Biochimica et Biophysica Acta 1798 (2010) 1812–1820

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta

j ourna l homepage: www.e lsev ie r.com/ locate /bbamem

Interaction of two intrinsically disordered plant stress proteins (COR15A andCOR15B) with lipid membranes in the dry state

Anja Thalhammer a, Michaela Hundertmark a,1, Antoaneta V. Popova a,2, Robert Seckler b, Dirk K. Hincha a,⁎a Max-Planck-Institut für Molekulare Pflanzenphysiologie, Am Mühlenberg 1, D-14476 Potsdam, Germanyb Institut für Physikalische Biochemie, Universität Potsdam, Karl-Liebknecht-Str. 24-25, D-14476 Potsdam, Germany

Abbreviations: CD, circular dichroism; COR, cold regulaglycero-3-phosphatidylcholine; DLnPE, 1, 2-dilinolenoyl-snamine;EPE, eggphosphatidylethanolamine;FTIR, Fourier-traphase; IDP, intrinsically disordered protein; LDH, lactembryogenesis abundant; MGDG, monogalactosyldiacyloleoyl-sn-glycero-3-phosphatidylcholine; THex, bilayer totemperature; Tm, gel to liquid-crystalline phase transition te⁎ Corresponding author. Tel.: +49 331 567 8253; fax

E-mail address: hincha@mpimp-golm.mpg.de (D.K.1 Present address: UMR 1191 Physiologie Molécu

Lavoisier, 49045 Angers Cedex 1, France.2 Permanent address: Institute of Biophysics, Bulgari

Sofia, Bulgaria.

0005-2736/$ – see front matter © 2010 Elsevier B.V. Adoi:10.1016/j.bbamem.2010.05.015

a b s t r a c t

a r t i c l e i n f o

Article history:Received 19 March 2010Received in revised form 27 April 2010Accepted 14 May 2010Available online 25 May 2010

Keywords:DesiccationFTIR spectroscopyIntrinsically disordered proteinsLEA proteinsLipid phase transitionsProtein secondary structure

COR15A and COR15B form a tandem repeat of highly homologous genes in Arabidopsis thaliana. Both genesare highly cold induced and the encoded proteins belong to the Pfam LEA_4 group (group 3) of the lateembryogenesis abundant (LEA) proteins. Both proteins were predicted to be intrinsically disordered insolution. Only COR15A has previously been characterized and it was shown to be localized in the solublestroma fraction of chloroplasts. Ectopic expression of COR15A in Arabidopsis resulted in increased freezingtolerance of both chloroplasts after freezing and thawing of intact leaves and of isolated protoplasts frozenand thawed in vitro. In the present study we have generated recombinant mature COR15A and COR15B for acomparative study of their structure and possible function as membrane protectants. CD spectroscopyshowed that both proteins are predominantly unstructured in solution and mainly α-helical after drying.Both proteins showed similar effects on the thermotropic phase behavior of dry liposomes. A decrease in thegel to liquid-crystalline phase transition temperature depended on both the unsaturation of the fatty acylchains and lipid headgroup structure. FTIR spectroscopy indicated no strong interactions between theproteins and the lipid phosphate and carbonyl groups, but significant interactions with the galactoseheadgroup of the chloroplast lipid monogalactosyldiacylglycerol. These findings were rationalized bymodeling the secondary structure of COR15A and COR15B. Helical wheel projection indicated the presence ofamphipathic α-helices in both proteins. The helices lacked a clear separation of positive and negative chargeson the hydrophilic face, but contained several hydroxylated amino acids.

ted; DLnPC, 1, 2-dilinolenoyl-sn--glycero-3-phosphatidylethanol-nsforminfrared;HII, hexagonal IIate dehydrogenase; LEA, lateglycerol; POPC, 1-palmitoyl-2-hexagonal II phase transitionmperature: +49 331 567 8250.Hincha).laire des Semences, 16 Bvd

an Academy of Sciences, 1113

ll rights reserved.

© 2010 Elsevier B.V. All rights reserved.

1. Introduction

Although it is generally assumed that protein function is based ona stable three-dimensional structure, it has more recently beenrecognized that a substantial part of all cellular proteomes consists ofproteins that either completely lack stable structure or that have largeunstructured domains [1]. These proteins are now mainly referred toin the literature as intrinsically disordered proteins (IDPs). Many ofthese proteins perform essential cellular functions, e.g. in signaltransduction and the regulation of transcription [2,3]. The ability of

IDPs to bind to their target molecules, such as RNA, DNA and other(structured) proteins [4] is of crucial importance for their functionand inmany cases it has been observed that binding induces increasedsecondary structure in IDPs [5,6], indicating their ability to fold underthe appropriate conditions.

Late embryogenesis abundant (LEA) proteins were first describedalmost 30 years ago as a group of proteins that are accumulated incotton seeds during the late stages of development, when the embryobecomes desiccation tolerant [7]. Subsequently, related proteins werefound not only in plant seeds, but also in other plant tissues, in somebacterial species and in animals such as nematodes, rotifers and brineshrimp (see [8] for a recent review). In all these cases the occurrenceof the proteins was related to environmental stress conditions such asfreezing, drought, or desiccation. Some LEA proteins have indepen-dently been described in the model plant species Arabidopsis thalianaas cold regulated or COR proteins because they are highly inducedupon cold treatment [9]. While a large body of knowledge is nowavailable about the cold regulation of the genes encoding CORproteins [10–12], the analysis of their structure and function islagging far behind.

One of the best characterized Arabidopsis COR proteins is COR15A.Its gene was first cloned and described as cold and drought induced

1813A. Thalhammer et al. / Biochimica et Biophysica Acta 1798 (2010) 1812–1820

20 years ago [13] and the cold induced accumulation of the protein inthe soluble chloroplast stroma fraction shortly thereafter [14]. TheCOR15A gene encodes a polypeptide of approximately 15 kDa that isprocessed to the mature form of 9.4 kDa upon import into thechloroplast [14,15]. In the Arabidopsis genome, the COR15A gene ispart of a tandem repeat pair with COR15B [16]. The coding regions ofthe two genes are 82% identical and both encode proteins with aputative N-terminal signal sequence for chloroplast import. Plastidlocalization, however, has not been experimentally verified forCOR15B. The only available information on COR15B is that theencoding gene is induced by cold and the phytohormone abscisic acid,similar to COR15A [16,17]. In addition, both proteins show homologyto the Pfam LEA_4 family of LEA proteins, previously described asgroup 3 [18], and both were predicted to be IDPs [17]. In contrast toCOR15A, no further functional or structural information is currentlyavailable for COR15B.

Constitutive overexpression of the COR15A gene in Arabidopsisleads to an accumulation of the mature COR15A protein in thechloroplast stroma [19], resulting in increased freezing tolerance ofboth chloroplasts after freezing and thawing of intact leaves [19] andof isolated protoplasts frozen and thawed in vitro [19,20]. Freeze-fracture electronmicroscopy of frozen protoplasts indicated a possibleinvolvement of COR15A in membrane protection by stabilizingbilayers against freeze induced formation of the nonbilayer HII

phase. This possible mechanism of COR15A function was furthercorroborated by 31P NMR measurements with lipid dispersionscontaining the prominent nonbilayer chloroplast lipid MGDG [20] inthe absence or presence of recombinant mature COR15A [21].

Further studies with recombinant COR15A indicated a protectivefunction for the freezing sensitive enzyme LDH of both the preprotein[22] and the mature protein [15]. COR15A has no chaperone activity,i.e. it is not able to refold or assist in the refolding of denaturedproteins. Instead, enzyme stabilization seems to result from theprevention of LDH aggregation during freezing [23], similar to thefunction shown for other LEA_4 proteins [24–26]. Whether COR15Afunctions as a stabilizer of chloroplast enzymes during freezing invivo, however, has not been reported.

In the present study we have generated recombinant matureCOR15A and COR15B for a comparative investigation of their structureand possible function as membrane protectants. CD spectroscopyshowed that both proteins are predominantly unstructured insolution and mainly α-helical after drying. Both proteins showedsimilar, but not identical, effects on the thermotropic phase behaviorof dry liposomes of different lipid compositions. FTIR spectroscopyindicated no strong interactions between the proteins and the lipidphosphate and carbonyl groups, but significant interactions with thegalactose headgroup of MGDG. These findings were rationalized bymodeling the secondary structure of COR15A and COR15B.

2. Materials and methods

2.1. Materials

POPC (16:0/18:1), DLnPC (18:3/18:3) and DLnPE (18:3/18:3)were obtained from Avanti Polar Lipids (Alabaster, AL). EPE and thechloroplast glycolipid MGDG were purchased from Lipid Products(Redhill, Surrey, UK).

2.2. Cloning

Full-length cDNA clones were obtained from the RIKEN (Tokyo,Japan) RAFL collection [27,28] for the Arabidopsis thaliana genesCOR15A (At2g42540; clone RAFL09-47-C04) and COR15B (At2g42530;clone RAFL05-20-N18). The cDNA sequences encoding the matureproteins lacking the N-terminal signal peptides were amplified by PCRand inserted into the Gateway pENTR.SD.D-TOPO vector (Invitrogen,

Karlsruhe, Germany). The identity of the inserts was checked bysequencing. The genes were transferred into the expression vectorpDEST17 (Invitrogen) to express the proteins with an N-terminal6xHis-tag under the control of the T7 expression system.

2.3. Expression and purification of recombinant proteins

The pDEST17.COR15A and pDEST17.COR15B constructs weretransformed into the Escherichia coli strain Rosetta (DE3) pLys S(Novagen, Madison, WI). Bacterial cell lysates containing therecombinant proteins were incubated in a boiling water bath for10 min. The COR15 proteins, like several other LEA proteins, are stableupon boiling and remain in solution [14]. Precipitated proteinswere removed by centrifugation at 4000g for 15 min at 4 °C. Thesupernatant was filtered through a 0.2 µm filter and applied to a 1-mlHisTrap HP column (GE Healthcare) equilibrated with 20 mM sodiumphosphate, 0.5 M NaCl, 20 mM imidazole (pH 7.4) with a flow rate of1 ml/min. The column was washed with increasing concentrations ofimidazole and COR15 proteins were eluted with 250 mM imidazole.Fractions were analyzed by SDS-PAGE and those containing recom-binant protein were pooled and dialyzed against ddH2O in QuixSepMicro Dialyzer capsules (Roth, Karlsruhe, Germany) with Spectra/Pordialysis membranes (3500 molecular weight cut-off). Purifiedproteins were lyophilized and stored at −20 °C.

2.4. SDS-PAGE and Western-blotting

SDS-PAGE was performed according to the method of Schäggerand von Jagow [29]. After electrophoresis proteins were stained withcolloidal Coomassie blue. For Western blotting proteins weretransferred from SDS-PAGE gels to nitrocellulose membranes (pora-blot NCP, Macherey-Nagel, Düren, Germany) in a semi-dry blottingsystem (Biometra, Fastblot B31). 5% non-fat dry milk was used toblock non-specific binding sites. Membranes were probed with eithera peroxidase-coupled anti-6xHis antibody (abcam, Cambridge, UK) oran antibody raised against recombinant COR15A (kindly provided byDr. Michael F. Thomashow, Michigan State University, USA). Mem-branes treated with the anti-COR15A antibody were incubated with aperoxidase coupled goat-anti-rabbit antibody (BioRad, Munich,Germany). Detection of peroxidase activity was performed with theOpti4-CN kit from BioRad according to manufacturer's instructions.

2.5. Circular dichroism spectroscopy

CD spectra were obtained with a Jasco-715 spectropolarimeter(Jasco Instruments). Protein solutions containing approximately0.75 mg ml−1 protein in H2O were measured in a 0.1 mm pathlengthcuvette. Four spectra were accumulated with a response of 4 s, 1 nmdata pitch, 1 nm band width from 240 to 180 nm. For themeasurement of dry samples, 50 µl of 2 mg ml−1 protein dissolvedin H2O were spread on CaF2 windows and dried in a desiccator oversilica gel over night at 28 °C. Windows were mounted in thespectropolarimeter which was continuously purged with N2. Asquantification of LEA proteins by standard colorimetric methods isunreliable because of the highly biased amino acid composition, weused the absorption at 193 nmmeasured in parallel with the CD signalto estimate the protein concentration [30]. The mean-residue circulardichroism was calculated as:

ΔεMR =ΔA ·mMW

d · c

ΔεMR=mean-residue circular dichroism [M−1 cm−1], ΔA=circular dichroism, mMW=mean molecular mass of an aminoacid in the investigated protein [g mol−1], d=pathlength [cm],c=concentration [g l−1].

1814 A. Thalhammer et al. / Biochimica et Biophysica Acta 1798 (2010) 1812–1820

Spectra were analyzed with the CDPro software [31] using threedifferent algorithms: CONTINLL, CDSSTR and SELCON3. Sets ofreference spectra containing denatured proteins were chosen forthe analysis. Since the results were similar in each case for the threeparallel samples and all algorithms, averages are shown.

2.6. Liposome preparation

Lipids dissolved in choroform were mixed at the appropriate massratios and the solvent was evaporated under a stream of N2. To ensurecomplete removal of the solvent, the lipids were stored under vacuumover night. Dry lipid films were hydrated in 200 µl of distilled water.Unilamellar liposomes were formed from the hydrated lipids using ahand-held extruder (Avestin, Ottawa, Canada) with two layers of100 nm pore filters [32].

2.7. Fourier-transform infrared spectroscopy

The liposome suspensions were mixed with the appropriateproteins in H2O to a final mass ratio of 25:1 (lipid:protein) and50 µl were spread on CaF2 windows and dried over silica gel indesiccators at 28 °C for 24 h in the dark [33]. Samples prepared in thisway contain below 0.02 g H2O g−1 dry weight [34]. A window withthe dry sample was placed in a cuvette holder, fixed in a vacuumchamber connected to a temperature control unit (Specac Eurotherm,Worthington, UK) and placed in the infrared beam. Temperature wasmonitored by a fine thermocouple fixed on the window next to thesample. Samples were kept in the sample holder under vacuum at30 °C for 30 min to remove residual moisture absorbed duringhandling. This was verified by the absence of a water band in theFTIR spectra at 1650 cm−1. Temperature was then decreased to –30 °Cand after 10 min equilibration, the temperature was increased at arate of 1 °C min−1. Two spectra with 4 cm−1 resolutionwere recordedand coadded every minute with a Perkin Elmer GX 2000 FTIRspectrometer [33,35]. Spectra were analyzed using Spectrum 5.0.1software. After normalization of absorbance and baseline correction,the wavenumber of the CH2 symmetric stretching vibration (νCH2s)around 2850 cm−1, the carbonyl stretching vibration (νC=O)between 1770 cm−1 and 1700 cm−1 and the P=O asymmetricstretching (νP=Oas) vibration (1300–1200 cm−1) were determinedby the automatic peak identification routine. In addition, in samplescontaining the galactolipid MGDG, the band attributed to the hydroxylstretching vibration (νOH) of the sugar headgroups (3100–3500 cm−1)was investigated [34]. The gel to liquid-crystalline phase transitiontemperature (Tm) was determined as the midpoint of the shift in νCH2swith temperature [36]. The νC=O and the νP=Oas vibrations wereadditionally analyzed by peak deconvolution and curve fitting using thepeak-fitting module of OriginPro 7.0 as described in detail recently[26,34,37]. Correlation coefficients for all fitted curves were higher than0.999.

Fig. 1. SDS-PAGE (A) and Western blot analysis of purified and dialyzed COR15A andCOR15B with anti-6xHis (B) and anti-COR15A (C) antibodies. The BSA standardconcentrations were (from left to right) 0.05 µg/µl, 0.075 µg/µl, 0.1 µg/µl, 0.15 µg/µl,0.2 µg/µl. Molecular masses of standard proteins are indicated on the right side ofpanel A.

2.8. Differential scanning calorimetry

Liposomes were prepared as above and the thermal behavior ofdry liposomes was investigated by DSC. Measurements wereperformed with a Netzsch (Selb, Germany) DSC 204. Samples werecooled from room temperature to −70 °C, equilibrated for 5 min,heated to 120 °C and equilibrated again for 5 minutes. This coolingand heating procedure was repeated three times at a cooling andheating rate of 20 °C/min. Phase transition temperatures (Tm or THex)were determined by the position of the peak maxima of the secondheating thermogram. THex and Tm values are themeans of two or threemeasurements.

2.9. Computational analysis of COR15A and COR15B

An alignment of the COR15A and COR15B amino acid sequenceswithout N-terminal signal peptides was performed using the Kaligntool [38,77] with the default settings and edited using Jalview [39].Secondary structure prediction of mature COR15A and COR15Bproteins was performed using the prediction tools NNpredict[40,78], SOPMA [41,79], SSPro8 [42,80], JPred [43,81] and Sable2[44,82] and the prediction programs included in the PELE tool on theSDSC BiologyWorkbench [83] (BPS [45], D_R [46], DSC [47], GGR [48],GOR [49], H_K [50] and K_S [51]). Creation of a consensus of allpredictions followed the “simple majority” principle. The resulting α-helical domains were visualized using helical wheel diagrams [52,84].

3. Results

We obtained full-length cDNA clones of the cold inducedArabidopsis genes COR15A and COR15B from the RIKEN RAFLcollection [27,28] and expressed the proteins without the transitpeptides in E. coli. The recombinant proteins carried a 6xHis tag andwere purified by metal chelate chromatography, yielding essentiallypure proteins as indicated by SDS-PAGE and Coomassie blue staining(Fig. 1A). While the molecular mass of the expressed proteins is about10 kDa, their apparent molecular mass in the SDS-PAGE gel wasapproximately 14 kDa. This aberrantmigration during electrophoresishas previously been shown to be common for IDPs [53]. The correctidentity of the purified proteins was established by Western blottingusing both an anti-6xHis (Fig. 1B) and an anti-COR15A (Fig. 1C)antiserum. In addition, both recombinant proteins were sequenced bymass spectrometry after tryptic digestion of the bands in SDS-PAGEgels, which confirmed their identity with a Probability Based MowseScore of 2797 for COR15A and 5174 for COR15B. The next best hit forthe COR15A sample was protease I precursor from Achromobactuslyticus (1041) and for the COR15B sample human keratin10 (335),both obviously minor nonplant contaminants.

The secondary structure of the purified recombinant proteins wasdetermined by CD spectroscopy (Fig. 2). In water, both proteinsexhibited far-UV CD spectra typical for unstructured proteins with aminimum around 200 nm. However, when the proteins were driedthey gained α-helical structure, as indicated by the double minimumat 208 nm and 222 nm. Secondary structure estimates using differentCD spectra analysis programs indicated that COR15A was approxi-mately 70% unstructured in solution, while the highly homologousCOR15B was about 60% unstructured. Upon drying, the two proteinsbecame about 65% and 57% α-helical, respectively.

Fig. 2. Secondary structure of COR15A and COR15B as analyzed by CD spectroscopy. Thetwo upper panels show CD spectra of the proteins either in the hydrated state or in thedry state. The bottom panel shows the relative content of different secondary structureelements in both proteins in either the hydrated or dry state as calculated from the CDspectra. Data represent the means from measurements on three different samples foreach protein and condition.

Fig. 3. Lipid melting curves of dry liposomes prepared from POPC (A) or 80% POPC/20%EPE (B). The temperature dependent increase in the position of the symmetric CH2

stretching band (νCH2s) of the fatty acyl chains was determined by FTIR spectroscopy.Samples contained either only liposomes (pure lipid) or in addition the proteinsCOR15A, COR15B, or RNaseA at a lipid:protein mass ratio of 25:1.

Fig. 4. Lipid phase transition temperatures (Tm) of dry liposomes prepared from POPCor 80% POPC and 20% of the indicated lipids in the absence or presence of COR15A orCOR15B. Tm was determined as the midpoint of melting curves as shown in Figs. 3 and5. The error bars indicate SE from measurements on 3 samples. The statisticaldifferences in Tm between samples containing one of the proteins and samplescontaining only the corresponding liposomes were evaluated using an unpaired t-test(*pb0.05; **pb0.01; ***pb0.001).

1815A. Thalhammer et al. / Biochimica et Biophysica Acta 1798 (2010) 1812–1820

We investigated possible interactions of the two proteins withmembranes in the dry state using FTIR spectroscopy. A lipid:proteinmass ratio of 25:1 was used in all experiments, although previousstudies had used significantly lower ratios from 2:1 [54] to 10:1 [20],to avoid the danger of nonspecific effects of the proteins onmembrane physical behavior. Fig. 3A shows that both COR15A andCOR15B shifted the melting curves of POPC membranes to lowertemperatures, while RNaseA induced a small shift to highertemperatures. RNaseA was used here as a nonspecific control, toexclude the possibility that any protein of similar size would have thesame effect as the COR15 proteins. The magnitude of the effect of theCOR15 proteins on lipid phase behavior was critically dependent onmembrane lipid composition. The presence of 20% of the nonbilayerlipid EPE strongly inhibited the effect of the proteins on membranephase transitions observed in pure POPC membranes (Fig. 3). Whileboth proteins induced a significant reduction in Tm in pure POPCmembranes, no significant differences were observed in POPC/EPEmixed membranes (Fig. 4).

COR15A [14,15] and presumably also COR15B are localized in thechloroplast stroma in vivo, where their potential target membranesare characterized by the presence of the nonbilayer glycolipid MGDG[55]. Membranes containing 80% POPC and 20%MGDG showed amorecomplex response to the presence of the COR15 proteins in the drystate (Fig. 5A) than membranes containing only phospholipids. Whileonly COR15B induced a significant reduction in Tm in POPC/MGDGmixed membranes (Fig. 4), both proteins induced an increase in the

wavenumber of νCH2s in the gel state (Fig. 5A), indicating a highermobility of the fatty acyl chains of the lipids in the presence of theproteins. A corresponding effect in the liquid-crystalline state was notobserved (Fig. 5A).

Since MGDG from plant leaves is not only a nonbilayer lipid, butalso highly unsaturated, containing mostly 18:3 fatty acids [55], we

Fig. 5. Lipid melting curves of dry liposomes prepared from 80% POPC/20% MGDG (A),80% POPC/20% DLnPE (B), or 80% POPC/20% DLnPC (C). See Fig. 3 for further details.

Fig. 6. DSC heating thermograms of dry samples. Thermograms are from the secondheating scan from −70 °C to 120 °C at a scan rate of 20 °C/min. Liposomes contained100% DLnPE (curve 1; THex=89.5 °C), 20% DLnPE/80% POPC (curve 2; Tm=29.9 °C),20% MGDG/80% POPC (curve 3; Tm=19.8 °C) or 20% EPE/80% POPC (curve 4;Tm=28.2 °C).

Fig. 7. Infrared spectra in the OH stretching vibration (νOH) region of dry liposomesamples containing the galactolipid MGDG at 30 ºC. The samples contained either onlyliposomes or liposomes and the indicated proteins at a lipid:protein mass ratio of 25:1.

1816 A. Thalhammer et al. / Biochimica et Biophysica Acta 1798 (2010) 1812–1820

further explored the roles of headgroup structure and fatty acidunsaturation in liposomes containg 80% POPC and 20% of the synthetic18:3 lipids DLnPE (nonbilayer, Fig. 5B) and DLnPC (bilayer, Fig. 5C).While in membranes containing DLnPE both proteins induced a small,but significant decrease in Tm, with DLnPC only COR15A had asignificant effect, although COR15B also slightly reduced Tm (Fig. 4).

Since some of themembranes contained nonbilayer lipids, theymayhave shown additional bilayer to HII transitions that would not bedetectable by FTIR spectroscopy [56,57]. Therefore, we used DSC to gaininformation about such transitions in the absence of proteins (Fig. 6).DSC experiments can distinguish between transitions fromgel to liquid-crystalline state and bilayer to nonbilayer transitions, because theformer are highly cooperative and strongly endothermic, while thelatter have a low transition enthalpy [58]. For pure dry DLnPE a lowenthalpy transition was registered at 91 °C, corresponding to atransition from liquid-crystalline to HII phase. For all mixed bilayer/nonbilayer membranes, however, no additional low enthalpy transi-tions indicative of HII transitionswere observed at higher temperatures.Tm values determined in these DSC experiments were lower comparedto the FTIR results. This ismost likely due to a small amountofwater thatthe samples absorbed during handling and that was subsequentlyremoved in the FTIR experiments in the vacuum cuvette.

Diacyl lipids contain different functional groups that can interactwithmolecules suchas sugars or proteins throughH-bondingor charge-pair interactions. Such interactions canbeobserved aspeak shifts in FTIRspectra. However, we were not able to find any evidence for suchinteractions of either COR15A or COR15B with any of the lipids in the

membranes described above (Figs. 3–5) at the levels of the carbonylester or phosphate groups. Also peak deconvolution [26,34,37] did notyield any clear evidence for interactions of the COR15 proteins withthese lipid moieties in the dry state (data not shown).

MGDG contains a galactose headgroup that is able to H-bond toother molecules through the sugar OH groups. The νOH presents abroad peak in the FTIR spectrum from about 3600 cm−1 to 3100 cm−1

(Fig. 7), indicating a wide range of H-bonding lengths and strengths.

1817A. Thalhammer et al. / Biochimica et Biophysica Acta 1798 (2010) 1812–1820

In the presence of both COR15A and COR15B the width of the νOHpeak was strongly reduced. This indicates a smaller population ofdifferent H-bonding lengths and strengths, in agreement withreduced motional freedom of the galactose headgroup due tointeractions with the proteins.

The mature proteins COR15A and COR15B used in the presentstudy are 70% identical in their amino acid sequences. However, analignment of the two sequences shows that the variable amino acidsare not uniformly distributed along the proteins (Fig. 8D). In the N-terminal half of the proteins, there is a stretch of 32 identical aminoacids, followed by the most variable part in the middle of the proteinsand again a more conserved C-terminal part. Fig. 8D also shows thatmost of the divergent amino acids have a high degree of conservation,i.e. very similar physico-chemical properties. Since CD spectroscopyindicated a high content of α-helical structure for both proteins in thedry state (Fig. 2), we used several different secondary structureprediction programs (see Section 2.9) to identify those parts of theproteins that may form α-helices. An amino acid was scored as part ofan α-helix if the majority of the programs identified it as part of thissecondary structure element. By this procedure an N-terminalα-helix(helix I) from amino acids 8 to 31 and a C-terminal α-helix (helix II)from amino acids 47 to 87were identified. This predicts a totalα-helixcontent of 66% for COR15A and 65% for COR15B, in excellentagreement with the α-helix content of 65% and 57% estimated byCD spectroscopy for the dry proteins (Fig. 2). Interestingly, allprediction programs predicted the secondary structure of the dryproteins rather than that of the fully hydrated proteins, which arelargely unstructured.

Fig. 8. Alignment of mature COR15A and COR15B amino acid sequences and visualization osequence domain from amino acid 8 to 31, which is identical in both COR15 proteins. Panelsin COR15B, respectively. Red circles indicate hydophobic, blue circles hydrophilic amino acisequence domain. Panel D shows the alignment of the amino acid sequences of COR15A andphysico-chemical properties in the alignment, with identity scored highest (10). Consensus

Since in particular amphipathic α-helices could play an importantrole in the interactions of the proteins with membranes [59] that weobserved by FTIR spectroscopy, we used helical wheel projections [60]to visualize the helices and to calculate their mean hydrophobicmoment (µH) and mean hydrophobicity (H). Helix I has an identicalamino acid sequence in COR15A and COR15B, resulting in the helicalwheel projection depicted in Fig. 8A, while helix II contained only 66%identical amino acids between the two proteins. Therefore, separatehelical wheels were constructed for helix II in COR15A (Fig. 8B) andCOR15B (Fig. 8C). All three helices contain between 38% (helix I) and49% (helix II in COR15A) hydrophobic amino acids and consequentlytheir mean hydrophobicity (H) is not far from zero. At the same timethey show a hydrophobic and a hydrophilic side in agreement withthe rather high mean hydrophobic moments (µH) indicative ofamphipathic α-helices that would localize to the membrane surface.

Positive (Lys/K) and negative (Asp/D and Glu/E) charges are notclearly separated along the helical wheels where negative chargespredominate (6 to 4 in helix I; 10 to 7 in helix II of both proteins). Inaddition, both helices contain hydroxylated amino acids (Asp/D; Glu/E; Ser/S; Thr/T; Tyr/Y), 10 in helix I and 12 in helix II of both proteinsthat could form H-bonds with the OH groups in the galactoseheadgroup of MGDG.

4. Discussion

In the fully sequenced model plant species Arabidopsis thaliana, 51genes encoding LEA proteins have been identified [17,61] and the vastmajority was predicted to be IDPs [17], including the mature forms of

f predicted α-helical domains by helical wheel diagrams. Panel A shows the α-helicalB and C depict the α-helical sequence domains from amino acid 47 to 87 in COR15A andds. µH indicates the mean hydrophobic moment and H the mean hydrophobicity of theCOR15B. Conservation is measured as a numerical index reflecting the conservation ofshows those amino acids that are identical between the two proteins.

1818 A. Thalhammer et al. / Biochimica et Biophysica Acta 1798 (2010) 1812–1820

COR15AandCOR15B. Experimental evidence fora lackof stable secondarystructure has only been published for a small number of these proteins[53,62] and for LEA proteins from other plant and animal species(reviewed in [8]). In addition, some LEA proteins have been shown toacquire mainly α-helical structure during drying [26,63–67]. CD spec-troscopy revealed that COR15A and COR15B behaved in the same way.They were about 70% unstructured in solution and became largely α-helical upon drying. The solution structure of COR15A reported here is inapparent disagreement with a recent study [15] that concluded from CDspectroscopy that the protein is only approximately 35% unstructuredunder fully hydrated conditions. The CD spectra reported in the study byNakayamaet al. [15], however, are also indicativeof a largelyunstructuredprotein, with a prominent minimum around 200 nm. It has been shownpreviously that the correct prediction of secondary structure content ofmainly unstructured proteins requires the inclusion of spectra fromdenatured proteins in the reference library to avoid erroneously lowpredictions of unstructuredness [31]. Also, the low amount of structure inour COR15A preparation was not due to the boiling step used in thepurification protocol, as Nakayama et al. ([15]) found no evidence forsignificant changes in secondary structure of the protein upon boiling.

CD spectroscopy showed that COR15A and COR15B are predom-inantlyα-helical in the dry state. The propensity of the proteins forα-helix formation was also indicated by secondary structure modeling,which interestingly predicted the structure of the dry, rather than thehydrated state. The amphipathic class A α-helix as examplified by themammalian apolipoprotein A-I has been shown to stabilize mem-branes by inserting into phospholipid bilayers near the surface,parallel to the membrane plane, with the nonpolar protein surfaceembedded in the membrane. The helix is further stabilized on themembrane by electrostatic interactions between positively chargedresidues of the protein and negatively charged moieties in thephospholipid headgroup [59,68,69]. The charges in class A-I helicesare distributed in such a way that the helix is flanked by positivecharges on two opposing sides at the polar-nonpolar interface for theinteraction with the lipid phosphate groups [69] and negative chargesat the center of the polar face, while the hydrophobic face inserts intothe membrane [59]. However, our helical wheel projections indicatethat unlike LEAM from pea [66], the helices of the two COR proteinsare not of the A-I type described for mammalian apolipoproteins [59],as they lack the separation between positive and negative charges onthe hydrophilic face. This is presumably also the reason why we werenot able to detect any interactions of the COR proteins with the lipidP=O groups that have been reported for A-I type proteins [69].

Structurally, the helical parts of COR15A and COR15B are mostsimilar to the class C apolipoproteins because of their intermediateµH, wide polar face and high charge density of the polar face without aclear separation between positive and negative charges [70]. Class Capolipoproteins are coiled-coil proteins and this structural motif hasalso been suggested for different LEA_4 proteins from plants andanimals from in-vacuo predictions [64,71] and molecular dynamicssimulations [72].

The structural differences between the two COR15 proteins areminor based on the CD data, the sequence comparison and thestructure predictions. Consequently, also the functional differencesjudged from their effects on lipid Tm values were only small and didnot present a coherent picture that would allow the prediction of anysignificant functional differences in vivo. Interestingly, the only stronginteractions with a lipid headgroup that were detectable by FTIRspectroscopy were H-bonding interactions with the OH groups in thegalactose moiety of the MGDG headgroups. It should be mentionedhere that the protein NH stretching vibration is located in the samespectral region as the sugar OH vibration. The presence of thisadditional vibrational peak in spectra from samples containingprotein could also contribute to the observed differences in peakshape. However, a comparison of the height of the NH peak in samplescontaining either of the COR15 proteins and pure POPC membranes

(that produce no OH peak) with the peak height of the combined NH/OH peaks as shown in Fig. 7, after normalizing all spectra to the heightof their respective C=O peaks to account for differences in samplethickness, revealed that the NH peak only contributed approximately10% of the height of the NH/OH peak. This small contribution of theprotein-derived NH peak to the NH/OH peaks in spectra from samplescontaining COR15 and MGDG/POPC membranes precludes that thespectral changes we observed in Fig. 7 are simply due to the presenceof the additional peak.

The interaction between MGDG and the COR15 proteins seemsbiologically significant, given the localization of the proteins in thechloroplast stroma, where the accessible membranes contain MGDG.In addition, only the FTIR spectra from membranes containing MGDGshowed evidence for increased fatty acyl chain mobility in the gelphase in the presence of the COR15 proteins, indicating strongerinteractions between membranes and proteins in the presence oftheir natural target lipid than in the presence of pure phospholipidbilayers. Even for membranes containing DLnPE, a phospholipid withsimilar structural properties (highly unsaturated nonbilayer lipid) asMGDG, no effect of the proteins on νCH2s was observed, emphasizingthe importance of the sugar headgroup for this interaction.

The present paper presents clear evidence for functional interac-tions of both COR15A and COR15B with membranes in the dry state.This is in accordance with an earlier NMR study on dry lipiddispersions containing either PE or MGDG [20], but in apparentcontradiction to the study by Webb et al. [54], who did not detect anyeffects of recombinant mature COR15A on the Tm of dry PC or PC/PEmembranes by DSC. The latter study, however, used multilamellarvesicles in contrast to our unilamellar liposomes which may havelimited the accessibility of the membranes for the protein.

Although it seems clear that COR15 proteins are able to interactwith membranes in the dry state, the reduction in Tm produced by theCOR15 proteins was at most 15–20 °C, far less than what has beenreported for sugars such as sucrose or trehalose [73,74]. However, theintention of the present study was to elucidate whether COR15A andCOR15B are able to interact with membranes and how suchinteractions are related to membrane lipid composition and proteinstructure, and not to investigate mechanisms of plant desiccationtolerance. The complete elucidation of the in vivo function of theCOR15 proteins obviously still requires additional experimentation.Since both proteins are cold induced in Arabidopsis leaves andpresumably involved in the stabilization of chloroplasts duringfreezing at rather mild temperatures (between −4 °C and −7 °C;[19,20]), it remains to be shown whether the proteins fold into α-helices already during such mild dehydration and whether theyinteract with and stabilize membranes under these conditions. It hasbeen proposed both from molecular dynamics simulations with afragment of a LEA_4 protein [72] and from experiments withArabidopsis dehydrins [75] that almost complete desiccation isnecessary to induce folding. However, sequence similarity to theCOR15 proteins is rather low in case of the LEA_4 protein and there isno similarity to the dehydrins. In addition, both simulations andexperiments were performed in the absence of possible targets of theLEA proteins that may increase the propensity of the proteins forfolding. Therefore, further experiments will be necessary to resolvethe questions of protein folding and interaction with membranesunder (partly) hydrated conditions.

In addition, it has been shown that COR15A is able to stabilize thefreezing sensitive enzyme LDH during freezing [15,22,23]. Whetherany of the enzymes localized in the chloroplast stroma are indeedfreezing sensitive and whether they can be stabilized by COR15proteins remains to be shown. Membrane protection and enzymestabilization are not necessarily mutually exclusive functions of theseproteins, as many IDPs have been suggested to bemultifunctional, dueto their highly flexible structure [76]. Our current research is focusedon elucidating these structural and functional complexities.

1819A. Thalhammer et al. / Biochimica et Biophysica Acta 1798 (2010) 1812–1820

Acknowledgments

We are thankful to Dr. Klaus Tauer (Max-Planck-Institute ofColloids and Interfaces, Potsdam, Germany) for making the DSCfacilities available, to Irina Shekova for excellent technical assistanceand to Dr. Wolfgang Engelsberger (Max-Planck-Institute of MolecularPlant Physiology, Potsdam, Germany) for mass spectral analysis. M.H.acknowledges financial support through a PhD fellowship from theUniversity of Potsdam.

References

[1] C.J. Oldfield, Y. Cheng, M.S. Cortese, C.J. Brown, V.N. Uversky, A.K. Dunker,Comparing and combining predictors of mostly disordered proteins, Biochemistry44 (2005) 1989–2000.

[2] H.J. Dyson, P.E. Wright, Intrinsically unstructured proteins and their functions,Nat. Rev. Mol. Cell Biol. 6 (2005) 197–208.

[3] P. Tompa, The interplay between structure and function in intrinsicallyunstructured proteins, FEBS Lett. 579 (2005) 3346–3354.

[4] P. Tompa, P. Csermely, The role of structural disorder in the function of RNA andprotein chaperones, FASEB J. 18 (2004) 1169–1175.

[5] C.J. Oldfield, Y. Cheng, M.S. Cortese, P. Romero, V.N. Uversky, A.K. Dunker, Coupledfolding and binding with α-helix-forming molecular recognition elements,Biochemistry 44 (2005) 12454–12470.

[6] P.E. Wright, H.J. Dyson, Linking folding and binding, Curr. Opin. Struct. Biol. 19(2009) 31–38.

[7] L. Dure III, S.C. Greenway, G.A. Galau, Developmental biochemistry of cottonseedembryogenesis and germination: changing messenger ribonucleic acid populations asshown by in vitro and in vivo protein synthesis, Biochemistry 20 (1981) 4162–4168.

[8] A. Tunnacliffe, M.J. Wise, The continuing conundrum of LEA proteins,Naturwissenschaften 94 (2007) 791–812.

[9] M.F. Thomashow, Plant cold acclimation: freezing tolerance genes and regulatorymechanisms, Annu. Rev. Plant Physiol. Plant Mol. Biol. 50 (1999) 571–599.

[10] V. Chinnusamy, J. Zhu, J.-K. Zhu, Cold stress regulation of gene expression inplants, Trends Plant Sci. 12 (2007) 444–451.

[11] H.A. van Buskirk, M.F. Thomashow, Arabidopsis transcription factors regulatingcold acclimation, Physiol. Plant. 126 (2006) 72–80.

[12] J.T. Vogel, D.G. Zarka, H.A. van Buskirk, S.G. Fowler, M.F. Thomashow, Roles of theCBF2 and ZAT12 transcription factors in configuring the low temperaturetranscriptome of Arabidopsis, Plant J. 41 (2005) 195–211.

[13] R.K. Hajela, D.P. Horvath, S.J. Gilmour, M.F. Thomashow, Molecular cloning andexpression of cor (cold-regulated) genes in Arabidopsis thaliana, Plant Physiol. 93(1990) 1246–1252.

[14] C. Lin, M.F. Thomashow, DNA sequence analysis of a complementary DNA for cold-regulated Arabidopsis gene cor15 and characterization of the COR15 polypeptide,Plant Physiol. 99 (1992) 519–525.

[15] K. Nakayama, K. Okawa, T. Kakizaki, T. Honma, H. Itoh, T. Inaba, ArabidopsisCor15am is a chloroplast stromal protein that has cryoprotective activity andforms oligomers, Plant Physiol. 144 (2007) 513–523.

[16] K.S. Wilhelm, M.F. Thomashow, Arabidopsis thaliana cor15b, an apparenthomologue of cor15a, is strongly responsive to cold and ABA, but not drought,Plant Mol. Biol. 23 (1993) 1073–1077.

[17] M. Hundertmark, D.K. Hincha, LEA (late embryogenesis abundant) proteins andtheir encoding genes in Arabidopsis thaliana, BMC Genomics 9 (2008) 118.

[18] E.A. Bray, Molecular responses to water deficit, Plant Physiol. 103 (1993)1035–1040.

[19] N.N. Artus, M. Uemura, P.L. Steponkus, S.J. Gilmour, C. Lin, M.F. Thomashow,Constitutive expression of the cold-regulated Arabidopsis thaliana COR15a geneaffects both chloroplast and protoplast freezing tolerance, Proc. Natl Acad. Sci.USA 93 (1996) 13404–13409.

[20] P.L. Steponkus, M. Uemura, R.A. Joseph, S.J. Gilmour, M.F. Thomashow, Mode ofaction of the COR15a gene on the freezing tolerance of Arabidopsis thaliana, Proc.Natl Acad. Sci. USA 95 (1998) 14570–14575.

[21] S.J. Gilmour, C. Lin, M.F. Thomashow, Purification and properties of Arabidopsisthaliana COR (Cold-Regulated) gene polypeptides COR15am and COR6.6expressed in Escherichia coli, Plant Physiol. 111 (1996) 293–299.

[22] C. Lin, M.F. Thomashow, A cold-regulated Arabidopsis gene encodes a polypeptidehaving potent cryoprotective activity, Biochem. Biophys. Res. Comm. 183 (1992)1103–1108.

[23] K. Nakayama, K. Okawa, T. Kakizaki, T. Inaba, Evaluation of the protective activitiesof a late embryogenesis abundant (LEA) related protein, Cor15am, during variousstresses in vitro, Biosci. Biotechnol. Biochem. 72 (2008) 1642–1645.

[24] S. Chakrabortee, C. Boschetti, L.J.Walton, S. Sarkar, D.C. Rubensztein, A. Tunnacliffe,Hydrophilic protein associated with desiccation tolerance exhibits broad proteinstabilization function, Proc. Natl Acad. Sci. USA 104 (2007) 18073–18078.

[25] K. Goyal, L.J. Walton, A. Tunnacliffe, LEA proteins prevent protein aggregation dueto water stress, Biochem. J. 388 (2005) 151–157.

[26] N.N. Pouchkina-Stantcheva, B.M. McGee, C. Boschetti, D. Tolleter, S. Chakrabortee,A.V. Popova, F. Meersman, D. Macherel, D.K. Hincha, A. Tunnacliffe, Functionaldivergence of former alleles in an ancient asexual invertebrate, Science 318 (2007)268–271.

[27] T. Sakurai, M. Satou, K. Akiyama, K. Iida, M. Seki, T. Kuromori, M. Itoh, A. Konagaya,T. Toyoda, K. Shinozaki, RARGE: a large-scale database of RIKEN Arabidopsis

resources ranging from transcriptome to phenome, Nucliec Acids Res. 33 (2005)D647–D650.

[28] M. Seki, M. Narusaka, A. Kamiya, J. Ishida, M. Satou, T. Sakurai, M. Nakajima, A. Enju, K.Akiyama, Y. Oono,M.Muramatsu, Y. Hayashizaki, J. Kawai, P. Carninci,M. Itoh, Y. Ishii, T.Arakawa, K. Shibata, A. Shinagawa, K. Shinozaki, Functional annotation of a full-lengthArabidopsis cDNA collection, Science 296 (2002) 141–145.

[29] H. Schägger, G. von Jagow, Tricine-sodium dodecyl sulfate-polyacrylamide gelelectrophoresis for the separation of proteins in the range from 1 to 100 kDa, Anal.Biochem. 166 (1987) 368–379.

[30] D.B. Wetlaufer, Ultraviolet spectra of proteins and amino acids, Adv. ProteinChem. 17 (1962) 303–390.

[31] N. Sreerama, R.W.Woody, Estimation of protein secondary structure from circulardichroism spectra: inclusion of denatured proteins with native proteins in theanalysis, Anal. Biochem. 287 (2000) 243–251.

[32] R.C. MacDonald, R.I. MacDonald, B.P.M. Menco, K. Takeshita, N.K. Subbarao, L. Hu,Small-volume extrusion apparatus for preparation of large, unilamellar vesicles,Biochim. Biophys. Acta 1061 (1991) 297–303.

[33] D.K.Hincha, E. Zuther, E.M.Hellwege, A.G.Heyer, Specific effects of fructo- and gluco-oligosaccharides in the preservation of liposomes during drying, Glycobiology12 (2002) 103–110.

[34] C. Cacela, D.K. Hincha, Low amounts of sucrose are sufficient to depress the phasetransition temperature of dry phosphatidylcholine, but not for lyoprotection ofliposomes, Biophys. J. 90 (2006) 2831–2842.

[35] A.V. Popova, D.K. Hincha, Intermolecular interactions in dry and rehydrated pureand mixed bilayers of phosphatidylcholine and digalactosyldiacylglycerol: aFourier-transform infrared spectroscopy study, Biophys. J. 85 (2003) 1682–1690.

[36] J.H. Crowe, A.E. Oliver, F.A. Hoekstra, L.M. Crowe, Stabilization of dry membranesby mixtures of hydroxyethyl starch and glucose: the role of vitrification,Cryobiology 35 (1997) 20–30.

[37] A.V. Popova, D.K. Hincha, Effects of cholesterol on dry bilayers: interactionsbetween phosphatidylcholine unsaturation and glycolopid or free sugar, Biophys.J. 93 (2007) 1204–1214.

[38] T. Lassmann, E.L.L. Sonnhammer, Kalign, Kalignvu and Mumsa: web servers formultiple sequence alignment, Nucleic Acids Res. 34 (2006) W596–W599.

[39] A.M. Waterhouse, J.B. Procter, D.M.A. Martin, M. Clamp, G.J. Barton, JalviewVersion 2 - a multiple sequence alignment editor and analysis workbench,Bioinformatics 25 (2009) 1189–1191.

[40] D.G. Kneller, F.E. Cohen, R. Langridge, Improvements in protein secondary structureprediction by an enhanced neural network, J. Mol. Biol. 214 (1990) 171–182.

[41] C. Geourjon, G. Deleage, SOPMA: significant improvements in protein secondarystructure prediction by consensus prediction from multiple alignments, Comput.Appl. Biosci. 11 (1995) 681–684.

[42] G. Pollastrii, D. Przybyski, B. Rost, P. Baldi, Improving the prediction of proteinsecondary structure in three and eight classes using recurrent neural networksand profiles, Proteins 47 (2002) 228–235.

[43] C. Cole, J.D. Barber, G.J. Barton, The Jpred 3 secondary structure prediction server,Nucleic Acids Res. 36 (2008) W197–W201.

[44] R. Adamczak, A. Porollo, J. Meller, Combining prediction of secondary structureand solvent accessibility in proteins, Proteins 59 (2005) 467–475.

[45] A.W. Burgess, P.K. Ponnuswamy, H.A. Sheraga, Analysis of conformations of aminoacid residues and prediction of backbone topography in proteins, Israel J. Chem.2 (1974) 239–286.

[46] G. Deleage, B. Roux, An algorithm for secondary structure prediction based onclass prediction, Protein Eng. 1 (1987) 289–294.

[47] R.D. King, M.J.E. Sternberg, Identification and application of the conceptsimportant for accurate and reliable protein secondary structure prediction,Protein Sci. 5 (1996) 2298–2310.

[48] J. Garnier, J.F. Gibrat, B. Robson, GOR method for predicting protein secondarystructure from amino acid sequence, Meth. Enzymol. 266 (1996) 97–120.

[49] J. Garnier, D.J. Osguthorpe, B. Robson, Analysis of the accuracy and implications ofsimple methods for predicting the secondary structure of globular proteins, J. Mol.Biol. 120 (1978) 97–120.

[50] L.H. Holley, M. Karplus, Protein secondary structure prediction with neuralnetworks, Proc. Natl Acad. Sci. USA 86 (1989) 152–156.

[51] R.D. King, M.J.E. Sternberg, Machine learning approach for the prediction ofprotein secondary structure, J. Mol. Biol. 216 (1990) 441–457.

[52] R. Gautier, D. Douguet, B. Antonny, G. Drin, HELIQUEST: a web server to screensequences with specific α-helical properties, Bioinformatics 15 (2008)2101–2102.

[53] D. Kovacs, E. Kalmar, Z. Torok, P. Tompa, Chaperone activity of ERD10 andERD14, two disordered stress-related plant proteins, Plant Physiol. 147 (2008)381–390.

[54] M.S. Webb, S.J. Gilmour, M.F. Thomashow, P.L. Steponkus, Effects of COR6.6 andCOR15am polypeptides encoded by cor (cold-regulated) genes of Arabidopsisthaliana on dehydration-induced phase transitions of phospholipid membranes,Plant Physiol. 111 (1996) 301–312.

[55] M.S. Webb, B.R. Green, Biochemical and biophysical properties of thylakoid acyllipids, Biochim. Biophys. Acta 1060 (1991) 133–158.

[56] C. Aurell Wistrom, R.P. Rand, L.M. Crowe, B.J. Spargo, J.H. Crowe, Direct transitionof dioleoylphosphatidylethanolamine from lamellar gel to inverted hexagonalphase caused by trehalose, Biochim. Biophys. Acta 984 (1989) 238–242.

[57] H.H. Mantsch, R.N. McElhaney, Phospholipid phase transitions in model andbiological membranes as studied by infrared spectroscopy, Chem. Phys. Lipids57 (1991) 213–226.

[58] J.M. Seddon, Structure of the inverted hexagonal (HII) phase, and non-lamellarphase transitions of lipids, Biochim. Biophys. Acta 1031 (1990) 1–69.

1820 A. Thalhammer et al. / Biochimica et Biophysica Acta 1798 (2010) 1812–1820

[59] J.P. Segrest, M.K. Jones, H. de Loof, C.G. Brouillette, Y.V. Venkatachalapathi, G.M. Anantharamaiah, The amphipathic helix in the exchangeable apolipopro-teins: a review of secondary structure and function, J. Lipid Res. 35 (1992)141–166.

[60] M. Schiffer, A.B. Edmundson, Use of helical wheels to represent the structures ofproteins and to identify segments with helical potential, Biophys. J. 7 (1967)121–135.

[61] N. Bies-Etheve, P. Gaubier-Comella, A. Debures, E. Lasserre, E. Jobet, M. Raynal, R.Cooke, M. Delseny, Inventory, evolution and expression profiling diversity of theLEA (late embryogenesis abundant) protein gene family in Arabidopsis thaliana,Plant Mol. Biol. 67 (2008) 107–124.

[62] J.-M. Mouillon, P. Gustafsson, P. Harryson, Structural investigation of disorderedstress proteins. Comparison of full-length dehydrins with isolated peptides oftheir conserved segments, Plant Physiol. 141 (2006) 638–650.

[63] J. Boudet, J. Buitink, F.A. Hoekstra, H. Rogniaux, C. Larre, P. Satour, O. Leprince,Comparative analysis of the heat stable proteome of radicles ofMedicago truncatulaseeds during germination identifies late embryogenesis abundant proteins associ-ated with desiccation tolerance, Plant Physiol. 140 (2006) 1418–1436.

[64] K. Goyal, L. Tisi, A. Basran, J. Browne, A. Burnell, J. Zurdo, A. Tunnacliffe, Transitionfrom natively unfolded to folded state induced by desiccation in an anhydrobioticnematode protein, J. Biol. Chem. 278 (2003) 12977–12984.

[65] M. Shih, S. Lin, J. Hsieh, C. Tsou, T. Chow, T. Lin, Y. Hsing, Gene cloning andcharacterization of a soybean (Glycine max L.) LEA protein, GmPM16, Plant Mol.Biol. 56 (2004) 689–703.

[66] D. Tolleter, M. Jaquinod, C. Mangavel, C. Passirani, P. Saulnier, S. Manon, E.Teyssier, N. Payet, M.-H. Avelange-Macherel, D. Macherel, Structure and functionof a mitochondrial late embryogenesis abundant protein are revealed bydesiccation, Plant Cell 19 (2007) 1580–1589.

[67] W.F. Wolkers, S. McCready, W.F. Brandt, G.G. Lindsey, F.A. Hoekstra, Isolation andcharacterization of a D-7 LEA protein from pollen that stabilizes glasses in vitro,Biochim. Biophys. Acta 1544 (2001) 196–206.

[68] K. Hristova, W.C. Wimley, V.K. Mishra, G.M. Anantharamaiah, J.P. Segrest, S.H.White, An amphipathic α-helix at a membrane interface: a structural study usinga novel X-ray diffraction method, J. Mol. Biol. 290 (1999) 99–117.

[69] V.K. Mishra, M.N. Palgunachari, J.P. Segrest, G.M. Anantharamaiah, Interactions ofsynthetic peptide analogs of the class A amphipathic helix with lipids, J. Biol.Chem. 289 (1994) 7185–7191.

[70] J.P. Segrest, H. de Loof, J.G. Dohlman, C.G. Brouillette, G.M. Anantharamaiah,Amphipathic helix motif: classes and properties, Proteins 8 (1990) 103–117.

[71] L. Dure III, A repeating 11-mer amino acid motif and plant desiccation, Plant J. 3(1993) 363–369.

[72] D. Li, X. He, Desiccation induced structural alterations in a 66-amino acid fragmentof an anhydrobiotic nematode late embryogenesis abundant (LEA) protein,Biomacromolecules 10 (2009) 1469–1477.

[73] D.K. Hincha, A.V. Popova, C. Cacela, Effects of sugars on the stability of lipidmembranes during drying, in: A. Leitmannova Liu (Ed.), Advances in Planar LipidBilayers and Liposomes, vol. 3, Elsevier, Amsterdam, 2006, pp. 189–217.

[74] A.E. Oliver, D.K. Hincha, J.H. Crowe, Looking beyond sugars: the role ofamphiphilic solutes in preventing adventitious reactions in anhydrobiotes atlow water contents, Comp. Biochem. Physiol. 131A (2002) 515–525.

[75] J.-M. Mouillon, S.K. Eriksson, P. Harryson, Mimicking the plant cell interior underwater stress by macromolecular crowding: disordered dehydrin proteins arehighly resistant to structural collapse, Plant Physiol. 148 (2008) 1925–1937.

[76] P. Tompa, C. Szasz, L. Buday, Structural disorder throws new light onmoonlighting, Trends Biochem. Sci. 30 (2005) 484–489.

Web references

[77] Kalign: http://www.ebi.ac.uk/Tools/kalign/index.html.[78] NNpredict: http://www.cmpharm.ucsf.edu/nomi/nnpredict.html.[79] SOPMA: http://npsa-pbil.ibcp.fr/cgi-bin/npsa_automat.pl?page=/NPSA/npsa_-

sopma.html.[80] SSPro8: http://scratch.proteomics.ics.uci.edu/.[81] JPred: http://www.compbio.dundee.ac.uk/www-jpred/.[82] Sable2: http://sable.cchmc.org/.[83] SDSC Biology Workbench: http://workbench.sdsc.edu/.[84] Helical wheel diagrams: http://heliquest.ipmc.cnrs.fr/cgi-bin/ComputParams.py.