Chapter Three Molecular modelling of the neuroglobin-cytochrome c complex Soft docking simulations of the neuroglobin-cytochrome c complex
Molecular modelling was performed using the Chemera/BiGGER software to simulate complexes of the neuroglobin-cytochrome c interaction. Initial simulations were carried out using wildtype human neuroglobin as the target, and equine cytochrome c as the probe. 5,000 complex structures were generated and ranked according to the parameters described (Chapter 2.6). The vast majority of the complex structures that were highly ranked by global score had cytochrome c targeting one surface around the exposed heme edge of neuroglobin (Figure 12), with only 2 complexes out of the top 100 having cytochrome c targeting alternative surfaces.
The position of the probe molecule in the complexes can be visualised by generating an ensemble of the geometric centres of the probe molecules around the target molecule. The geometric centre of cytochrome c in the majority of the top 100 globally ranked complexes in this simulation form an arc over the E and F helices of neuroglobin (Figure 13). Only a very small proportion of the predicted complexes have cytochrome c positioned outside of this arc elsewhere on the neuroglobin surface (4 out of 500).
In these simulations, the cytochrome c molecule was positioned in a range of different orientations relative to the neuroglobin molecule. This observation is best represented by reversing the orientation of the simulation so that equine cytochrome c is the target and neuroglobin is the probe (the opposite orientation to the previous simulation). In this simulation, neuroglobin populated several major surfaces on cytochrome c in the top 100 globally ranked structures, and a large proportion of the complexes positioned neuroglobin near the loops that surround the heme edge of cytochrome c (Figure 14). Thus, in the neuroglobin-cytochrome c soft docking simulations, cytochrome c is predicted to interact with a specific surface on neuroglobin, but the interaction could take place on several different surfaces of cytochrome c.
Previous studies have shown that a small iron-iron distance (< 16 Å) is critical in electron transfer complexes (Moser, et al., 1992). Thus, the iron-iron distance was measured in the top 100 globally ranked complexes (Figure 15). Many these complexes had iron -iron distances greater than this, with just 24 % having a distance of less than 15 Å.
In order to choose a putative specific complex for the neuroglobin-cytochrome c interaction, the complexes were filtered to an iron-iron distance of 15 Å, and complexes within 2 Å of each other were clustered. The structure of the complex with the highest global score shows very close association between the two heme groups, and an iron-iron distance of 13.4 Å (Figure 16).
In addition to global ranking, the complexes can be ranked by the individual parameters (electrostatics, hydrophobics, sidechains, geometric). This may be useful to see if highly ranked complexes in individual parameters are being averaged out and lost in the global score, which may be important if one parameter is dominant in the specific complex.
In the simulations of the neuroglobin-cytochrome c complexes, the distribution of the geometric centre of the probes around the target do not change significantly if the complexes are ranked by individual parameters compared to the global score. The exception is the hydrophobics parameter, in which the heme edge of cytochrome c is the dominant surface in the highly ranked complexes, and the surface area that forms the interaction decreases significantly on both molecules compared to the complexes ranked by global score (Figure 17). Thus, ranking the complexes by global score reduces the perceived specificity of the cytochrome c interface because the global score takes an average across all of the parameters.
In addition to the soft docking simulations, electric dipoles were calculated for neuroglobin and cytochrome c using Chemera. Both proteins have strong electric dipole, with neuroglobin and cytochrome c having dipoles of 171 and 282 Debye, respectively (Figure 18). Furthermore, the dipoles are aligned such that the heme edge of each protein would be likely to come close to each other during complex formation.
Based on the results from the simulations and electric dipole calculations, the interaction between neuroglobin and cytochrome c is likely to take place on surfaces of the proteins such that the hemes come as close as possible in the complex.
The hypothesised interacting surface of neuroglobin is composed of the E and F helices, and the heme group. The two helices and the connecting loop contain 7 negatively charged aspartic acid and glutamic acid residues (out of 41). By close inspection of the neuroglobin surface, several candidate amino acids can be identified that project out into the interface that could potentially interact with amino acids on cytochrome c (Figure 19, Figure 20).
The hypothesised interacting surface of cytochrome c has 6 positively charged lysine and histidine residues that project into the interface of the putative complex. These could form intermolecular electrostatic interactions with the negatively charged neuroglobin residues to stabilise the complex in the neuroglobin-cytochrome c interface. The putative specific complex with the lowest iron-iron distance (Figure 16) was used to identify several candidate interactions (Table 4, Figure 21).
Other residues projecting out into the interface on neuroglobin are aspartic acid 63 and serine 91; however these have no apparent contacts. In addition aspartic acid 73, threonine 77, asparagine 78, glutamic acid 80, aspartic acid 81 and serine 84 project outwards into the interface, but these are further away from the heme and cannot make contacts with cytochrome in the putative complex structure. Thus, the five most promising candidate neuroglobin residues were investigated further (Table 4).
Soft docking simulations were performed to simulate the interaction between cytochrome c and the neuroglobin proteins with single mutations as above (after energy minimisation using Yasara). An ensemble of the top 100 globally ranked structures showed that there was no significant change in the distribution of cytochrome c molecules around the exposed heme edge of neuroglobin compared to wildtype neuroglobin (Figure 22). However, the simulations for mutant 60 and 60,66,87 neuroglobin, and to a lesser extent, mutant 87 neuroglobin showed the appearance of a second, minor binding interface on the opposite side of the protein to the heme edge. This new binding site is introduced by complexes that are highly ranked by the hydrophobics parameter (Figure 23). Thus, mutation of the glutamic acid 60 residue to lysine may form a new binding site on the surface of neuroglobin for cytochrome c. All of the other mutants have just one major binding site observed under all of the parameters viewed individually
Soft docking simulations of other neuroglobin complexes
Soft docking simulations were performed to investigate if the interaction between neuroglobin and cytochrome c was more or less specific, compared to the interaction with other hypothesised binding partners of neuroglobin. These simulations showed that in a large proportion of the top 100 globally ranked complexes the binding partner formed a complex on the surface over the exposed heme edge of neuroglobin, for all of the binding partners tested (Figure 24, Figure 25).
By global score ranking, neuroglobin does not interact with a single specific surface on any of the binding partners for which simulations were performed. When the rankings are divided into the individual parameters, the distribution of the complexes in the ensembles does not change significantly in any of these simulations (not shown). The exception is the simulation for the AIF-neuroglobin complex which is predicted to form a complex between the specific surfaces around the exposed heme edge of neuroglobin and the surface around Gly 506 on AIF in the top 100 complexes when only the sidechains parameter is ranked (Figure 26).
Other soft docking simulations
To test the ability of the Chemera/BiGGER software to discriminate between specific complex and non-specific structures, several simulations were performed on protein partners that are known to interact and others that are known to not interact with each other.
Myoglobin and cytochrome c are not known to form a physiologically relevant complex. Soft docking of myoglobin and cytochrome c showed that these two proteins do not interact with each other on a specific interface, and in particular the surfaces around the exposed heme edge on both proteins are largely unpopulated by the respective partner in the ensembles (Figure 27). Thus, the interaction between neuroglobin and cytochrome c observed in the previous simulations is not globin-specific, and is not expected to be an artefact caused by globin heme-edge geometry.
The interaction of the well characterised cytochrome c peroxidise (CcP)-cytochrome c complex was simulated. An ensemble of the top 100 ranked complexes based on a global score failed to predict a specific interface for the CcP and cytochrome c interaction (Figure 28).
However, when the ensemble is ranked on electrostatic interactions which are known to drive complex formation in this interaction, a specific interface between CcP and cytochrome c is shown that is similar to previous simulations (Bashir, et al., 2010). Additionally, when known interacting residues (Bashir, et al., 2010) are mutated to an oppositely charged residue (Gly 290 on CcP, or Lys 72 on cytochrome c), the distribution of the ensemble on this surface becomes more disperse.
Summary of main findings
Molecular modelling predicts that cytochrome c binds to a specific surface on neuroglobin, around the exposed heme edge.
Neuroglobin is predicted to bind to a specific surface over the exposed heme edge of cytochrome c in complexes ranked highly by the hydrophobics parameter.
Neuroglobin and cytochrome c have very strong charge dipoles.
A putative specific complex with an iron-iron distance of less than 15 Å was identified for the neuroglobin cytochrome c interaction.
Key residues on the interface were predicted as Glu 60, Arg 66, Lys 67, Glu 87 and Arg 94 of neuroglobin.
Molecular modelling of complex formation between these mutant proteins and cytochrome c show changes in the distribution of cytochrome c over the surface of neuroglobin for some mutants.
Molecular modelling predicts that other hypothesised neuroglobin binding partners may bind to the same surface as indicated in the neuroglobin-cytochrome c complex.
Molecular modelling can be applied to other known complexes to identify known binding sites.
Chapter Four Production of recombinant wildtype neuroglobin
Expression and purification of wildtype neuroglobin
Human wildtype neuroglobin was produced in E. coli BL21(DE3) grown for 2 days at 28 °C under leaky
expression. The bacterial cells were reddish in colour compared to untransformed cells which are pale brown. The reddish colour was attributed to the presence of recombinant neuroglobin. The yield of recombinant neuroglobin in E. coli cells was approximately 60-80 mg/L of culture as assessed by difference spectrum (Chapter 2).
After sonication and centrifugation of the cell lysate, a two-step ammonium sulphate precipitation was performed. After 20 % saturated ammonium sulphate, the supernatant remained a reddish brown colour, indicating that neuroglobin was still soluble. However, after 60 % saturated ammonium sulphate, neuroglobin precipitated out of solution as a red precipitate, leaving the supernatant a milky colour. The red neuroglobin precipitate was rapidly resolubilised upon addition of lysis buffer.
Neuroglobin is highly stable at high temperatures, with a melting temperature of 100 °C (Hamdane, et al., 2005), thus the neuroglobin solution was heat-treated after dialysis to remove contaminants that are denatured by high temperatures. The heat-treated neuroglobin sample was then further purified by ion-exchange chromatography (Figure 29).
Chapter One: Introduction
1.1 Introduction to the globins
Chapter Two: Materials and Methods
2.2 Buffers used in this study
2.3 Media used in this study
2.4 General methods
2.5 Cloning and expression of human neuroglobin
2.7 Neuroglobin mutagenesis
2.8 Isothermal Titration Calorimetry (ITC)
2.9 Surface Plasmon Resonance
2.10 Caspase activity in a cell-free system
2.11 Apoptosome assembly
Chapter Three: Molecular modelling of the neuroglobin-cytochrome c complex
3.1 Soft docking simulations of the neuroglobin-cytochrome c complex
3.2 Soft docking simulations of other neuroglobin complexes
3.3 Other soft docking simulations
3.4 Summary of main findings
Chapter Four: Production of recombinant wildtype neuroglobin
4.1 Expression and purification of wildtype neuroglobin
4.2 Characterisation of purified recombinant wildtype neuroglobin protein
Chapter Five: Characterisation of mutant neuroglobin proteins
5.1 Purification of mutant neuroglobin proteins
5.2 Characterisation of mutant neuroglobin proteins Spectral properties of mutant neuroglobin proteins
5.3 Molecular dynamics of the mutant neuroglobin proteins
5.4 Secondary structure of neuroglobin proteins Secondary structure composition of apo-neuroglobin
5.5 The three-dimensional structure of cysteine-free mutant neuroglobin
5.6 Summary of main findings
Chapter Six: Characterisation of the neuroglobin-cytochrome c interaction
6.1 Isothermal Titration Calorimetry
6.2 Surface Plasmon Resonance
6.3 Summary of main findings
Chapter Seven: Investigations of the downstream effects of the neuroglobin-cytochrome c interaction
7.1. Cytochrome c-mediated caspase-9 activation in cytosolic extracts
7.2 Direct assay of apoptosome assembly in vitro
7.3 Summary of main findings
Chapter Eight: Discussion
8.1 The neuroglobin-cytochrome c interaction
8.2 The redox state of cytochrome c regulates apoptosome formation
8.3 Concluding remarks
GET THE COMPLETE PROJECT
A Role for Neuroglobin in the Inhibition of Cytochrome c-Mediated Apoptosome Formation