Why?

LovTAP-VP16 is a new fusion protein, designed by Nicolas Gobet in 2010 at EPFL. It has never been characterized or expressed before. The LOV2-TrpR part (LovTAP) has already been characterized by its author, Strickland et al (2008) and the 2009 EPFL iGEM team. Therefore, our main concerns are the addition of a NLS and a VP16 activation domain. The main questions were:

• Will the VP16 activator be placed in a position that allows it to bind the transcription cofactors?
• Will the VP16 prevent LovTAP-VP16 from dimerizing?

What?

We built LovTAP-VP16 by putting together the different parts, for which X-ray or NMR structures have been published.

Building LovTAP-VP16

Fig. 1: A reconstruction of how LovTAP-VP16 might look like in the active state, binding a Trp operator. Red: TrpR domain, blue: LOV2 domain, yellow: Nuclear localization signal, green: VP16 domain

According to the plasmid designed by Nicolas Gobet, LovTAP-VP16 consists of the following domains (from N to C terminal):

• Residues 456 to 490 of the Herpes virus VP16 protein. It corresponds to one of it's activation domains.
• Nuclear Localization Sequence (NLS): PKKKRKV
• Residues 401 to 543 of phot1 LOV2 from Avena sativa.
• Residues 22 to 106 of Trp Repressor from E. coli.

In order to build the molecule, we used PyMOL. All of the genetic elements in LovTAP-VP16, except for the NLS, were taken from the RCSB Protein Data Bank. The TrpR dimer and the DNA in the binding position were taken from the file 1TRR, the LOV2 domain from 2V0U and VP16 from 2K2U.

To fuse the TrpR and LOV2 domains, we followed the instructions given by Strickland et al (2008). Since the fusion happens at residue 543 of LOV2 and residue 22 of TrpR, we applied PyMOL's command "align" to residues 542, 543 and 544 of LOV2 and 21, 22 and 23 of TrpR. The resulting pdb files were then truncated and fused into a single file with a text editor. Back in PyMOL, the NLS sequence was appended to the LOV2 domain. To finish, the VP16 domain was positioned by aligning its C terminus with the N terminus of LovTAP-NLS, and then fused by applying the "fuse" command to the same 2 atoms. In order to illustrate the photoinduced conformation of LovTAP-VP16 we rotated the LOV2 domain, leaving the α-helix in its place, from residue 522, until the clashes with the TrpR on the other monomer disappeared. The resulting activated dimer, bound to DNA, can be seen in fig. 1.

Light and dark conformations

In fig. 1, we have seen how a LovTAP-VP16 dimer in the light-induced state can be compatible with a high DNA affinity conformation.

What happens in the dark? As Strickland et al proposed, we can see, in fig. 2, how maintaining the high DNA affinity conformation while the LOV2 domains are bound to their respective Jα-helices would imply very important steric clashes. Therefore the conformation will change to another with a lower DNA affinity.

Fig. 2: In the dark conformation, with the LOV2 domain bound to its Jα-helix, there would be important steric clashes between the LOV2 domain in one monomer (dark blue) and the TrpR in the other monomer (light red).

Fig. 3: A - If both monomers are in the light activated state, the TrpR domains can adopt the right conformation to bind DNA. B - If only one monomer is active, the conformation of the TrpR domains has to change to avoid the clashes on one side. This conformation will have to be further modified if both monomers get deactivated.

What if only one of the monomers of the dimer are activated? Rotating and moving one of them allows them to find a conformation free of steric clashes, but no longer has the right shape to bind DNA. However, if the active monomer gets deactivated, there will again appear new clashes. This means the dimeric conformation will change again, probably further reducing the DNA binding affinity.

Conclusions

We can see that the α-helical subdomain (residues 465 to 490 according to Hendrik et al) of VP16 that is supposed to bind TFIIB, one of the cofactors that VP16 is thought to bind to trigger the activation cascade, seems to be free of any sort of steric constraints. This allowed us to decide that an additional linker between the LOV2 domain and the VP16 domain is not required.

On the other hand, the geometry of the dimer suggests that a half activated LovTAP-VP16 would have a binding affinity somewhere between an activated and a deactivated one. This can be useful to model the average binding affinity of LovTAP-VP16 given the average proportion of photoactive monomers.