The competitive interface for LOV-LOV dimerization and interdomain

In many sensory proteins with the conserved PAS fold, both, a homo- or hetero- dimerization state and also ligand binding are important processes to affect inter/intra-protein interactions, which, in turn, regulate protein functions15. Besides the light induced conformational changes in the LOV core, currently the dimerization states of different LOV domains are controversially under debate, especially when we consider those LOV domains in the signal transmission processes that affect their downstream partners.

As described in chapter 2.1, phot1-LOV1 tends to form a dimer, whereas phot1- LOV2 prefers to stay as a monomer16. Combined with the fact that phot1-LOV2 has a higher quantum yield in its photo-cycle than phot1-LOV117, it is then plausible that LOV1 might be responsible for dimerization and LOV2 might act as the predominant light sensing domain, providing a possible explanation for the organization of two tandem LOV domains in phototropin16,17. In contrast with the result mentioned above, the small angle X-ray scattering experiments showed that phot1-LOV2 has a tendency to dimerize, whereas only phot2-LOV2 is monomeric. Recently, with the help of time-resolved techniques such as pulsed thermal grating, a transient dimerization state of phot1-LOV2 (including the N- terminal helical cap and the Ja-linker) with a time constant 300 μs upon light activation has been detected18. The authors from the same group suggested that the presence of the Ja-linker inhibited the dimer formation in dark state. A light induced unfolding of the LOV2- Jα-linker helical construct with a time constant of 1 ms has also been observed18,19. A further transient thermal lensing (TrL) experiments demonstrated that this unfolding is caused by a dissociation of the Jα-linker from the LOV core, and the time constant of this dissociation has been

determined as 300 μs20. The similar time constants between the transient

dimerization state and the dissociation of Jα-linker might imply that the blue-light induced conformational changes in the LOV core can cause unfolding of the Jα-

linker and affect the possible hydrophobic interactions between Jα-linker and the LOV core21.

The sequence identity level among different LOV domains is quite high, but there are still some differences at several crucial positions. This may account for their diverse signal transmission mechanisms as well as for the different oligomerization modes. As discussed in Chapter 2.1, the central β-scaffold might be involved in both the LOV-LOV dimerization and the signal transmission from the LOV core to the effector domain. This has partially been confirmed by the crystal structure of a YtvA-LOV dimer, in which YtvA also displayed an extended β-scaffold (Hβ and Iβ) with a more hydrophobic patch (Figure 5-4)21

. On the other hand it was previously noticed that in YtvA the Jα-helix has more polar residues than other LOV domains and that this helix is connected to the LOV core by a shorter loop21-23. It is therefore conceivable that the Jα-helix can not readily cover the central β-scaffold and the hydrophobic residues on the outer β-sheets are exposed for the inter/intra-protein interactions.

Figure 5-4: Sequence alignment of YtvA-LOV domain and other LOV domains with known 3-D

structure. The sequence of α-helices is highlighted in yellow boxes and β-strands in blue boxes. Residues in bold red are involved in homodimeric contacts in the crystal structure; residues in bold blue are involved in intra-protein interactions between the LOV core and C-terminal or N- terminal extensions24-28. Abbreviations: B. subtilis (B. s.), Avena sativa (A. s., oat), Adiantum capillus-veneris (A. c., fern), Chlamydomonas reinhardtii (C. r.), Neurospora crassa (N. c.)21.

In the full length YtvA protein, as shown in Chapter 2.1, the protein prefers to stay as a monomer at least in vitro. A possible interpretation might be that the Jα- linker together with the C-terminal STAS domain acts as a competitor for the LOV-LOV dimerization by blocking the exposed β-sheets. Thus, the dissociation or unfolding of the Jα-helix, caused by conformational changes in the LOV core, might make the central β-scaffold accessible for the interactions with the STAS domain. However, unfolding of the Jα-helix could not be observed through the CD experiments as shown in Chapter 2.1. In the crystal structure21 the Jα-helix points to an opposite direction rather than being packed underneath the LOV- core, and also the loss of helical content could not be clearly revealed either when applying circular dichroism measurements21.

In summary, the available data from diverse experiments suggest different mechanisms for the signal transmission from the FMN-binding pocket to effector domains, as discussed in Chapter 1.5.2. On the other hand, the central β-scaffold is probably the common interface for different signal transduction processes. In YtvA from B. subtilis, the Jα-helical linker may also play an important role to regulate the possible interaction between the central β-scaffold of the LOV domain and the STAS domain. Thus, a further mutational study concerning several key positions in both the central β-scaffold and the Jα-linker will provide more experimental evidences for the assumption mentioned above. This is discussed in Chapter 2.3 and Chapter 5.1.3. Before the mutational study is presented, a crystal structure of a similar YtvA-LOV domain from Bacillus