Báo cáo khoa học: Effect of mutations in the b5–b7 loop on the structure and properties of human small heat shock protein HSP22 (HspB8, H11) pptx

To understand the role of this part of molecule in the structure and func-tion of small heat shock proteins, we mutated two highly conservative resi-dues K137 and K141 of human HSP22 and

and properties of human small heat shock protein HSP22 (HspB8, H11)

Alexei S Kasakov1,*, Olesya V Bukach1,*, Alim S Seit-Nebi1, Steven B Marston2and

Nikolai B Gusev1

1 Department of Biochemistry, School of Biology, Moscow State University, Russia

2 National Heart and Lung Institute, Imperial College London, UK

Small heat shock proteins (sHsp) form a large

super-family of ubiquitous proteins detected in all

organ-isms, except for some bacteria [1–3] The members

of this family range in size from 12–42 kDa and

contain a conservative so-called a-crystallin domain

consisting of 80–100 residues that is located in the

C-terminal part of the polypeptide chain [1–3] This

conservative domain is flanked by the N-terminal domain and short C-terminal extension with a differ-ent size and structure [4,5] All sHsp tend to form flexible oligomers, ranging from a dimer to more than 40 subunits, exchanging their subunits [6,7], and some sHsp are able to form mixed oligomers consist-ing of subunits of different natures [8,9] Crystal

Keywords

chaperone-like activity; intrinsically

disordered regions; oligomeric structure;

small heat shock proteins

Correspondence

N B Gusev, Department of Biochemistry,

School of Biology, Moscow State

University, Moscow 119991, Russia

Fax ⁄ Tel: +7 495 939 2747

E-mail: nbgusev@mail.ru

*These authors contributed equally to this

work

(Received 17 June 2007, revised 30 July

2007, accepted 3 September 2007)

doi:10.1111/j.1742-4658.2007.06086.x

The human genome encodes ten different small heat shock proteins, each

of which contains the so-called a-crystallin domain consisting of 80–

100 residues and located in the C-terminal part of the molecule The a-crystallin domain consists of six or seven b-strands connected by different size loops and combined in two b-sheets Mutations in the loop connecting the b5 and b7 strands and conservative residues of b7 in aA-, aB-crystallin and HSP27 correlate with the development of different congenital diseases

To understand the role of this part of molecule in the structure and func-tion of small heat shock proteins, we mutated two highly conservative resi-dues (K137 and K141) of human HSP22 and investigated the properties of the K137E and K137,141E mutants These mutations lead to a decrease in intrinsic Trp fluorescence and the double mutation decreased fluorescence resonance energy transfer from Trp to bis-ANS bound to HSP22 Muta-tions K137E and especially K137,141E lead to an increase in unordered structure in HSP22 and increased susceptibility to trypsinolysis Both mutations decreased the probability of dissociation of small oligomers of HSP22, and mutation K137E increased the probability of HSP22 crosslink-ing The wild-type HSP22 possessed higher chaperone-like activity than their mutants when insulin or rhodanase were used as the model substrates Because conservative Lys residues located in the b5–b7 loop and in the b7 strand appear to play an important role in the structure and properties of HSP22, mutations in this part of the small heat shock protein molecule might have a deleterious effect and often correlate with the development of different congenital diseases

Abbreviations

bis-ANS, 4,4¢-bis(1-anilinonaphtalene-8-sulfonate); DMS, dimethylsuberimidate; FRET, fluorescent resonance energy transfer; GuCl,

guanidinium chloride; sHSP, small heat shock protein.

structures are described in the literature for the

hyperthermophile Methanococcus jannaschii Hsp16.5

[10] and wheat (Triticum aestivum) Hsp16.9 [11], each

containing a single a-crystallin domain, and the

par-asitic flatworm Taenia saginata Tsp36, containing

two a-crystallin domains in the single polypeptide

chain [12]

Ten different sHsp are encoded in the human

gen-ome and are differently expressed in human tissues

[13,14] None of these proteins has been crystallized;

however, different experimental approaches

(cryo-electron microscopy, (cryo-electron spin resonance

spectros-copy, protein pin array, etc.) [15–17] and protein

modeling were used to reconstruct the structure of

mammalian aB-crystallin and Hsp27 (HspB1) [16–

18] According to these models, the a-crystallin

domain of both proteins consists of seven b-strands

packed into two b-sheets [16–18] The loop

connect-ing b5 and b7 and the N-terminal part of b7

appears to play an important role in the structure of

sHsp monomers [12,18] and intermonomer

interac-tions [12,18,19], as well as in the binding of protein

substrates to sHsp [17] The importance of this part

of molecule of the sHsp is supported by the fact

that mutations in the loop connecting the b5 and b7

strands, or in the b7 strand of sHsp, often correlate

with the development of certain congenital diseases

(congenital cataract, desmin related myopathy, distal

hereditary motor neuropathy, amongst others)

[20,21]

A recently described protein with an apparent

molecular mass of 22 kDa, denoted as HSP22,

HspB8 or H11 kinase, shares structural properties

typical to all members of the family of sHsp [22]

HSP22 possesses chaperone-like activity [23–25] and

appears to be involved in the regulation of many

processes such as proliferation, myocardium

hyper-trophy and apoptosis [26] Missense mutations of

K141 (K141E, K141N) located at the beginning

of b7 of HSP22 correlate with the development of

motor neuropathy and Charcot–Marie–Tooth disease

[27,28] Another conservative residue of HSP22,

namely K137, presumably located in the b5–b7 loop,

is homologous to R136 of human HSP27 that is

mutated in the case of Charcot–Marie–Tooth type 2

disease [20,21] Previously, we compared the structure

and properties of the wild-type HSP22 and its

K141E mutant [29] The present study analyses the

structure and properties of K137E and the

K137,141E mutant of human HSP22, aiming to

pro-vide new information on the structure of sHsp and

to shed new light on their role in the development

of human congenital diseases

Results

Peculiarities of HSP22 structure

Up to now, all attempts to crystallize mammalian sHsp have been unsuccessful Therefore, all structural information derives from a comparison of human sHsp with the crystal structures of M jannaschii Hsp16.5 [10] and T aestivum Hsp16.9 [11] The 3D structure of the monomer of T aestivum Hsp16.9 is presented in Fig 1A (protein databank accession code 1GME) and,

as shown in Fig 1B, we aligned the structures of

M jannaschii Hsp16.5 and T aestivum Hsp16.9 with the corresponding structures of three human sHsp [30] The elements of the secondary structure of M janna-schii Hsp16.5 and T aestivum Hsp16.9, as determined

by X-ray crystallography, are indicated by solid blue (a-helices) or solid red (b-strands) lines above and below the corresponding sequences (Fig 1A) Both these proteins contain a large number of well preserved b-strands that are predominantly (with the exception

of the b10 strand) located in the a-crystallin domain [10–12]

The models built for two mammalian sHsp (aB-crys-tallin [16] and HSP27 [18]) predict that both these pro-teins contain short a-helices in the N-terminal part of molecule (dashed blue lines denoted a1–a3 above the aB-crystallin and below the HSP27 sequences in Fig 1B) According to these models, both aB-crystallin and HSP27 contain seven b-strands (b2–b9) (dashed red lines) located in positions homologous to the cor-responding strands of two crystallized nonmetazoan sHsp Two predictions slightly differ with respect to the location and length of specific b-strands For example, in the model of aB-crystallin, the b7 strand is only four residues long [16] whereas, in the model of HSP27, the same strand is ten residues long [18] How-ever, the overall structures of aB-crystallin and HSP27 predicted by these two models are very similar, and the positions of the b-strands correlate well with the corresponding positions of the b-strands in M janna-schii Hsp16.5 and T aestivum Hsp16.9 (Fig 1B) Predictions of the secondary structure of HSP22 per-formed with the jpred program (http://www.combio dundee.ac.uk) indicate that this protein contains very small quantities of a-helices and is enriched in unor-dered structure and b-strands The residues of HSP22 that are predicted to form b-strands are indicated by wide dashed red lines in Fig 1B and are located in positions corresponding to the b3, b4, b5, b7 and b9 strands jpred failed to predict the formation of a b2 strand in the HSP22 structure According to this prediction, residues 153–155 of HSP22 tend to form an

B

Fig 1 Comparison of the structure of human HSP22 and other sHsp (A) Ribbon diagram of T aestivum Hsp16.9 monomer (protein data-bank accession code 1GME) The N- and C-terminal domains are indicated by N and C correspondingly All b-strands are numbered and the b5 and b7 strands are shown in red and blue, respectively G104 (equivalent to K137 of human HSP22) and R108 (equivalent to K141

of human HSP22) are shown in purple and grey, respectively (B) Alignment of human HSP22 with human aB-crystallin and HSP27 and

M jannaschii Hsp16.5 and T aestivum Hsp16.9 made with CLUSTALW [30] using the default settings The residues shown in black are identical in at least four sequences; residues in dark grey are conservative in at least four or identical in at least three sequences; resi-dues in light grey are homologous at three or identical in at least two sequences Solid blue and red lines above M jannaschii Hsp16.5 and below T aestivum Hsp16.9 sequences indicate a-helices and b-strands detected in the crystal structure of the corresponding proteins [10,11] Dashed blue and red lines above human aB-crystallin and below human HSP27 sequences indicate a-helices and b-strands pre-dicted in the models of the corresponding proteins [16,18] Residues of HSP22 prepre-dicted to form b-strands according to JPRED are indi-cated by wide dashed red lines and K137 and K141 are shown in red Numbers in parenthesis correspond to NCBI-Entrez-Protein database accession numbers.

a-helix, whereas residues 156 and 157 tend to form a

very short b-strand that might correspond to the b8

strand of the other sHsp

The primary structure of the a-crystallin domain of

human sHsp is very conservative and the loop

connect-ing the b5 and b7 strands is shorter than the

corre-sponding loop connecting the b5 and b7 strands of

nonmetazoan sHsp (Fig 1B) Moreover, the structure

of human sHsp lacks the b6 strand that is involved in

dimer formation of nonmetazoan sHsp (Fig 1A)

Although the b5–b7 loop is very short, it is not

com-pletely deleted in any human sHsp This part of the

molecule has a very conservative primary structure and

appears to play a diverse and important role For

example, mutation of a highly conservative positively

charged residue (R116 of aA-crystallin, R120 of

aB-crystallin or K141 of HSP22 located in homologous

position; Fig 1B) correlates with the development of

congenital cataract and⁄ or desmin related myopathy

[20,21], whereas mutations of R127, S135 and R136 of

human HSP27 are associated with distal hereditary

motor neuropathy and Charcot–Marie–Tooth disease

[20,21] Therefore, it is advisable to analyze the effect

of a mutation in this part of the molecule on the

struc-ture and properties of human sHsp

Oligomeric structure of HSP22 and its mutants

All samples of recombinant HSP22 and its mutants

purified by the method described previously [23] were

homogeneous according to SDS gel electrophoresis

(Fig 2) HSP22 and its mutants are highly susceptible

to proteolysis [23,24,29] and occasionally contained small quantities of proteolytic fragments Under the conditions used, the wild-type HSP22 and its K137E and K141E mutants migrated on the SDS gel electro-phoresis [31] as a band with an apparent molecular mass of 25.4 kDa, whereas the apparent molecular mass of the double mutant K137,141E was 30.4 kDa The calculated molecular mass of human wild-type HSP22 is close to 21.6 kDa [22] The unusually high apparent molecular mass determined by SDS gel elec-trophoresis can be due to anomalous binding of SDS

to acidic HSP22 and this effect is especially pro-nounced in the case of the particularly acidic double mutant K137,141E of HSP22 On native gel electro-phoresis performed both at neutral [32] and alkaline

pH [33], the wild-type HSP22 and its mutants migrated

as a single band with an apparent molecular mass of approximately 60 kDa (data not shown), thus indicat-ing that, under these conditions, HSP22 and its mutants form small oligomers

Size-exclusion chromatography was used for further investigation of the quaternary structure of HSP22 and its mutants When 200 lg of the wild-type HSP22 was loaded on the column, a single peak was detected with

a Stokes radius equal to 26.2 A˚, corresponding to an apparent molecular mass of 36.1 kDa (Fig 3A) These data agree well with the previously published data [23,24,29] On size-exclusion chromatography, both K141E and K137,141E were eluted as symmetrical peaks and the width at the respective half-height of their peaks was similar to that of the wild-type HSP22 The Stokes radii and apparent molecular masses of the K141E and K137,141E mutants were similar: 26.7 A˚ and 37.9 kDa (Fig 3A) [29] At the same time, the K137E mutant of HSP22 was eluted as a broad peak with a trailing end, with a Stokes radius and apparent molecular mass of 28.2 A˚ and 43.9 kDa, respectively (Fig 3A) Taking into account that the molecular mass

of HSP22 monomer is 21.6 kDa [22], it might be assumed that, under conditions of size-exclusion chro-matography, HSP22 and its mutants are either highly asymmetric (or intrinsically unfolded) or presented in the form of a mixture of monomers and dimers The data presented indicate that mutations in the b5–b7 loop (and especially K137E) affect either folding or extension of oligomerization of HSP22

To test this suggestion, we performed size-exclusion chromatography on the Superdex 200 HR10⁄ 30 col-umn in the presence of 6 m guanidinium chloride (GuCl) and, under these conditions, calibrated the col-umn with a set of protein standards (BSA, ovalbumin, chymotrypsin A and RNAse) [34] (Fig 3B) Under denaturating conditions, all samples of HSP22 were

Fig 2 SDS electrophoresis of the wild-type HSP22 (1) and its

K137E (2), K141E (3) and K137,141E (4) mutants The positions of

the molecular mass standards (in kDa) are indicated by arrows.

eluted in the form of symmetrical peaks with an

appar-ent molecular mass of 22.8 kDa, which is close to the

calculated value of the HSP22 monomer (21.6 kDa)

The data presented agree with the suggestion that,

under native conditions, HSP22 and its mutants form

dimers that dissociate to monomers in the presence of

6 m GuCl

If this suggestion is correct, we might assume that a

decrease in protein concentration will result in the

dis-sociation of small HSP22 oligomers and the formation

of protein species with smaller apparent molecular mass Indeed, if the quantity of the wild-type HSP22 loaded on the column was decreased from 200 lg to

10 lg, the elution volume of the protein peak was increased from approximately 11.3 mL to 11.8 mL (Fig 3C) This increase in elution volume corresponds

to a decrease in the apparent molecular mass from approximately 36.9 kDa to 29.3 kDa A similar decrease in the apparent molecular mass was observed for the K137,141E mutant of HSP22; however, at all concentrations, the apparent molecular mass of this mutant was slightly larger than the molecular mass of the wild-type protein (Fig 3C) At high concentration, the K137E mutant formed oligomers with an apparent molecular mass of approximately 44 kDa whereas, at very low concentration, the molecular mass of oligo-mers formed by this mutant was close to 32 kDa (Fig 1C) The data presented mean that mutations of K137 and K141 might affect either folding or dissocia-tion of HSP22 oligomers

There are many examples indicating that certain point mutations do not dramatically affect the quater-nary structure but, at the same time, induce destabili-zation of the overall structure of the sHsp [35,36] Therefore, we analyzed the effect of point mutations in the linker connecting the b5 and b7 strands of HSP22

on its thermal stability The wild-type protein or its mutants were heated for 30 min at 70C and, after

Fig 3 Size-exclusion chromatography of the wild-type HPS22 and its point mutants (A) Size-exclusion chromatography of the wild-type HSP22 (1, 2) and its K137E (3, 4) and K137,141E (5, 6) mutants on Superdex 75 column under native conditions The sam-ples were either kept on ice (solid curves 1, 3, 5) or heated for

30 min at 70 C (dashed curves 2, 4, 6) Equal volumes (150 lL) of each protein (210 lg) were subjected to chromatography on a Su-perdex 75 HR10 ⁄ 30 column For clarity, elution profiles of unheated and heated proteins are shifted from each other by 10 mAu and elution profiles between different proteins are shifted from each other by 30 mAu Arrows above the panel indicate the elution vol-ume of protein standards and their apparent molecular masses (B) Size-exclusion chromatography of the wild-type HSP22 (1) and its K137E (2), K141E (3) and K137,141E (4) mutants on the Super-dex 200 HR10 ⁄ 30 column in the presence of 6 M GuCl Equal volumes (150 lL) of each protein (150 lg) were subjected to chromatography For clarity, elution profiles are shifted from each other by 20 mAu Arrows above the panel indicate the elution vol-ume of protein standards and their apparent molecular masses (C) Dependence of elution volume on the quantity of protein loaded on

a Superdex 75 HR10 ⁄ 30 column Equal volumes (150 lL) contain-ing 10–200 lg of the wild-type protein (1) and its K137E (2) or K137,141E (3) mutants were subjected to chromatography under native conditions The data are representative of three independent experiments.

cooling for 20 min and centrifugation, were subjected

to size-exclusion chromatography (Fig 3A) Prolonged

heating at 70C did not affect the elution profile of

any of the proteins analyzed The amplitude, position

and the width of the protein peaks were not dependent

on the transient heating These data suggest that the

wild-type HSP22 and its mutants belong to the group

of the so-called intrinsically disordered proteins with

long stretches of unordered structure [37] and this

is one of the reasons for their unusual high thermal

stability

To further investigate the oligomeric structure of

HSP22, we employed chemical crosslinking HSP22

and its mutants at three different concentrations (0.1,

0.5 and 2.0 mgÆmL)1) were incubated in the presence

of 3.5 mm dimethylsuberimidate (DMS) for 1 h at

37C and the protein composition of the sample thus

obtained was analyzed by means of SDS gel

electro-phoresis In good agreement with the previously

pub-lished data [23,29], we found that incubation of the

wild-type HSP22 with the bifunctional reagent resulted

in the formation of an additional protein band with an

apparent molecular mass of 50 kDa, which presumably

corresponds to the HSP22 dimer (Fig 4A) Similar

results were observed in the case of the K137E mutant

of HSP22 (Fig 4B); however, in this case, the intensity

of the band corresponding to the HSP22 dimer was

more intense than in the case of the wild-type protein

Thus, although mutation K137E eliminates one

poten-tial site of chemical modification, the probability of

crosslinking of the K137E mutant by DMS is higher

than the probability of crosslinking of the wild-type

protein This fact agrees well with the size-exclusion

chromatography data indicating that the K137E

mutant forms larger oligomers than the wild-type

pro-tein (Fig 3C) If the double mutant K137,141E was

subjected to crosslinking, we detected only a very faint

band corresponding to dimer and this band was

detected only at a rather high protein concentration

(Fig 4C) The decreased probability of crosslinking of

the K137,141E mutant might be due to replacements

of Lys residues being potential sites of crosslinking or,

more likely, to the overall changes in the structure of

HSP22 that are induced by replacing two closely

sepa-rated positively charged Lys residues by negatively

charged Glu (see below)

Effect of K137E and K137,141E mutations on the

structure of HSP22

The data presented might indicate that the analyzed

mutations affect the secondary and tertiary structure

of HSP22 To check this suggestion, we analyzed some

spectral properties of the wild-type protein and its two mutants

The maximum of intrinsic Trp fluorescence of the wild-type HSP22 was located at 342 nm and the posi-tion of this maximum was not changed by mutaposi-tions K137E or K137,141E (Fig 5) Similar results were obtained previously with the K141E mutant of HSP22 [29] The fluorescence spectrum of HSP22 was decom-posed into discrete components characteristic of Trp located in different environments [38] For this

A

B

C

Fig 4 Crosslinking of the wild-type HSP22 (A) and its K137E (B) and K137,141E (C) mutants by DMS HSP22 was incubated either

in the absence of DMS (0), or in the presence of 3.5 m M of DMS (1–3) The protein concentration was equal to 0.10 (1), 0.50 (2) or 2.0 (3) mgÆmL)1 and, after incubation, equal quantities (2.5 lg) of protein were loaded onto the gel The positions of the molecular mass standards (in kDa) are indicated by arrows on the right.

purpose, the fluorescence spectra were fitted as a sum

of three polynomial distributions of the fourth of fifth

order, corresponding to three classes of Trp residues

differing in their environment, accessibility to solvent

and position of the fluorescent spectrum Using this

approach, we estimated the portion of each class of

fluorophores in the protein spectrum and found that

HSP22 contains Trp residues belonging to the so-called

classes I, II and III Class I corresponds to indole

located inside the protein globule, forming a 2 : 1

exci-plex with neighboring polar groups and having

maxi-mum fluorescence at 330–332 nm Class II corresponds

to Trp at the protein surface in contact with bound

water molecules (maximum fluorescence at 340–

342 nm) Finally, class III corresponds to indole

located at the protein surface in contact with free

water molecules (maximum fluorescence at 350–

355 nm) Approximately 44% of Trp residues of

HSP22 belong to class I, approximately 18% belong to

class II and approximately 38% belong to class III

Point mutations K137E or K137,141E do not

signifi-cantly affect the distribution of Trp residues between

these classes (data not shown) This may be due to

the fact that three out of four Trp residues are located

in the N-terminal end (Trp48, Trp51, Trp60) and the

fourth Trp residue (Trp96) are located at the very

beginning of the a-crystallin domain, far apart from

the mutated Lys residues Although the point

muta-tions do not affect the position of maximum

fluorescence, they slightly decrease the amplitude of

fluorescence and this decrease was more pronounced

for the K141E [29] and K137,141E mutants than for

the K137E mutant (Fig 5) The small decrease in

the amplitude of fluorescence detected for the point

mutants of HSP22 might reflect small changes in

structure, leading to an altered Trp environment or their accessibility to quencher or water molecules Hydrophobic interactions appear to play an impor-tant role in oligomer formation and in the interaction

of sHsp with their protein substrates [1,10–12] Hydro-phobic surfaces of HSP22 and its mutants were probed

by using bis-ANS In the isolated state, this hydropho-bic probe has a very low quantum yield that is dramat-ically increased after its binding to hydrophobic sites

on the protein molecules [24,29] Titration of HSP22 with bis-ANS was accompanied by an increase in fluo-rescence at 495 nm, indicating binding of the fluores-cence probe to the protein [24,29] In agreement with the previously published data [29], we were unable to achieve saturation and, in the range of 0–10 lm bis-ANS, the fluorescence at 495 nm was approximately proportional to the concentration of the fluorescent probe added These data indicate that HSP22 contains many low affinity bis-ANS binding sites that cannot

be completely saturated in the range of bis-ANS con-centrations used This is to be expected if HSP22 belongs to the group of intrinsically disordered pro-teins lacking well-organized hydrophobic sites To obtain more information on the structure, we analyzed fluorescence resonance energy transfer (FRET) from Trp residues of HSP22 and its mutants to the bound bis-ANS As indicated in Fig 6, titration of the wild-type HSP22 and its K137,141E mutant with bis-ANS was accompanied by a decrease in intrinsic Trp fluo-rescence at 342 nm and a concomitant increase in the fluorescence of bis-ANS at 495 nm Because, at any bis-ANS concentration, the ratio of fluorescence at 342

to fluorescence at 495 nm (F342⁄ F395) was lower for the wild-type protein than for its K137,141E mutant, we conclude that the probability of FRET is higher for the wild-type protein than for its mutant This may indicate that the mutation K137,141E affects the mutual orientation, overall flexibility and⁄ or distances between Trp and bis-ANS bound to HSP22

To obtain more detailed information on the struc-ture of HSP22 mutants, we employed CD spectros-copy The far-UV CD spectra of the wild-type protein has a negative maximum at 208 nm and its molar ellip-ticity at this wavelength is rather low (Fig 7) This spectrum is characteristic for proteins with a low a-helix content and a high content of unordered and b-structures Mutation K137E had no dramatic effect

on the far-UV CD spectra and a blue shift of only 2–

3 nm was observed in the position of the negative maximum (Fig 7) Previously, we have found that mutation K141E induces a rather large increase in the amplitude of the negative maximum on the far-UV

CD spectrum of HSP22 [29] Even larger changes were

Fig 5 Intrinsic Trp fluorescence of the wild-type HSP22 (1) and its

K137E (2) and K137,141E (3) mutants Fluorescence was excited at

295 nm The protein concentration was 0.1 mgÆmL)1.

observed in the case of the double K137,141E mutant.

Indeed, the double mutation results in a blue shift of

5–6 nm with respect to the position of negative

maxi-mum and a significant increase in the amplitude of this

maximum This change of the far-UV CD spectra can

reflect pronounced changes in the secondary structure

Using the approach developed by Sreerama and

Woody [39], we attempted to estimate the changes

induced by the point mutations in the secondary

struc-ture of HSP22

According to this estimation, the a-helix content is

equally low (approximately 5–6%) in the structure of

both the wild-type HSP22 and its two mutants As

expected, the secondary structure of HSP22 and its

mutants was characterized by a high content of

b-strands (approximately 31–37%) and turns and

unordered structures (approximately 58–63%)

Muta-tion K137E induced only very moderate changes in the

secondary structure At the same time, mutation K141E [29] and especially double mutation K137,141E were accompanied by a simultaneous decrease in the content of b-structure (from 37% to 31%) and an increase in the content of turns and unordered structure (from 58% to 63%) These data might indicate that mutations in the b5–b7 loop and in the N-terminal part

of the b7 strands destabilize the structure of HSP22

Limited trypsinolysis of the wild-type HSP22 and its K137E and K137,141E mutants

The method of limited trypsinolysis was used to check the suggestion that the analyzed mutations affect the stability of HSP22 The available literature [23,24,29] indicate that HSP22 is highly susceptible to proteoly-sis Indeed, even at a weight ratio for HSP22⁄ trypsin equal to 12 000 : 1, the sHsp was rapidly hydrolyzed (Fig 8A) Trypsinolysis of the wild-type HSP22 was accompanied by disappearance of the band corre-sponding to intact protein that migrated with an apparent molecular mass of 25.4 kDa and accumula-tion of peptides with apparent molecular masses equal

to 16.5, 18.0, 19.0, 22.0 and 23 kDa, respectively (Fig 8A) The same set of peptides was observed if K137E and K137,141E mutants were subjected to trypsinolysis To compare the apparent rates of tryp-sinolysis of the wild-type HSP22 and its mutants, we plotted ln(At⁄ Ao) (where Ao and At are the intensities

of the band of intact protein at the beginning of tryp-sinolysis and at the fixed time of tryptryp-sinolysis) against the time of incubation (Fig 8D) The apparent rate constants of trypsinolysis under these conditions were equal to 0.0496 ± 0.0027, 0.068 ± 0.026, 0.0863 ± 0.0039Æmin)1 (n¼ 7) for the wild-type HSP22, and its K137E and K137,141E mutants, respectively The data

Fig 6 Fluorescence resonance energy transfer from Trp residues of the (A) wild-type HSP22 and (B) its K137,141E mutant to the bound bis-ANS All experiments were performed at a protein concentration of 0.03 mgÆmL)1(1.5 l M of HSP22 monomer) and bis-ANS (in l M ) indi-cated above each spectrum.

Fig 7 Far-UV CD spectra of the wild-type HPS22 (1) and its K137E

(2) and K137,141E (3) mutants The spectra were recorded at the

concentration 0.65 mgÆmL)1 of each species with a cell path of

0.05 cm The spectra reported are the average of eight

determina-tions.

presented mean that K137E and especially K137,141E

mutants were more susceptible to proteolysis than the

wild-type HSP22 Mutations K137E and K137,141E

should eliminate one or two potential sites of

trypsin-olysis and, in this way, were expected to decrease the

rate of proteolysis Instead of decreasing susceptibility,

these mutations increased the susceptibility of HSP22

to trypsinolysis and this finding agrees well with the data of far-UV CD indicating that the ana-lyzed mutations induce destabilization of the HSP22 structure

Chaperone-like activity of wild-type HSP22 and its mutants

The data presented indicate that the point mutations

of residues 137 and 141 affect the structure and sta-bility of HSP22 Therefore, it can be expected that these mutations might change the chaperone-like activity of HSP22 To investigate this idea, we used two different model protein substrates Reduction of the disulfide bonds of insulin results in dissociation of its peptide chains and aggregation of chain B Addi-tion of the wild-type HSP22 retarded the onset of aggregation and decreased the amplitude of light scat-tering induced by insulin aggregation (Fig 9A, curves 3 and 3¢) K137E (Fig 9A, curves 1 and 1¢) and K137,141E (Fig 9A, curves 2 and 2¢) also retarded the onset of insulin aggregation and decreased the amplitude of light scattering; however, their effects were less pronounced than the corre-sponding effects of the wild-type protein For exam-ple, the aggregation curve in the presence of 0.2 mgÆmL)1 of K137E was comparable to the aggre-gation curve observed in the presence of 0.1 mgÆmL)1

of the wild-type HSP22 (compare curves 1¢ and 3 in Fig 9) The double mutant (K137,141E) possessed higher chaperone-like activity than the K137E mutant However, both at low and high concentra-tions, the double mutant possessed slightly lower chaperone-like activity than the wild-type protein (curves 2 and 3 in Fig 9)

Heating of rhodanase at 43C induces its denatur-ation, which is followed by aggregation The wild-type HSP22 and its mutants decreased the rate of rhodanase aggregation and the amplitude of light scattering (Fig 9B) In good agreement with the data obtained for insulin, we found that the K137E mutant was much less effective than the wild-type protein in preventing rhodanase aggregation (com-pare curves 1 and 3 in Fig 9B) The chaperone-like activity of K137,141E mutant was lower than, but comparable to, the chaperone activity of the wild-type protein Thus, on two different protein sub-strates, the chaperone-like activity of the wild-type HSP22 was higher than the corresponding activity of the two mutants analyzed and, among these mutants, the chaperone activity of the K137E mutant was especially low

A

B

C

D

Fig 8 Limited trypsinolysis of the wild-type HSP22 and its K137E

and K137,141E mutants Kinetics of trypsinolysis of the wild-type

HSP22 (A) and its K137E (B) and K137,141E (C) mutants The time

of incubation (in min) is indicated below each track and the arrows

show the positions of molecular mass markers (D) Determination

of apparent rate constants of trypsinolysis of the wild-type HSP22

(1, squares), K137E mutant (2, circles) and K137,141E mutant (3,

triangles) The data are representative of three independent

experi-ments.

All sHsp are characterized by the presence of a highly

conservative a-crystallin domain consisting of six or

seven b-strands combined in two b-sheets [1,10–12,40]

The detailed location and orientation of these

b-strands is known only for Hsp16.5 of M jannaschii

[10], Hsp16.9 of wheat [11] and Tsp36 of T saginata

[12] that were all obtained in crystallized form The

tertiary structure of other sHsp (including all human

proteins) is unknown; however, several models of their

structure have been proposed in the literature [16–

18,40] These models are based on a comparison of the

primary structure of different sHsp, predictions of

their secondary structure and on superposition of the mammalian protein sequence on the 3D structure of already crystallized sHsp [16–18,40,41] The a-crystallin domain of human sHsp lacks the b6 strand detected in the structure of Hsp16.5 of M jannaschii and wheat Hsp16.9 and the loop connecting b5 and b7 is much shorter than the corresponding loop of bacterial or plant sHsp (Fig 1) [10–12,40,41] This loop appears to play an important role in the structure and properties

of the sHsp [10–12,17,19,40,41] and mutations inside this loop or in the b7 strand correlate with the devel-opment of different congenital diseases [20,21,27,28] Because, at present, only two mammalian sHsp (a-crystallin and HSP27) have been investigated in detail, we were interested in analyzing the role of the b5–b7 loop in the structure of recently described human HSP22

The data obtained with respect to far-UV CD (Fig 7), extra high susceptibility to proteolysis (Fig 8) and resistance to thermal denaturation (Fig 3) indicate that HSP22 has a predominantly unordered structure Therefore, we might assume that HSP22 belongs to the group of intrinsically disordered proteins Accord-ing to the predictions, K137 is located either in the C-terminal part of the b5–b7 loop or in the N-terminal part of the b7 strand, whereas K141 is located inside the b7 strand (Fig 1A) Predictions of disordered regions using two different programs (http://www strubi.ox.ac.uk/RONN and http://iupred.enzim.hu) indi-cate that residues 137–141 of HSP22 are loindi-cated on the border of the unordered and ordered regions of HSP22 in the so-called downward spike [37] (Fig 10) Very often, these parts of the molecules are involved in inter- or intramolecular interactions and play an important role in recognition and cell signaling [37]

Fig 9 Chaperone activity of the wild-type HSP22 and its K137E and

K137,141E mutants using insulin (A) and rhodanase (B) as a model

protein substrates (A) Reduction induced aggregation of insulin

(0.2 mgÆmL)1) in the absence of HSP22 (curve 0) or in the presence

of 0.1 mgÆmL)1(empty symbols) or 0.2 mgÆmL)1(filled symbols) of

HSP22 (curves 3 and 3¢), or its K137E (curve 1 and 1¢) and

K137,141E mutant (curves 2 and 2¢) (B) Heat-induced aggregation of

rhodanase (0.14 mgÆmL)1) in the absence of HSP22 (curve 0) or in

the presence of 0.07 mgÆmL)1(empty symbols) or 0.14 mgÆmL)1

(filled symbols) of HSP22 or its mutants Curve numbers and

symbols are same as given in (A).

Fig 10 RONN plot of disorder probability of the wild-type HSP22 (NCBI-Entrez-Protein database accession number Q9UKS3) The horizontal line indicates the threshold for disorder prediction Posi-tions of K137 and K141 are marked by black dots.