Evidence for an essential function of the N terminus of a small heat shock protein in vivo, independent of in vitro chaperone activity

Abstract

To investigate the mechanism of small heat shock protein (sHsp) function, unbiased by current models of sHsp chaperone activity, we performed a screen for mutations of Synechocystis Hsp16.6 that reduced the ability of the protein to provide thermotolerance in vivo. Missense mutations at 17 positions throughout the protein and a C-terminal truncation of 5 aa were identified, representing the largest collection of sHsp mutants impaired in function in vivo. Ten mutant proteins were purified and tested for alterations in native oligomeric structure and in vitro chaperone activity. These biochemical assays separated the mutants into two groups. The C-terminal truncation and six mutations in the α-crystallin domain destabilized the sHsp oligomer and reduced in vitro chaperone activity. In contrast, the other three mutations had little effect on oligomer stability or chaperone activity in vitro. These mutations were clustered in the N terminus of Hsp16.6, pointing to a previously unrecognized, important function for this evolutionarily variable domain. Furthermore, the fact that the N-terminal mutations were impaired in function in vivo, but active as chaperones in vitro, indicates that current biochemical assays do not adequately measure essential features of the sHsp mechanism of action.

Small heat shock proteins (sHsps) are a ubiquitous family of chaperones defined by their conserved, ≈90-aa α-crystallin domain (1), which is flanked by a variable N-terminal region and a short C-terminal extension. In their native state the majority of sHsps are found as large oligomers of 9 to >30 subunits, depending on the sHsp (2–4). Biochemical studies with purified components have demonstrated that sHsps act as chaperones by binding and holding denaturing proteins in refoldable states until ATP-dependent chaperones, primarily DnaK/Hsp70 (but enhanced by ClpB/Hsp100), are supplied to refold them (5–12).

Evidence that sHsps bind denaturing proteins in vivo comes from a number of studies. In the cyanobacterium Synechocystis sp. strain PCC 6803 (Synechocystis), Hsp16.6 interacts during heat stress in vivo with a large number of heat-labile proteins that can subsequently be released by DnaK in cell extracts (13). sHsps in Escherichia coli, plants, and yeast enter an insoluble fraction of the cell during heat stress, presumably bound to substrates, and release to the soluble fraction is dependent on the ClpB/Hsp100 chaperones (14–16). In human diseases that are linked to sHsp gene mutations, sHsps have been found colocalized with misfolded cellular proteins (17–19). However, sHsps have been shown to interact with an increasing number of diverse cellular proteins, implicating them in everything from ubiquitin-mediated proteolysis to modulation of translation (4). sHsps can also interact with lipids (20). Thus, it remains to be demonstrated that sHsp chaperone activity, as currently defined biochemically, fully accounts for the role of sHsps in vivo.

Definition of sHsp function has suffered from lack of a genetic model system in which to test sHsp action in vivo. In both E. coli and Saccharomyces cerevisiae, phenotypes associated with sHsp deletions are difficult to measure (21, 22). However, the single sHsp in Synechocystis, Hsp16.6, is essential for tolerance to high temperature (23, 24), affording a unique genetic model for sHsp studies. The effects of specific mutations on sHsp activity in vivo have been tested, and results have indicated a link between the requirements for cellular thermotolerance and chaperone activity as measured in vitro by established techniques (24, 25). However, these previous studies targeted mutations that altered the oligomeric stability of Hsp16.6, leaving other essential properties of the protein untested and undefined.

Here, we screened for random mutations of hsp16.6 that cause a loss of thermotolerance in Synechocystis, to define features of Hsp16.6 that are essential for in vivo function. The screen recovered point mutations in 17 codons, located in both conserved and variable regions of hsp16.6, and a C-terminal truncation of 5 aa. All of the mutant proteins accumulated in vivo. When 10 of these mutant Hsp16.6 proteins were purified, 7 reduced both in vitro chaperone activity and oligomeric stability, similar to other mutations of sHsp oligomeric contacts (24, 25). A second group of three mutations, all located in the N terminus, had little effect on in vitro chaperone activity or oligomeric stability, although they significantly decreased sHsp function in vivo. These mutations point to an important function for the evolutionarily variable sHsp N terminus in cellular thermotolerance, which would not have been identified by assaying mutants solely by standard chaperone assays in vitro.

Materials

Plasmids. For transformation of Synechocystis with randomly mutagenized hsp16.6, pNaive/Xho was created by inserting an oligonucleotide (TACCTCGAGG) into the NdeI site of pNaive (24) to create a unique XhoI site, which is lost when hsp16.6 is ligated into its HpaI and ApaI sites. pJC20/Hpa hsp16.6 (24) was randomly mutagenized by degenerate PCR (26). Plasmids were amplified in E. coli DH5α before excising hsp16.6 as HpaI/ApaI fragments. Pools of fragments were ligated into pNaive/Xho. The ligation reaction was digested with XhoI before E. coli transformation to reduce the population of empty vector.

Screen for Thermotolerance-Defective Mutants in Vivo. Pools of pNaive containing randomly mutated hsp16.6 were isolated from E. coli and transformed into the Synechocystis strain HK-1/ΔClpB1, which is null for both hsp16.6 and clpB1 (24). Clonal populations were replica plated to BG-11/agar (27) with 140 mM MgSO₄ and 5 mM glucose, and heated at 44°C for 8 h. Colonies that did not survive were isolated from duplicate, unstressed plates and screened for Hsp16.6 accumulation by Western blotting. Genomic DNA was isolated from cells that were positive for Hsp16.6 and used to sequence hsp16.6. Alleles containing more than one missense mutation or large truncations of hsp16.6 were not characterized. Alleles with single amino acid changes were recloned into pNaive and transformed into Synechocystis HK-1/ΔClpB1 cells. To quantify thermotolerance, serial dilutions of cells were heat stressed as above (24).

Protein Purification. Hsp16.6 and mutant variants were inserted into pJC20 and purified from the E. coli strain BL21 (24). The concentration of WT Hsp16.6 was determined by using an extinction coefficient of ε₂₈₀ = 5,960 M^–1·cm^–1 (24). Concentrations of mutant Hsp16.6 proteins were assayed by the Bradford method (28), with WT Hsp16.6 as standard. Expected molecular mass of the mutant proteins was confirmed by ion trap mass spectrometry (Finnigan Classic, Thermo Electron, San Jose, CA) after separation by HPLC (Michrom BioResources, Auburn, CA) (Magic C18, 5-μm particles, 1 × 150 mm column) using an acetonitrile/water gradient.

Luciferase Protection Assays. Firefly luciferase (Luc) reactivation was performed with reticulocyte lysate components as described previously in ref. 24.

Size-Exclusion Chromatography (SEC). We injected 100 μlof24 μM sHsp onto a Bio-Sil SEC 400 column (Bio-Rad) equilibrated in 20 mM sodium phosphate, pH 7.3/20 mM NaCl, flow rate 1 ml/min. Buffer and column were both at room temperature or at 44°C, as described. Samples were heated as indicated and centrifuged 15 min at 16,000 × g before separation on the column.

Results

Mutations Throughout Synechocystis Hsp16.6 Disrupt in Vivo Activity. We screened for thermotolerance-defective strains of Synechocystis, to identify residues critical for sHsp function, without bias from current models of chaperone activity. Random mutants of hsp16.6 were transformed into a strain of Synechocystis that carries deletions of both clpB1 and hsp16.6, because of an enhancement of the dependence on Hsp16.6 activity for thermotolerance in the absence of the ClpB1 chaperone (24). Transformation places alleles of hsp16.6 under the control of the hsp16.6 promoter. Transformed colonies were heated for 8 h at 44°C, then returned to 30°C for growth. The +Hsp16.6 strain, which expresses WT Hsp16.6, survives this treatment, whereas an isogenic null strain (ΔHsp16.6) does not (24). Colonies of mutants that did not survive were isolated from duplicate, unheated plates. To focus our studies on mutations that reduce sHsp activity, mutant strains were tested for Hsp16.6 accumulation. Only the genes from heat-stress-sensitive cells that accumulated detectable levels of Hsp16.6 were sequenced and studied further.

Using this screen, we identified 18 missense mutations at 17 codons positioned throughout the gene that reduced thermotolerance (Fig. 1A). Fourteen mutations are in the conserved α-crystallin domain (residues 44–137), whereas four are in the nonconserved N terminus (residues 1–43). A nonsense mutation of Lys-142, which causes a 5-aa truncation of the C-terminal extension, was also identified. While some mutations were isolated twice, most were found only once (Table 2, which is published as supporting information on the PNAS web site), indicating that the screen for thermotolerance-defective mutants is not saturated.

Fig. 1.

Thermotolerance mutants of Hsp16.6. (A) Position of thermotolerance mutations relative to the sequences and structural elements of other sHsps. An alignment of the protein sequences of Synechocystis Hsp16.6 (Syn), Methanococcus jannaschii Hsp16.5 (Mj), Pisum sativum Hsp18.1 (Ps), and Triticum aestivum Hsp16.9 (Ta) is shown with identical residues shaded (for details, see Supporting Text, which is published as supporting information on the PNAS web site). The secondary structure, β-strands (B) and 3₁₀-helices (H), of the α-crystallin domain of MjHsp16.5 (2) is below. Mutations isolated in the thermotolerance screen are shown above the Hsp16.6 sequence in regular type, and the site-directed mutation R104G is in boldface. (B) Thermotolerance of Synechocystis strains expressing WT Hsp16.6 (+Hsp16.6), no sHsp (ΔHsp16.6), and mutants. Means and standard errors of survival (%) after8hat 44°C are shown. Each strain was tested in at least three independent experiments. (C) Accumulation of Hsp16.6 after 2 h at 42°C was determined by Western blotting of 1 μg of cellular protein with anti-Hsp16.6.

The mutant genes were each retransformed into Synechocystis to verify that the loss of thermotolerance was the result of the hsp16.6 allele. Survival after a 44°C, 8-h heat treatment of the retransformed strains is shown in Fig. 1B. The least defective mutants, D13G, R72G, R94P, F97S, and I112T, provided survival from 4% to 10% of unheated cells, only slightly less than the ≈30% viability of WT. Mutants with intermediate activity (0.3–2.0% viability) were comparable to previously characterized thermotolerance mutants, L66A and V143A (Fig. 1B, two lanes on the right), which also impair the chaperone activity of Hsp16.6 in vitro (24, 25). The most severe effects (viabilities from 0.02% to 0.2%) were caused by F102S and two N-terminal mutations, E25K and L9P. The F102S and L9P strains survived better than ΔHsp16.6 (0.006%), demonstrating that these mutants provided some thermotolerance. Survival of the E25K strain was not significantly different from that of ΔHsp16.6 after 8 h at 44°C, but shorter stresses demonstrated some activity of E25K (data not shown).

The effect of the mutations on Hsp16.6 accumulation in the retransformed strains was examined by inducing hsp16.6 with a 42°C heat treatment for 2 h. This heat treatment results in high Hsp16.6 accumulation in WT cells, and it is not lethal even for cells with no sHsp function (24). As shown in Fig. 1C, the majority of the mutant proteins accumulated to WT levels, whereas four mutations of conserved hydrophobic residues reduced Hsp16.6 accumulation. Cells expressing F102S and I106N, mutations of β7, and I112T in the turn after β7, had <20% of WT Hsp16.6 levels. Nevertheless, I112T-expressing cells had only a small thermotolerance defect, and survived >10-fold better than many mutants with WT levels of Hsp16.6. L130S in β9 led to a low level of sHsp that was just detectable in 1 μg of cellular protein, but this mutant also supported thermotolerance better than other mutants that accumulated Hsp16.6 to WT levels. Thus accumulation of an Hsp16.6 mutant is not a good predictor of the severity of its thermotolerance defect.

Mutations of both conserved and nonconserved amino acids were identified, but the severity of the defects did not correlate with conservation (Fig. 1 A). P59H, R81H, and G124S changed conserved residues, but had relatively mild effects on Synechocystis thermotolerance (2% viability). Mutations of conserved hydrophobic residues caused severe heat sensitivity in the case of F102S, but a mild effect for I112T. Mutations of nonconserved residues also resulted in moderate to severe defects. Interestingly, two of the most deleterious mutations that accumulated WT levels of sHsp were E25K and L9P, residues in the evolutionarily variable N-terminal region. These mutations provide direct genetic evidence for an important role of this variable domain in sHsp function in vivo.

Effects of the Mutations on Hsp16.6 Oligomeric Structure. As for other sHsps, the ability of Hsp16.6 to protect heat-denaturing proteins is correlated with the ability of the protein to oligomerize (24, 25). We therefore reasoned that the newly identified thermotolerance mutants might have oligomeric defects. Ten mutations, which were distributed throughout the protein and reduced cell viability to ≤2% after heat stress, were selected for biochemical analysis. The mutant proteins listed in Table 1 were successfully expressed in E. coli and purified. We also attempted to purify E44V, but it aggregated during purification and was not investigated further.

Table 1. Biochemical properties of mutants.

Hsp16.6	Oligomerization after heating*	Refolding,† %	Refolding t1/2,‡ min
WT	Oligomer	49 ± 1	28
BSA	Not applicable	3.5 ± 0.4	<10
L9P	Oligomer	54 ± 3	15
Q16R	Oligomer	41 ± 2	34
E25K	Aggregate/oligomer	70 ± 3	20
E57G	Mostly dimer	20 ± 3	<10
R81H	Dimer	19 ± 2	<10
F102S	Dimer	2.55 ± 0.04	<10
I106N	Mostly dimer	4.4 ± 0.3	12
G124S	Intermediate and dimer	14.6 ± 0.7	11
L130S	Mostly dimer	6.2 ± 0.4	<10
K142stop	Dimer	4.8 ± 0.4	<10

^*Apparent structure of 24 μM sHsp was determined as described in Fig. 2.

^†Luc reactivation by reticulocyte lysate, after heating 1 μM Luc with 24 μM sHsp or an equivalent weight of BSA (0.4 mg/ml). Means and standard deviations from three experiments are given.

^‡Time for half of final amount of luc to refold estimated from three experiments. Error is ± 2 min.

The oligomeric states of the mutant proteins were examined by SEC under conditions where WT Hsp16.6 is most distinct from mutants that destabilize the oligomer (25). When 24 μM sHsp is heated at 42°C for 7.5 min, and then kept at 4°C for 20 min before injection onto the column, WT Hsp16.6 elutes in an asymmetric peak of ≈400 kDa (elution time 8.5 min, Fig. 2), consistent with an oligomer of ≥20 subunits. Dimers of the previously characterized L66A and V143A mutants elute later than WT, and also migrate slightly differently from one another (10.2 and 9.9 min, respectively) (25).

Fig. 2.

Comparison of the native structures of Hsp16.6 mutants. Hsp16.6 (24 μM) was heated for 7.5 min at 42°C, incubated at 4°C for 20 min, then injected onto a room-temperature SEC column. Top Left shows WT, V143A, and L66A run independently. Arrowheads above are elution times of molecular mass standards: thyroglobulin (670 kDa) at 7.7 min, β-amylase (200 kDa) at 9.1 min, BSA (66 kDa) at 9.6 min, carbonic anhydrase (29 kDa) at 10.6 min, and myoglobin (17 kDa) at 11.1 min.

When analyzed by SEC under these conditions, the Hsp16.6 mutants could be divided into two groups: those that significantly disrupted oligomerization and those that did not. The first group included R81H, F102S, and K142stop, which eluted similarly to L66A and V143A, consistent with dimeric structures (Fig. 2). In support of this interpretation, crosslinking has shown that K142stop is fully dimeric after heat treatment (25). The peak elution times of the mutants varied from 9.9 to 10.3 min, suggesting that they may have slightly different shapes, or that interactions with the column vary. E57G, I106N, and L130S also eluted as dimers, but, in addition, some fraction of each also eluted as intermediately sized structures, smaller than WT oligomers. G124S was approximately equally distributed between dimer and intermediately sized species.

In contrast, the N-terminal mutants L9P, Q16R, and E25K behaved as oligomers after heat treatment. L9P and Q16R eluted predominantly as oligomers, although a small proportion also eluted as dimers (Fig. 2). The major peak of Q16R resembled that of unheated WT (data not shown), with an elution time of 8.2 min. The E25K mutant also formed a large structure, but eluted in a broad peak that migrated somewhat faster than the WT oligomer. A significant amount of E25K failed to elute from the column. Thus E25K either assembles oligomers with more subunits than WT or it is susceptible to aggregation.

The relative stabilities of oligomers formed by L9P, Q16R, and E25K were also examined by SEC at high temperature. At 24 μM and 44°C, these mutations of the N terminus have different effects. L9P slightly destabilizes the oligomer, whereas Q16R and E25K increase the stability relative to WT (Fig. 3). Thus we have identified mutants that significantly impair the function of Hsp16.6 in vivo without greatly disrupting oligomerization.

Fig. 3.

Oligomeric structures of N-terminal mutants at high temperature. Hsp16.6 (24 μM) was incubated at 44°C for 35 min before being injected onto a 44°C SEC column. Proteins examined were as follows: WT (•), L9P (♦), Q16R (▪), and E25K (▴). Peak elution times were 8.7, 8.9, 8.0, and 7.3 min for WT, L9P, Q16R, and E25K, respectively. Arrowheads show elution of oligomer (7.2 min) and dimer (9.0 min), as determined by using the S2Y mutant, which forms oligomers that are stable at 44°C (25), and the dimeric K142stop.

Mutants in the N-Terminal Region Have Little Effect on in Vitro Chaperone Activity. Hsp16.6 can protect Luc from heat-induced aggregation and allows protected Luc to refold in the presence of ATP-dependent chaperones in vitro (24). To test further the relationship of sHsp function in vivo to this in vitro chaperone activity, we determined the effects of the new thermotolerance-defective mutants on substrate protection in vitro. Chaperone activity was tested by heating 1 μM Luc with 24 μM sHsp, then assaying Luc refolding in reticulocyte lysate, a rich source of the Hsp70 chaperone system. Luc protected by WT Hsp16.6 under these conditions is reactivated to nearly 50% of its initial activity, whereas an equal weight of BSA results in 4% Luc activity (Table 1).

Notably, the two groups of mutant sHsps, as defined by oligomeric stability, provided different levels of Luc protection. Mutations that significantly reduced oligomeric stability impaired Hsp16.6 chaperone activity. The least defective mutants of this group, E57G, R81H, and G124S, allowed 15–20% Luc reactivation (Fig. 4A and Table 1). The other mutants were even less effective chaperones, allowing ≤6% of Luc to refold. F102S had the worst activity; it protected Luc less well than an equal weight of BSA. As seen previously (24), mutants that destabilized the oligomer also allowed faster refolding than WT (Table 1).

Fig. 4.

Reactivation of Luc protected by Hsp16.6. Luc (1 μM) was heated at 42°C for 7.5 min with sHsp and then allowed to refold in reticulocyte lysate. Luc activity was compared to unheated samples. (A) Representative refolding time courses are shown after heating with 24 μM sHsp. (B) Concentration-dependent protection of Luc by sHsp, or an equivalent weight of BSA. Averages and standard deviations of maximum reactivation from three experiments are shown. White, gray, and black bars represent 24, 96, and 192 μM sHsp, or 0.4, 1.6, and 3.2 mg/ml BSA, respectively.

L66A, which also fails to protect Luc under these conditions, is more active at higher ratios of sHsp to substrate (24). Similarly, the protection of 1 μM Luc by 96 μM I106N, L130S, or K142stop was improved so that at least 50% of Luc was refolded (Fig. 4B and data not shown). Increasing F102S to 192 μM had only a small effect on its chaperone activity (Fig. 4B).

Despite their loss of function in vivo, at 24 μM the N-terminal Hsp16.6 mutants showed good protection of 1 μM Luc (Fig. 4A, Table 1). Luc activities recovered after protection by L9P (54 ± 3%) and E25K (70 ± 3%) were actually better than WT (49 ± 1%). Q16R was almost as active as WT, allowing 41 ± 2% Luc reactivation. Ability of each of these mutants to protect Luc from insolubilization was further tested at decreasing molar ratios of sHsp monomer to Luc (12, 6, and 3 μM sHsp to 1 μM Luc). L9P and E25K behaved essentially like WT, giving full protection at 6 μM. Q16R was slightly less effective, giving full protection between 6 and 12 μM for 1 μM Luc (Fig. 5, which is published as supporting information on the PNAS web site). Complexes between WT Hsp16.6 or the N-terminal mutant sHsps and Luc, as detected by Western blotting after nondenaturing gel electrophoresis, remained unchanged 24 h after formation, further indicating that WT and these mutant Hsp16.6 proteins interact similarly with Luc (data not shown).

To determine whether the inactivity of these N-terminal sHsp mutants in vivo might be related to some type of instability in vivo, we determined the distribution of WT, L9P, Q16R, and E25K in the soluble and detergent-insoluble fractions of cells after1hat 42°C, followed by 0, 1, or2hat 44°C. These conditions are analogous to the viability assay, but before significant cell death. The majority of each sHsp was soluble at all three time points. Like WT, a fraction of the mutant proteins became less soluble with time at 44°C, although to a smaller extent than WT (Fig. 6, which is published as supporting information on the PNAS web site).

Altogether, the N-terminal mutants are excellent chaperones for the model substrate Luc, and are soluble and stable in vivo, despite their failure to provide thermotolerance in vivo. These data indicate that these standard biochemical chaperone assays do not reflect an essential in vivo activity of Hsp16.6.

Discussion

Although the function of sHsps in vivo has been related to their chaperone activities in vitro, details of their role within the cell remain to be defined. We have taken a genetic approach, unbiased by current models of sHsp chaperone activity, to investigate the mechanism of Hsp16.6 action in vivo. By screening for random mutants of Hsp16.6 that could not provide thermotolerance, we identified 18 missense mutations and one 5-aa C-terminal truncation that disrupt the essential function(s) of Hsp16.6 in vivo. This represents the largest collection of sHsp mutants linked to thermotolerance in vivo. The mutations reduce viability of Synechocystis after heat stress to as little as 0.02%, where +Hsp16.6 and ΔHsp16.6 strains survived to 30 and 0.006%, respectively (Fig. 1B). Significantly, three mutations in the N-terminal domain of Hsp16.6 fail to confer full thermotolerance in vivo, yet still retain oligomeric structure and in vitro chaperone activity. These data point to at least one novel function for the evolutionarily variable sHsp N terminus, and indicate that current biochemical assays do not measure some critical feature(s) of the sHsp mechanism of action in vivo.

Mutants That Impair Oligomerization of Hsp16.6. Of the 10 thermotolerance mutants that were characterized biochemically, all 6 mutations in the α-crystallin domain and the C-terminal truncation reduced both the oligomeric stability and the in vitro chaperone activity of Hsp16.6. Previous biochemical studies with a number of different sHsp family members have shown a strong relationship between oligomerization and chaperone activity (29–31). Furthermore, the correlation between the ability to form oligomers and the ability to confer thermotolerance in vivo has been shown with the L66A and V143A mutants of Hsp16.6, which are dimeric. Intragenic suppressor analyses of these two mutants demonstrated that second-site mutations that restore function in vivo and in vitro also restabilize the oligomer (24, 25). Therefore the new mutants, along with L66A and V143A, appear to define regions of the protein that are involved directly or indirectly in oligomerization and that are also essential for function.

On the basis of similarity of the dimer structures of Methanococcus jannaschii Hsp16.5 (2) and Triticum aestivum Hsp16.9 (3), along with the similar sequence characteristics of Hsp16.6 (Fig. 1 A), it is likely that the Hsp16.6 oligomer is composed of dimers. This possibility is also supported by crosslinking analysis of Hsp16.6 (25). The locations of the seven mutations that destabilize the Hsp16.6 oligomer were estimated by the positions of homologous residues (Fig. 1 A) in the TaHsp16.9 dimer (Fig. 7, which is published as supporting information on the PNAS web site). The L130S and K142stop mutants likely disrupt the same oligomeric contacts between a hydrophobic patch in the α-crystallin domain and the C-terminal extension as do L66A and V143A (2, 3). Other mutations, for example E57G, R81H, and G124S, lie near the interface between monomers, suggesting they could destabilize the dimer. However, even after a heat treatment, the Hsp16.6 mutants we analyzed eluted from SEC as dimers or larger (Fig. 2). In contrast, mutation of Bradyrhizobium japonicum HspH Phe-94, which is equivalent to Phe-102 in SynHsp16.6, disrupts the dimer, making the protein monomeric and inactive (31). It is possible that by requiring Hsp16.6 accumulation in vivo we selected against mutations that disrupt the dimer.

The data also indicate that the β7 strand in the α-crystallin domain is critical for oligomerization and sHsp function. Of the three mutations identified in β7 (G100S, F102S, and I106N), the two proteins we purified were both dimeric. F102S also caused severe activity defects in vivo and in vitro. Another residue on the same face of β7, Arg-104, is homologous to Arg-120 in αB-crystallin. R120G is a dominant mutation in αB-crystallin associated with desmin-related myopathy in humans (17), and it reduces the chaperone activity of αB-crystallin in vitro (18, 19). Consistent with an important role for β7, site-directed mutation R104G in Hsp16.6 reduced the survival of Synechocystis to 1% after an 8-h 44°C treatment without affecting Hsp16.6 accumulation (data not shown).

The seven mutants with reduced oligomeric stability were also impaired in protection of the model substrate Luc (Table 1). Overall, the data are consistent with earlier results showing that mutations that disrupt the Hsp16.6 oligomer have activity defects in both thermotolerance in vivo and protection of a model substrate in vitro (24, 25). However, the effects of mutations on oligomer stability did not always directly correlate with the reduction in sHsp chaperone activity. R81H was ≈4-fold more active than F102S and K142stop in chaperone assays, but all three proteins behaved essentially as dimers (Table 1). R81H was also more active in vivo than F102S and K142stop, indicating its oligomerization defect has a relatively minor impact on sHsp activity. It is possible that while all three of these residues contribute to forming stable sHsp oligomers, Phe-102 and the five C-terminal residues are more essential than Arg-81 for interactions in sHsp/substrate complexes or interactions with other cellular components.

N-Terminal Mutants Act as Chaperones in Vitro but Do Not Support Cellular Thermotolerance. Three mutations of the N terminus, L9P, Q16R, and E25K, showed distinctly different levels of activity between in vivo and in vitro assays for function. These mutants are striking, as they indicate that the chaperone activity of sHsps as measured in vitro does not fully reflect sHsp activity in vivo.

The N terminus is known to play a role in sHsp oligomerization (29, 32, 33), and has also been implicated as a potential site for substrate binding (3, 9, 34). The N-terminal mutations had only small effects on the Hsp16.6 oligomer. L9P was somewhat less stable than WT, whereas at elevated temperatures, E25K and Q16R formed more stable oligomers than WT (Fig. 3). All three N-terminal mutants also protected Luc well in vitro (Fig. 4A). This protection is especially notable for L9P and E25K, which provided better Luc protection than WT, but reduced cell viability relative to WT after heat stress by 100- and 1,000-fold, respectively. We found no evidence for significantly altered accumulation or stability of these three mutant sHsps in vivo that would correlate with their reduced activity (Fig. 6). Although these mutant sHsps retain the ability to protect a model substrate, it is possible that a site crucial for binding cellular substrates in vivo has been affected. This possibility can now be tested, and could lead to identification of critical Hsp16.6 substrates.

Another explanation for the failure of the N-terminal mutants to provide thermotolerance in vivo is that an interaction with a cochaperone has been affected. sHsps and Hsp100/ClpB interact genetically, the deletion of one causing a greater requirement for the other (15, 21, 24). Although our experiments were performed in ΔclpB1 strains, the essential clpB2 gene remains (24), making it possible that the interdependence of Hsp16.6 and ClpB2 has been affected by the N-terminal mutations. Alternatively, interactions with unidentified cochaperones or with lipids could be affected.

Demonstrating the importance of the N-terminal arm for sHsp activity in vivo, independent of oligomerization and in vitro chaperone activity with a model substrate, provides insight into the importance of this domain. Whereas the primary sequence of the sHsp N-terminal arm is highly variable between taxonomic groups, different, but specific, N-terminal motifs have been conserved over 400 million years in diverse sHsps (4, 35). Variation in the N-terminal arm may be key to the many roles proposed for sHsps in vivo. The mutants of the N terminus of Synechocystis Hsp16.6 should help define these roles.