watermap: using water energetics for lead … using water energetics for lead optimization...
TRANSCRIPT
WaterMap: Using Water Energetics for Lead Optimization
Schrodinger Bootcamp
14.6.16
WHAT WILL BE COVERED?
What Will Be Covered
• What is a WaterMap?
• Generating and Analyzing WaterMaps
• The theory behind WaterMap
• A survey of protein-water thermodynamics
• Druggability analysis using WaterMap
• Understanding a binding-site’s requirements with WaterMap
Methods and Applications of WaterMap
• Projects & Collaborations • Enzymes
• Factor Xa • HIV-RT & HIV-PR • PDE4 • CDK2 • Thrombin • HCV NS5B polymerase • Kinases (selectivity)
• GPCRs • Bromodomains • Protein-Protein Interactions
• PDZ Domain • Bcl-XL/Mcl • Zip-A
• Nucleic Acids
• Young, T., et al. 2007 PNAS 104:808-813
• Abel, R., et al. 2008 J Am Chem Soc 130:2817-2831
• Beuming, T., et al. 2009 Protein Science 18:1609-1619
• Robinson, D., et al. 2010 ChemMedChem 5:618-627
• Higgs, C. et al. 2010 Med Chem Lett 1:160-164
• Abel, R, et al. 2010 J Chem Theory Comput , 6(9):2924-2934
• Abel, R., et al. 2011 ChemMedChem 6(6):1049-1066
• Wang L et al. 2011 PNAS 108:1326-1330
• Snyder, P, et al. 2011 PNAS 108: 17889-17894
• Beuming, T, et al. 2012 Proteins 80: 871-883
• Shah, F, et al. 2012 J Chem Info Model 52: 696-710
• Myrianthopoulos, V, et al. 2013 ACS Med Chem Lett: 4, 22-26
• Breiten, B, et al. 2013 PNAS 135: 15579-15584
• Pearlstein, R, et al. 2013 Proteins (2013) 81:1509-1526
• Weldon, DJ et al. 2014 Bioorg Med Chem Lett 24: 1274-1279
• Ask for more references, and works in progress…
BEFORE WE START...
Why Is Water Important?
• Water is everywhere in biology
• Protein binding sites are mostly filled with water
• Water is a direct competitor in ligand and substrate binding
• Displacement of unhappy waters can lead to big potency gains
• But…water energetics cannot be determined from structure alone
Thermodynamic Decomposition of Ligand/Protein Binding
Solvated Ligand Solvated Apo
Protein
Desolvated Ligand
Solvated Ligand in Bioactive
Conformation
Ligand-Induced Desolvated Protein
Binding Site
Solvated Protein in Ligand-Induced
Conformation
Protein/Ligand/Water Complex
DGbind = DGi=1
5
å i( )
∆G( 5 ) ∆H reward ∆Srot / t rans penalty
∆G( 1 ) ∆Hconf penalty ∆Sconf penalty
∆G( 3 ) ∆H penalty ∆S reward
∆G( 2 ) ∆Hconf penalty ∆Sconf penalty
∆G( 4 ) ∆H ? ∆S ? WaterMap
WaterMap Visualization
• WaterMap computes the entropy and enthalpy of “hydration sites”
• These can be used to rationalize SAR, drive potency, and tune selectivity – Green = stable – Red = unstable
• Provides a “map”, not a GPS
Stable
(happy)
waters
Unstable
(unhappy)
water
Thrombin
S1 pocket
A General Comment About WaterMap
• WaterMap is not ‘magic’
• WaterMap does not design molecules for you
• WaterMap requires a lot of thought
• WaterMap requires a lot of imagination and insight – The good news is that this becomes easier the more you use
WaterMap
• Drugs are designed by humans – The computers and software merely assist
A General Comment About the Demo
• We will use the new Maestro 11 GUI
• See the second Bootcamp talk later today
An Example WaterMap
• The figure shows an example WaterMap – This WaterMap shows the hydration-
structure around the ligand-binding-site of 3RLP (HSP-90)
• Each of the spheres is a ‘hydration-site’ – A hydration-site is a region of space
where water-molecules tend to aggregate • The term ‘hydration-site’ and ‘water’
tend to be used interchangeably – Each hydration-site has a number of
associated thermodynamic-properties
THE THEORY BEHIND WATERMAP
This section looks at some of the theory behind WaterMap: What are the calculations being run? What do the various thermodynamic quantities actually mean? It also attempts to dispel some common misconceptions or misinterpretations of the quantitative WaterMap results
What Happens During a WaterMap Calculation?
• WaterMap generation is a multi-stage process: – Stage 1 – Simulation
• The system in question is simulated for 2ns (at 300K, 1atm) – The simulation is run with full-explicit solvent (TIP4P)
– During the simulation the majority of the system is subjected to fairly heavy restraints
– Stage 2 – Clustering • The position of every water-molecule in the ‘region of interest’ is then accumulated
– This provides a measure of the water-density at each position in space
• The location with the highest water-density shows the location of the first hydration-site – The clustering-algorithm then continues, searching for each spatially distinct peak in the water-density,
generating hydration-sites
– The clustering-algorithm terminates when there are no more spatially distinct peaks with a density > 2x the bulk density of water
– Stage 3 – Thermodynamic Property Generation and Further Analysis • The thermodynamic properties of each hydration-site are then calculated based on the
behaviour of each water-molecule that has been observed to occupy that hydration site
• Other calculations are also run at this time
Why Restrain the Protein?
• As noted, the majority of the system is heavily restrained during the simulation – By default only a few terminal rotors – mainly hydroxyls – are allowed
to move
• The thermodynamic quantities calculated by WaterMap are only valid for a particular conformation – Restraining the system prevents any re-arrangement polluting the
values calculated by WaterMap with those from other conformations • This is particularly the case when considering an apo-WaterMap
– An apo-structure formed by deleting a bound ligand is a very high-energy system, it will want to collapse if left unrestrained
Aside: The Effect of Restraints on WaterMap Transferrability
• Clearly the restraints are needed to give a particular protein-ligand complex WaterMap meaning
• However, often we are using a WaterMap generated on one complex to score non-native ligands or generate new ideas – This brings into question the transferability of the WaterMap results across
different ligands
• Our experience with this is broadly in line with expectations: – In reasonably rigid (normally somewhat buried) regions of the protein, where re-
organisation is low, WaterMap results are highly transferrable across ideas – However, in mobile (normally solvent exposed) regions of the protein, where re-
organisation is trivial, WaterMap results are less transferrable across ideas • In these areas, fully dynamics-based approaches such as FEP are seen to be much more
reliable
Advanced Feature: Modifying the WaterMap Simulation
• The WaterMap simulation is controlled by a Multisim-workflow file – This is a standard text-based input file for the molecular-dynamics engine
Desmond
• The file contains various sections controlling different stages of the simulation – These can be easily edited – For example, the restraints are specified using Schrödinger’s ASL syntax:
restrain = { atom = "asl: (NOT ((SMARTS. \"[H]O[H]\") AND NOT (atom.i_wmap_restrain 1) ) ) AND NOT (atom.ele H)" force_constant = 5.0 reference_position = "reset" }
• It should be noted that experiences with modified parameters are rather limited
What Do DH, -TDS and DG Correspond To? – 1
• The values calculated by WaterMap correspond to the average excess enthalpy, entropy and free-energy that a water-molecule, located at the hydration-site, would possess – The excess energies are measured relative to bulk water
• This means that: – A hydration-site with a negative DH-value is making stronger interactions
with the surrounding protein than it would with surrounding water-molecules in solution e.g. near a charged group
– A hydration-site with a positive DH-value is making weaker interactions with the surrounding protein than it would with surrounding water-molecules in solution e.g. near a hydrophobic-residue
What Do DH, -TDS and DG Correspond To? – 2
• Every interaction with the protein has a corresponding entropic-penalty – The magnitude of this entropic-penalty relative to the DH value gives us the
overall excess free-energy, the DG-value
• It is favourable to displace a water-molecule from a hydration-site with DG>0
• Similarly it costs energy to displace a water-molecule from a hydration-site with DG<0
• This displacement is brought about by the ligand binding to the protein – In this light we can see that the DG-values calculated for an apo-WaterMap
correspond to the ligand-induced protein desolvation energies • This link is vitally important for (semi)-quantitative WaterMap analyses a bit later on
Why Do We See Hydration-Sites With DG>0?
• One frequent point of confusion regarding WaterMap results is the presence of hydration-sites with DG>0 – Many people feel that these should ‘spontaneously’ leave the binding-cavity
• However, this ignores the fact that we are in a condensed phase – If the hydration-site is vacated, we are left with a vacuum – It has been known for many years that the formation of a vacuum, within a cavity
is a highly unfavourable event • Estimates suggest that the penalty for a single water-sized vacuum, in a cavity would be
about 6-8kcal/mol
• Consequently, there is a 6-8kcal/mol ‘bias’ holding the DG>0 water-molecules in-place – Only when the environment becomes even more unfavourable than 6-8kcal/mol
do we see spontaneous evacuation of the binding-site – These are referred to as de-wetted or evacuated regions
How Does WaterMap Calculate DH and –TDS?
• The DH-values calculated by WaterMap come directly from the force-field energies
• The –TDS-values come from a fairly complicated bit of statistical-thermodynamics known as ‘Inhomogeneous Solvation Theory’ – This takes the form of a rather ugly series of integrals – Each integral measures the entropy of the position/orientation of the
water-molecules that have occupied a given hydration-site
Se kbw
gsw r, lngsw r, drd
kbw
2
22gsww r
2,2 lngsww r2,2 dr2d2 ...
r = position = orientation = integral over = bulk water density g = correlation function sw = solute-solvent terms sww = solute-solvent-solvent terms
How Does WaterMap Calculate DH and –TDS?
• In reality the offensive mathematics is doing something that’s visually quite obvious – It’s just quantifying the ‘randomness’ the water-molecules that have
occupied each hydration-site
Disordered water-molecules
Favourable entropy -TDS=0.99kcal/mol
Highly ordered water-molecules
Unfavourable entropy -TDS=5.24kcal/mol
Continuous WaterMaps
• The generation of continuous WaterMaps follows basically the same procedure, however the clustering stage is removed and replaced with a set of calculations on a high-resolution grid (0.5Å) covering the region of interest: – The thermodynamic properties are calculated for each lattice point
• The enthalpies are averaged over water-molecules found inside each cell
• The entropies are calculated for all water molecules within a 1Å sphere, centered on each lattice point – 30o bins are used for rotational entropy calculation
Analysing WaterMaps Within Maestro
• The WaterMap ‘Examine Results’ GUI is the main interface for importing and navigating WaterMap data
• The interface is accessible from two places: – The Applications->WaterMap menu
item – By selecting the ‘W’ icon next to a
WaterMap entry in the Project Table • Naturally this only applies to previously
loaded WaterMap data
The ‘Examine Results’ GUI
The first section of the GUI deals with importing data:
As expected the ‘Import Files...’ button will bring up a standard file dialog where the user can select the WaterMap .zip archive to be imported
The ‘Analyze Workspace’ button is a somewhat more advanced feature. It is useful when the Project Table already contains a number of WaterMaps and the user wishes to switch between them
Tip: The ‘Adjust view’ option can be really annoying as it changes the orientation of the structure in the Workspace. Deselecting the option prevent this from happening. The ‘Views’ functionality in Maestro is really useful for maintaining a consistent view of various aligned WaterMaps
The ‘Reset Panel’ button is really useful for preventing Maestro getting ‘confused’. Press this button everytime you make a significant change to the Workspace. Then hit ‘Analyze Workspace’.
The ‘Examine Results’ GUI The second section of the GUI deals with the representation of the WaterMap in the Workspace
A number of extra properties are calculated in addition to the basic WaterMap data. This section controls whether these are displayed
The ‘Shape by entry’ option is vital when analysing multiple WaterMaps simultaneously
The ‘Cavity map’ displays regions which are not hydrated
The ‘Free energy density’ is used to display the continuous WaterMap
Site-labels allow many of the calculated properties to be overlaid with the hydration-sites in the Workspace
The DG(simple) colour scheme is the easiest to interpret – red-hydration sites have a +ve DG, green a negative DG. The DH and –TDS colour scheme is somewhat harder to understand - Cyan to green hydration-sites have DH values less than –TDS while Green to yellow hydration-sites have DH values greater than -TDS
The ‘Examine Results’ GUI
The third section of the GUI deals with the underlying WaterMap data and methods for filtering what’s displayed in the Workspace
The bulk of the ‘Examine Results’ GUI is taken up with a table showing all of the properties of each of the hydration-sites contained in the selected WaterMaps – this table can be quite long
The ‘Pick to select sites’ option is really useful as it enables the user to ‘click’ on a hydration-site in the Workspace and be taken to the corresponding entry in the table
The various filters are extremely useful for removing clutter from the display allowing the user to focus on the most significant hydration-sites
The ‘Examine Results’ GUI
The fourth section of the ‘Examine Results’ GUI deals with (semi)-quantitative WaterMap analyses
The ‘Score Ligands’ button allows the molecules selected in the Project Table to be scored against the currently loaded WaterMap
Tip: The ‘Score only selected sites’ option is useful for filtering docking results based on what molecules enter a region occupied by an ‘interesting’ hydration-site
‘Perform WM/MM Scoring’ starts the more detailed (and time-consuming) WM/MM calculation – this combines WaterMap and MM-GBSA to give a reasonably thorough model of the protein-ligand binding process
The ‘Examine Results’ GUI
The fifth section of the ‘Examine Results’ GUI contains some miscellaneous functionality
The ‘Import Trajectory’ and ‘Import Water Frames’ allow the user to interact with the raw molecular-dynamics data used to create the WaterMap. These are rarely required
The ‘Archive’ option is also rarely required – the user will already have a .zip archive containing all of the WaterMap data
The most useful option is the ‘Open Ligand Interaction Diagram’, this generates a 2D-representation showing how the current ligand interacts with the WaterMap. This is a very useful option for discussing WaterMap results with chemists
Continuous WaterMaps
• Continuous WaterMaps are a relatively new feature of WaterMap – These eschew the hydration-site
based interpretation of the WaterMap in favour of a continuous grid of ‘free-energy’
• This form of WaterMap has applications for advanced scoring of protein-ligand complexes
HOW TO GENERATE A WATERMAP?
This section explores the steps needed to generate a WaterMap for a given protein-structure: The main purpose is to introduce the ‘WaterMap Calculation GUI’ But other aspects, such as the effect of protein preparation on the final WaterMap, will also be covered
Generating WaterMaps Within Maestro
• The WaterMap ‘Perform Calculation’ GUI is the main mechanism for creating a new WaterMap – The interface is accessible from the
Applications->WaterMap menu item
The ‘Perform Calculation’ GUI
The first section of the ‘Perform Calculation’ GUI deals with the ‘region of interest’
The simplest method of specifying the region of interest is to have a ligand present – this is similar to the way a Glide docking-box is specified. Any ligand can be used as the default calculation produces an ‘apo-WaterMap’, where the ligand is deleted prior to calculation
If it is not possible to use a ligand to specify the region of interest, one can use ASL. The ASL specification can also be used to run WaterMap on a complete protein
TIP: The ‘Retain ligand’ toggle forces WaterMap to generate a ‘holo-WaterMap’. In this case the ligand is not deleted and is present during the WaterMap calculation. Holo-WaterMaps are useful when investigating the effect of a ligand on the behaviour of the water-molecules within the binding-site
The ‘Perform Calculation’ GUI
The second section of the ‘Perform Calculation’ GUI deals with the simulation settings
The ‘Truncate protein’ option deletes parts of the protein that are too far from the region of interest to have an influence on the WaterMap. This is done for speed. The vast majority of systems should have this option turned off. WaterMap calculations require ~1-day on 8-16CPUs for all but the largest systems.
TIP: If truncation is used, great care should be taken to ensure that no holes have been created in the protein structure that allows bulk water access to buried pockets
“Treatment of existing waters” should be set to “as solvent” or “delete” in most cases. If the “as solvent” option is used, any existing crystal-waters will join the solvent added by WaterMap prior to the simulation. “Deleting” the crystal-structure waters will remove any possible bias from the crystal-structure. This is useful if there are reasons to doubt the veracity of the crystal-waters
Advanced Feature: Generating WaterMaps From the Command Line
• It is also possible to generate new WaterMaps from the command line – This is useful when a large number of WaterMaps need to be run on a given target
• For example a whole set of holo-WaterMaps
• The required utility is: $SCHRODINGER/utilities/create_wm_job (-help)
• Most of the options have a direct equivalent in the ‘Perform Calculation’ GUI – However the –extended_gcmc option is unique to the command line
• This option performs considerably more exhaustive solvation of the pocket • This is sometimes useful when running holo-WaterMaps, or other circumstances where there are
small, tight, pockets that need to be correctly solvated • The option is very computationally expensive, generating 20-independent starting points for the
main WaterMap simulation run
Preparing a System for a WaterMap Calculation
• The preparation of a system can have a profound influence on the quality of results coming from a WaterMap calculation – Far more so than, for example, a docking calculation
• It goes without saying that the Protein Preparation Wizard should be used to get the protein ready for the calculation – But more than ever, it is critical to check what PPrep has done
• Unusual charges on residues, or bad ‘flips’ can drastically alter the hydration-structure seen in the final WaterMap
• Holo-WaterMaps are also influenced by the charge and/or tautomerisation state of any ligands present
AN ANALYSIS OF MANY WATERMAP CALCULATIONS
In this section we explore the general properties of water-molecules within a protein-cavity: This is based on the work by Beuming et al. Proteins. 2012, 80, 871-883
A Survey of Hydration-Site Thermodynamics
• 27-different proteins were analysed with WaterMap across a range of different families – The result was a dataset of ~32,000 hydration-sites
• Each hydration-site having an associated DH, -TDS and DG-value
• The statistics from this dataset give some idea of the range of different water-behaviour that will be encountered
The Distribution of DH, -TDS and DG
• Careful analysis of the DH-values reveals a tri-modal distribution – The modes are centred on:
• -4.0kcal/mol for interactions with acidic-groups
• -2.0kcal/mol for interactions with basic-groups
• 0.0kcal/mol for interactions with uncharged-groups
The Distribution of DH, -TDS and DG
• The –TDS value are all >0.0kcal/mol – This comes from our definition of
excess entropy • Any interactions between the hydration-
site and the protein will yield some protein-water correlation entropy
• The –TDS values fall off asymptotically towards 6kcal/mol – This maximum value is reasonable
• The entropy loss of transferring a water-molecule from the gas-phase to an ice-crystal at 298K is estimated to be 6.3kcal/mol
The Distribution of DH, -TDS and DG
• The DG-values are, of course, a superposition of the DH and –TDS-values
• Here we can see that few hydration-sites have DG>8kcal/mol – This is in line with earlier
statements concerning vacuum-formation
The Effect of SASA
• The maximum SASA for a 1Å-radius hydration-site in 1.4Å-radius water is ~72Å2
– Such hydration sites are ‘fully’ solvent-exposed and should have DG=0.0kcal/mol
• As the hydration-site gets more buried we see a divergence in the energy – Some become more stable, some less
• The majority of profoundly unstable water-molecules (DG>2.0kcal/mol) have a SASA<20Å2
– i.e. they are buried
• However, a significant proportion of water-molecules with SASA<20Å2 are quite stable – Being buried in a protein does not necessarily
imply that a water-molecule is unstable – Determining the water-molecule’s stability is
where WaterMap comes in
USING WATERMAP FOR DRUGGABILITY ASSESSMENT
In this section we look at how WaterMap can be used to assist in the assessment of druggability for a given site: WaterMap results are compared with results from more familiar tools such as SiteMap As well as a general comparison with other established methods of ranking targets for druggability
What Constitutes a Druggable Binding-Site?
• Generally we’re looking for a binding-site that can be targeted with a small-molecule – The binding-site needs to be large enough to accommodate a
reasonable-sized ligand – It needs to have some degree of enclosure to allow the ligand to
survive the continual ‘buffeting’ of solvent – And present a balance of hydrophobicity and polarity to ensure that
the resulting molecule does not have ‘extreme’ physical-properties
Schrödinger’s SiteMap Tool
• Schrödinger’s SiteMap tool is an established program for assessing druggability – It proves a SiteScore-metric that considers the effects of:
• Volume
• The degree of enclosure
• Balance between hydrophobicity/polarity
• The SiteScore is calculated using the following equation: Where n is the number of site points (effectively volume), where e is a measure of the enclosure of the site and where p is a measure of the hydrophobicity/hydrophilicity of the site
penSiteScore 20.06688.0094.0
Druggability and WaterMap
• The hydration-site thermodynamics of a binding-site actually provide very similar information – The number of hydration-sites gives a measure of the volume – The energies of the hydration-sites give a very detailed description of the
enclosure and hydrophobic/hydrophilic balance of the binding-site
• Basically what we are looking for is a binding-site with a small
(drug-sized) cluster of highly unstable hydration-sites – This implies that a drug-sized molecule will receive significant binding from
simply ‘occupying’ the binding-site • Fine tuning its overall pharmacophore is likely to yield the desired potency
• Binding-sites with few (or no) unstable hydration-sites are likely to be either too shallow, or too polar to bind a drug-like molecule
Druggability – Gleevec/Abl
• Abl is obviously a highly druggable target
• The WaterMap shows a complete chain of highly unstable hydration-sites – These actually provide a useful
indication of the shape of an ‘ideal’ molecule
– This is reinforced by looking at the continuous WaterMap, which shows an almost unbroken red-region throughout the binding-site
Druggability – HIV-Protease
• HIV-protease is a moderately druggable-site
• The WaterMap shows a number of highly unstable hydration-sites – But as underlined by the continuous
WaterMap, these aren’t as well connected as they were in Abl • Consequently we’ll need a large molecule to
span them all – And the ligand will probably have a low atom
efficiency – There are also two stable-hydration-sites in
the core of the binding-site • These surround the catalytic Asp-diad
– Care will need to be taken in this region to avoid losing large amounts of binding-energy
Druggability – PTP-1b
• PTP-1b provides an excellent example of an undruggable binding-site
• The core of the binding-site contains a large cloud of stable water-molecules – Hydrophobic atoms in this region are
actually detrimental to binding – Only powerfully ionic atoms are capable of
replacing some of these water-molecules • This naturally limits the druglikeness of any
ligand – There are a scattering of hydrophobic-
regions away from the main binding-site • But reaching these requires large, inefficient,
ligands
Druggability – COX-2
• COX-2 is clearly a highly-druggable binding-site
• As with Abl, the WaterMap reveals a tight cluster of unstable hydration-sites and continuous high free-energy density
• The WaterMap also shows a region that is devoid of hydration-sites – This actually corresponds to a de-
wetted/evacuated region • The cavity map option allows this region to
made more explicit
Druggability – COX-2
• COX-2 is clearly a highly-druggable binding-site
• As with Abl, the WaterMap reveals a tight cluster of unstable hydration-sites and continuous high free-energy density
• The WaterMap also shows a region that is devoid of hydration-sites – This actually corresponds to a de-
wetted/evacuated region • The cavity map option allows this region to
made more explicit
Cavity Maps
• There are two reasons for a WaterMap to show no hydration-sites: – The water in that region behaves like bulk-water
• And is essentially uninteresting
– There is no water in that region
• This is a highly interesting phenomenon, de-wetted regions are highly unfavourable – Filling them will yield a lot of binding energy
• The cavity map feature of WaterMap allows us to distinguish those two possibilities
Cavity Maps - Theory
• Cavity maps are generated using the same simulation as the rest of the WaterMap data – The cavity is detected via test-particle insertion
• We attempt to add 1Å test particles to the frames of the main WaterMap trajectory – If a region contains protein/ligand/water, we will not be able to insert any particles as
they will collide with other atoms
– If a region is evacuated, the test particles can be added easily
• The cavity map shows the probability (ease) of inserting a test particle at a given location in space – The higher the probability, the more evacuated the region
USING WATERMAP TO UNDERSTAND A BINDING-SITE’S REQUIREMENTS
In this section we start to look at how the WaterMap thermodynamics can guide us when designing ligands: The concepts of displacement, replacement, interaction and avoidance will be introduced along with some simple examples based on literature compounds
Displace, Replace, Interact, or Avoid?
• For any hydration-site within the binding-cavity we have four choices when it comes to designing our ligand: – Displace – Occurs when we position a ligand atom in the same space as
the hydration-site. Such interactions are essentially hydrophobic in nature – Replace – Occurs when we position a ligand atom in the same space as the
hydration-site with functionality that can mimic the interactions of the occupying water. These interactions have a degree of polar character
– Interact – Occurs when we decorate our ligand with a group that can interact (favourably) with the waters occupying a nearby hydration-site. These are polar interactions
– Avoid – Occurs when there is a region of the binding-site we particularly need to steer clear of
• The thermodynamics of the individual hydration-sites and their response to the changes in ligand structure guide us between these options
Displacement
• When a hydration-site has both DG and DH>>0kcal/mol it is generally favourable to displace it with an atom from the ligand – Hydration-sites with DG and DH>>0kcal/mol are hydrophobic-regions • Consequently Me-groups or halogens are
ideal for this task
• The effect of displacing a single, highly unstable, water-molecule from the binding-site gives rise to many of the ‘magic-methyl’ effects
DH=2.58kcal/mol -TDS=0.99kcal/mol DG=3.57kcal/mol
DH=5.26kcal/mol -TDS=1.03kcal/mol DG=6.29kcal/mol
Replacement
• Hydration-sites with DH<<0kcal/mol but with DG>~0kcal/mol are candidates for replacement – Water-molecules in these regions are making
strong interactions with the protein • But are paying a substantial entropic-penalty for doing
so
• A suitably chosen (polar) functional-group on our ligand, if positioned carefully, can make the same strong enthalpic interactions with the protein – Thereby negating the enthalpic-penalty of
removing the water-molecule
• But such a group will yield a net-benefit to binding as the liberated water will be entropically stabilised by returning to bulk
DH=-4.58kcal/mol -TDS=4.04kcal/mol DG=-0.54kcal/mol
Interaction
• Some hydration-sites are clearly highly-conserved – Displacing or replacing these
hydration-sites is frequently very difficult
• Either energetically DG<<0kcal/mol
• Or geometrically
• These sites can be treated as part of the protein – Forming bridging waters
DH=-5.97kcal/mol -TDS=5.24kcal/mol DG=-0.73kcal/mol
Interaction – Holo-WaterMaps
• A holo-WaterMap can show the influence of the bridging interaction on the affected water-molecule – It’s worth checking that the new
bridging interaction doesn’t actually destabilise the water-molecule
• In this case the holo-WaterMap shows the hydration-site being stabilised by ca. 3kcal/mol
DH=-9.68kcal/mol -TDS=6.08kcal/mol DG=-3.60kcal/mol
Avoidance
• It is very hard to do anything productive when DG<<0kcal/mol – Such water-molecules are in highly stable locations and virtually all
attempts at removing or interacting with them fail
• In such circumstances it can be easier to avoid that region of the binding-site entirely
SUMMARY
Summary
• By now you should: – Be familiar with the concept of a WaterMap
• And the theory behind its generation
– Be able to generate new WaterMaps – Be able to analyse WaterMaps
• For a variety of different purposes – Druggability
– Activity/potency
– Selectivity
Summary
• This is, by and large, the ‘state of the art’ in terms of WaterMap usage – However, as witnessed by some of the ‘Advanced’ and ‘Experimental’
features, different methods of using WaterMap are still developing
• End-users play an important role in developing these new methodologies
• Please keep Schrödinger up-to-date with your progress with WaterMap – And do not hesitate to contact us if there are any questions or issues