Our most striking findings are that wild type Tsp36, depending on the re- duction status, forms dimers or tetram- ers (Figures 3 and 4), the latter being stabilized by a disulfide bond (Figure 2), and that the tetrameric form is a better chaperone than the dimer (Figure 6B). A maximum complex size of four subu- nits, or eight α-crystallin domains in the case of Tsp36, is exceptional amongst sHsps. Only the C. elegans Hsp12 pro-
teins are known to form at most tetram- ers (Leroux et al. 1997; Kokke et al. 1998; van Montfort et al. 2002). Other sHsps that are found as dimers and/or
tetramers are in equilibrium with larg- er oligomers, like 470 kDa complex- es for rat Hsp20 (van de Klundert et al. 1998), 400 kDa for murine Hsp25 (Ehrnsperger et al. 1999), and 550 kDa for S. cerevisiae Hsp26 (Haslbeck et al.
1999).
The fact that a disulfide can form between the first α-crystallin domains of two Tsp36 monomers indicates that these domains are in close proximity in the tetrameric configuration. Studying interactions between the separate α- crystallin domains of Tsp36 in the yeast two-hybrid system, we also observed in- teractions between first domains, as well as between first and second domains, but not between second domains (G.K. and W.B., unpublished data). The find- ing that even after reoxidation a large amount of non-crosslinked monomers remains (Figure 2, lane 4) indicates that not all cysteines in the native tetramer complex (Figure 3) can be readily in- volved in disulfide formation, and that other residues of the first α-crystallin domains are also important for tetramer formation.
Intersubunit disulfide bridges have also been found between the single cysteines present in human and rat Hsp20 (Kato et al. 1994; van de Klundert et al. 1998) and in murine Hsp25 (Zavialov et al. 1998). Importantly, in these cases it has been shown that the disulfide bridges can occur intracellu- larly. In contrast to Tsp36, the oligomer size of Hsp25 is not affected by reduc- tion. Also, for human Hsp27 there is evidence that its cysteine is subjected to reversible oxidation by S-thiolation (Eaton et al. 2002). In other stress-relat- ed proteins, i.c. the oxidative stress tran- scription factors Yap1 and OxyR (Zheng
et al. 1998; Delaunay et al. 2002), the molecular chaperone Hsp33 (Jakob et al. 1999) and the heat shock factor 1 (Ahn and Thiele 2003), the forma- tion of disulfide bonds induced by oxi- dative stress appears to be a strategy to adjust protein activity. One wonders whether the occurrence of reversible di- sulfide formation in various sHsps might be a functionally important feature, too. Since the position of the cysteine resi- dues is not conserved between the dif- ferent sHsps, such a putative function might be specific for every sHsp.
A C-terminal extension with an in- tact I-x-I/V motif is thought to play an important role in the oligomerization of sHsps, both within and between dimers (Kim et al. 1998a; van Montfort et al. 2001; van Montfort et al. 2002), and may also be involved in the regulation of substrate binding by interacting with a hydrophobic groove in the α-crystallin domain (van Montfort et al. 2001). It is therefore noteworthy that in Tsp36, which does form dimers and tetramers, a C-terminal extension is lacking. It might be possible that in Tsp36 the con- necting peptide between the two do- mains takes over the function of the C-terminal extension in this respect, considering that the connecting peptide is predicted to contain a β-strand (“β10” in Figure 1) which resembles in se- quence and location - relative to the first domain - the β10 strand in the extension of other sHsps (de Jong et al. 1998; van Montfort et al. 2002; Studer et al. 2002). The fact that the sequence following this β10-like strand comprises the im- munodominant epitope in S. mansoni
p40 (Figure 1, positions 236-248) (Hernandez and Stadecker 1999) may suggest that this region is surface ex-
posed and available for intersubunit contacts in Tsp36.
It is generally agreed that subunit exchange, rearrangements and desta- bilization of the quaternary structure are essential for effective binding of non-na- tive substrate proteins by sHsps (Bova et al. 2002; Sobott et al. 2002; Putilina et al. 2003). Our finding that the tetra- meric form of Tsp36 is a better chaper- one than the dimeric form indicates that the potential of the tetramer to dissoci- ate reversibly into dimers enhances the chaperoning capacity. The fact that the equilibrium between dimers and tetram- ers is sensitive for temperature, and probably for other conditions such as pH and concentration as well, makes it suitable for modulating chaperone- like activity. It is interesting to note that the presence of disulfide-bonded mon- omers is apparently required to enable the formation and stabilization of tetram- ers, yet making it a better chaperone.
It is clear that Tsp36 differs in im- portant aspects from the other known sHsps, notably concerning its redox state dependant complex size in rela- tion to its chaperone-like activity. Tsp36 is therefore an interesting tool for broad- ening and improving our understanding of the functional and structural diversity of the sHsps.
Acknowledgements
We thank L.M. González for his con- tribution to constructing the T. saginata
genomic library, J.L.P. Benesch for assistence with mass spectrom- etry analysis, and N.H. Lubsen for useful comments. Part of the work was supported by grants EU-INCO (IC18CT96002) and ISCIII (1114/03).