Oestrogen Beta (1qkm) Receptor Protein Research Project

Introduction

The protein that is used for the modelling process in this research project is an Oestrogen Beta (1QKM) receptor. The structure of 1QKM is shown in figure 1. and from this we can see the more complex tertiary and quaternary structures. The oestrogen receptor is a ligand-inducible intracellular transcri ption factor that mediates most of the biological effects of oestrogens at the level of gene regulation [1]. 1QKM has a similar structure to that of the Oestrogen Alpha receptor. Each ligand interacts with a unique set of residues within the hormone-binding cavity of the 1QKM receptor and induces a distinct positioning in the AF-2 helix [2].

Genistein is a phytoestrogen, also known as an Isoflavoid which is a plant derived xenoestrogen that is not produced with in the endocrine system. It inhibits topoisomerase-II and protein-tyrosine kinase activity and is used as a treatment for breast cancer. At present Genistein is in the process of clinical trials as a treatment for prostate cancer [3]. There are a few different isoflavones including: Genistein, Biochanin A, Formononetin, and Daidzein. Isoflavones are almost solely produced by plants from the bean family, such as the soy plant. Biochanin A is an insulin mimetic flavonoid that agonises the nuclear receptor which is currently the pharmacological target for the treatment of type 2 diabetes [4]. Daidzein, like genistein, can be found in the soy plant. In some humans it can be metabolised to S-equol [5]. S-equol is beneficial in the treatment of menopause, and the maintenance of breast, bone, cardiovascular and prostate health. [6]

Ligand Calculated Energy (Kcalmol-1)

ZINC01383169 -21.8202

ZINC00153513 -21.3560

ZINC00266424 -20.8599

ZINC00169867 -20.5939

ZINC00158691 -20.4670

Genistein -19.1527

Using Chemcomp MOETM a thousand ligands were run through a docking process into the 1QKM receptor. The energies of each ligand were calculated by the software, from that the five most suitable ligands were further analysed. Tools such as electrostatic mapping, ligand interactions and interaction potential mapping were used. Being able to use software to model ligands like this allows the analysis of a lot more ligands in much greater detail than would be physically possible in a laboratory in the same period of time. The modelling software also drastically cuts the price of this project by eliminating the need for lots of expensive equipment and man hours. On top of this. It also eliminates human error to an extent and increases accuracy.

The main aims of this project were to use Chemcomp’s MOETM to screen part of a database of drug-like compounds, in order to identify potential candidate molecules and use these to try to generate a new drug, which would not be metabolised as easily as genistein for the treatment of breast cancer.

Results (Allocated Ligands)

In order to find the most suitable ligand, the 1000 potential candidate compounds that were allocated to myself (12001 – 13000) were docked and the most favourable energies were recorded, see table 1. For comparison, genistein was also docked and it’s most favourable energy was recorded. As you can see from table 1 all of the compounds from the data base were more favourable than genistein itself.

By using the ligand interaction tool I produced a 2D image of the molecular interactions between the compounds and the protein’s receptor pocket, see figures 3, 4, 5, 6, 7 and 8. Using this tool allowed me to observe which functional groups interacted well with the amino acid residues in the 1QKM receptor site, this was useful when thinking about how to alter the candidate ligands to make them more suitable.

For ZINC01383169 the 2D model shows a hydrogen bond between the centre of the five member ring and Phe 356, see figure 3. There is also an interaction between the hydroxy group and Glu 305.

Figure 4 for compound ZINC00153513 shows an interaction between Glu 305 and NH2. This is the only interaction displayed in this model.

The 2D model of ZINC00266424 exhibits no interactions between the ligand and the receptor pocket, see figure 5.

ZINC00169867 appears to have only the one interaction between the ligands own OH group and the amino acid residue Glu 305, see figure 6.

From figure 7 we can observe no interactions between ligand ZINC00158691 and the 1QKM receptor.

Genistein displays 3 interactions with the protein pocket, see figure 8. There are two hydrogen bonds between a six member carbon ring and Phe 256, and a six member carbon substituted ring and Leu 295. There is also an interaction between a hydroxy group and Glu 305.

To gain a better understanding of the electrophilicity of each of the candidate compounds I produced maps of both the receptor pocket and the ligands. This made it possible for me to see the areas of the pocket that were empty, hydrophilic (red) and hydrophobic (blue). This information was necessary for the later step of ligand alteration, the electrophilicity of the area dictated what functional groups could be added to certain points of the ligands. For example if an area was hydrophilic and was calling for a lone pair, a nitrogenous or hydroxy group could be added to the ligand at that point.

The mapping for compound ZINC01383169, see figure 9, shows that there is quite a bit of empty space with in the receptor pocket. There is less space available for ZINC01383169, shown in figure 9, compared to genistein in figure 14. This suggests a better fit for ZINC01383169, this theory is backed up by the energies calculated from the docking process, see table 1, where ZINC01383169 has an energy of -21.8202 Kcalmol-1 and Genistein has an energy of -19.1527 Kcalmol-1. This ligand shows the best initial improvement on genistein without any alterations being made.

For compound ZINC00153513 the mapping indicates a good fit within the pocket and a relatively even balance between hydrophilic and hydrophobic areas, see figure 11. There appears to be more space in the map of genistein, see figure 14, than in the mappings of ZINC00153513 and the space seems more evenly distributed in the ligand ZINC00153513. The difference in available space points towards a better fit with ZINC00153513 than for genistein. This again is supported by the results from the calculated energies, where ZINC00153513 has an energy of -21.3560 Kcalmol-1 and genistein has a calculated energy of -19.1527 Kcalmol-1.

In the case of ZINC00266424, see figure 11, there is a considerable area of empty space surrounding the ligand itself, but it still has a better calculated energy than genistein, -20.8599 Kcalmol-1 to genistein’s -19.1527 Kcalmol-1.

ZINC00169867’s mapping in figure 13 shows a lot less empty space within the 1QKM receptor pocket than genistein. This is once again reflected in the calculated energies in table 1, where ZINC00169867 shows a docked energy of -20.5939 Kcalmol-1 compared to genistein’s -19.1527 Kcalmol-1.

Genistein has quite a bit more empty space in its mapping, refer to figure 14, than ZINC00158691, see figure 12. From the trend being produced by these observations I would say that the tighter the fit of the ligand within the receptor pocket the better the calculated energy is when it is docked.

The lipophilic mappings, shown in figures 16, 17, 18, 19, 20, 21, for the candidate compounds and for genistein appear to present a link with empty pocket space and better energies. For these mappings: pink represents hydrophilic areas, white represents neutral areas and green represents lipophilic areas.

For figure 16, ZINC01383169 the colours of the pocket and the ligand appear to be pretty well matched up, i.e. there appears to be pink areas on the pocket’s map where there are pink areas on the ligand’s map.

For the rest of the ligands (ZINC00153513, ZINC00266424, ZINC00169867 and ZINC00158691) there appears to be a similar pattern where the lipophilic mappings are concerned, see figures 17, 18, 19, 20, 21. However, there does not appear to be a direct relationship between lipophilicity and the calculated docking energies in table 1.

Results (Designed Ligands)

For compound ZINC01383169 we can see that there is a hydrophilic area towards the bottom left hand corner of figure 22 this corresponds with an area of blue gridding, indicating the possibility of a functional group with a lone pair being added at this point of the ligand. It is important however, that there aren’t too many functional groups with lone pairs added to the ligand. Even though

a nitrogenous or hydroxy group could make the ligand more compatible with the receptor, it could increase the chances of the ligand being metabolised by glucuronidation like genistein, see figure 15. In this case I decided to add an OH group, see figure 23, to the carbon in the specified blue gridded area depicted in figure 22. I then docked the new ligand and collected the results, see table 2. The calculated energy had been improved by the addition of the new functional group from -21.8202 Kcalmol-1 to -22.4760 Kcalmol-1.

I then analysed this new ligand further by using the ligand interaction tool to produce a 2D model, see figure 24. The model does not show any new interactions or any loss of interactions between the ligand and the amino acid residues.

By producing an electrostatic map of the ligand and receptor pocket, it is possible to see the difference between the original ZINC01383169’s and the altered version’s electrophilicity and the empty space surrounding the ligands.

The addition of the OH group has induced a much better electrostatic mapping which is shown in figure 25. The ligand and receptor maps now correlate in hydrophilic and hydrophobic areas, red with red and blue with blue. This is a good improvement to the complex and appears to have had a direct effect on the energy as it has reduced the energy from -21.8202 to-22.4760, see table 2.

Original molecule Alteration New calculated energy (Kcalmol-1) Original energy (Kcalmol-1)

ZINC01383169 OH group added to benzene ring -22.4760 -21.8202

ZINC00169867 CH3 group removed from 5 member ring -20.2292 -20.5939

ZINC00153513 OH group added to benzene ring -22.3896 -21.3560

From this new mapping, see figure 25, it is clear to see that there is less empty space surrounding the ligand within the receptor pocket. This therefore suggests that my original theory that the energy calculated was directly related to the size of the space that was unoccupied now has evidence to support it.

A mapping for the lipophilicity of the new altered ZINC01383169 was also constructed with a small alteration in the colours compared with the original mapping for the unaltered version of ZINC01383169, see figure 16. The result shows a better correlation of lipophilicity of the ligand and pocket.

In the case of compound ZINC00169867 the electrostatic mapping, shown in figures 13 and 29, revealed that there was an area of the ligand map protruding from the receptor pocket. I therefore decided that it was likely that the removal of this group would allow the ligand to fit better with in the pocket and, in turn, improve the energy calculated from docking it into the 1QKM protein. I removed the CH3 group from the 5 member ring, see figure 30, and docked the new altered ligand. The new calculated energy was not improved as is shown by the records in table 2. By using the mapping tools I was able to see that the removal of this one functional group had not eradicated the problem with the protrusion. The ligand had simply rotated and the other CH3 group on the five member ring had taken its place. I believe that the energy did not improve because the removal of the group prompted the rotation and created more empty space with in the pocket.

By using the ligand interaction tool it is clear to see that there are no changes to the interactions between the now altered ZINC00169867 and the amino acid residues of the receptor pocket.

My next ligand to be altered was the compound ZINC00153513. To this ligand I added an OH group to the benzene ring, see figure 33. This improved the energy from -21.3560 Kcalmol-1 to -22.3896 Kcalmol-1. The reason that I added the OH group to this particular point was because of the empty space visible in the pocket in figure 10. My theory that the less empty space there was, the better the energy, prompted me to try and fill that space with an OH group as the addition of the OH group to ZINC01383169 had improved its energy.

I used the ligand interaction tool to produce figure 34. This model shows that there are no new interactions between the ligand and the receptor pocket, despite the alterations that have been made to the compound.

The electrostatic mapping produced in figure 35, shows the reduction in empty space with in the receptor pocket compared to the original unaltered ZINC00153513 compound in figure 10. This is due to the addition of the OH group and I believe that this is the main reason for the improved energy, shown in table 2.

Conclusions

From my research I have found that the compound ZINC01383169 has the best calculated energy when docked into the 1QKM receptor protein. This compound has the IUPAC name [3-(2-chlorophenyl)oxazol-5-yl]methanol and is not available for purchase, meaning that ZINC01383169 is not a suitable replacement for genistein. My designed version of this compound, with the additional OH group, showed an improvement in the calculated docking energy. However, this is no longer of any benefit to finding a replacement for genistein.

The second best compound was ZINC00153513, which has the IUPAC name of 2-amino-4-(4-fluoro-3-methylphenyl)-1,3-thiazole. This compound is available for purchase at £118 per 100mg [7] making it considerably more expensive than Genistein, which is marketed at £53 per 100mg [8] by the same company. In order to improve this compound I added an OH group to its benzene ring and this improved the docking energy.

I would suggest trialling the altered version of the compound ZINC00153513 with the additional OH group in the place of genistein. Even though the altered ZINC00153513 is more expensive, if it is metabolised more slowly than genistein, there could still be an improvement, and the therapeutic treatments of patients could be enhanced.

The results produced in this report are of course theoretical, and further analysis, clinical trials and tests on animals would have to take place before any of these compounds could be used therapeutically. There is no evidence that the altered versions of my designed compounds would be non-toxic or even therapeutically viable.

After completing my research my theory is that the less empty space there is within the protein receptor pocket i.e. the better the ligand fits within the pocket and fills it, the better the calculated docking energy will be. The electrostatics of the ligand also make a big difference to the calculations and therefore a balance must be reached through the further design and analysis of the compounds. The lipophilic mappings have provided me with the theory that as long as there are no drastic differences between the ligand’s lipophilicity and the pocket’s lipophilicity, then there isn’t much effect on the calculated energies of the compounds.

References

[1] Raj Kumar, Mikhail N. Zakharov, Shagufta H. Khan. The Dynamic Structure of the Estrogen Receptor. Journal of Amino Acids.20112011(2011):7

[2] Pike A.C.W., Brzozowski, A.M., Carlquist, M. . HUMAN OESTROGEN RECEPTOR BETA LIGAND-BINDING DOMAIN IN COMPLEX WITH PARTIAL AGONIST GENISTEIN . http://www.rcsb.org/pdb/explore/explore.do?structureId=1qkm (accessed 19th February 2015).

Figure 1. Pike A.C.W., Brzozowski, A.M., Carlquist, M. . HUMAN OESTROGEN RECEPTOR BETA LIGAND-BINDING DOMAIN IN COMPLEX WITH PARTIAL AGONIST GENISTEIN . http://www.rcsb.org/pdb/explore/explore.do?structureId=1qkm (accessed 19th February 2015).

[3] Open Chemistry Database. Genistein. http://pubchem.ncbi.nlm.nih.gov/compound/genistein#section=Top (accessed 8th February 2015).

[4] Author Unknown. Biochanin A. http://en.wikipedia.org/wiki/Biochanin_A (accessed 27th February 2015).

[5] Author Unknown. Daidzein. http://en.wikipedia.org/wiki/Daidzein (accessed 27th February 2015).

[6] Author Unknown. S-equol. http://en.wikipedia.org/wiki/S-equol#Pharmacokinetics (accessed 27th February 2015).

Figure 16. Bursztyka J, Perdu E, Tulliez J, Debrauwer L, Delous G, Cravedi J, et al. Comparison of genistein metabolism in rats and humans using liver microsomes and hepatocytes. Food And Chemical Toxicology . http://www.sciencedirect.com/science/article/pii/S0278691507004978 (accessed 5th March 2015).

[7] Key Organics Ltd. 2-Amino-4-(4-fluoro-3-methylphenyl)-1,3-thiazole. http://www.keyorganics.net/bionet/catalog/product/view/id/408863 (accessed 5th March 2015).

[8] Key Organics Ltd. Genistein. http://www.keyorganics.net/bionet/genistein-mfcd00016952-446-72-0-c15h10o5-1.html (accessed 5th March 2015).

Appendix