not studied this step.
ss
18aa
Post-ER GO
PRO-GO
(Aspergillus
639aa
? pro
mature
6aa
17aa
PRE-MAT GO
(Pichia
MAT GO
Active MAT-GO
Cys22
Tyr27
Y
Tyr272 radical
To investigate step (2) we have performed crystallographic
analyses of proGO that reveal important interactions between
the prosequence and residues in the mature portion of the
protein. Regions of the protein involving residues (1 to 5, 189
to 200, 216 to 227, 254 to 261, 290 to 296) show a rmsd of
5.1Å compared with the mature enzyme structure. When the
prosequence is removed, there are substantial changes in
these regions of the protein with some residues that make
interactions with the prosequence now interacting with other
residues in the mature protein. There are major effects on the
active site. For example Trp290 Cα moves by 6.3Å.
Soaking experiments where copper was introduced into
proGO crystals under anaerobic conditions reveal a bound
copper with His496, His581 and Tyr495 as ligands, but the
prosequence is retained and the thioether bond does not form.
This suggests that processing can not occur in this crystal
form under these conditions.
We have recently generated an antibody that is specific
for the prosequence by a peptide synthesis route. We have demonstrated that this antibody will cross react only
with proGO and not with other forms of the protein lacking the prosequence. Use of this antibody to analyse
samples of Aspergillus expressing GO reveal detection of GO both within the mycelium and in the supernatant.
Notably the form of GO identified within the mycelium is the mature protein We see no proGO within the
mycelium. This provides the first direct evidence for prosequence processing within the cell. It seems likely
that the processing occurs quite rapidly following synthesis, accounting for undetectable steady-state levels of
proGO. Speculatively we propose that processing is most likely to occur in the Golgi and experiments with
secretion inhibitors, such as brefeldin A, are important to determine whether
we can detect accumulation of proGO following imposition of secretion
blocking agents. The anti-proGO antibody also provides a tool for the
development of techniques to monitor prosequence cleavage in solution.
To study steps (3) and (4), we have purified a premature form of GO
(prematGO) from metal-free growth conditions, that does not contain the
prosequence. We have previously used Aspergillus nidulans as the
heterologous expression host for GO, which is naturally produced by the
filamentous fungus Fusarium graminearum. The generation of transformants and the relative difficulty of handling
Aspergillus and the time consuming nature of grow-ups led us to explore alternate expression systems for the
processing studies. We therefore established an efficient Pichia pastoris system through subcloning the gene inframe with the mating factor secretion sequence in the vector pPICαZ (Invtirogen). This system relies upon
induction of the aox promoter upon methanol addition to drive expression of the cloned transgene. We generated
two version of the GO gene in this vector, one carrying only the mature sequence and used for the premature GO
studies described below, and one carrying the mature sequence plus the 17 amino acid N-terminal prosequence
(proGO). The proteins were produced and characterised by a range of techniques including ES-MS, N-terminal
sequencing, ICP-MS for metal analysis, kinetics, UV-vis spectroscopy and crystallographic studies. The proteins
behaved as anticipated with the exception of prosequence processing in the proGO. ES-MS and N-terminal
sequencing indicated that prosequence processing did not occur in the same clean processing manner as in A.
nidulans. The major species was the mature protein, but there is also a
substantial contribution from a form with the residues -3 to -1 (SLR) of the
prosequence. This finding has limited our ability to study prosequence
processing in Pichia expressed enzyme; we have had to rely on the use of
A. nidulans and so have not investigated the range of mutational variants
we had intended to study. Most recently we have developed an expression
system based on E. coli, in which a directed evolution form of GO with
enhanced expression has been coupled with the inclusion of N-terminal
synonymous mutations to provide very high level expression (~20 mg/L).
This system produces the proGO intracellularly and therefore provides a
good expression system for producing protein for studying processing steps.
For our studies of step 4, we used the prematGO construct in Pichia under normal copper supplemented
growth conditions and demonstrated that the 3d structure of the mature form is identical to that previously
determined for other preparations of GO. We also performed spectroscopic and kinetic measurements that show the
protein has essentially the same properties as protein produced in our original expression host Aspergillus nidulans.
Defining appropriate metal-free purification conditions for the protein was not trivial and led to the understanding
that protein quality can vary. By this we mean that premat GO can be purified that behaves normally in terms of
thioether bond formation, but that will not oxidise when exposed to oxygen. This protein will however oxidise with
other oxidants such as ferricyanide. This presumably reflects some subtle difference within the enzyme. It thus
proved time-consuming to generate high quality copper-free enzyme for
the various studies reported here.
For the pro-form of GO, it is known that aerobic incubation in the
presence of Cu(II) leads to pro-sequence cleavage, thioether bond
formation and oxidation of the protein to the radical state. We have
shown that prematGO can similarly process under aerobic conditions to
give the active radical state consistent with it being a processing
intermediate. In addition, under anaerobic conditions, the thioether
bond forms efficiently but the enzyme is not oxidised to the radical
state; the resulting protein can however be oxidised to the radical state
by reaction with oxygen or Fe(CN)63-. Under anaerobic conditions
addition of Cu(II) leads to the formation of a 410 nm species that may
represent a processing intermediate. Exposure of this protein species to
oxygen leads to loss of the 410nm feature and rapid production of the
spectrum of oxidised GO (445 nm and broad 800 nm feature). We have
also determined 3d structures under cryocrystallography conditions for
the prematGO without Cu(II) and following brief (3min) soaking of apoform crystals with Cu(II) under anaerobic conditions.
This reveals an unusual copper (II) co-ordination with Cys228, His496 and His581. This SN2 co-ordination of
Cu(II) is supported by a range of other studies, including EXAFS, rR, EPR and comparision with a model Mop5
(Schnepf et al 2001 JACS 123, 2186) Longer incubation of the premat GO crystals with Cu(II) (24hr) reveals
formation of the thioether bond. This supports the solution studies and provides further evidence that thioether
bond formation can occur under anaerobic conditions.
Recently Whittaker and Whittaker (2003, JBC 278, 22090–101) have
reported processing of premat GO by Cu(I) and we have been able to
reproduce these results. We have also shown that Cu(I) supports anaerobic
formation of the thioether bond at a rate similar to Cu(II). These
observations indicate that an alternate electron acceptor must exist for
formation of the thioether bond since the exogenous electron acceptors
(Cu(II) & oxygen) are absent. The bond is also able to form at
stoichiometric and sub-stoichiometric concentrations of Cu(II) further
implicating an alternate electron acceptor, most likely residues on the
protein. One obvious potential site is the disulphide C515/C518 quite close
to the active site. However, titration of cysteine residues within GO before
and after processing provides no evidence for alteration of their redox state.
At present we do not know the identity of the endogenous electron
acceptor; however, in experiments performed under conditions of excess
Cu(II), the rate of thioether bond formation increases with copper concentration suggesting that Cu(II) can act as
an additional electron acceptor. We have recently solved the crystal structure of prematGO soaked with Cu(I)
under anaerobic conditions and the thioether bond has formed within 30 min. Earlier time points will reveal
important information about the initial Cu(I) co-ordination to prematGO and whether this differs from the SN2 coordination of Cu(II).
None of the crystallographic evidence so far reveals copper co-ordination to Tyr272 making the involvement
of a tyrosyl radical mechanism unlikely. We considered a sulphenic acid mechanism for bond formation since an
oxidised Cys228 was seen during the structural studies of the proGO, but processing is still supported albeit more
slowly, at pH9, while sulphenate mechanisms are not expected to operate above pH8.5. The observed interaction
between Cys228 and Cu leads us to propose a thiyl radical mechanism for the formation of the thioether bond.
B
We have generated a number of mutational variants of
D
A
C
active site residues, in particular Y495A, H496A, H581A,
Y272A to investigate the importance of copper binding at the
active site on the processing of GO. One difficulty we have faced
relates to the production of the proform of GO in Pichia pastoris
(see above)
The Y272A variant displays altered properties and Nterminal sequencing reveals the additional 6 amino acids that
E
precede the 17 amino acid region we call the prosequence. It is
not yet clear why this mutation results in this failure to process in
H
G
the presence of copper and further studies are necessary.
F
We have also
investigated the role of
the only free cysteine
in GO, C383 that lies at the back of the active site. Mutation of this
residue to serine leads to a 5-fold reduction in KM for GO, but the
enzyme still form the thioether bond normally. Interestingly a C383A
variant has a similar KM to wild type. We have determined the 3d
structure of this mutational variant to investigate the cause of the
reduction in KM but the structure does not provide a satisfactory
explanation. Remarkably however, the structure was determined in the
presence of –methyl-1,3-propandiol as cryoprotectant and this compound
410 nm Complex
S
Tyr 272
S-
HS
OH
OH
OH
Cys 228
S
OH
H
Cu(I)
Cu(II)
O2
Cu(II)
OR
HO2
X
Cu(I)/H+
Seen in
EPR
O-
H
S
Cu(II)
e-/H+
S
O
Cu(II)
S
O
OH
Cu(I)
S
H
Cu(I)
is observed in the active site of the enzyme as a potential substrate. This is accompanied by the presence of an
additional water in C383S compared to wild-type, that interacts with the OH of Tyr405 and therefore suggests an
altered pattern of H-bonds that may be responsible for the enhanced substrate binding to this mutant enzyme.
Objective 3 : to explore the nature and role of a putative electron transfer pathway between Cys515/Cys518
and the radical site in galactose oxidase by a combination of mutagenesis, biochemistry and spectroscopy.
The studies on the processing aspects of the enzyme proved more time consuming than anticipated, and due to
their exciting and topical nature, we focused more intently on these aspects than on the electron transfer aspect.
Nonetheless, we have generated mutations at Cys515 and Cys518 to explore their effects on radical decay.
Sykes and colleagues have performed pulse radiolysis experiments in which they claim to have identified a
transient cysteinyl radical. They ascribed the site involved to be Cys515 or Cys518, but provided no direct
evidence for this. They titrated the protein with thiophosphorylchloride and concluded that the modification
(presumed to be at C515/518) stabilsed the radical species and prevented the normal rate of radical decay. This
would be consistent with Cys515/Cys518 acting as an electron source during the decay process. We took a
different, more direct approach and mutated these residues to Ala. The enzyme processes normally and retains
enzyme activity, but the radical decays much more rapidly than in wild-type GO, and yet the disulphide bond is
absent. In a single mutant C518S, the rapid also decay more rapidly than wild-type, but more slowly than in the
double mutant. These data suggest that Cys515 and Cys518 are not involved in the radical decay pathway..
This conclusion contradicts that by Sykes and colleagues, although the experiments were admittedly different.
Perhaps they were titrating a different cysteine, although this seems unlikely. Cys383 should not be accessible, and
in any case we have shown by mutation, has no effect on radical stability. The second disulphide in GO is on
domain 1 (Cys18/Cys27), a long way from the active site and therefore an unlikely source of electrons. Finally it is
very unlikely that they would have modified Cys228 since this residue is involved in a thioether linkage.
We can not exclude a role for C515/C518 in enzyme processing. Under physiological conditions there may be
a wealth of exogenous oxidants capable of supporting thioether formation and radical oxidation. However, when
we force the reaction to occur under non-physiological in vitro anaerobic conditions, then perhaps the process
occurs by another pathway, perhaps involving these cysteines or other residues within the protein.
A new galactose oxidase
We have cloned and are beginning to characterise the GO homologue from F. veneatum that shows distinct
differences in substrate specificity and spectral properties. We are going to investigate the presence of a thioether
bond and whether this enzyme also undergoes similar post-translational cofactor processing.
Issues relating to the grant
The award of a single PDRA precluded any work on model compounds and therefore the work focused on protein
studies. The project started in Nov 1999 with the appointment of Dr C Allardyce. She worked on the grant for 8
months before leaving to move to London for personal reasons. The grant was put into abeyance for 6 months
pending the appointment of a new PDRA.
We believe that we have met many of the objectives set out in the original application and given the constraints
of reduced funding level and staff retention and reappointment issues, and the technical difficulties associated with
production of high quality copper-free precursor protein. These have led to some delay in publication of our work that
does not reflect the achievements made during this grant. The work advances our understanding of the posttranslational modification of an amino acid derived cofactor involved in the expansion of biological catalytic
capability. Such understanding is important for future manipulation of biocatalysts and should allow us to tailor
enzymes to perform new chemistries that are currently difficult with enzyme catalysts. The general area is one in
which there is considerable international interest and we regularly participate in the Gordon Conference dedicated to
this scientific area. The project has extended our collaboration with Prof. D M Dooley in Bozeman, Montana and we
have also has some collaboration with scientists at Hercules Inc. (Delaware and Barnveld, The Netherlands) who are
interested in industrial applications of GO as a biocatalyst. We have recently prepared manuscripts describing both
solution and structural studies of processing in prematGO and the characterisation of the C383S variant.
• D. Wilkinson, N. Akumanyi, R. Hurtado Guerrero, H. Dawkes, P. F. Knowles, S. E. V. Phillips and M.
J.McPherson Structural and kinetic studies of a series of mutants of galactose oxidase identified by directed
evolution, Submitted to Protein Engineering.
• R. Hurtado-Guerrero, M.S. Rogers, S. J. Firbank, D.M. Dooley<, M.A. Halcrow, S. E. V. Phillips, P.F. Knowles
and M.J. McPherson Formation of the thioether bond in galactose oxidase does not require oxygen. To be
submitted to Journal of Biological Chemistry
The work has been presented at several international meetings.
• Galactose oxidase post-translational processing. 54th Harden Conference on “Enzymology: emerging trends and
future prospects”, Ambleside, UK, 2002. y Cofactor processing and catalysis in galactose oxidase,
CERLIB,
Chamonix, France, 2003 y Cofactor processing and catalysis in galactose oxidase, Enzymology: a structural
perspective, Biochemical Society Focus Meeting, St. Andrews University, 2003 y Cofactor processing in
galactose oxidase, Biochemical Society Annual Symposium, Essex, 2003.