Báo cáo khoa học: Two short protein domains are responsible for the nuclear localization of the mouse spermine oxidase l isoform pdf

By amino acid sequence comparison and molecular mode-ling of mSMO proteins, we identified a second domain that is necessary for nuclear localization of the mSMOl splice variant.. Abbrevia

localization of the mouse spermine oxidase l isoform

Marzia Bianchi1, Roberto Amendola2, Rodolfo Federico1, Fabio Polticelli1and Paolo Mariottini1

1 Dipartimento di Biologia, Universita` ‘Roma Tre’, Roma, Italy

2 Istituto per la Radioprotezione, ENEA, CR Casaccia, Roma, Italy

The polyamines putrescine (Put), spermine (Spm) and

spermidine (Spd) are aliphatic amines that are

posi-tively charged under physiological conditions and

have been shown to be involved in major cellular

pro-cesses such as cell growth and proliferation [1,2] The

concerted actions of Spd⁄ Spm N1-acetyl-transferase,

vertebrate polyamine oxidase (PAO) (EC 1.5.3.11)

and spermine oxidase (SMO) are involved in

main-taining polyamine homeostasis in mammalian cells

The cytosolic Spd⁄ Spm N1-acetyl-transferase enzyme

is responsible for adding N1-acetyl groups to both

Spm and Spd [3] The N1-acetylated Spm and Spd are

oxidized by the peroxisomal FAD-containing enzyme,

PAO, to yield stoichiometric amounts of

3-acetamido-propanal and H2O2, plus Spd and Put, respectively

[4–6] The last enzyme involved in the mammalian

polyamine homeostasis is the flavoprotein SMO,

which preferentially oxidizes Spm, producing Spd,

3-aminopropanal and H2O2 [7–9]

Analysis of the expression of the mouse SMO gene (mSMO), encoding at least nine splice variants, as well

as biochemical characterization of the canonical alfa isoform (mSMOa), have been reported recently [10,11] The subcellular localization of the catalytically active isoforms mSMOa and mSMOl has been investigated

in the transiently and stably transfected murine neuro-blastoma cell line, N18TG2 Interestingly, mSMOl is present in both nuclear and cytoplasmic compart-ments, while mSMOa is cytosolic The only structural difference between the two isoforms is the presence of

an extra protein domain in mSMOl, encoded by the exon VIa [10]

Comparative analysis of the amino acid sequence

of the vertebrate members of the SMO family has revealed a region that is extremely conserved in mam-mals, highly variable and⁄ or reduced in length in non-mammalian vertebrates, and absent in the aligned PAO sequences Molecular modeling of mSMO

Keywords

mouse; nuclear localization; polyamine

oxidase; polyamines; spermine oxidase

Correspondence

P Mariottini, Dipartimento di Biologia,

Universita` degli Studi ‘Roma Tre’, Viale

Guglielmo Marconi 446, 00146 Roma, Italy

Fax: +39 06 55176321

Tel: +39 06 55176359

E-mail: mariotpa@bio.uniroma3.it

(Received 18 February 2005, revised 7 April

2005, accepted 13 April 2005)

doi:10.1111/j.1742-4658.2005.04718.x

In mouse, at least two catalytically active splice variants (mSMOa and mSMOl) of the flavin-containing spermine oxidase enzyme are present We have demonstrated previously that the cytosolic mSMOa is the major iso-form, while the mSMOl enzyme is present in both nuclear and cytoplasmic compartments and has an extra protein domain corresponding to the addi-tional exon VIa By amino acid sequence comparison and molecular mode-ling of mSMO proteins, we identified a second domain that is necessary for nuclear localization of the mSMOl splice variant A deletion mutant enzyme of this region was constructed to demonstrate its role in protein nuclear targeting by means of transient expression in the murine neurobla-stoma cell line, N18TG2

Abbreviations

MPAO, maize polyamine oxidase; mPAO, mouse polyamine oxidase; hSMO, human spermine oxidase; mSMO, mouse spermine oxidase; NDA, nuclear domain A; NDB, nuclear domain B; PAO, polyamine oxidase; Put, putrescine; SMO, spermine oxidase; Spd, spermidine; Spm, spermine.

proteins, based on the 3D structure of maize

poly-amine oxidase (MPAO), indicated that this region is

localized on the tip of the FAD-binding domain, in

close spatial proximity to the protein region encoded

by exon VIa of the mSMOl isoform This observation

has led us to hypothesize that these two protein

domains, named nuclear domain A and nuclear

domain B (NDA and NDB, respectively), may have

coevolved in mammalian SMOs and that they may

cooperate in targeting the mSMOl isoform to the

nuc-leus By means of transient expression of the deletion

mutant, mSMOlD, in the murine neuroblastoma cell

line, N18TG2, we demonstrated that removal of the

NDA amino acid region abolishes proper nuclear

targeting of the mSMOl isoform

Results

Structural analysis and modeling of vertebrate

SMO proteins

Comparison of the derived amino acid sequences of

vertebrate SMO proteins has revealed that their overall

primary structure is well conserved Taking the

sequence of the human SMO (hSMO) as the reference

point, the amino acid identity ranges from  99%

(chimpanzee) to  67% (pufferfish); as expected, the

identity decreases to 40% when compared to

the mouse PAO (mPAO) primary sequence (Fig 1A)

The only region that shows a low degree of

conserva-tion among SMO proteins, when comparing mammals

to other vertebrates, is the central part of the primary

sequence, located between positions 277 and 307 in the

mSMO sequence (Fig 1B)

This region, of 31 amino acids, has not been shown

to contain any residue involved in either the catalytic

site or the FAD-binding domain [9,10,12,13]

Interest-ingly, this 31 amino acid region is highly conserved

among mammals (human, chimpanzee, dog, cow and

rodents), with an identity ranging from 82 to 95%,

while there is little, if any, conservation with chicken,

frog or fish counterparts It is interesting to note

that the sequence analysis of the mammalian genes

encoding SMO (AL121675, human; NW120319,

chim-panzee; AF498364, mouse; NW0436471, rat;

AAEX01031426, dog; AAFC01101092, cow, partial

gene sequence) has revealed the presence of the extra

exon VIa [10] (Fig 1B) By contrast, the same analysis

performed on the homolog SMO genes of chicken

(M_420872) and pufferfish (http://www.ensembl.org/

Fugu_rubripes/) shows the lack of this extra domain

This observation suggests that the presence of the extra

exon VIa is a mammalian feature that is strictly related

to the high homology displayed by the 31 amino acid region (residues 277–307; numbering of the human SMO enzyme) The two protein domains may have coevolved, conferring novel properties to mammalian SMOs

Molecular modeling of the 3D structure of the mSMOl isoform was thus carried out in order to test the hypothesis that the 277–307 region and the protein domain encoded by the exon VIa could be spatially contiguous and represent a functional epitope involved

in a mammalian-specific function of SMO (e.g nuclear targeting of the mSMOl isoform) Inspection of the mSMOl modelled structure (Fig 2) indicates that both regions are located on the tip of the FAD-binding domain, with residues 300–307 located in close spatial proximity to the extra domain of mSMOl

Hence, we postulated that both regions could be involved in the nuclear targeting of the mSMOl enzyme With this rationale, we made a deletion mutant of the mSMOl isoform, deleting exactly the region 277–307, as described in the Experimental pro-cedures (Fig 3A)

Expression and purification of mSMOlD protein

in Escherichia coli cells The recombinant cDNA construct, pmSMOlD-HT, and the controls pmSMOa-HT and pmSMOl-HT, were used to transform E coli BL21 DE3 cells After induction and over-expression, the proteins were puri-fied by using a His-Bind chromatography kit (Novagen, Darmstadt, Germany) The SDS⁄ PAGE electrophoretic analysis performed on purified recombinant mSMO proteins is shown in Fig 3B The enzyme activities were measured spectrophotometrically and the catalyti-cally active proteins were expressed at levels ranging from 5 to 15 IUÆL)1of culture broth

Kinetic properties of the mSMOlD protein The biochemical properties of mSMOa and mSMOl have been reported previously [9,10] The recombinant mSMOlD isoform also shows catalytic activity The substrate specificity of mSMOlD for Spm, Spd and

N1-acetylpolyamines has been investigated under stand-ard conditions at pH 8.5 Purified mSMOlD specifically oxidizes Spm and is not active on Spd, N1-acetylSpd

or N1-acetylSpm Values of Km, Vmaxand pH optimum were determined by using Spm as the substrate The purified mSMOlD exhibited biochemical properties very similar to those of mSMOa and mSMOl, in par-ticular a pH optimum of 8.5 in 0.1 m NaPibuffer, a Km value of 220 lm and a kcatvalue of 1.25 s)1

B

Fig 1 Amino acid sequence comparison of members of the spermine oxidase (SMO) and polyamine oxidase (PAO) family (A) Amino acid sequence alignment of the SMO and PAO proteins (B) Alignment of the deduced amino acid sequences corresponding to nuclear domain A and nuclear domain B (exon VIa) Multi-alignment was performed by using the program CLUSTAL W SEQUENCE ALIGNMENT HsSMO, Homo sap-iens (AAN77119); PtSMO, Pan troglodytes (NW120319); CfSMO, Canis familiaris (AAEX01031426); BtSMO, Bos taurus (AAFC01101092); MmSMO and MmPAO, Mus musculus (AAN32915) and (AAN40705), respectively; RnSMO, Rattus norvegicus (XP_218704); GgSMO, Gallus gallus (XP_420872.1); XlSMO, Xenopus laevis (Q6INQ4); DrSMO, Danio rerio (Q6NYY8); and FrSMO, Fugu rubripes (http://www.ensembl org/Fugu_rubripes/).

Cell localization of mSMOlD protein in murine

neuroblastoma N18TG2 cells

The mSMOlD mutant protein was transiently

expres-sed in the neuroblastoma cell line, N18TG2, to

investi-gate its subcellular localization Augmented transcript

levels for each recombinant protein were detected in

transiently transfected neuroblastoma cells, using

b-actin as a control housekeeping gene to monitor RNA

stability, and amount of processed RNA for each

sample (Fig 3C)

To establish where each tagged protein was

locali-zed, a confocal microscopy investigation was carried

out, using the V5-TAG as epitope to direct primary

mAbs As shown in Fig 4, in N18TG2⁄ pcDNA3 ⁄

mSMOa-V5 and N18TG2⁄ pcDNA3 ⁄ mSMOlD-V5

trans-iently transfected cells, we observed a cytoplasmic

localization of the tagged recombinant proteins By

contrast, in N18TG2⁄ pcDNA3 ⁄ mSMOl-V5 transiently

transfected cells, we confirmed a nuclear localization

for the mSMOl isoform (Fig 4)

Taken together, these results consistently

substanti-ate the hypothesis that these two structural regions are

mandatory for the nuclear localization of mSMOl, as

the only difference between mSMOl and mSMOlD

proteins consists of the lack of the amino acid

sequence region 277–307 (Figs 1,2)

Discussion

In the murine polyamine homeostasis at least two

cata-lytically active splice variants of the spermine oxidase

enzyme are involved The cytosolic mSMOa is the

major isoform, while the mSMOl enzyme, displaying

an extra protein domain corresponding to the addi-tional exon VIa, is localized in both the cytoplasm and the nucleus The overall primary structure of verteb-rate SMO enzymes is well conserved, with the excep-tion of a region comprising 31 residues (amino acids 277–307) Molecular modeling of the 3D structure of mSMOl indicates that this region (NDA) is localized

on the tip of the FAD-binding domain and is located near the protein region encoded by exon VIa (NDB) This amino acid region is highly conserved in mam-mals, while it is highly variable and⁄ or reduced in length in nonmammalian vertebrates, indicating that

a selective evolutionary constrain is operating on it Interestingly, the presence of exon VIa in vertebrate SMO gene sequences is also a unique mammalian feature

These data suggest that the two domains NDA and NDB, not involved in enzyme activity or FAD bind-ing, could be responsible for the interaction with the nuclear targeting machine With this hypothesis in mind, we constructed a deletion mutant lacking the amino acid region 277–307, named mSMOlD We expressed this mutant in E coli cells and, as expected, the purified recombinant protein showed a catalytic activity comparable to that of the wild-type mSMOl [10] Notably, by means of transient expression of mSMOlD in the murine neuroblastoma cell line, N18TG2, we demonstrated that deletion of the 277–

307 region abolished nuclear targeting The presence of the translated region encoded by exon VIa in mSMOl

is thus necessary, but not sufficient, for the correct localization of this isoform within the nucleus In con-clusion, the mSMOl enzyme needs at least two domains to be nuclear localized

Fig 2 Stereo representation of the

mod-elled 3D structure of the mouse spermine

oxidase catalytically active splice variant,

mSMOl The molecular surface of the

pro-tein is shown in a ‘mesh’ representation.

The backbone and the molecular surface of

nuclear domains A and B (see the text) are

coloured green and blue, respectively The

FAD cofactor is shown as red sticks The

figure was produced by using GRASP [21].

Experimental procedures

Chemicals

Spd, Spm, N1-acetylspermidine, N1-acetylspermine and Put

were purchased from Sigma (Milan, Italy) Restriction

enzymes and DNA-modifying enzymes were purchased

from MBI Fermetas Taq polymerase and M-MLV reverse

transcriptase enzymes were from Promega (Milan, Italy)

Other chemicals were from Sigma, Bio-Rad (Milan, Italy)

and J T Baker (Milan, Italy)

DNA methodology and construction of the

mSMO expression plasmid

DNA manipulation was carried out by using standard

techniques [12] The absence of errors in DNA products

generated by the PCR was verified by sequence analysis The deletion mutant of the mSMOl protein was con-structed by the PCR following the method described by Horton [13] and by using the mSMOl cDNA as a template The mutagenic primer sequences used are avail-able on request from the first author (M.B.) The intro-duction of the deletion was confirmed by sequence analysis

Amino acid sequence analysis and molecular modeling

Overall and local amino acid sequence identity between SMOs and other proteins belonging to the PAO family has been determined from multiple sequence alignments obtained using clustal w [14] The molecular model of

A

B

C

Fig 3 Amino acid sequence alignment, protein purification and RT-PCR transcript analysis of the mouse spermine oxidase catalytically active splice variants mSMOa, mSMOl and mSMOlD (A) Amino acid sequence alignment of the region enclosing nuclear domains A and B (exon VIa) of mSMOa, mSMOl, mSMOlD and mouse polyamine oxidase (mPAO) isoforms Dele-ted residues are marked by dots; the mPAO gap is represented by a dashed region Amino acid numbering is shown on the right side of the figure (B) SDS ⁄ PAGE analysis

of the recombinant mSMOa, mSMOl and mSMOlD proteins (5–10 lg of the purified enzyme) after staining the gel with Coomas-sie Brilliant Blue MW, protein molecular mass markers (MBI Fermentas) (C) Total RNA extracted from different homogenates was analyzed by RT-PCR within the linear range A representative RT-PCR experiment from three independent experiments is shown M, GeneRuler 1 kb DNA ladder (MBI Fermentas); /, /X174-HaeIII digested DNA marker (MBI Fermentas); NT, untrans-fected cells; P, cells transuntrans-fected with pcDNA 3 -V5-TAG; Ta, l and lD, cells trans-fected with pcDNA3⁄ mSMOa, l and

lD, ⁄ V5-TAG plasmids; C, no-template control.

mSMOl was built by using the crystal structure of MPAO

as a template (PDB code: 1B37) [12] Given the fairly low

sequence identity between mSMOl and MPAO (26.5%), a

reliable alignment between the two protein sequences was

derived from the multiple sequence alignment between

mSMOs, MPAO and other PAOs with known amino acid

sequence, obtained by using clustal w In addition, the

alignment was manually refined on the basis of mSMOl

secondary structure prediction, obtained using the Predict

Protein server [17] (available online at http://cubic.bioc

columbia.edu/predictprotein), to avoid the unlikely

occur-rence of insertions and deletions within secondary structure

elements Based on this alignment, the 3D structure of

mSMOl was then built by using nest, a fast

model-build-ing program that applies an ‘artificial evolution’ algorithm

to construct a model from a given template and alignment

[18]

Expression of mSMOa, mSMOl and mSMOlD isoforms in E coli cells

E coli BL21 DE3 (Novagen) cells transformed with the pmSMOa and pmSMOl plasmids, as described previously [10], and with the pmSMOlD plasmid, were cultured at

30
C in Luria–Bertani (LB) medium, containing 50 lgÆmL)1 ampicillin, to an attenuance (D) of 0.6 at 600 nm, and then induced with isopropyl thio-b-d-galactoside (0.4 mm final concentration), followed by further culture for 5 h at 30
C The E coli BL21 DE3 cells were harvested by centrifugation

at 4
C for 10 min at 10 000 g, washed with 0.4 culture vol-umes of 30 mm Tris⁄ HCl, pH 8.0, containing 20% (w ⁄ v) sucrose and 1 mm EDTA, and then incubated for 5–10 min

at room temperature Each suspension was centrifuged at

10 000 g for 10 min at 4
C and then the pellets were resus-pended in 0.05 culture volumes of ice-cold 5 mm MgSO4,

Fig 4 Subcellular localization of the mouse

spermine oxidase catalytically active splice

variants mSMOa, mSMOl and mSMOlD

in transiently transfected neuroblastoma

N18TG2 cells Transiently transfected cells

are indicated on the left side of the figure.

Anti-V5 and propidium iodide (PI) dye

col-umns indicate the secondary

immuno-fluorescence detection and nuclei

counterstaining, respectively Merge

column is the result of overlapping images.

with vigorous shaking, for 10 min on ice The resuspended

pellets were then centrifuged at 10 000 g for 10 min at

4
C The supernatant, corresponding to the periplasmic

fraction, was collected

Rapid affinity purification of mSMOa, mSMOl

and mSMOlD isoforms with pET His Tag systems

The supernatant from E coli BL21 DE3 cells transformed

with the plasmids pmSMOa-HT, pmSMOl-HT or

pmSMOlD-HT was applied to a column (3 mL) with

Ni2+ cations immobilized on the His-Bind resin

(Nov-agen), equilibrated with Binding Buffer (5 mm imidazole,

0.5 m NaCl, 20 mm Tris⁄ HCl pH 7.9) The column was

washed with 20 mm Tris⁄ HCl, pH 7.9, containing 60 mm

imidazole and 0.5 m NaCl, and then eluted with 20 mm

Tris⁄ HCl, pH 7.9, containing 750 mm imidazole and 0.5 m

NaCl

Determination of the enzyme activity and kinetic

constants of recombinant mSMO

Enzyme activity was measured by using the

spectrophoto-metric assay previously described by Cervelli et al [10]

The measurements were performed in 0.1 m sodium

phos-phate (NaPi) buffer, pH 8.5, with different substrates at

various concentrations Km and kcat values were

deter-mined using Spm as a substrate at concentrations ranging

from 50 to 500 lm, at a constant mSMO isoform

concen-tration of 2.0· 10)8m Enzyme activities were expressed

in international units (IU: the enzyme concentration that

catalyzed the oxidation of 1 lmol of substrateÆmin)1) per

litre of culture broth Protein content was estimated by

the method of Markwell et al [19] with BSA as a

stand-ard SDS⁄ PAGE was performed according to the method

of Laemmli [20]

Expression of mSMOa, mSMOl and mSMOlD

isoforms in murine neuroblastoma N18TG2 cells

All experiments were performed using a pool isolated from

three separate transient transfections mSMOa, -l and -lD

cDNA coding sequences were cloned into directional

pcDNA3-V5-TAG plasmid (Invitrogen, Milan, Italy),

accord-ing to the manufacturer’s instructions, to produce

recom-binant V5-tagged proteins Cell culture conditions and

transfection procedures of the murine neuroblastoma

N18TG2 cell line have been described previously [10]

Aliquots of selected N18TG2 cells were seeded on chamber

slides and, 24 h later, fixed with fresh 3.7% (v⁄ v)

parafor-maldehyde in NaCl⁄ Pi(15 min at 4
C) to evaluate the

sub-cellular localization of the various isoforms Determination

of the subcellular localization of mSMOa, -l and -lD ⁄

V5-tagged proteins was carried out by indirect

immunoflures-cence experiments with mouse anti-V5 mAb (Sigma) [1 lgÆmL)1, 1% (w⁄ v) BSA in NaCl ⁄ Pi], followed by secon-dary detection using fluorescein isothiocyanate (FITC)-con-jugated goat polyclonal anti-mouse IgG (Sigma) [diluted

1 : 60; 1% (w⁄ v) BSA in NaCl ⁄ Pi] Nuclei were counter-stained with propidium iodide and digital images were taken with a LSM510 confocal microscope (Carl Zeiss, Milano, Italy)

The transfection efficiency was verified by RT-PCR ana-lysis, utilizing the same experimental conditions as des-cribed previously [10] The mSMOa-specific primer pairs used were as follows: mSMOa1 forward 5¢-GTACCTGAA GGTGGAGAGC-3¢ and mSMOa2 reverse 5¢-TGCATG GGCGCTGTCTTGG-3¢; mSMOl and mSMOlD specific primer-pairs: mSMOl1 forward 5¢-GATGAGCCGTGG CCTGT-3¢ and mSMOl2 reverse 5¢-CTTTATGGAGCC CCTACTAG-3¢; murine rpS7 control specific primer-pairs: rpS7-forward 5¢-CGAAGTTGGTCGG-3¢ and rpS7-reverse 5¢-GGGAATTCAAAATTAACATCC-3¢; b-actin control specific primer pairs: b-actin-forward 5¢-TGTTACCAACT GGGACGACA-3¢ and b-actin-reverse 5¢-AAGGAAGGC TGGAAAAGAGC-3¢ Three separate experiments were performed from each RNA preparation

Acknowledgements

This research was partially supported by the grant PRIN 2003 from ‘Ministero Istruzione, Universita` e Ricerca’ (MIUR)

References

1 Wallace HM, Fraser AV & Hughes A (2003) A perspec-tive of polyamine metabolism Biochem J 376, 1–14

2 Seiler N (2003) Thirty years of polyamine-related approaches to cancer therapy Retrospect and prospect Part 2 Structural analogues and derivatives Curr Drug Targets 4, 565–585

3 Casero RA Jr & Pegg AE (1993) Spermidine⁄ spermine

N1-acetyltransferase – the turning point in polyamine metabolism FASEB J 7, 653–661

4 McIntire WS & Hartman C (1993) Copper containing amine oxidases In Principle and Application of Quino-proteins(Davison, VL, ed.), pp 97–171 Marcel Dekker Inc., New York

5 Seiler N (1995) Polyamine oxidase, properties and func-tions Prog Brain Res 106, 333–344

6 Van den Munckhof RJ, Denyn M, Tigchelaar-Gutter W, Schipper RG, Verhofstad AA, Van Noorden CJ & Fre-deriks WM (1995) In situ substrate specificity and ultra-structural localization of polyamine oxidase activity in unfixed rat tissues J Histochem Cytochem 43, 1155–1162

7 Wang Y, Devereux W, Woster PM, Stewart TM, Hacker A & Casero RA Jr (2001) Cloning and

charac-terization of a human polyamine oxidase that is

induci-ble by polyamine analogue exposure Cancer Res 61,

5370–5373

8 Vujcic S, Diegelman P, Bacchi CJ, Kramer DL & Porter

CW (2002) Identification and characterization of a novel

flavin-containing spermine oxidase of mammalian cell

origin Biochem J 367, 665–675

9 Cervelli M, Polticelli F, Federico R & Mariottini P

(2003) Heterologous expression and characterization of

mouse spermine oxidase J Biol Chem 278, 5271–5276

10 Cervelli M, Bellini A, Bianchi M, Marcocci L, Nocera

S, Polticelli F, Federico R, Amendola R & Mariottini P

(2004) Mouse spermine oxidase gene splice variants:

nuclear sub-cellular localization of a novel active

iso-form Eur J Biochem 271, 760–770

11 Bellelli A, Stefano Cavallo S, Nicolini L, Cervelli M,

Bianchi M, Mariottini P, Zelli M & Federico R (2004)

A model of the catalytic cycle and its inhibition by

N,N1-bis (2,3-butadienyl)-1,4-butanediamine Biochem

Biophys Res Commun 322, 1–8

12 Binda C, Coda A, Angelini R, Federico R, Ascenzi P &

Mattevi A (1999) A 30 A˚ long U-shaped catalytic tunnel

in the crystal structure of polymine oxidase Structure 7,

265–276

13 Binda C, Newton-Vinson P, Hubalek F, Edmondson

DE & Mattevi A (2002) Structure of human monoamine

oxidase B, a drug target for the treatment of

neurologi-cal disorders Nat Struct Biol 9, 22–26

14 Sambrook J, Fritsch EF & Maniatis T (1989) Molecular

Cloning: a Laboratory Manual, 2nd edn Cold Spring

Harbor Laboratory, Cold Spring Harbor, NY

15 Horton RM (1993) In vitro recombination and mutagen-esis of DNA In Methods in Molecular Biology, Vol 15: PCR Protocols: Current Method and Applications (White BA, ed.), pp 251–261 Humana Press Inc., Totowa, N J

16 Thompson JD, Higgins DG & Gibson TJ (1994) CLUS-TAL W: improving the sensitivity of progressive multi-ple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice Nucleic Acids Res 22, 4673–4680

17 Rost B (1996) PHD: predicting one-dimensional protein structure by profile-based neural networks Methods Enzymol 266, 525–539

18 Petrey D, Xiang Z, Tang CL, Xie L, Gimpelev M, Mitros

T, Soto CS, Goldsmith-Fischman S, Kernytsky A, Schles-singer A et al (2003) Using multiple structure alignments, fast model building, and energetic analysis in fold recog-nition and homology modeling Proteins 53, 430–435

19 Markwell MA, Haas SM, Bieber LL & Tolbert NE (1978) A modification of the Lowry procedure to simplify protein determination in membrane and lipoprotein samples Anal Biochem 87, 206–210

20 Laemmli UK (1970) Cleavage of structural proteins during assembly of the head of bacteriophage T4 Nature 277, 680–685

21 Nicholls A, Sharp K & Honig B (1991) Protein folding and association: insights from the interfacial and ther-modynamic properties of hydrocarbons Proteins 11, 281–296