Download Structural adaptation to selective pressure for altered ligand

Protein Engineering vol.13 no.2 pp.129–132, 2000
Structural adaptation to selective pressure for altered ligand
specificity in the Pseudomonas aeruginosa amide receptor, AmiC
Bernard P.O’Hara1,2, Stuart A.Wilson1,4, Alan W.L.Lee1,
S.Mark Roe1,3,5, Giuliano Siligardi6, Robert E.Drew1 and
Laurence H.Pearl1,3,5,7
1Department
of Biochemistry and Molecular Biology and 3Joint UCL/LICR
X-Ray Crystallography Laboratory, University College London, Gower
Street, London WC1E 6BT and 6Pharmaceutical Optical Spectroscopy
Centre, Department of Pharmacy, King’s College London, Manresa Road,
London SW3 6LX, UK
2Present
address: Department of Crystallography, Birkbeck College, Malet
Street, London WC1E 7HX, UK
4Present
address: Department Biomolecular Sciences, UMIST, PO Box 88,
Manchester M60 1QD, UK
5Present
address: Section of Structural Biology, Institute of Cancer
Research, Chester Beatty Laboratories, 237 Fulham Road, London SW3
6JB, UK
7To
whom correspondence should be addressed; email: [email protected]
Bernard P.O’Hara and Stuart A.Wilson contributed equally to this work
The AmiC protein in Pseudomonas aeruginosa is the negative regulator and ligand receptor for an amide-inducible
aliphatic amidase operon. In the wild-type PAC1 strain,
amidase expression is induced by acetamide or lactamide,
but not by butyramide. A mutant strain of P.aeruginosa,
PAC181, was selected for its sensitivity to induction by
butyramide. The molecular basis for the butyramide inducible phenotype of P.aeruginosa PAC181 has now been
determined, and results from a Thr→Asn mutation at
position 106 in PAC181-AmiC. In the wild-type PAC1AmiC protein this residue forms part of the side wall of
the amide-binding pocket but does not interact with the
acetamide ligand directly. In the crystal structure of
PAC181-AmiC complexed with butyramide, the Thr→Asn
mutation increases the size of the ligand binding site such
that the mutant protein is able to close into its ‘on’
configuration even in the presence of butyramide. Although
the mutation allows butyramide to be recognized as an
inducer of amidase expression, the mutation is structurally
sub-optimal, and produces a significant decrease in the
stability of the mutant protein.
Keywords: crystallography/ligand specificity/mutation/protein
evolution/small molecule binding protein
Introduction
AmiC is the negative regulatory component of a ‘molecular
switch’ which controls amide-inducible expression of the
amidase operon of Pseudomonas aeruginosa (Wilson and
Drew, 1991). Functionally, AmiC controls the activity of a
transcription anti-termination factor, AmiR, which is required
for transcription of the amidase operon past a ρ-independent
terminator occurring downstream of the constitutive amidase
promoter (Drew and Lowe, 1989). In the presence of an
inducing amide, inhibition of AmiR by AmiC is relieved, and
expression of amidase and other genes of the operon (including
© Oxford University Press
amiC and amiR) is induced. Sensitivity to amides in vivo is
conferred by AmiC, which functions as a cytoplasmic amide
receptor, and has been shown to bind acetamide with micromolar affinity in vitro (Wilson et al., 1993).
Structurally, AmiC is a two-domain protein related to
periplasmic small-molecule binding proteins (SMBPs) such as
the Escherichia coli branched amino acid binding protein, LivJ
(Sack et al., 1989). The crystal structure of an AmiC–acetamide
complex determined at 2.1 Å resolution, precisely defined the
amide binding site in AmiC (Pearl et al., 1994), which lies at
the interface of the two domains, which are stabilized in a
‘closed’ conformation by contacts with the bound ligand. This
closed-down acetamide-bound form of AmiC represents the
‘on’ configuration of the regulatory system, in which AmiR is
free to interact with the leader RNA of the operon transcript,
and permit full expression. The mechanism of amide-switched
regulation of AmiR by AmiC remains to be determined, but
it probably involves disruption of a silencing AmiC–AmiR
complex on binding of inducing amides to AmiC (Wilson
et al., 1996).
Amidase expression in the PAC1 strain of P.aeruginosa is
strongly induced by acetamide (one carbon chain) and lactamide or propionamide (two carbon chain) but not by butyramide
(three carbon chain), which instead acts as an anti-inducer
in vivo and competes for binding to AmiC in vitro. Thus,
lengthening of the aliphatic chain by a single carbon converts
a strong agonist into an antagonist. We have now cloned and
sequenced AmiC from the PAC181 strain of P.aeruginosa,
which was generated by classical in vivo selection for its ability
to induce amidase expression in the presence of butyramide
(Turberville and Clarke, 1981). The crystal structure of the
AmiC protein from this mutant strain reveals the subtle
structural adaptation that enables the butyramide-inducible
phenotype, again demonstrating the remarkable adaptability of
the ‘small molecule binding protein’ fold to the specific
recognition of ligands.
Material and methods
Cloning and characterization of PAC181 amidase regulatory
genes
The genes amiC and amiR were amplified by PCR from
PAC181 chromosomal DNA using the oligonucleotides CRA
(ATCCGAATTCTCACAGGAGAGGAAACGGATG)
and
CRL (CCGATGCCGAAGCCTTGATGACGATACCCCTCTT), (Genosys, Cambridge, UK) and Taq DNA polymerase.
The amplified DNA fragment was isolated from a preparative
agarose gel, cloned initially into pUC19 (pSW181) and characterized by restriction enzyme mapping. The 1.8 kb amiCR
fragment was then subcloned into the broad host range vector
pMMB66EH (Temple et al., 1990) (plasmid pSW1181) and
mobilized from E.coli S17-1 into P.aeruginosa PAC327 (amiC–,
amiR). Amidase activity was determined in intact cells under
non-inducing, inducing and repressing conditions as described
previously (Drew, 1984) and the results presented are the mean
129
B.P.O’Hara et al.
Table I. Amidase activity of P.aeruginosa strains
Strain
Phenotype
Plasmid
Succinate
Succinate ⫹
lactamide
Succinate ⫹
butyramide
PAC1
PAC181
PAC327
PAC327
Inducible Ami⫹
Butyramide inducible
amiCR–
amiCR–
—
—
—
pSW181
0.1
0.1
0.2
0.1
8.1
0.8
0.3
9.7
0.2
1.58
0.2
5.6
Activities are in µM acethydroxamate formed per min in a standard amidase assay (Drew, 1984)
Fig. 1. Structure of AmiC-PAC181 butyramide complex. Secondary
structure cartoon of AmiC-PAC181, with the N-terminal domain on the
right and the C-terminal domain on the left. The positions of the butyramide
ligand bound between the two domains, and the mutation at residue 106, are
indicated.
values of duplicate assays carried out on at least three separate
occasions. One unit represents 1 µM acethydroxamate formed
per min. Specific activities are units per mg of bacterial cells.
The nucleotide sequence of the amiC and amiR genes was
determined by double-stranded sequencing using pSW181 as
template as described previously (Wilson and Drew, 1991).
Identification of mutations was facilitated by running parallel
sequencing reactions of the PAC1 amiC and amiR genes.
Protein expression and purification
The mutant AmiC protein was isolated from P.aeruginosa
strain PAC452 harbouring the plasmid pSW1181. Protein was
overexpressed and purified essentially as described (Wilson
et al., 1991), but with 5 mM butyramide (Sigma) present in
the culture medium and maintained in all buffers during
purification. The mutant PAC181 AmiC protein is much less
easily handled than the wild type, and the omission of
butyramide from the growth medium and isolation buffers
significantly decreased the solubility of the mutant AmiC and
prevented effective purification.
Crystallization of PAC181-AmiC–butyramide complex
The solubility profile of the PAC181-AmiC protein in the
presence of butyramide proved to be significantly different
from that for the PAC1-AmiC with acetamide (Wilson et al.,
1991), and a new crystallization condition was identified using
a sparse matrix screen (Jancarik and Kim, 1991). Crystals of
diffraction quality were eventually obtained from micro-batch
experiments containing 12.8 mg/ml PAC181-AmiC, 5 mM
butyramide, 1.36 M sodium citrate and 100 mM HEPESNaOH (pH 7.5), implemented under paraffin oil in Terasaki
dishes (Chayen et al., 1992).
130
X-Ray data collection, processing and refinement
Diffraction data to 2.7 Å resolution were collected at 100 K
from a single PAC181-AmiC–butyramide co-crystal, cryoprotected by soaking in 1.5 M sodium citrate, 100 mM HEPESNaOH and 25% (v/v) glycerol, on a 30 cm MAR Image Plate
detector mounted on a Rigaku RU200 rotating anode Xray generator. Diffraction images were integrated using the
MOSFLM package (Leslie, 1995) and reduced using the
SCALA, AGROVATA and TRUNCATE programs of the CCP4
Suite (CCP4, 1994). The scaled and merged data consist of
10 102 unique reflections, 97.8% complete in the range 38 to
2.7 Å (97.2% in the outer shell), with an average of 2.7
observations per reflection, and a merging R-factor of 0.098
(0.185 in the outer shell). Although the crystallization conditions for PAC181-AmiC with butyramide differ significantly
from those for PAC1-AmiC with acetamide, the space group
of the crystals is the same (P42212), and unit cell parameters
(a, 104.15 Å; c, 65.68 Å) are very similar. Initial models were
obtained by simulated annealing refinement of the coordinates
(PDB code: 1PEA) for the PAC1-AmiC–acetamide complex
(Pearl et al., 1994) against the PAC181-AmiC–butyramide
complex data, using X-PLOR (Brunger, 1992), but with
coordinates for the bound acetamide and solvent molecules
omitted from the start model, and residue 106 modelled as
alanine. Difference electron density maps were calculated with
σA-weighted coefficients to minimize model bias (Read, 1986),
and examined using ‘O’ (Jones et al., 1991). Subsequent
simulated annealing and conjugate gradient refinement produced the current model consisting of 370 residues, 57 solvent
molecules and a butyramide ligand. The crystallographic Rfactor is 0.24 and the free-R factor with 5% of the data omitted
from refinement is 0.30. The geometric parameters of the
model are within the ranges expected at this resolution (Laskowski et al., 1993). Coordinates have been deposited in the
Protein Data Bank with accession code 1QNL.
CD spectroscopy
Near-UV circular dichroism signals were measured on a
nitrogen flushed Jasco J720 spectrapolarimeter. Measurements
were made with protein concentrations of 0.4 mg/ml and
ligand concentrations of 10 mM using a 1 cm pathlength
cuvette. The CD signal at 289 nm was measured as a function
of temperature and fit to a van’t Hoff equation (Freeman et al.,
1998). The PAC1-AmiC–acetamide complex displayed a Tm
of 84°C; the PAC1-AmiC–butyramide complex a Tm of 79°C
and the PAC181-AmiC–butyramide complex a Tm of 59°C.
Results
Characterization of PAC181 amidase regulatory genes
Amidase activities were determined for P.aeruginosa strains
PAC1 (wild-type), PAC181 (butyramide inducible), PAC327
Altered specificity of amide receptor AmiC
Fig. 2. Electron density for the 106Thr→Asn mutation. Electron density from an Fo–Fc map with structure factors calculated from the wild-type PAC1-AmiC
(PDB code: 1PEA) refined against the PAC181-AmiC diffraction data, with the ligand omitted, and only a Cα included for the side chain of residue 106. The
resultant density (contoured at 1.25 Å) confirms the replacement of Thr by Asn at 106, as shown by sequencing of the PAC181-amiC gene, and the presence
of butyramide.
Fig. 4. Melting curves for AmiC–ligand complexes. Denaturation of AmiC–
ligand complexes as a function of temperature, followed by change in the
near-UV circular dichroism signal at 289 nm. Curves are for PAC1-AmiC–
acetamide complex (d), PAC1-AmiC–butyramide complex (m) and
PAC181–butyramide complex (s).
Fig. 3. Detailed interactions in the AmiC binding site. (a) Acetamide bound
to PAC1-AmiC. (b) Butyramide bound to PAC181-AmiC. The loss of the
Cα of threonine, which delimits the side of the ligand-binding pocket in
PAC1-AmiC, provides a space for the extra two carbons of the butyramide
aliphatic chain. Hydrogen bonds are shown as broken cylinders.
(amiCR–) and for PAC327 harbouring the plasmid pSW1181,
which contains the 1.8 kb amiCR fragment amplified from
PAC181 (see Materials and methods). PAC1 displayed characteristic lactamide inducibility, and PAC181 displayed lactamide
and butyramide inducibility as previously described (Turberville and Clarke, 1981), but with lactamide inducibility at a
lower level than in PAC1 (Table I). Despite the regulatory
negative phenotype, there is some residual background amidase
activity detectable in PAC327, but no induction is observed in
the presence of lactamide or butyramide. PAC327 harbouring
pSW1181 shows substantial induction of amidase activity in
the presence of lactamide or butyramide, confirming that the
butyramide inducible phenotype was retained in the cloned
PAC181 amiCR fragment. Nucleotide sequencing of the entire
fragment revealed only a single mutation from the wild-type
PAC1 sequence, a C→A transversion at position 317 in the
amiC gene.
Molecular basis of butyramide inducibility of PAC181-AmiC
Earlier studies of the amidase operon had demonstrated that
the responsiveness to amide inducers was a function of the
AmiC protein (Wilson et al., 1993). Phenotypic changes in
the response to specific amides by mutant strains would
therefore be expected to be accompanied by mutations in the
AmiC protein rather than in other components of the operon.
The nucleotide sequence of the amiC gene from the butyramideinducible strain P.aeruginosa PAC181 revealed a single base
change relative to the wild-type amiC gene sequence, resulting
in the replacement of a threonine residue at 106 in the PAC1AmiC sequence (SwissProt entry AMIC_PSEAE) with an
asparagine. As residue 106 is located close to the amide
binding site at the interface of the N- and C-terminal domains
of AmiC (Pearl et al., 1994), a mutation in this residue
would be consistent with a change in amide ligand specificity
(Figure 1).
To define the structural effect of the Thr106→Asn mutation
131
B.P.O’Hara et al.
in PAC181, the crystal structure of PAC181-AmiC was determined in complex with butyramide. Difference Fourier maps
calculated after refinement with a model in which only a Cα
side chain was included for residue 106, clearly show electron
density for asparagine rather than threonine (Figure 2), confirming the nucleotide sequence. Clear positive difference
electron density is also present for a bound butyramide
molecule whose amide head group and α-carbon are in
essentially identical positions to the equivalent atoms of
acetamide bound to PAC1-AmiC.
The mutation of Thr106→Asn removes the γ-methyl group
of threonine, which forms part of the side wall of the amide
binding pocket in PAC1-AmiC, and thereby increases the
volume of the amide-binding site. Butyramide binds to
PAC181-AmiC in an unfavourable folded-over conformation
in which the α and β methylene protons are eclipsed, and the
γ-methyl group is in van der Waal’s contact with the plane of
the amide head group. The β methylene group is in van der
Waal’s contact with the γ-methyl of Thr 233, and the γ-methyl
packs against the planes of the side chains of Asn106 and
Tyr83 (Figure 3).
Adaptation to binding butyramide destabilizes AmiC
In the wild-type PAC1-AmiC, the side-chain γ-hydroxyl group
of Thr106 is hydrogen bonded to the peptide oxygen of Cys82
and weakly (3.2 Å), to the side-chain amide of Gln27. The γmethyl of Thr106 is directed towards the amide-binding pocket,
but does not make any direct contact with the bound acetamide
in that complex. In the mutant PAC181-AmiC, the side chain
of the Asn106 is orientated so that its Cβ–Cγ bond bisects the
direction of the γ-hydroxyl and γ-methyl of the Thr106 in the
PAC1-AmiC. The amide nitrogen in the head group of Asn106
picks up the hydrogen bond to the peptide oxygen of Cys82,
and makes a strong hydrogen bond (2.7 Å) to the amide head
group of Gln27, which changes conformation to make this
interaction. Thus, the substitution of Asn for Thr at 106 retains
the hydrogen bonding interactions made by Thr106 in the
wild-type PAC1-AmiC. However, the amide oxygen of Asn106
in the PAC181 mutant is left buried in a hydrophobic environment, packed against the side chain of Tyr83, and with no
hydrogen bonds to the rest of the protein, or to any solvent
molecules. The burial of this polar group without any compensating interactions would be expected to be highly unfavourable.
Consistent with this, the melting temperature (Tm) measured
by near-UV CD (see Materials and methods) for denaturation
of PAC181-AmiC with butyramide bound is down-shifted
20 K compared with the Tm of the PAC1-AmiC with butyramide, which is a poor ligand for the wild-type AmiC. Compared
with the Tm for PAC1-AmiC with bound acetamide, an optimal
ligand, the Tm for PAC181-AmiC with butyramide is decreased
by 25 K (Figure 4).
Discussion
Before the advent of recombinant DNA technology, biological
selection and screening of mutagenized cultures was the only
way in which molecular evolution could be studied. With the
P.aeruginosa amidase, an advantage at the outset was the
differing inducer and substrate specificities of the system
(Kelly and Clarke, 1962). A large number of both regulatory
and structural gene mutations were isolated which allowed
growth on novel amide substrates (Clarke and Drew, 1988).
These novel abilities have been attributed to changes in the
amidase itself and to those in the negative regulator, ligand
132
receptor, AmiC. AmiC, a member of the SMBP family, contains
a widely used and adaptable ligand binding site. It appears
that a single amino acid substitution in a residue close to the
binding site in the AmiC of the butyramide inducible strain
PAC181 is solely responsible for the new phenotype. PAC181
also contains a single residue substitution within the amidase
enzyme itself; Ser→Phe at position 7. This gives an enzyme
with higher activity towards butyramide hydrolysis (Turberville
and Clarke, 1981), although the structural basis for this remains
to be determined. PAC181 thus represents a new strain able
to use butyramide, a repressor of the wild-type system, to
induce expression of a new butyramide hydrolysing enzyme.
In adapting to butyramide induction via the Thr106→Asn
mutation, PAC181-AmiC undergoes a significant decrease in
its thermostability compared with wild type, and its half-life
in vivo would be expected to be correspondingly shorter.
However, the selective pressure to which the PAC181-AmiC
sequence is a response, requires only that the amidase operon
be inducible by butyramide. Thus, so long as the mutant AmiC
can be synthesized at a sufficient level, fold and fulfil its
biological role, its absolute structural stability need not be
optimal. It is protein function that is selected in evolution, not
structural stability, a fact often neglected in theoretical analyses
of protein structure.
Acknowledgements
We are very grateful to Prof. Patricia H.Clarke FRS for her continuing interest.
Work on the amidase operon at UCL has been supported by project grants
from the BBSRC and Wellcome Trust to R.E.D. and L.H.P.
References
Brunger,A. (1992) X-PLOR Version 3.1. A System for X-Ray Crystallography
and NMR. Yale University Press, New Haven.
CCP4 (1994) Acta Crystallogr., D50, 760–763.
Chayen,N.E., Stewart,P.D.S. and Blow,D.M. (1992) J. Cryst. Growth, 122,
176–180.
Clarke,P.H. and Drew,R. (1988) Bioscience Rep., 8, 103–120.
Drew,R. and Lowe,N. (1989) J. Gen. Microbiol., 135, 817–823.
Drew,R.E. (1984) J. Gen. Microbiol., 130, 3101–3111.
Freeman,D.J., Pattenden,G., Drake,A.F. and Siligardi,G. (1998) J. Chem. Soc.
Perkin, 2, 129–135.
Jancarik,J. and Kim,S.H. (1991) J. Appl. Crystallogr., 24, 409–411.
Jones,T.A., Zou,J.-Y., Cowan,S.W. and Kjeldgaard,M. (1991) Acta Crystallogr.,
A47, 110–119.
Kelly,M. and Clarke,P.A. (1962) J. Gen. Microbiol., 27, 305–316.
Laskowski,R.A., Macarthur,M.W., Moss,D.S. and Thornton,J.M. (1993)
J. Appl. Crystallogr., 26, 283–291.
Leslie,A.G.W. (1995) MOSFLM Users Guide. MRC Laboratory of Molecular
Biology, Cambridge.
Pearl,L.H., O’Hara,B.P., Drew,R.E. and Wilson,S.A. (1994) EMBO J., 13,
5810–5817.
Read,R.J. (1986) Acta Crystallogr., A42, 140–149.
Sack,J.S., Saper,M.A. and Quiocho,F.A. (1989) J. Mol. Biol., 206, 171–191.
Temple,L.,
Cuskey,S.M.,
Perkins,R.E.,
Bass,R.C.,
Morales,N.M.,
Christie,G.E., Olsen,R.H. and Phibbs,P.V. (1990) J. Bacteriol., 172, 6396–
6402.
Turberville,C. and Clarke,P.H. (1981) FEMS Microbiol. Lett., 10, 87–90.
Wilson,S. and Drew,R. (1991) J. Bacteriol., 173, 4914–4921.
Wilson,S.A., Chayen,N.E., Hemmings,A.M., Drew,R.E. and Pearl,L.H. (1991)
J. Mol. Biol., 222, 869–871.
Wilson,S.A., Wachira,S.J., Drew,R.E., Jones,D. and Pearl,L.H. (1993) EMBO
J., 12, 3637–3642.
Wilson,S.A., Wachira,S.J.M., Norman,R.A., Pearl,L.H. and Drew,R.E. (1996)
EMBO J., 15, 5907–5916.
Received May 10, 1999; revised November 10, 1999; accepted November
16, 1999