Stability of recombinant wild-type StP and TOWER helix mutants thereof as revealed in urea and thermal denaturation experiments To compare the stabilities of StP and Argfi Ala mutants the
Trang 1Tracking interactions that stabilize the dimer structure of starch
Roles of Arg234 and Arg242 revealed by sequence analysis and site-directed
mutagenesis
Richard Griessler1,2,3, Alexandra Schwarz1,3, Jan Mucha2and Bernd Nidetzky1,3
1
Institute of Food Technology and2Centre of Applied Genetics, University of Agricultural Sciences, Vienna, Austria;
3
Institute of Biotechnology, Graz University of Technology, Austria
Glycogen phosphorylases (GPs) constitute a family of
widely spread catabolic a1,4-glucosyltransferases that are
active as dimers of two identical, pyridoxal
5¢-phosphate-containing subunits In GP from Corynebacterium callunae,
physiological concentrations of phosphate are required
to inhibit dissociation of protomers and cause a 100-fold
increase in kinetic stability of the functional quarternary
structure To examine interactions involved in this large
stabilization, we have cloned and sequenced the coding gene
and have expressed fully active C callunae GP in Escherichia
coli By comparing multiple sequence alignment to
structure-function assignments for regulated and nonregulated GPs
that are stable in the absence of phosphate, we have
scru-tinized the primary structure of C callunae enzyme for
sequence changes possibly related to phosphate-dependent
dimer stability Location of Arg234, Arg236, and Arg242
within the predicted subunit-to-subunit contact region made
these residues primary candidates for site-directed
muta-genesis Individual Argfi Ala mutants were purified and characterized using time-dependent denaturation assays in urea and at 45C R234A and R242A are enzymatically active dimers and in the absence of added phosphate, they display a sixfold and fourfold greater kinetic stability of quarternary interactions than the wild-type, respectively The stabilization by 10 mMof phosphate was, however, up
to 20-fold greater in the wild-type than in the two mutants The replacement of Arg236 by Ala was functionally silent under all conditions tested Arg234 and Arg242 thus parti-ally destabilize the C callunae GP dimer structure, and phosphate binding causes a change of their tertiary or quartenary contacts, likely by an allosteric mechanism, which contributes to a reduced protomer dissociation rate Keywords: interface; oxyanion; phosphate; stabilization; subunit dissociation
Glycogen phosphorylases (GPs) catalyse degradation of
glycogen and structurally related reserve polysaccharides
in the cytosol to provide energy via the branch point
metabolite a-D-glucose-1-phosphate All known GPs are
functional homodimers composed of 90-kDa subunits
and require pyridoxal 5¢-phosphate (PLP) cofactor for
activity [1–7] Although a very low basal activity may be
present in the holoenzyme protomer, quarternary
inter-actions clearly determine physiological levels of
phos-phorylase activity and are a prerequisite for the regulatory
properties of eukaryotic GPs [8–10] Forces that stabilize the dimer structure of GP are therefore essential to optimal enzyme function under physiological boundary conditions GPs are a/b proteins that display a two-domain fold in which the N-terminal two-domain and the C-terminal domain are separated by a catalytic site cleft The structural elements that comprise the subunit–subunit interface are located in the N-terminal domain The dimer contact regions of regulated and nonregulated GPs share structural similiarity overall, but differ on the molecular level [3–7]
Starch phosphorylase (StP) from the soil bacterium Corynebacterium callunaeis a member of the GP family [11] Its activity is not under control of common allosteric effectors of mammalian GPs such as AMP orD -glucose-6-phosphate The enzyme differs from structurally well-characterized GPs [3–7] because it requires physiological concentrations of phosphate ( 5 mM) for stability of the functional oligomeric structure [12] Binding of phosphate
to a protein site different from the site where the substrate phosphate binds causes apparent tightening of quarternary interactions present in StP and leads to a 100-fold increase
in kinetic stability of the active dimer [12] The very low in-vitrolifetime of StP activity in the absence of phosphate ( 30 min [12]) suggests that this stabilizing effect might be
Correspondence to B Nidetzky, Institute of Biotechnology, Graz
University of Technology, Petersgasse 12/I, A-8010 Graz, Austria.
Fax: + 43 316 873 8434, Tel.: + 43 316 873 8400;
E-mail: bernd.nidetzky@tugraz.at
Abbreviations: GP, glycogen phosphorylase; rmGP, rabbit muscle GP;
MalP, maltodextrin phosphorylase; EcMalP, MalP from Escherichia
coli; PLP, pyridoxal 5¢-phosphate; StP, starch phosphorylase.
Enzymes: a-glucan (glycogen, starch, maltodextrin) phosphorylase
(1,4-a- D -glucan:phosphate a- D -glucosyltransferase) (EC 2.4.1.1).
Note: The genomic sequence of C callunae that comprises the entire
structural gene of starch phosphorylase is available under the
GenBank accession number: AY102616.
(Received 28 November 2002, accepted 7 March 2003)
Trang 2important in the physiology of C callunae In light of the
fact that interactions apparently critical to a stable and
active protein conformation converge at the dimer interface
of GP, we considered oxyanion-dependent stability of the
StP dimer to be a significant problem and hence worth
examining We thus turned our attention to the primary
structure of StP and report here the cloning, sequencing and
heterologous expression in Escherichia coli of the gene
encoding this enzyme In an effort to identify sequence
changes relevant to GP dimer stability, we compared
multiple sequence alignment with secondary structure
assignments and dimer contacts in structurally characterized
GPs [3–7,13–15] Through this process, the main element of
the dimer interface in GPs, the so-called TOWER helix, was
allocated to the linear sequence of StP In rabbit muscle GP
(rmGP) and likewise other GPs [3–7,13–15], this helixforms
intimate contacts with its counterpart helixon the opposite
subunit Three arginine residues are located within the
predicted TOWER helixof StP Because arginine are
common components of interfaces of oligomeric proteins
and frequently show interaction with oxyanion ligands such
as phosphate or sulfate, the TOWER arginines of StP were
selected as candidates to be replaced by alanine-scanning
site-directed mutagenesis We have assessed the role of each
arginine for oxyanion-dependent stability of the StP dimer
Experimental procedures
Materials
Natural StP from C callunae DSM 20145 was produced
and purified by reported procedures [11] Materials and
assays for measuring enzyme activities in the directions of
a-glucan degradation and synthesis have been described
elsewhere [11,12] Restriction endonucleases, T4DNA ligase
and Taq DNA polymerase were obtained from Promega
PfuDNA polymerase, alkaline phosphatase, RNase, and
positively charged nylon membranes were from Roche The
expression vectors pQE 30 and pQE 70, the gel extraction
kit QiaexII, and the plasmid purification kit were from
Qiagen
Preparation of an oligonucleotide probe
for theStP gene
Chromosomal DNA from C callunae DSM 20145 was
prepared by incubating approximately 200 mg of wet cell
mass suspended in 10 mM Tris/HCl buffer, pH 7.6,
con-taining 1 mMEDTA and 15 mg lysozyme (Sigma) for 3 h
at 37C To this mixture were added 3 mL of a solution of
0.4MNaCl, 0.7% (w/v) SDS and 1 mgÆmL)1proteinase K
(Sigma) in 10 mMTris/HCl buffer, pH 8.2 After incubation
at 50C for 5 h, protein was precipitated with 1 mL of 6M
NaCl and removed by centrifugation at 13 000 g The DNA
in the supernatant was precipitated with ethanol and
purified by standard protocols [16]
From a comparison of GP sequences in the GenBank
database, two well-conserved peptides, GNGGLGRL
(res-idues 131–138 in rmGP) and TNHTLMPEAL (res(res-idues
374–383 in rmGP), were chosen and reverse translated into
a pair of degenerated PCR primers: forward (p1), 5¢-GG
(ACT)AA(CT)GG(GCT)GGT(CT)T(AG)GG(ACT)CG
(GT)CT(AGT)GC-3¢; and reverse (p2), 5¢-GC(CT)TC (AGT)GGCA(AGT)(AC)AC(AGC)GT(AG)TGGTT(AGT) GT-3¢ Polymerase chain reactions (50 lL) were carried out with a Hybaid thermocycler (Thermo Life Sciences) and used 200 ng of chromosomal template DNA, 40 pmol sense and antisense primers, 150 lmol dNTPs in PCR buffer (Promega) PCRs consisted of 30 cycles of 30 s denaturation
at 95C, followed by 20 s primer annealing at 55 C and
40 s elongation at 72C The final extension step was carried out at 72C for 7 min The resulting PCR product was gel-purified and placed into a pUC 18 vector (Life Technologies) using SmaI cleavage and blunt-end cloning This recombinant vector was transferred by electroporation into competent cells of E coli DH5a (Stratagene), and after plasmid purification the insert was subjected to dideoxy sequencing with an ABI Prism 310 Genetic Analyzer (Applied Biosystems) using Universal Primer (Amersham Pharmacia Biotech) The PCR product was used as specific probe for Southern blot hybridization experiments Frag-ments of interest were cloned into pBluescript II SK(+/–) (Life Technologies) via BamHI and HindIII restriction sites and used to generate a partial genomic library of C callunae DNA Colony hybridization with the PCR probe was used
to screen this library Positive clones were sequenced in both senses of the DNA at the VBC Genomics Sequencing Service Facility of the University of Vienna using the primer walking method
Construction of expression plasmids The following primer set was designed to amplify the entire open reading frame of the StP gene by using PCR: pNter (5¢-CGCGCATGCAGCCCTGAAAAACAGCC-3¢) derived from the authentic N-terminus of StP [11], and
GAGTTG-3¢) where SphI and SalI restriction sites are underlined The 50-lL reaction mixture contained 200 ng of C callu-naeDNA, 50 pmol of pNter and pCter, 0.1 mM dNTPs, Pfupolymerase buffer, 5 units of Pfu DNA polymerase and was subjected to 35 cycles of 1 min denaturation at 95C,
1 min annealing at 52C and 3 min elongation at 72 C The resulting PCR product was blunt-end subcloned into
a SmaI-digested pUC 19 vector, yielding pUC 19-StP pUC 19-StP was then digested with SphI and SalI and cloned in-frame into the SphI/SalI site of the pQE 30 vector
to produce a fusion protein bearing an N-terminal metal affinity tag (RGSHHHHHHGSA) Competent cells of
E coli XL1 Blue (Stratagene) were transformed with the pQE 30 vector containing the DNA insert (pQE 30-StP) In order to obtain nontagged recombinant StP, the StP gene was cloned in the SphI site of the expression vector pQE 70 The C-terminal His-tag provided by the plasmid was deleted
by inserting a stop codon in the C-terminal primer The following primers were synthesized to amplify the open reading frame,
pNter (5¢-CGCGCATGCCTGAAAAACAGCCACTCC-3¢), pCter (5¢-ACGGCATGCTTAAACAGCAGGAGTTGG-3¢), where restriction sites are underlined The resulting recom-binant StP lacked Ser1
Trang 3Site-directed mutagenesis
The single point mutations were introduced by the
PCR-based overlap extension method [17] The following
mutagenic oligonucleotide primers were used where the
mismatched bases are underlined: 5¢-ATCGAAGCC
GAGCGCGTTTCC-3¢ (R234A); 5¢-GAACGCGAGGCC
GTTTCC-3¢ (R236A); and 5¢-GATATCTGCGCCGT
GCTC-3¢ (R242A)
A 1400-bp fragment of the StP gene, obtained by
digestion of pQE 30-StP with SphI and Eco91I, was used
as a template The flanking primers were pNter and pEco91I
(5¢-CCAGATCGGTTACCCAATCATCGGAACCG-3¢)
PCR conditions were as described above except for the
annealing temperature which was 50C Plasmid mini-prep
DNA was subjected to dideoxy sequencing to verify that the
desired mutation had been introduced and that no
misin-corporation of nucleotides had occurred as a result of the
DNA polymerase Each mutagenized fragment was then
cloned into the residual pQE 30-StP vector
Expression of theStP gene in Escherichia coli
Cells of E coli XL1 Blue harbouring pQE 30-StP (or pQE
70-StP) were grown in media that contained tryptone
(10 gÆL)1), yeast extract (5 gÆL)1), and NaCl (10 gÆL)1) and
ampicillin (100 lgÆmL)1) After the optical density at
600 nm had reached a value of approximately 1, the initial
temperature of 37C was reduced to 25 C, and gene
expression was induced with 0.5 mMof isopropyl thio-b-D
-galactoside for 12 h Cells were harvested by centrifugation
(2000 g for 15 min) and diluted approximately twofold with
50 mM potassium phosphate buffer, pH 7.0 The
suspen-sion was passed three times through a 1-inch French
pressure cell (Aminco), and cell debris was removed by
centrifugation (10 000 g for 30 min) The resulting
super-natant was used further Expression of the mutagenized StP
genes, placed into the pQE 30 expression vector, was
performed in exactly the same way as just described for the
wild-type
Purification and characterization of recombinant StP
and mutants thereof
Recombinant wild-type StP was purified by a reported
protocol [18] The following procedure was used to purify
His-tagged StP and mutants thereof The E coli cell extract
(100 mg protein) was applied to a 10-mL copper-loaded
chelating Sepharose fast flow resin column (Amersham
Pharmacia Biotech; 16 mm diameter) equilibrated with a
50 mMtriethanolamine buffer, pH 7.0, containing 20 mMof
sodium sulfate Bound protein was eluted with a linear
gradient from 0 to 250 mM imidazole in the same buffer
(pH 7.0) Fractions containing phosphorylase activity were
pooled and brought to 65% saturation in ammonium
sulfate The protein pellet obtained after centrifugation
(10 000 g for 30 min) was dissolved in a small volume of
300 mMpotassium phosphate buffer, pH 7.0, and
incuba-ted at 60C for 40 min Note that heat treatment
inacti-vates any remaining endogenous E coli maltodextrin
phosphorylase [19] After centrifugation and
concentra-tion using 30-kDa Microsep tubes (Pall Filtron), further
purification was carried out by size exclusion chromato-graphy on Superose 12 Prep Grade (Amersham Pharmacia Biotech; 16 mm diameter, 140 mL) equilibrated with
50 mMphosphate buffer, pH 7.0, containing 0.2MNaCl The methods used for the characterization of the activity and the stability of the recombinant enzymes were those described in detail elsewhere for natural StP [11,12,18] Unless mentioned otherwise, a continuous coupled enzyme assay at 30C was used to measure phosphorylase activity using as the substrate 30 gÆL)1 of maltodextrin (dextrin equivalent 19.4; Agrana [11]) dissolved in a 50 mM
potassium phosphate buffer, pH 7.0, containing 5 lM of glucose 1,6-bisphosphate, 2.5 mMof NAD+, rabbit muscle phosphoglucomutase (8 units; Boehringer), and glucose 6-phosphate dehydrogenase (3 units; Sigma) One unit of activity (1 U) refers to 1 lmol NADH produced under the given conditions Binding of phosphate and sulfate to StP
or mutants thereof was determined by using a previously reported procedure in which inhibition of quenching of cofactor fluorescence by iodide was measured [12] CD spectroscopic measurements were carried out as described recently [12] using protein solutions (0.1 mgÆmL)1± 10%)
in a 20 mMMops buffer, pH 7.0 CD data are expressed in terms of molar ellipticity
Results
Cloning and sequencing of theStP gene Using the PCR primers p1 and p2, a 710-bp fragment was amplified from chromosomal C callunae DNA, blunt-end cloned into pUC 18, and sequenced The sequence similar-ity search clearly indicated that this fragment was a part of a putative phosphorylase gene The [a-P32dCTP]-labeled PCR fragment was used as a probe for Southern blot hybridization to C callunae genomic DNA that had been exhaustively digested with different endonucleases A strong hybridization to a BamHI fragment of approximately 2.9 kb and a HindIII fragment of approximately 4.2 kb was found (Fig 1, panel A) These fragments were cloned into pBluescript II SK(+/–), and positive clones were identified by colony hybridization The sequence of the HindIII fragment comprised the entire StP gene except for
158 nucleotides corresponding to the N-terminal part of
C callunaeStP (Fig 1) The BamHI fragment included this part of the open reading frame, as shown in Fig 1 (panel B) Further gene sequencing revealed the absence of another open reading frame 1000 bp upstream of the start codon and 2000 bp downstream of the stop codon of the StP gene The entire open reading frame for StP consisted of
2388 bp encoding a protein of 796 amino acids The calculated molecular mass of the StP subunit is 90 603 Da,
in good agreement with the value of 88 000 obtained from protein characterization [11]
Identification of dimer contact regions in StP from the structural alignment of StP with other GPs
An alignment of the amino acid sequences of StP, rmGP and maltodextrin phosphorylase from Escherichia coli (EcMalP) is shown in Fig 2 rmGP and EcMalP were chosen for structure-based sequence comparison because
Trang 4they represent prototypes of regulated and nonregulated
GPs, respectively, and both have well-established
structure-function relationships StP is 41% identical to rmGP and
42% identical to EcMalP Figure 2 maps structural and
functional elements of rmGP [4] and EcMalP [3] onto the
linear sequence of StP and thus measures the extent of
conservation of the respective sites in StP As in other GPs,
the catalytic site and the PLP binding site of StP are virtually
identical to the corresponding sites in rmGP and EcMalP
By contrast, the regulatory sites of rmGP are almost
completely lost in StP Interestingly, there is only small
sequence identity between StP and EcMalP in the segments
of the sequence that correspond to regulatory sites of rmGP
Hudson et al [14] have classified dimer contacts in rmGP
into three relatively independent networks of
interact-ing groups The first two networks, often dubbed the
cap¢-a2-b7 interface, are mediated by residues in the ultimate
N-terminal part of the rmGP protomers and are associated
with control by allosteric effectors and covalent
phosphory-lation, as shown in Fig 2 These dimer contact pairs of
rmGP are conserved to a very low degree in EcMalP and
likewise StP The third network constitutes the major dimer contact region in GP and involves the so-called TOWER (a7) helices (residues 266–277 in rmGP) and the subsequent gate loops of adjacent subunits While specific interactions between the subunits at the TOWER interface vary considerably among different GPs [3–7], the position of the a7 helixin the primary structures of rmGP, human liver
GP, yeast GP, and EcMalP is very well conserved Therefore, the TOWER-GATE region could be easily assigned to the sequence of StP, as shown by underlining in Fig 2 (StP residues 231–242) Considering the involvement
of the TOWER interface of rmGP in signal transmission from regulatory sites into the active centre [4–6,15], it was interesting that residues in StP corresponding to the a7 helix
of rmGP exhibited a higher degree of similarity to the mammalian enzyme than to EcMalP Furthermore, the occurrence of three arginine residues within the TOWER helixof StP at positions 234, 236 and 242 was interesting (Fig 2) Arginines are common components of protein interfaces and occur frequently at oxyanion-binding protein sites [20,21] Residues Arg234, Arg236, and Arg242 were thus targets for site-directed mutagenesis, and their PCR-based replacement by alanine was chosen to eliminate all electrostatic interactions at the respective position It is worth noting that Arg234 and Arg242 are positionally conserved in all mammalian a-glucan phosphorylases while the same positions show considerable variation in the related enzymes from bacteria, fungi and plants
Expression of the wild-type and mutagenizedStP genes
inE coli, and purification and characterization
of recombinant enzymes Following induction with isopropyl thio-b-D-galactoside using the conditions described in Experimental procedures,
a specific phosphorylase activity of approximately
10 UÆmg)1(± 15% SD) was measured in cell extracts of
E coliXL1 Blue cells transformed with either pQE 30-StP
or pQE 70-StP Comparison of this figure with the known specific activity of 30 UÆmg)1 for pure natural StP [11] shows that recombinant StP corresponded to 30% of the total soluble E coli protein Recombinant wild-type and His-tagged StP, and likewise StP mutants were purified to apparent homogeneity and all were recovered in approxi-mately 25 ± 5% yield Like native StP, His-tagged StP and all StP mutants contained 0.8–1.0 mol of PLP per mol of 90-kDa protomer, as expected if incorporation of cofactor during folding of the recombinant proteins had taken place correctly Circular dichroism (Fig 3) and Fourier-trans-form infrared spectra (not shown) of the purified recom-binant proteins were recorded in an effort to identify alterations in structure, relative to the natural wild-type enzyme [12,18], as a result of recombinant protein produc-tion and mutagenesis There were no traceable differences between CD spectra of natural and recombinant StP (not shown) The CD spectrum of His-tagged StP and CD spectra of R234A and R242A mutants were not super-imposable (Fig 3), but the overall picture is one of close structural similarity among the wild-type and the two mutants Therefore, site-specific replacements of Arg234 and Arg242 did not cause gross changes in the composition
of secondary structural element in the two mutants, relative
Fig 1 Southern blot analysis for C callunae genomic DNA (A) and
results of DNA library screening (B) using a 0.71 kb PCR probe for the
StP gene Lanes 1–4 of the autoradiogram in panel A show the
hybridization patterns of the32P-labeled PCR probe with C callunae
DNA (50 lg) digested for up to 3 days with different endonucleases
(10–20 U) as indicated DNA fragments were separated on a 0.8%
agarose gel and after transfer to a nylon membrane (Roche) allowed to
hybridize with the PCR probe overnight at 60 C Arrows on the right
and left of the blot indicate the sizes (in kb) of the main hybridizing
DNA fragments of which the BamHI and HindIII fragments were
cloned to give a partial genomic DNA library After screening using
colony hybridization with the 710-bp PCR probe, positive clones were
selected and the inserts sequenced The bottom of the figure (B)
indi-cates the positions of BamHI and HindIII fragments, relative to the
entire StP structural gene.
Trang 5to the wild-type The specific activities of recombinant StP,
His-tagged StP, and the R234A and R236A mutants were
identical within the experimental error of ± 10% to the
specific activity of StP isolated from C callunae [11] In the
standard assay of phosphorylase activity (Experimental
section), the R242A mutant displayed only 10% of
wild-type activity However, under conditions of saturation in
a-glucan substrate (30 gÆL)1 of maltodextrin) and
phos-phate (500 mM), the R242A mutant had a specific activity
approximately 40% that of the wild-type (A discontinuous
assay was used here because the high phosphate
concentra-tion interferes with coupled enzyme measurements [11])
The result reveals that maximum reaction rate and substrate
affinity are both decreased in the R242A mutant, compared
to the wild-type Although this implies that the replacement
Arg242 by alanine is not without effect on steps involved in enzymic catalysis, we point out that in wild-type StP, loss of active site integrity and subunit dissociation occur as, clearly, kinetically uncoupled events at an elevated tem-perature [12] Therefore, the analysis of steady-state kinetic data for the R242A mutant and the wild-type must not be interpreted to weaken the comparative evaluation of stabilities of the same enzymes, which follows later
The N-terminal His-tag causes formation
of an active StP tetramer Preparations of His-tagged StP that were > 98% pure by the criterion of a single protein band in SDS/PAGE (not shown) eluted from a Superose 12 size exclusion column in
Fig 2 Comparison of the StP amino acid sequence with the sequences of EcMalP and rmGP The alignment was performed with the
MEGALIGN program using CLUSTALW with standard settings Amino acids conserved are shaded in black Catalytic and regulatory sites
of rmGP [14] are marked above the sequence,
as follows: a, AMP-binding site; c, caffeine/ purine inhibitor site; g, active site residues; p, residues involved in covalent phosphorylation;
s, glycogen storage sites; v, pyridoxal phos-phate binding site Residues contributing to the dimer interface of rmGP are indicated using the letter, d The primary structure of StP is 41% and 42% identical to the sequences
of rmGP and EcMalP, respectively, indicating overall conservation of the structural fold [3,4] The positions of the TOWER helices in EcMalP and rmGP are underlined by a thick line, and the mutations (to be reported later) are indicated by arrows Also, note that the natural enzyme isolated from C callunae [11] lacks Ser1.
Trang 6two fractions of well-defined apparent molecular masses, as
shown in Fig 4: a major 180-kDa fraction corresponding to
the dimer and containing approximately 85% of the total
protein, and another fraction that accounted for the
remainder protein and displayed a molecular mass of
360 kDa, as expected for a StP tetramer The minor protein
fraction had the same specific enzyme activity as the dimeric
wild-type A monomer fraction was not observed The StP
mutants gave elution profiles that were superimposable to that of the wild-type The StP tetramer was of interest because mammalian GP is known to form tetramers at high protein concentrations These tetramers are inactive but dissociate into active dimers when glycogen is present [4] To determine whether dissociation of the StP tetramer could be induced in the presence of substrate, we subjected the purified tetramer fraction to size exclusion chromatography (SEC) under conditions where the elution buffer contained a saturating concentration of maltohexaose (20 mM) or a-D -glucose-1-phosphate (20 mM) The tetramer was completely stable for the time of the experiment ( 2 h) when one of the above ligands was present In marked contrast to observations made with His-tagged StP, the recombinant StP lacking the metal affinity fusion eluted as a single protein peak from the Superose 12 column Its estimated molecular mass was 180 kDa Automated Edman degra-dation of this recombinant StP yielded the sequence, Pro-Glu-Lys-Gln, for the N-terminal tetrapeptide of the recombinant wild-type, which is in accordance with the authentic N-terminal sequence of native StP [11] Therefore, the N-terminal metal affinity peptide appears to be respon-sible for the observed tetramer : dimer ratio of 0.2 in His-tagged StP, and likewise the Argfi Ala mutants thereof The results suggest that if the occurrence of tetrameric and dimeric forms of His-tagged StP truly represents an altered oligomerization equilibrium, relative to wild-type StP, and is not an artifact of the protein folding process in the E coli cytosol, the conversion of the tetramer into its constituent dimers must take place at a slow rate The data suggest that the use of amino-terminal affinity tags may not be ideal for studies of GP structure However, we emphasize that dimer : tetramer heterogeneity of His-tagged wild-type StP was not changed in the His-tagged mutants and did thus not affect the conclusions of this work
Determination of dissociation constants of binary enzyme–oxyanion complexes
Fluorescence titration assays [12] were carried out with His-tagged StP and the R234A and R242A mutants and yielded dissociation constants for enzyme–sulfate (KdSO4) and enzyme–phosphate (KdPi) complexes These Kdvalues are summarized in Table 1 The Kdvalues for the His-tagged wild-type enzyme agree closely with the corresponding values measured recently for native StP (KdSO4¼ 4;
KdPi¼ 16) [12] The data also reveal that the replacement
of the guanidinium side chain of arginine by a methyl side chain of alanine in the R234A and R242A mutants caused only a small effect on the binding of sulfate An approxi-mately 2.5-fold increase in KdSO4 was observed for the R234A mutant, compared to the wild-type The KdSO4value for the R242A mutant was very similar to that of the wild-type These observations are not consistent with a scenario
in which the original side chains of Arg234 and Arg242 participate in binding the sulfate dianion If these side chains provided direct interactions with sulfate, a much larger increase in KdSO4 would be expected for the mutants in comparison to wild-type We did not observe any significant inhibition of quenching of PLP fluorescence in the R234A and R242A mutants in the presence of phosphate at levels
of 10 m and 20 m , relative to a control that did not
Fig 4 Elution profile of purified recombinant His-tagged StP upon
analytical gel filtration on a Superose 12 column Approximately 100 lg
of protein in 100 lL triethanolamine buffer (50 m M , pH 7.0) were
applied to the column (20 mL; 1.6 cm diameter) equilibrated with
50 m M potassium phosphate buffer, pH 7.0, containing 0.2 M NaCl.
Elution was carried out with the same phosphate buffer at a flow rate
of 0.25 mLÆmin)1 using an A¨ktaexplorer 100 system (Amersham
Pharmacia Biotech) and detection at 280 nm Calibration of the sizing
column was performed using appropriate protein standards of known
molecular mass Apparent molecular masses of 180 kDa and 360 kDa
were determined for the eluting protein fractions in this figure.
Fig 3 Comparison of CD spectra of wild-type StP (solid line), R234A
mutant (dotted line), and R242A mutant (dashed line) Spectra were
recorded in a 20 m M Mops buffer, pH 7.0, not containing oxyanion.
Note that the value of protein concentration (0.1 mgÆmL)1) contained
10% error and may be partly responsible for observed differences in
molar ellipticity at 222 nm.
Trang 7contain the oxyanion This could result if the site-directed
replacement of Arg234 and Arg242 strongly weakened
binding of phosphate or if it altered the conformational
change in response to phosphate binding Considering that
values of KdSO4are not very sensitive to the mutations, the
latter interpretation would seem to be more likely, but the
relatively high KdPivalue for the wild-type prevents any firm
conclusion on the mutants
Stability of recombinant wild-type StP and TOWER helix
mutants thereof as revealed in urea and thermal
denaturation experiments
To compare the stabilities of StP and Argfi Ala mutants
thereof, we carried out urea denaturation assays in which
protein concentration and incubation time were constant
parameters, and [urea] was varied in steps of 0.25M
between 0.0 and 6.0M The chosen assay monitors enzyme
inactivation that is completely irreversible and thus
provides a measure of the kinetic stability of the respective
enzyme under the conditions used For each protein, the
dependence of percentage of remaining enzyme activity on
[urea] was analyzed under conditions in which either no
oxyanion was present, or phosphate or sulfate was added
in a concentration corresponding approximately to the
dissociation constant (Kd) of the respective binary enzyme–
phosphate or enzyme–sulfate complexat 30C (Table 1)
Saturation in oxyanion was not attempted to avoid
possible interferences of stability measurements by a
lyotropic anion effect in the presence of high
concentra-tions of phosphate or sulfate Using nonlinear least squares
regression analysis with the SIGMAPLOT 2000 programme
(SPSS Inc.), data were fitted to Eqn (1), which describes a
sigmoidal decrease of enzyme activity (EA) with increasing
concentration of denaturant,
EA (urea)¼ a=½1 þ expðb½urea cÞ ð1Þ
where a, b, and c are parameters (which are not derived
from any formal mechanism of denaturation of StP) The
apparent denaturation midpoint (Cm) is calculated by using
Eqn (1) and the respective parameter estimates, and
corresponds to the urea concentration where half the
original enzyme activity has been lost
Cmvalues for wild-type StP and two Argfi Ala mutants thereof are summarized in Table 1 Results for the R236A mutant are not shown in the Table because the stabilities of this mutant and the wild-type were identical within limits of experimental error (DCm ± 0.15M) under all conditions examined The Cmvalues in Table 1 reveal large stabilizing effects of the Argfi Ala replacements at positions 234 and
242 for conditions in which no oxyanion was present Note that values of the parameter b, which is a measure of the slope of the decrease in EA as [urea] increases, showed little variation in dependence of the enzyme or the reaction conditions and were in the range)0.43 to )0.49 The extra stability brought about by the mutations is reflected by significant shifts of the Cmvalues for the mutants, relative to that for the wild-type, to higher urea concentrations by
1M or greater The stabilization of wild-type StP by a half-saturating concentration of phosphate can be expressed quantitatively by a dramatic up-shift in Cmvalue by 4.0M, compared to the control reaction lacking phosphate The observed increase in Cm value effected by a sulfate level matching KdSO4 was 1.2M, suggesting that under the conditions used, the StP–sulfate complexdisplays a much smaller kinetic stability than the StP–phosphate complex The DCm-values for the wild-type serve as a frame of reference for analyzing the stabilities of the mutants Taking into account the large stabilization of wild-type StP by bound phosphate, it was unfortunate that KdPivalues were not accessible for the R234A and R242A mutants and so defined conditions with regard to saturation in oxyanion were possible only for sulfate
Irrespective of the added oxyanion, observed DCm-values for the R242A mutant were smaller than corresponding values for the wild-type (Table 1) Considering that sulfate binding takes place with almost identical affinities in wild-type StP and the R242A mutant and assuming that this reflects similar sulfate binding modes in both proteins, the results show that the binding event as such is not sufficient for sulfate to induce a large stabilization, which in turn is mirrored in the value of DCm It is important to recognize therefore that denaturation midpoints in the presence of half-saturating levels of sulfate were identical within the experimental error for the wild-type and the R242A mutant The simplest explanation of this finding is that
Table 1 Comparison of stabilities of recombinant wild-type StP and two enzyme variants in urea and thermal denaturation experiments at pH 7.0 The experiments were carried out in 50 m M triethanolamine buffer, pH 7.0, and used 200 lgÆmL)1of protein in each assay Other conditions and procedures were as reported previously [11,12,18] n.a., not applicable because no significant change in iodide quenching of cofactor fluorescence occurred in the presence of phosphate up to a concentration of 20 m M
Enzyme K dSO4 (m M )/K dPi (m M )
C m at 30 C ( M )/t 1/2 at 45 C (min)
No oxyanion added + sulfate a + phosphate a Wild-type 4.5 ± 0.5/18 ± 4 1.17 ± 0.03/3.2 ± 0.1 2.95 ± 0.10/stableb 5.2 ± 0.2c/stableb,c
(3.45 ± 0.03d/stableb,d) R234A 9 ± 3/n.a 2.60 ± 0.06/20 ± 1 4.45 ± 0.05/stable b 3.55 ± 0.02 d /stable b,d R242A 3.8 ± 0.4/n.a 2.00 ± 0.02/12 ± 0.5 2.93 ± 0.02/stable b 2.27 ± 0.02 d /43 ± 5 d a
Potassium phosphate and ammonium sulfate were used Unless indicated, the oxyanion concentrations matched the respective K d values It was shown in separate control experiments that the cation, K + or NH 4+, had no influence on stabilities of wild-type StP and mutants thereof b Being stable means that no significant inactivation occurred during a 0.5-h long incubation at 45 C c,d Data obtained in the presence ofc20 m M andd10 m M phosphate.
Trang 8the enzyme-sulfate complexes of wild-type and R242A
mutant share similar kinetic stabilities; and that the
Argfi Ala replacement at position 242 offsets the
stabi-lizing effect of sulfate binding in the wild-type to the extent
that this mutation stabilizes the enzyme when no sulfate is
present (Fig 5B) Interestingly therefore bound sulfate
stabilized the R234A mutant and the wild-type equally
Hence, although Arg234 is clearly destabilizing in unligated
StP, site-directed mutagenesis of the side chain of Arg234 into the methyl side chain of alanine did not diminish the stabilizing effect of sulfate binding in comparison to wild-type, as it was observed for the R242A mutant This result is interesting because it leads to a different interpretation of the role of Arg234 and Arg242 for oxyanion-dependent stability of StP The stabilization brought about by the presence of 10 mMof phosphate was substantially smaller for the R234A mutant (DCm1M) than the wild-type (DCm2.3M) Even in the absence of a KdPi value for R234A (and likewise R242A), the comparison at a fixed phosphate level is relevant It shows that site-specific replacement in each mutant either decreases the affinity for phosphate, relative to the wild-type, or lowers the kinetic stability of the mutant-phosphate complex, relative to the same wild-type complex Figure 5 illustrates this point by comparing the dependence of DCm-values on the concen-trations of phosphate and sulfate for wild-type StP and the R242A mutant The results show a marked preference for stabilization by sulfate over stabilization by phosphate
in the mutant, which is clearly different to what was observed for the wild-type Note that the separ-ation of the parallel lines in panel B of Fig 5 corresponds to the difference in Cm-values for the R242A mutant and the wild-type under conditions in which no sulfate was added The data in Fig 5 can be used to roughly estimate the apparent half-saturation constants (app K) for the stabil-ization of the wild-type (app KSO417 mM; app
KPi28 mM) and the R242A mutant (app KSO412 mM; app KPi 130 mM)
Thermal stabilities of wild-type and Argfi Ala mutants were determined at 45C and are shown in Table 1 In the absence of added oxyanion, the R242A and R234A mutants were 3.8- and 6.2-fold more stable than the wild-type, respectively No significant inactivation of StP and the two enzyme variants was seen over an incubation time of 30 min
in the presence of sulfate concentration matching KdSO4 In the presence of 10 mMphosphate, wild-type and the R234A mutant were stable while the R242A mutant displayed significant loss of activity
Discussion
The goal of the present paper was to advance the relationships between structure and oxyanion-dependent stability of StP from Corynebacterium callunae Cloning, sequencing, and heterologous expression of the gene encoding StP were essential requirements for the utilization
of site-directed mutagenesis to examine the functional roles
of potentially important amino acid residues that were identifiable through analysis of the StP primary structure The results have revealed clearly that Arg234 and Arg242 of the TOWER interface region of StP partially destabilize the dimer structure of the unligated enzyme so that loss of these residues in the Argfi Ala mutants leads to significantly higher kinetic stability Phosphate binding appears to cause
a change in interactions of these arginines, most probably by
an allosteric mechanism as discussed below, contributing to the observed stabilization An unexpected finding was that replacements of Arg242 and Arg234 induced a large apparent preference for sulfate over phosphate with regard
to the stabilizing effect
Fig 5 Stabilization of wild-type StP and the R242A mutant by
phos-phate (A) and sulfate (B) against urea denaturation Results show DC m
-values, which report the difference between C m at the shown oxyanion
concentration and the C m measured in buffer lacking oxyanion The
data are presented as a double reciprocal plot to emphasize the
satu-ratable dependence of DC m on [oxyanion] However, extrapolation to
infinite [oxyanion] must be made with caution (hence, the broken lines)
because of the additional lyotropic anion effect Also note that in panel
A, lines do not have identical intercept values Experiments were
car-ried out at 30 C in 50 m M triethanolamine buffer, pH 7.0, using
conditions reported in the text.
Trang 9Relationships between StP structure and
oxyanion-dependent kinetic stability
Previous studies have shown that subunit dissociation
occurs as an early step during denaturation of StP at
elevated temperatures (30C) [12] or in urea (R Griessler,
& B Nidetzky, unpublished observations) Under
condi-tions of dilute protein and in the absence of free PLP, loss of
oligomer structure is accompanied by immediate release of
cofactor from the StP subunit Therefore, it is not detectably
reversible on the time scale of the assay for phosphorylase
activity ( 1–2 min) [12] Measurement of irreversible
inactivation of StP can thus serve as a useful reporter of
the protomer dissociation event It would seem likely
therefore that observed changes in Cmand t1/2-values for
irreversible inactivation in urea and at elevated
tempera-tures, brought about by site-specific amino acid
replace-ments in the dimer contact region of StP and likewise,
oxyanion bound at the enzyme oxyanion site, result from
altered kinetic barriers for subunit dissociation, relative to
unliganded wild-type StP We stress, however, that based on
the available data, it is not possible to rule out completely a
contribution of thermodynamic effects to the measured
kinetic stabilities
Arginine residues are known for their prevalence in both
intra- and inter-chain interfaces [22,23] where the charged
guanidinium group is often involved in formation of strong
intermolecular hydrogen bonds Such non-covalent
inter-actions have been hypothesized to stabilize multidomain
and oligomeric proteins by strengthening either the network
of interfacial contacts or the tertiary bonds that prevail in
the segment of the interface Site directed mutagenesis has
been used, in a few instances though, to verify the role of
arginines as stabilizing elements of dimer contact regions
[24,25] Therefore, irrespective of the exact orientation of
Arg234, Arg236, and Arg242 at the TOWER interface
region of StP, it was unexpected that two out of three
enzyme variants harboring the Argfi Ala substitution
exhibited a considerably greater kinetic stability in thermal
and urea denaturation studies than the wild-type
Hydro-gen-bonding or other electrostatic interactions involving the
TOWER arginines are obviously not optimized for kinetic
stability In this scenario, oxyanions could have a stabilizing
effect if their binding was capable of either decreasing
nonfavorable contacts between protomers or increasing the
favorable ones This could occur by various mechanisms,
but likely an allosteric one in which oxyanion binding
affects the tertiary and/or quarternary interactions involving
Arg234 and Arg242 thus leading to a greater stability
For the interpretation of the kinetic stabilities of the
R234A and R242A mutants, it is most useful to first
consider the effects of phosphate and sulfate on
conforma-tion and stability of wild-type StP It was shown here that
under conditions of half-saturation in oxyanion, phosphate
stabilizes the wild-type much more efficiently against
denaturation by urea than sulfate, the difference in DCm
-value (which is the increase in Cm compared to the
unliganded enzyme when oxyanion is present) being as
large as 2.2M A greater stability of the enzyme-phosphate
complexthan the enzyme–sulfate complexcorrelates well
with a greater compactness of the former complex,
as revealed recently by comparing iodide quenching of
cofactor fluorescence in StP saturated with phosphate and sulfate [12] The results for wild-type StP imply that a conformational change in protein structure accompanies the oxyanion-binding event and is required for kinetic stability The extent of the structural rearrangement is larger for a phosphate than a sulfate ligand, suggesting that more binding energy from the StP–oxyanion interaction can be translated into a stabilized protein conformation when phosphate is bound
The comparison of Kdvalues for enzyme-sulfate com-plexes of wild-type and the two Argfi Ala mutants reveals that a direct participation of the side chains of Arg234 or Arg242 in binding of sulfate is not likely However, both arginines, clearly, take part in the just described oxyanion-dependent conformational relay of wild-type StP, and analysis of R234A and R242A mutants serves to emphasize the differential effect of bound phosphate and sulfate in the wild-type In both mutants, however, mainly R242A, phosphate has lost much of the stabilizing potential originally present in the wild-type Expressed as the ratio
of DCm (M) and [phosphate] (M), the phosphate-specific stabilization is 30 for the wild-type, but only 10 and
1 for the R234A and R242A mutant, respectively The situation is different for sulfate, which stabilizes wild-type StP and the R234A mutant to approximately the same extent In the R242A mutant, the stabilizing effect of the Argfi Ala replacement in the unligated protein is offset by the smaller stabilization when sulfate is bound, compared to wild-type StP In conclusion, these data can be summarized
to yield the following hypothetical model of the stabilization
of StP by oxyanions Phosphate binding at an allosteric site, perhaps within the subunit, leads to propagation of a conformational change into the dimer contact region of the protein Arg242 is a key residue implicated in this structural rearrangement and may even have an active role in relaying the phosphate-dependent and to a lesser extent though, the sulfate-dependent conformational switches Arg234 appears
to be part of the relay when phosphate is bound, but not when sulfate is bound
Comparison of StP with rmGP and other a-glucan phosphorylases
Structure-function studies of rmGP are highly relevant for the interpretation of results for StP First of all, a dissociative mechanism of thermal denaturation of rmGP, similar to that proposed for StP, has been reported recently [26] A major difference between rmGP and StP, however, pertains to the moderate effect that oxyanions have on rmGP stability [26] Secondly, Fletterick and coworkers have mutated TOWER helixresidues of rmGP, among them Arg277, which is the rmGP counterpart of Arg242, into alanine and characterized the variant enzymes struc-turally and with respect to allosteric activation by AMP [27] Their conclusion from a detailed comparison of intersubunit contacts in X-ray structures of wild-type rmGP and R277A was that the Argfi Ala replacement would destabilize significantly the quaternary interactions originally present in the muscle enzyme Keeping in mind the limitations of using irreversible inactivation as a measure of global protein stability, our results then suggest that Arg242 in StP must participate in interactions clearly different from those of the
Trang 10corresponding residue in rmGP Interestingly, mutating
Arg242 and Arg277 had similar effects on the catalytic
competence of StP and rmGP, respectively, resulting in each
case, in a significant decrease in specific activity, compared
to the wild-type level The side chains of Arg269 and Arg277
make direct hydrogen bonds across the dimer interface of
activated rmGP with side chains of Asn250¢ and Asn270¢,
respectively, on the adjacent subunit Each of the two
asparagines of rmGP is replaced positionally by a glutamate
in StP Considerations of charge and packing arrangements
suggest that if Arg234 and Arg242 were truly involved in
inter-subunit interactions analogous to those seen in rmGP
structures, bonding across the interface of StP should be
stronger, compared to rmGP, which is unlikely in light of
the experimental evidence for StP Another interesting
difference between StP and rmGP revealed by
structure-based sequence comparison pertains to contacts of Arg277
(and likewise Arg242) within the subunit In rmGP, this
arginine forms a charged hydrogen bond with the
carboxy-late group of Glu162 [4,27] In StP, Glu162 is positionally
replaced by an arginine (Arg142), and this is a likely reason
for different atomic environments of Arg242 in StP and
Arg277 in rmGP
Multiple alignment of the first one-hundred a-glucan
phosphorylase sequences identified through screening of the
nonredundant data bases with the StP primary structure
using the BLAST2 program (not shown) revealed an
interesting conservation pattern for positions equivalent
to Arg142 and Arg242 of StP In all but two cases, namely
a-glucan phosphorylases from Corynebacterium glutamicum
(Q8NQW4) and Fusobacterium nucleatum (Q8RF61), the
pair of arginine residues found in StP is not observed A pair
of amino acids with oppositely charged side chains,
glutamate (or aspartate) at position 142 and arginine (or
lysine) at position 242, occurs most frequently in the aligned
sequences Several other pairwise combinations of amino
acids are possible, but bulk and charge at a certain position
appear not to be conserved across all organisms and cell
types To give two examples for structurally characterized
enzymes, EcMalP has a glutamine-lysine pair whereas
Saccharomyces cerevisiae glycogen phosphorylase has a
glutamate-alanine pair However, an interesting
generaliza-tion is that enzymes from (hyper)thermophilic bacteria and
archaea contain a conserved pair of lysine (position 142)
and glutamic acid (position 242) It would be interesting
therefore to examine if positional charge reversal for
extremophilic structures, compared to most other a-glucan
phosphorylase sequences including the mammalian ones, is
related to increased stability [28–30] Furthermore, our
comparisons show that if an arginine residue occurs at
position 142 in a-glucan phosphorylases, position 242 is
generally taken by an alanine, serine, or threonine These
residues whose side chains are uncharged and sterically less
demanding than the side chain of arginine may be primed
to avoid unfavorable (destabilizing) interactions with or
relayed to counterpart Arg142 This interpretation of the
sequence changes among aligned a-glucan phosphorylases
is in excellent agreement with the observed kinetic stability
of R242A mutant of StP, compared to wild-type enzyme It
also provides a rational for an allosteric mechanism of
stabilization of StP by oxyanion binding In light of the fact
that Arg242 is conserved in all mammalian GPs and
considering that the R242A mutant shows only 40% of wild-type activity, it seems probable that Arg242 has been selected in StP for so far unknown reasons of enzyme function
Acknowledgements
Financial support from the Austrian Science Funds (P-15118-MOB to B.N.) is gratefully acknowledged.
References
1 Palm, D., Klein, H.W., Schinzel, R., Buehner, M & Helmreich, E.J (1990) The role of pyridoxal 5¢-phosphate in glycogen phos-phorylase catalysis Biochemistry 29, 1099–1107.
2 Schinzel, R & Nidetzky, B (1999) Bacterial a-glucan phos-phorylases FEMS Microbiol Lett 171, 73–79.
3 Watson, K.A., Schinzel, R., Palm, D & Johnson, L.N (1997) The crystal structure of Escherichia coli maltodextrin phosphorylase provides an explanation for the activity without control in this basic archetype of a phosphorylase EMBO J 16, 1–14.
4 Johnson, L.N (1992) Glycogen phosphorylase: control by phos-phorylation and allosteric effectors FASEB J 6, 2274–2282.
5 Rath, V.L & Fletterick, R.J (1994) Parallel evolution in two homologues of phosphorylase Nat Struct Biol 1, 681–690.
6 Newgard, C.B., Hwang, P.K & Fletterick, R.J (1989) The family
of glycogen phosphorylases: structure and function Crit Rev Biochem Mol Biol 24, 69–99.
7 Buchbinder, J.L., Rath, V.L & Fletterick, R.J (2001) Structural relationships among regulated and unregulated phosphorylases Annu Rev Biophys Biomol Struct 30, 191–209.
8 Tu, J.-I & Graves, D.J (1973) Association-dissociation properties
of sodium borohydride-reduced phosphorylase b J Biol Chem.
248, 4617–4622.
9 Feldmann, K., Zeisel, H & Helmreich, E (1972) Interactions between native and chemically modified subunits of matrix-bound glycogen phosphorylase Proc Natl Acad Sci USA 69, 2278– 2282.
10 Tagaya, M., Shimomura, S., Nakano, K & Fukui, T (1982)
A monomeric intermediate in the reconstitution of potato apophosphorylase with pyridoxal 5¢-phosphate J Biochem 91, 589–597.
11 Weinha¨usel, A., Griessler, R., Krebs, A., Zipper, P., Haltrich, D., Kulbe, K.D & Nidetzky, B (1997) a-1,4- D -glucan phosphorylase
of gram-positive Corynebacterium callunae: isolation, biochemical properties and molecular shape of the enzyme from solution X-ray scattering Biochem J 326, 773–783.
12 Griessler, R., D’Auria, S., Tanfani, F & Nidetzky, B (2000) Thermal denaturation pathway of starch phosphorylase from Corynebacterium callunae: oxyanion binding provides the glue that efficiently stabilizes the dimer structure of the protein Protein Sci.
9, 1149–1161.
13 Watson, K.A., McCleverty, C., Geremia, S., Cottaz, S., Driguez,
H & Johnson, L.N (1999) Phosphorylase recognition and phos-phorolysis of its oligosaccharide substrate: answers to a long outstanding question EMBO J 18, 4619–4632.
14 Hudson, J.W., Golding, G.B & Crerar, M.M (1993) Evolution of allosteric control in glycogen phosphorylase J Mol Biol 234, 700–721.
15 Lin, K., Hwang, P.K & Fletterick, R.J (1997) Distinct phos-phorylation signals converge at the catalytic center in glycogen phosphorylases Structure 5, 1511–1523.
16 Sambrook, J., Fritsch, E.F & Maniatis, T (1989) Molecular Cloning: A Laboratory Manual, 2nd edn Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, USA.