Final Report Summary - VDAC2 (Investigation of the voltage-dependent anion channel 2 by solid-state NMR and molecular dynamic simulations)
Voltage-dependent anion channels (VDACs) are mitochondrial porins (pore-forming proteins) found in the mitochondrial outer membrane (MOM) of eukaryotes. The first mitochondrial voltage-dependent anion channel (VDAC)-porin was discovered by Schein, Colombini, and Finkelstein in 1976. To date, multiple VDAC isoforms have been identified in a variety of organisms, including yeast, plants, mouse, and humans. In mammals, three VDAC isoforms (VDAC1, VDAC2, and VDAC3) have been identified, with variable expression in all tissues. The fundamental properties of all VDACs in all eukaryotic kingdoms are highly preserved although some studies revealed important differences in the regulatory functions within the different cell types. These studies strongly indicate functional specialisation rather than redundancy between the three VDAC isoforms. For example, biochemical studies of the mammalian VDACs have revealed that all three isoforms have similar channel-forming activity, and each can compensate for deficiency of the others. However, genetic studies have also shown that VDAC2 knockout mice are embryonic lethal, whereas both VDAC1 and VDAC3 knockout mice are viable, which might suggest that VDAC2 has some different functions from VDAC1 and VDAC3. When reconstituted into planar lipid membranes, a VDAC monomer forms an aqueous pore of 2.5 - 3 nm in diameter.
Mitochondrial membrane proteins play a crucial role in cell apoptosis (suicide of the cell), mechanism which is necessary for normal functioning of the cell, but is absent in cancerous cells. Among mitochondrial proteins also VDAC protein family was found to have influence in apoptosis activation mechanisms. Better understanding of these mechanisms is needed for anti-cancer drug development as the apoptosis in cancerous cells is either hampered or sizably stopped. The direct involvement of VDAC2 in anti-cancerous activity was demonstrated by Yagoda et al (N. Yagoda et al, Nature 2007, 447, 864-868).
Amongst the three mammalian VDACs, VDAC1, the most abundant isoform, was continuously investigated in the past decades. Only in the last few years biological data on VDAC2 became available and it was shown that it possesses a different functionality than VDAC1. However, so far neither extensive experimental data nor computational investigations have been made which could provide structural information on the VDAC2 isoform.
Our multidisciplinary investigation with molecular dynamics (MD) and solid-state nuclear magnetic resonance (ss-NMR) was focused on structural differences between VDAC2 and VDAC1 at a molecular level and functional differences which arise from the different structures. The main interest of this project was the determination of the structure of the N-terminal part of VDAC2 and the study of its dynamics in a lipid bilayer. Although the N-teminal part of the VDAC2 is 11 amino acids longer than of VDAC1, the part where the alpha-helix is expected in VDAC2 has more than 90 % similarity with VDAC1 so we expect that the N-terminal part of VDAC2 will be similar to the N-terminal part of VDAC1 due to this high-sequence similarity. The importance of the N-terminal part of VDACs was demonstrated with deletion experiments, where after deletion of the first 20 to 30 amino-acids, VDAC protein lost its voltage gating ability and ability to incorporate in the lipid bilayer.
For a detailed characterisation of folded state, we prepared VDAC2 samples with various isotope labelling schemes, uniformly (13C, 15N), reverse labelling (where few amino acids were added in natural abundance (12C and 13N) and forward labelling (where few amino acids were added with 13C, 15N labelling). VDAC2 molecules were prepared using the same preparation scheme as for VDAC1. In every preparation VDAC2 molecule was reconstituted in DMPC lipid bilayer environment ensuring the investigation of VDAC2 in nature like environment.
The first proton driven spin diffusion (PDSD) ssNMR spectra of VDAC2 differed from VDAC1 spectra in a way which we did not initially expected. Although VDAC2 and VDAC1 have more than 75 % sequence identity there was a clear difference in their spectra. Where VDAC1 showed isolated peaks which were assigned to amino acids from N-terminal part and from a specific part of the beta-barrel, for full length VDAC2 we did not see so many isolated peaks. This lack of isolated peaks and overall peak broadening of VDAC2 could be due to more dynamic nature of VDAC2 compared to VDAC1.
To be able to better tackle the broad spectra of VDAC2, we have prepared and measured VDAC2 samples as isotope diluted systems where we used reverse (F, K, L and V amino acids were not labelled) or forward (only M, A, L and V amino acids were labelled) labelling schemes. Unfortunately this approach did not decrease overall broad spectra as some of the amino-acids appear in the same part of the spectra. This result prompted us to use cysteine, isoleucine, phenylalanine and proline (CIFP) amino-acids in forward labelling scheme as they not only are a part of N-terminal region but also do not share big parts of the same spectral space.
To test whether we have a functional form of wild-type VDAC2 (WT-VDAC2) and to see the deletion effect of delta31-VDAC2 (where the first 31 amino-acid was deleted from the sequence) we performed an electrophysiological study of VDAC2 (WT-VDAC2) and of deletion mutant (delta31-VDAC2).
Parallel to the ssNMR experiments, MD simulations of full VDAC2 in DMPC lipid bilayer was simulated where starting structure was a VDAC2 homology model (where mouse VDAC1 was used as the template).
Our results showed that there must be more than one open (anion gating) state of VDAC2 in contrary to VDAC1 which has only one open state. This finding was also suggested by Yu lab. As with ss-NMR study we investigated only the open state(s) the resulted spectra was broad but still comprising the overall beta-barrel features found for VDAC1 isoform. Unfortunately till now we did not manage to separate the different open states of VDAC2 which made the analysis of fully labelled VDAC2 very complicated. We managed though with the help of the CIFP forward labelled sample to show that VDAC2 shares same fold in the N-terminal region as VDAC1.
As for the MD simulations we were using as starting structure the homology model of VDAC2 based on mouse VDAC1 structure, we sampled only one open state of VDAC2, the one similar to VDAC1 open state. That we did not get any significant structural changes during 1-µs MD simulation we could conclude that these two open states do not co-exist in steady equilibrium.
Mitochondrial membrane proteins play a crucial role in cell apoptosis (suicide of the cell), mechanism which is necessary for normal functioning of the cell, but is absent in cancerous cells. Among mitochondrial proteins also VDAC protein family was found to have influence in apoptosis activation mechanisms. Better understanding of these mechanisms is needed for anti-cancer drug development as the apoptosis in cancerous cells is either hampered or sizably stopped. The direct involvement of VDAC2 in anti-cancerous activity was demonstrated by Yagoda et al (N. Yagoda et al, Nature 2007, 447, 864-868).
Amongst the three mammalian VDACs, VDAC1, the most abundant isoform, was continuously investigated in the past decades. Only in the last few years biological data on VDAC2 became available and it was shown that it possesses a different functionality than VDAC1. However, so far neither extensive experimental data nor computational investigations have been made which could provide structural information on the VDAC2 isoform.
Our multidisciplinary investigation with molecular dynamics (MD) and solid-state nuclear magnetic resonance (ss-NMR) was focused on structural differences between VDAC2 and VDAC1 at a molecular level and functional differences which arise from the different structures. The main interest of this project was the determination of the structure of the N-terminal part of VDAC2 and the study of its dynamics in a lipid bilayer. Although the N-teminal part of the VDAC2 is 11 amino acids longer than of VDAC1, the part where the alpha-helix is expected in VDAC2 has more than 90 % similarity with VDAC1 so we expect that the N-terminal part of VDAC2 will be similar to the N-terminal part of VDAC1 due to this high-sequence similarity. The importance of the N-terminal part of VDACs was demonstrated with deletion experiments, where after deletion of the first 20 to 30 amino-acids, VDAC protein lost its voltage gating ability and ability to incorporate in the lipid bilayer.
For a detailed characterisation of folded state, we prepared VDAC2 samples with various isotope labelling schemes, uniformly (13C, 15N), reverse labelling (where few amino acids were added in natural abundance (12C and 13N) and forward labelling (where few amino acids were added with 13C, 15N labelling). VDAC2 molecules were prepared using the same preparation scheme as for VDAC1. In every preparation VDAC2 molecule was reconstituted in DMPC lipid bilayer environment ensuring the investigation of VDAC2 in nature like environment.
The first proton driven spin diffusion (PDSD) ssNMR spectra of VDAC2 differed from VDAC1 spectra in a way which we did not initially expected. Although VDAC2 and VDAC1 have more than 75 % sequence identity there was a clear difference in their spectra. Where VDAC1 showed isolated peaks which were assigned to amino acids from N-terminal part and from a specific part of the beta-barrel, for full length VDAC2 we did not see so many isolated peaks. This lack of isolated peaks and overall peak broadening of VDAC2 could be due to more dynamic nature of VDAC2 compared to VDAC1.
To be able to better tackle the broad spectra of VDAC2, we have prepared and measured VDAC2 samples as isotope diluted systems where we used reverse (F, K, L and V amino acids were not labelled) or forward (only M, A, L and V amino acids were labelled) labelling schemes. Unfortunately this approach did not decrease overall broad spectra as some of the amino-acids appear in the same part of the spectra. This result prompted us to use cysteine, isoleucine, phenylalanine and proline (CIFP) amino-acids in forward labelling scheme as they not only are a part of N-terminal region but also do not share big parts of the same spectral space.
To test whether we have a functional form of wild-type VDAC2 (WT-VDAC2) and to see the deletion effect of delta31-VDAC2 (where the first 31 amino-acid was deleted from the sequence) we performed an electrophysiological study of VDAC2 (WT-VDAC2) and of deletion mutant (delta31-VDAC2).
Parallel to the ssNMR experiments, MD simulations of full VDAC2 in DMPC lipid bilayer was simulated where starting structure was a VDAC2 homology model (where mouse VDAC1 was used as the template).
Our results showed that there must be more than one open (anion gating) state of VDAC2 in contrary to VDAC1 which has only one open state. This finding was also suggested by Yu lab. As with ss-NMR study we investigated only the open state(s) the resulted spectra was broad but still comprising the overall beta-barrel features found for VDAC1 isoform. Unfortunately till now we did not manage to separate the different open states of VDAC2 which made the analysis of fully labelled VDAC2 very complicated. We managed though with the help of the CIFP forward labelled sample to show that VDAC2 shares same fold in the N-terminal region as VDAC1.
As for the MD simulations we were using as starting structure the homology model of VDAC2 based on mouse VDAC1 structure, we sampled only one open state of VDAC2, the one similar to VDAC1 open state. That we did not get any significant structural changes during 1-µs MD simulation we could conclude that these two open states do not co-exist in steady equilibrium.