Hydrophilic protein associated with desiccationtolerance exhibits broad protein stabilization functionSohini Chakrabortee*, Chiara Boschetti*, Laura J. Walton*†, Sovan Sarkar‡, David C. Rubinsztein‡,and Alan Tunnacliffe*§

*Institute of Biotechnology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QT, United Kingdom; and ‡Department of Medical Genetics,Cambridge Institute for Medical Research, Addenbrooke’s Hospital, Hills Road, Cambridge CB2 0XY, United Kingdom

Edited by David L. Denlinger, Ohio State University, Columbus, OH, and approved September 25, 2007 (received for review July 25, 2007)

The ability of certain plants, invertebrates, and microorganisms tosurvive almost complete loss of water has long been recognized,but the molecular mechanisms of this phenomenon remain to bedefined. One phylogenetically widespread adaptation is the pres-ence of abundant, highly hydrophilic proteins in desiccation-tol-erant organisms. The best characterized of these polypeptides arethe late embryogenesis abundant (LEA) proteins, first described inplant seeds >20 years ago but recently identified in invertebratesand bacteria. The function of these largely unstructured proteinshas been unclear, but we now show that a group 3 LEA proteinfrom the desiccation-tolerant nematode Aphelenchus avenae isable to prevent aggregation of a wide range of other proteins bothin vitro and in vivo. The presence of water is essential for main-tenance of the structure of many proteins, and therefore desicca-tion stress induces unfolding and aggregation. The nematode LEAprotein is able to abrogate desiccation-induced aggregation of thewater-soluble proteomes from nematodes and mammalian cellsand affords protection during both dehydration and rehydration.Furthermore, when coexpressed in a human cell line, the LEAprotein reduces the propensity of polyglutamine and polyalanineexpansion proteins associated with neurodegenerative diseases toform aggregates, demonstrating in vivo function of an LEA proteinas an antiaggregant. Finally, human cells expressing LEA proteinexhibit increased survival of dehydration imposed by osmoticupshift, consistent with a broad protein stabilization function ofLEA proteins under conditions of water stress.

anhydrobiosis � late embryogenesis abundant protein

Water is essential for life, but some organisms survive desic-cation and the dry state for long periods during which

metabolism and life processes come to a halt, but resume onrehydration. Desiccation tolerance, or anhydrobiosis (‘‘life withoutwater’’), is found across all biological kingdoms, including animalsand plants such as the nematode Aphelenchus avenae and theresurrection plant Craterostigma plantagineum (1–3). Investiga-tions into the molecular mechanisms of desiccation tolerancehave highlighted the importance of various hydrophilic proteins,chief among which are the late embryogenesis abundant (LEA)proteins (4).

LEA proteins have been known for many years to accumulate inmaturing plant seeds as they acquire desiccation tolerance (5, 6), buttheir discovery in invertebrates (7–13) suggests that similar mech-anisms govern anhydrobiosis in both animals and plants. LEAproteins are known to be largely unstructured in solution, probablybecause their extreme hydrophilicity favors association with waterover intrachain interactions, but they can show increased foldingwhen dried or when associated with phospholipid bilayers (14–16).Although LEA proteins are widely held to protect cells againstwater stress, their precise role has been a puzzle since they were firstdescribed. Recently, evidence supporting possible functions hasbeen obtained (6), including in vitro data that some LEA proteinsact to prevent other proteins aggregating during water loss, al-though not under heat stress. The enzymes citrate synthase andlactate dehydrogenase form insoluble aggregates when dried or

frozen but are prevented from doing so in the presence of Aav-LEA1, a group 3 member from the nematode A. avenae, or thegroup 1 LEA protein, Em, from wheat (17). Probably because oftheir hydrophilic, unstructured nature, LEA proteins themselvesare not susceptible to aggregation on desiccation. Their antiaggre-gation activity could mean they represent a novel form of dehy-dration-specific molecular chaperone. An alternative, perhaps sim-pler, explanation is that LEA proteins behave as molecular shields,which prevent the approach and interaction of aggregation-proneprotein species by steric or electrostatic repulsion, analogous topolymer stabilization of colloidal suspensions (6).

Whether LEA proteins are molecular chaperones or molecularshields, we hypothesize that they have a physiological role inprotecting a wide range of proteins against aggregation. We askedwhether AavLEA1 is able to prevent desiccation-induced aggrega-tion of a complex mixture of proteins, represented by the water-soluble proteomes of several species. Further, we tested whetherthis group 3 LEA protein showed antiaggregation activity in a livingcell. To achieve this outcome, we used a spontaneously aggregatingpolyglutamine (polyQ) expansion protein derived from the hun-tingtin protein associated with Huntington’s disease (HD). HD is aneurodegenerative disorder characterized by expansion of a trinu-cleotide (CAG) repeat in the huntingtin gene beyond the asymp-tomatic limit of 35 repeat units. Most adult-onset HD patients haveCAG repeats in the range 40–50 whereas, in the juvenile form ofHD, a CAG repeat number in excess of 55 is frequently observed.Higher order repeat lengths result in aggregation of N-terminalfragments of huntingtin, which form inclusion bodies in neuraltissue (18). Finally, we investigated whether expression of an LEAprotein in mammalian cells was able to increase resistance to waterstress. We will therefore present evidence for a protein stabilizationactivity for AavLEA1 both in vitro and in vivo, in both the dry andhydrated states, and highlight LEA proteins as key elements of themolecular toolkit required for desiccation tolerance.

ResultsLEA Protein Protects Water-Soluble Proteomes from Aggregation onDesiccation. Subjection of the water-soluble proteomes from thehuman cell line T-REx293 (derived from the embryonal kidney cellline HEK293) and the nematode A. avenae to in vitro desiccation

Author contributions: A.T. designed research; S.C., C.B., and L.J.W. performed research; S.S.and D.C.R. contributed new reagents/analytic tools; S.C., C.B., L.J.W., S.S., D.C.R., and A.T.analyzed data; and S.C. and A.T. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Abbreviations: LEA, late embryogenesis abundant; EGFP-HDQ74, EGFP-tagged huntingtinexon 1 protein with 74 polyQ repeats; HD, Huntington’s disease; 3MA, 3-methyladenine;polyQ, polyglutamine.

†Present address: Abcam plc, 332 Cambridge Science Park, Cambridge CB4 0FW, UnitedKingdom.

§To whom correspondence should be addressed. E-mail: at10004@biotech.cam.ac.uk.

This article contains supporting information online at www.pnas.org/cgi/content/full/0706964104/DC1.

© 2007 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0706964104 PNAS � November 13, 2007 � vol. 104 � no. 46 � 18073–18078

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Oct

ober

4, 2

020

stress by vacuum drying led to significant aggregation, as measuredby a light scattering assay (Fig. 1 A and B). To determine whethera small subset or a broad range of proteins was susceptible toaggregation on desiccation, we separated the insoluble, aggregatedproteins from the soluble fraction by centrifugation and thenanalyzed both fractions by SDS/PAGE. Very little protein wasapparent in the pellet in control samples not subjected to desicca-tion stress, consistent with the light scattering results, but desicca-tion resulted in part of the protein sample becoming insoluble. Theelectrophoresis pattern was similar in both soluble and insolublefractions, suggesting that most proteins form aggregates, but thatnot all molecules of any particular protein do so (Fig. 1C). Thisresult has also been observed previously when a single proteinspecies, e.g., citrate synthase, is dried (17, 19). Addition of the group3 LEA protein at an approximate molar ratio of AavLEA1 toproteome of 5:1 reduced this aggregation to levels comparable withcontrols (Fig. 1). Similar results were obtained with water solubletotal protein extracts from African green monkey COS-7 cells andthe desiccation-sensitive nematode Caenorhabditis elegans [support-ing information (SI) Fig. 7 A and B]. Other proteins do not preventproteome aggregation; indeed some, such as histone 2A andalpha-casein, are aggregation-prone themselves (SI Fig. 7C). Pre-vention of desiccation-induced aggregation by LEA protein wasapparent by SDS/PAGE, which revealed little protein in the insol-uble fraction (Fig. 1C) and this observation could be confirmed byquantitating the amount of protein in each fraction (SI Fig. 7D).LEA protein itself fractionated in the supernatant in the desiccation

experiment, i.e., it was not significantly incorporated into aggre-gates, as was apparent from the Coomassie-stained gel (Fig. 1C)and also from immunoblotting experiments (data not shown). Thus,LEA protein is able to reduce desiccation-induced aggregation ofwater-soluble proteomes from mammalian cell cultures and wholenematodes.

We tested LEA protein at molar ratios of 0:1, 1:1, 2:1, 5:1, and10:1 to the human water-soluble proteome to determine theoptimum amount required to prevent aggregation. Although amolar ratio of 1:1 did not offer appreciable protection, LEA proteinat a ratio of 2:1 provided significant reduction in aggregation, andlight scattering was reduced to control levels by AavLEA1 toproteome ratios of 5:1 or greater (Fig. 2A). These ratios are similarto those required for effective suppression of client protein aggre-gation by molecular chaperones in vitro (e.g., ref. 20).

We next investigated whether the LEA protein exerts its protec-tive effect during drying or during the rehydration process. Maxi-mum reduction in protein aggregation was obtained when Aav-LEA1 was present during desiccation, but some protection was alsoafforded when the LEA protein was added during the rehydrationperiod, at least up to 5 min after water was added to the driedprotein (Fig. 2B). This result indicates that a large fraction of theaggregation observed takes place during desiccation itself, but alsothat some aggregation must occur during rehydration and that thisaggregation can be prevented by LEA protein in aqueous condi-tions. This observation is consistent with the interpretation that,although desiccation drives unfolding of many proteins, time isrequired for sufficient encounter events to take place to allow buildup of large aggregates capable of scattering light. It is likely that thedry state is reached in drying experiments before adequate time has

Fig. 1. LEA protein prevents aggregation of water-soluble proteomes sub-jected to desiccation. (A and B) Light scattering measured as apparent absor-bance (A340) of human (T-REx293) and nematode (A. avenae) water-solubleproteomes, respectively, in the hydrated state (white), after vacuum dryingand rehydration (black), and after drying with AavLEA1 (gray). **, significanceat P � 0.01; ns, not significant. (C) Coomassie blue-stained polyacrylamide gelof soluble and insoluble fractions of the water-soluble proteome of A. avenae.Control, not dried; desiccated, vacuum-dried and rehydrated; desiccated �LEA, dried with AavLEA1. AavLEA1 is indicated by an arrow. sup, Solublefractions; pel, insoluble fractions. Size markers (M) range from 250 to 10 kDa.

Fig. 2. LEA protein reduces aggregation of water-soluble proteome duringboth dehydration and rehydration. Light scattering measured as apparentabsorbance (A340) of human water-soluble proteome, with addition of vary-ing molar ratios of AavLEA1 (A) and with LEA protein added at different timepoints, as indicated (B). *, significance at P � 0.05; **, significance at P � 0.01;

***, significance at P � 0.001; ns, not significant.

18074 � www.pnas.org�cgi�doi�10.1073�pnas.0706964104 Chakrabortee et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

4, 2

020

elapsed for all ‘‘preaggregates’’ to form large particles, but that thisprocess continues once water is reintroduced.

In Vivo Antiaggregant Role of LEA Protein. The ability of AavLEA1to prevent aggregation during rehydration of dried proteomessuggests that the LEA protein is also active in the hydrated state,and not just on desiccation. If so, this property should allow us toanswer a key question about LEA protein function, namely,whether it has similar activity in living cells. We have approachedthis question by developing several inducible human cell linesexpressing the A. avenae LEA protein. Aav-lea-1 gene expression isblocked by the TetR protein in these cell lines, but repression isremoved by addition of tetracycline to the medium. Five differentstable transformants were characterized (SI Table 1), and of theseT-REx293-LEA15 showed the highest expression and was chosenfor further studies. Immunofluorescence confocal microscopy usingan affinity-purified polyclonal antibody on tetracycline-inducedT-REx293-LEA15 cells showed AavLEA1 to be localized in thecytoplasm (Fig. 3A). No or very little AavLEA1 was observed in theuninduced T-REx293-LEA15 cell line, and none in the parentT-REx293 cell line, with or without tetracycline; immunoblottingconfirmed these results (Fig. 3B). According to both rocket immu-noelectrophoresis and ELISA analyses of the T-REx293-LEA15cell line, AavLEA1 represents �1.4% of total soluble protein oninduction. In the absence of tetracycline, the LEA protein repre-sents only 0.02% of total soluble protein (SI Fig. 8 and SI Table 1).

We tested for antiaggregation activity in T-REx293-LEA15 usinga model protein, EGFP-HDQ74 (EGFP-tagged huntingtin exon 1protein with 74 polyQ repeats), derived from the exon 1-encodedN-terminal region of huntingtin; this protein contains a 74-residuepolyQ sequence, making it prone to spontaneous aggregation, andis fused to EGFP to allow visualization. When uninduced

T-REx293-LEA15 cells were transiently transfected with a con-struct encoding EGFP-HDQ74, 13% of cells contained largeaggregates after 24 h. However, if cells were induced to expressAavLEA1 immediately after the transfection, only 5% developedEGFP-HDQ74 aggregates over the same time period, demonstrat-ing a marked antiaggregation effect of the LEA protein in vivo (Fig.4 A and B). Odds ratio analysis from multiple experiments showedthat the observed difference was highly significant (P � 0.0001; SIFig. 9A). In control experiments involving transfection of T-

Fig. 3. Expression of LEA protein in a human cell line. (A) Confocal micros-copy of T-REx293-LEA15 cells and T-REx293 cells with (Induced) or without(Uninduced) addition of tetracycline. AavLEA1 is immunostained green, andnuclei are stained blue with DAPI. (Scale bar: 5 �m.) (B) Immunoblotting usingimmunopurified antibody against AavLEA1. A nonspecific protein is shown asa loading control.

Fig. 4. LEA protein reduces protein aggregation in vivo. (A) Confocalmicroscopy of T-REx293-LEA15 cells after transfection with EGFP-HDQ74 ex-pression construct, with (Induced) or without (Uninduced) expression of LEAprotein. AavLEA1 is immunostained red, EGFP-HDQ74 is green, and nuclei arestained blue with DAPI. Indicative EGFP-HDQ74 aggregates are shown byarrowheads. (Scale bar: 40 �m.) Percentage of T-REx293-LEA15 cells withEGFP-HDQ74 aggregates with (filled) or without (open) LEA protein inductionwere assessed where both AavLEA1 and EGFP-HDQ74 were expressed for 24 h(B), 48 h (C), or where EGFP-HDQ74 is expressed for 48 h and AavLEA1 for thelast 24 h (D). (E) Inhibition of protein degradation pathways with 10 �Mlactacystin and 10 mM 3MA does not affect LEA protein antiaggregationactivity. Standard deviation of multiple replicates is shown as vertical bars. Allresults are highly significant as evaluated by logistic regression analysis (SIFig. 9).

Chakrabortee et al. PNAS � November 13, 2007 � vol. 104 � no. 46 � 18075

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Oct

ober

4, 2

020

REx293 with the EGFP-HDQ74 construct, there was no differencein numbers of aggregate-forming cells with or without sham induc-tion by tetracycline (SI Fig. 10).

Because polyQ aggregates continue to develop over at least 48 hafter transfection, we next asked whether AavLEA1 would alsoprotect over this longer time frame. Indeed, whereas 24% ofuninduced T-REx293-LEA15 cells contain large aggregates whenviewed after 48 h, this proportion was reduced to 15% when LEAprotein expression was induced (Fig. 4C and SI Fig. 9B). In a furthervariation of this experiment, a similar reduction in the proportionof cells containing aggregates was obtained, i.e., from 26% to 17%,even when LEA protein expression was induced for only the last24 h of the 48 h expression period of EGFP-HDQ74 (Fig. 4D andSI Fig. 9C). This result suggests that AavLEA1 has an inhibitoryeffect on the probability of EGFP-HDQ74 aggregate formation incells over the 48-h period assessed.

An alternative possibility, however, is that the LEA proteininstead accelerates clearance of aggregate-prone proteins from thecell, through one or both of the two main routes, i.e., the ubiquitin-proteasome and autophagosome-lysosome pathways, which effectremoval of EGFP-HDQ74 aggregates (21). To test this hypothesis,we used pathway-specific inhibitors to block both routes, when anyincrease in aggregate clearance by AavLEA1 should be abrogated.T-REx293-LEA15 cells were transfected with the EGFP-HDQ74construct and then LEA protein expression induced (or shaminduced in the control) immediately afterward. At the same time,proteasome and autophagosome function were inhibited simulta-neously by addition of lactacystin and 3-methyladenine (3MA),respectively (22). In controls where LEA protein expression was notinduced the clearance inhibitors led to an increase in numbers ofaggregate forming cells, as expected. However, even in the presenceof lactacystin and 3MA, LEA protein expression resulted in mark-edly reduced aggregation in the cell population (Fig. 4E and SI Fig.9D). This observation is consistent with AavLEA1 preventingEGFP-HDQ74 aggregate formation, rather than enhancing re-moval of the aggregate-prone protein.

Finally, to determine whether aggregation of other proteinscould be diminished, we introduced a construct encoding a poly-alanine-GFP fusion protein, EGFP-A37, into T-REx293-LEA15cells, with and without LEA protein induction. Expression ofAavLEA1 was able to reduce the proportion of aggregate-containing cells from 15% to 4% over the 48 h experimental period(Fig. 5), showing that the LEA protein is also active againstEGFP-A37 aggregation.

In Vivo Osmoprotective Role of LEA Protein. Given that AavLEA1can act as a protein antiaggregant both in vitro and in vivo, and thatthe level of expression in the T-REx293-LEA15 cell line is com-parable with the quantities of LEA proteins previously measured incotton seeds (23), we asked whether this activity could improve thetolerance of T-REx293-LEA15 cells to water stress. After inducingAavLEA1 expression for 24 h, the cells were subjected to desicca-tion stress at 90% relative humidity (R.H.), and the survival ratedetermined over time. The levels of viability quickly decreased to�40% within 2 h, and all cells were dead after 8 h (SI Fig. 11A).With a milder drying stress of 98% R.H., �40% survival wasobserved after 24 h, but there were no live cells after 48 h (data notshown and ref. 24). Identical results were obtained whether or notAavLEA1 expression was induced and therefore the LEA proteindid not increase the tolerance of human cells to desiccation. Thisresult is not particularly surprising, because it is probable thatdesiccation stress in human cells is too severe to be countered by asingle modification.

To assess whether LEA protein would offer protection against amilder form of water stress, T-REx293-LEA15 cells were exposedto hyperosmotic shock by addition of 100 mM NaCl to the medium.Tetracycline-induced cells expressing AavLEA1 did indeed show abetter survival rate compared with uninduced cells of the same cell

line over a 48 h period (Fig. 6A). In a control experiment, we alsoexposed T-REx293 (with and without tetracycline induction) to thesame stress conditions and no difference in survival rate wasobserved (SI Fig. 11B). To confirm that LEA protein expression inT-REx293-LEA15 cells improved osmotolerance, we also investi-gated the effect of other osmolytes such as mannitol, NaNO3,trehalose and sorbitol, with an osmotic upshift of �400 mOsm overa 24 h period, on the survival of induced and uninduced cells, andin all cases AavLEA1 expression significantly increased cell viability(Fig. 6B). Osmolyte concentrations, which increased osmotic po-tential by �1,000 mOsm, led to cell death even when LEA proteinexpression was induced, however (data not shown). These datatherefore show that although expression of a single LEA proteinalone is insufficient to protect cells against extreme water loss,nevertheless it can improve viability in cases of moderate stress.

DiscussionSurvival of desiccation and the dry state is one of the mostintriguing phenomena in nature, but is far from fully understood.

Fig. 5. LEA protein reduces aggregate formation by EGFP-A37 in vivo. (A)Confocal microscopy of T-REx293-LEA15 cells after transfection with EGFP-A37expression construct, with (Induced) or without (Uninduced) expression ofLEA protein. AavLEA1 is immunostained red, EGFP-A37 is green, and nuclei arestained blue with DAPI. Indicative EGFP-A37 aggregates are shown by arrow-heads. (Scale bar: 40 �m.) (B) Percentage of T-REx293-LEA15 cells with EGFP-A37 aggregates with (filled) or without (open) LEA protein induction wereassessed where both AavLEA1 and EGFP-A37 were expressed for 48 h. ***,significant at P � 0.0001 by logistic regression analysis (SI Fig. 9E).

Fig. 6. LEA protein expression improves osmotolerance of human cells.Shown are T-REx293-LEA15 cells subjected to 100 mM NaCl over 48 h (A), andvarious osmolytes at 400 mOsm for 24 h (B), either with (filled) or without(open) induction of AavLEA1 expression. Viability is assessed by measuringmetabolic rate compared with untreated controls. *, significant differencesbetween paired values (induced vs. uninduced) at P � 0.05 using the Studentt test.

18076 � www.pnas.org�cgi�doi�10.1073�pnas.0706964104 Chakrabortee et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

4, 2

020

The nonreducing disaccharides, trehalose in animals and fungi, andsucrose (with various oligosaccharides) in plants, accumulate indiverse anhydrobiotic organisms before desiccation, and arethought to have a protective function either as water replacementmolecules or vitrification agents in the dry state (1, 3). Trehaloseand other organic osmolytes also exert thermodynamic pressure tomaintain proteins in their native conformation under hydratedconditions (25) and, recently, trehalose has been shown to stimulateautophagy (22). Therefore, under stress conditions such as thoseimposed by water loss, trehalose could have a role in both preven-tion of protein denaturation and aggregation, and also in clearanceof any aggregates that do form. However, nonreducing disacchar-ides are not an absolute requirement for anhydrobiosis, beingabsent from bdelloid rotifers (26, 27), and apparently unnecessaryin yeast (28). Therefore, discovery programmes have been launchedin a number of anhydrobiotic organisms to determine key genes andproteins important in desiccation tolerance. One recurring theme isthe presence of a number of abundant, highly hydrophilic proteinsin desiccation-tolerant organisms (e.g., ref. 29), chief among whichare the LEA proteins (5, 6).

With one exception, all invertebrate LEA proteins identified todate are from group 3 (6) and the current picture therefore suggeststhat the other main classes of LEA proteins, group 1 and group 2,have functions of most benefit to plants. For example, group 1proteins are highly seed-specific and might act as hydration buffersto slow dehydration rates during seed development (30). However,because group 3 LEA proteins are common to both plants andanimals, it is likely they confer a function essential for desiccationtolerance across a wide phylogenetic spectrum. Two putative roleshave been attributed to group 3 LEA proteins: a membraneprotection activity, because a pea seed mitochondrial protein is ableto decrease fusion of liposomes subjected to drying in vitro (16); andthe prevention of protein aggregation, first reported with singlemodel proteins (citrate synthase, lactate dehydrogenase) exposed todesiccation or freezing (17).

The demonstration here that complex mixtures of proteins(water-soluble proteomes from mammalian cell lines and nema-todes) can also be protected from dehydration-induced aggregationin vitro shows that the effect is not restricted to a small subset oftargets, i.e., that it is not obviously substrate-specific. This obser-vation is consistent with a similar role for LEA proteins in anhy-drobiotic organisms, where a large fraction of the proteome is likelyto be susceptible to denaturation and aggregation due to desicca-tion. The data of Fig. 2B provide another previously undescribedfinding, namely that a proportion of such aggregation takes placeafter rehydration, and that LEA proteins continue to offer protec-tion after water is reintroduced. Therefore, LEA proteins have abroad protein stabilization function in both the dry and hydratedstates.

The two roles of group 3 LEA proteins (prevention of bothprotein aggregation and membrane fusion) might not be mutuallyexclusive and could, indeed, be undertaken by a single polypeptide.Performance of more than one function, known as ‘‘moonlighting,’’is not uncommon: e.g., a cyanobacterial small heat shock proteinhas both protein and membrane stabilization activities (31). LEAproteins are examples of the large number of unstructured proteinsthat comprise 6–17% of the proteome and in which it has beenargued that moonlighting is more likely to evolve than in conven-tional, folded proteins (32). Indeed, AavLEA1 has some propensityto associate with phospholipid bilayers in vitro, which might indicatemultifunctionality (33).

Although folding of AavLEA1 increases on desiccation in vitro,it remains unstructured in solution (15). Therefore, it is likely thatthe function of AavLEA1 in living cells is governed by the unfoldedstate. In turn, this property is consistent with the LEA proteinbehaving as a molecular shield rather than a molecular chaperone,because chaperones usually have defined secondary and tertiarystructure. Both the protein and membrane stabilization functions of

group 3 LEA proteins could be related if they behave as molecularshields: in the former case, they would prevent interaction betweenaggregation-prone proteins; and in the latter case, they would blockclose approach of fusogenic lipid bilayers, perhaps after associationwith membrane surfaces.

The availability of aggregation-prone proteins such as huntingtin-derived EGFP-HDQ74 has allowed us to demonstrate antiaggre-gation activity of an LEA protein in a living system. Experimentswith specific inhibitors confirm that this activity is as a result ofreduced formation of aggregates, rather than increased clearancevia the proteasome or autophagy. Previously, we had speculatedthat this activity would only be apparent in vivo under the excessivelycrowded conditions in a dehydrating cell (17). However, the abilityof a nematode LEA protein to reduce the likelihood of EGFP-HDQ74 and EGFP-A37 aggregate formation in fully hydrated cellsshows that AavLEA1 is not a dehydration-specific antiaggregant,and neither does it require water stress to develop antiaggregationactivity. Nevertheless, in desiccation-resistant organisms, LEAproteins might effectively only operate under water stress condi-tions because the expression of their genes is strictly regulatedin many species either by developmental cues or by environmen-tal conditions (5, 6). In the nematode A. avenae, from whichAavLEA1 derives, the cognate Aav-lea-1 gene is switched on bydesiccation or hyperosmotic stress but is largely inactive under otherconditions (34).

When expressed in a human cell line, AavLEA1 is found in thecytoplasm, consistent with its location in the nematode (35), andthis expression pattern is sufficient, under the mild water stressconditions of up to a 400 mOsm increase in medium osmolarity, toimprove cell viability by �10% over controls. Protection againstmore extreme water loss imposed by a �1,000 mOsm upshift or bydesiccation might result from extending LEA protein expression toother compartments in the cell, or by combination with, forexample, the capacity to synthesize trehalose (36). Thus, our resultsoffer insight into one of the protective mechanisms in anhydrobio-sis, namely protein stabilization, but also highlight a potentiallyimportant component for anhydrobiotic engineering of desiccation-sensitive cell types to full desiccation tolerance (37).

Materials and MethodsMammalian Cell Lines. T-REx293 (Invitrogen), T-REx293-LEA15,and COS7 were grown at 37°C in a 5% CO2, 100% relative humidityatmosphere in DMEM with 10% FBS, 5 mM glutamine, 500units/ml penicillin, and 0.5 mg/ml streptomycin (Sigma, Poole,U.K.). For T-REx293, 5 �g/ml blasticidin (Invitrogen, Paisley,U.K.) was added. The T-REx293-LEA15 cell line was constructedafter cloning an Aav-lea-1 cDNA with C-terminal myc and His6 tagsinto a modified pcDNA4/TO vector (Invitrogen) with the zeocinresistance cassette replaced by hygromycin resistance. Stable trans-fection was performed by using GeneJammer (Stratagene, La Jolla,CA) according to the manufacturer’s instructions. Selection andmaintenance of T-REx293-LEA15 was in 5 �g/ml blasticidin and 50�g/ml hygromycin. For transient transfections, 0.75 �g of pEGFP-HDQ74 (38) or pEGFP-A37 (39) plasmid was introduced into cellsby using GeneJammer, and expression was allowed to proceed for24 or 48 h. For Aav-lea-1 induction, 10 �g/ml tetracycline, orequivalent volume of 100% EtOH as control, was added. To inhibitthe ubiquitin-proteasome and autophagosome-lysosome pathways,10 �M lactacystin and 10 mM 3MA were added, respectively.

Protein Extraction from Mammalian Cell Cultures. PBS-washed pel-lets from cell lines were lysed on ice in lysis buffer (10 mM Tris�HCl,pH 6.8/68.5 nM NaCl/0.5 mM EGTA/0.5% Triton X-100/5%glycerol) for 30 min in the presence of Complete protease inhibitor(Roche Diagnostics, Burgess Hill, U.K.). The lysate was dialyzedovernight in distilled water at 4°C and centrifuged at 15,000 � g (30min, 4°C) before use. Protein concentration was determined spec-trophotometrically (A280).

Chakrabortee et al. PNAS � November 13, 2007 � vol. 104 � no. 46 � 18077

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Oct

ober

4, 2

020

Protein Extraction from Nematodes. A. avenae and C. elegans weregrown as described (15), harvested, suspended in 100 �l of lysisbuffer (50 mm Tris, pH 7.5/75 mM NaCl/15 mM EGTA/15 mMMgCl2/1 mM DTT/0.1% Tween 20/60 mM �-glycerophosphate/1�phosphatase inhibitor/Complete protease inhibitor) and sonicated(30 times, each 10 s; 20 s on ice between each sonication). Lysateswere centrifuged at 15,000 � g (30 min, 4°C). Supernatants werecollected and dialyzed as above before use. Water soluble proteinconcentration was determined by using a DC kit (Bio-Rad, HemelHempstead, U.K.).

In Vitro Protein Desiccation. Recombinant AavLEA1 protein wasmade as described (15) and dialyzed into water before use. Forvacuum drying, 0.15 mg of water-soluble proteomes in 200-�lvolumes, with or without AavLEA1 (at molar ratio of 5:1 unlessstated otherwise; molarity of proteomes was estimated assumingaverage Mr 50,000), were first degassed for 10 min in an Eppendorf5301 vacuum concentrator and then dried without freezing in avacuum tray drier, and aggregation was assessed by measuringapparent absorption at 340 nm in a spectrophotometer as described(17). Controls were not dried. Average and standard deviations areshown, and statistical relevance was determined by one-wayANOVA and Tukey post hoc test using InStat3 (GraphPad Soft-ware, San Diego, CA). For gel analysis, rehydrated samples andcontrols were centrifuged at 15,000 � g (30 min, 4°C), supernatantwas removed, and the pellet was dissolved in 11 �l of water; 9 �lfrom each fraction was used for SDS/PAGE as described (15).

Immunochemistry. Cells grown on coverslips in six-well Petri disheswere washed once in PBS, fixed in 4% paraformaldehyde, washedagain twice in PBS, blocked with 3% BSA/PBS for 30 min, andincubated with anti-AavLEA1 rabbit polyclonal antibody (15) at1:50 in 3% BSA/PBS for 1 h. Controls without primary antibodywere also performed. After a further three PBS washes, cells wereincubated in secondary antibody [goat anti-rabbit IgG, Cy3 conju-gate (Amersham, Little Chalfont, U.K.), or goat anti-rabbit IgG,FITC conjugate (Pierce, Rockford, IL)] at 1:100 in 3% BSA/PBS

for 1 h. Cells were washed three times more in PBS and mountedin 15 �l of Vectashield containing 1.5 �g/ml DAPI (VectorLaboratories, Peterborough, U.K.). Immunoblotting with anti-AavLEA1 polyclonal antibody was as described (15).

Cell Stress and Viability. Osmotic stress was imposed by addition ofosmolytes to growth medium at concentrations indicated. Viabilitywas assayed by using CellTiter 96 AQueous One Solution CellProliferation Assay (Promega, Southampton, U.K.), according tothe manufacturer’s instructions. Typically, six replica wells wereused for each experimental combination. Triplicate wells contain-ing only DMEM and cell proliferation reagent were used todetermine average background A490 of growth medium, which wassubtracted from other readings. Viability was expressed as a per-centage of the metabolic rate of a control population of cells notsubjected to stress. Each viability experiment was repeated at leasttwice. Significance was determined by using the Student t testbetween paired values.

In Vivo Aggregation Assay. Approximately 200 EGFP-positiveT-REx293-LEA15 cells were counted in each uninduced sample,whereas �200 EGFP-positive cells coexpressing AavLEA1 werecounted in each induced sample. Images were acquired with a Zeiss(Welwyn Garden City, U.K.) LSM510 META confocal microscope(�63, 1.4 N.A. PlanApochromat objective) with Zeiss LSM 510 v3.2software. Aggregate-containing cells were scored as percentagesand P values were determined by logistic regression analysis (esti-mated odds ratio and 95% confidence intervals) using SPSSsoftware (SPSS, Chicago, IL) as described (21). Also see SI Fig. 9legend.

This research was supported by Leverhulme Trust Grant F/09 717/B,Isaac Newton Trust Grant 6.20, Biotechnology and Biological SciencesResearch Council (BBSRC) Grant BB/D001307/1, and a BBSRC In-dustrial CASE studentship with Merck Chemicals Ltd. D.C.R. is aWellcome Trust Senior Fellow in Clinical Science.

1. Crowe JH, Carpenter JF, Crowe LM (1998) Annu Rev Physiol 60:73–103.2. Perry RN (1999) Parasitology 119:S19–S30.3. Clegg JS (2001) Comp Biochem Physiol B 128:613–624.4. Baker J, Steele C, Dure L, III (1988) Plant Mol Biol 11:277–291.5. Cuming AC (1999) in Seed Proteins, eds Shewry PR, Casey R (Kluwer

Academic, Dordrecht, The Netherlands), pp 753–780.6. Tunnacliffe A, Wise M (2007) Naturwissenschaften 94:791–812.7. Solomon A, Salomo R, Paperna I, Glazer I (2000) Parasitology 121:409–416.8. Browne J, Tunnacliffe A, Burnell A (2002) Nature 416:38.9. Gal TZ, Glazer I, Koltai H (2003) J Parasitol 89:761–766.

10. Gal TZ, Glazer I, Koltai H (2004) FEBS Lett 577:21–26.11. Tunnacliffe A, Lapinski J, McGee B (2005) Hydrobiologia 546:315–321.12. Kikawada T, Nakahara Y, Kanamori Y, Iwata KI, Watanabe M, McGee B,

Tunnacliffe A, Okuda T (2006) Biochem Biophys Res Commun 348:56–61.13. Hand SC, Jones D, Menze MW, Witt TL (2006) J Exp Zool 305A:1–5.14. Wolkers WF, McCready S, Brandt WF, Lindsey GG, Hoekstra FA (2001)

Biochim Biophys Acta 1544:196–206.15. Goyal K, Tisi L, Basran A, Browne J, Burnell A, Zurdo J, Tunnacliffe A (2003)

J Biol Chem 278:12977–12984.16. Tolleter D, Jaquinod M, Mangavel C, Passirani C, Saulnier P, Manon S,

Teyssier E, Payet N, Avelange-Macherel M-H, Macherel D (2007) Plant Cell19:1580–1589.

17. Goyal K, Walton LJ, Tunnacliffe A (2005) Biochem J 388:151–157.18. Rubinsztein DC (2002) Trends Genet 18:202–209.19. Walton LJ (2005) PhD thesis (Univ. of Cambridge, Cambridge, UK).20. Haslbeck M, Walke S, Stromer T, Ehrnsperger M, Whit HE, Chen S, Saibil HR,

Buchner J (1999) EMBO J 18:6744–6751.21. Ravikumar B, Duden R, Rubinsztein DC (2002) Hum Mol Genet 11:1107–1117.

22. Sarkar S, Davies JE, Huang Z, Tunnacliffe A, Rubinsztein DC (2007) J BiolChem 282:5641–5652.

23. Roberts JK, DeSimone NA, Lingle WL, Dure L, III (1993) Plant Cell 5:769–780.24. Huang Z, Tunnacliffe A (2004) J Physiol (London) 558:181–191.25. Tunnacliffe A, Lapinski J (2003) Philos Trans R Soc London B 358:1755–1771.26. Lapinski J, Tunnacliffe A (2003) FEBS Lett 553:387–390.27. Caprioli M, Krabbe Katholm A, Melone G, Ramløv H, Ricci C, Santo N (2004)

Comp Biochem Physiol A 139:527–532.28. Ratnakumar S, Tunnacliffe A (2006) FEMS Yeast Res 6:902–913.29. Tyson T, Reardon W, Browne JA, Burnell AM (2007) Int J Parasitol 37:763–776.30. Manfre AJ, Lanni LM, Marcotte WR, Jr (2006) Plant Physiol 140:140–149.31. Torok Z, Goloubinoff P, Horvath I, Tsvetkova NM, Glatz A, Balogh G,

Varvasovszki V, Los DA, Vierling E, Crowe JH, et al. (2001) Proc Natl AcadSci USA 98:3098–3103.

32. Tompa P, Szasz C, Buday L (2005) Trends Biochem Sci 30:484–489.33. Pouchkina-Stantcheva NN, McGee BM, Boschetti C, Tolleter D, Chakrabortee

S, Popova AV, Meersman F, Macherel D, Hincha DK, Tunnacliffe A (2007)Science 318:268–271.

34. Browne JA, Dolan KM, Tyson T, Goyal K, Tunnacliffe A, Burnell AM (2004)Eukaryot Cell 3:966–975.

35. Goyal K, Pinelli C, Maslen SL, Rastogi RK, Stephens E, Tunnacliffe A (2005)FEBS Lett 579:4093–4098.

36. Garcıa de Castro A, Tunnacliffe A (2000) FEBS Lett 487:199–202.37. Kanias T, Acker JP (2006) Cell Preserv Technol 4:253–277.38. Narain Y, Wyttenbach A, Rankin J, Furlong RA, Rubinsztein DC (1999) J Med

Genet 36:739–746.39. Bao YP, Cook LJ, O’Donovan D, Uyama E, Rubinsztein DC (2002) J Biol

Chem 277:12263–12269.

18078 � www.pnas.org�cgi�doi�10.1073�pnas.0706964104 Chakrabortee et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

4, 2

020