HINT 3.00 Manual: Chapter 5S

Usage and Strategy

Objectives

This chapter describes the basic steps you should follow to use the SYBYL version of HINT. It contains two sections related to the application of HINT: an overview of the HINT strategy and several specific tutorials that cover most of the commonly used features of HINT.

After studying the descriptive material in the HINT Strategy section and performing the later tutorials you should comfortable with the following operations of HINT:

Partitioning both small molecules and biological macromolecules.
Tabulating InterMolecular and IntraMolecular interactions.
Mapping hydropathic surfaces and interactions.
Using HINT fields and parameters in CoMFA.

Methodology

First we will outlines the basic steps involved in a HINT calculation. Commands provided in HINT are discussed in general terms in the order in which they are used in a typical calculation. Also included are suggestions about when certain parameter values should or should not be used. For a more detailed discussion of each command, refer to the Graphical Interface chapter which has a detailed description of each parameter.

The second portion presents general scenarios where using HINT would be appropriate. Much of this information is background material for the HINT tutorials, so it is recommended that you read this section before starting the tutorial lessons.

Basic Steps for a HINT Calculation

To summarize the basic steps involved in the HINT calculation, the process consists of six steps:

Create and modify the input molecule(s).
Partition the molecule(s).
Set up calculation parameters.
Run the calculation.
Analyze the results.

Step 1: Defining the Molecule(s)
A HINT calculation can be performed on any molecule that exists in SYBYL. (Some HINT calculations are run on two molecules simultaneously.) Molecules can be read from files, created within SYBYL, or may be the result of work performed in another SYBYL application module. Note that HINT does not modify the molecule(s) in any way. Therefore, any desired modifications, such as adding hydrogens to a molecule read from a PDB file, must be done prior to starting the HINT calculation. As HINT relies on the SYBYL atom types for hydrophobic atom constant assignment, these must be properly set for most molecules before initiation of the calculation. In the sections below, the molecule(s) upon which the calculation is performed are referred to as the input molecule(s).
Step 2: Partitioning the Molecule(s)
The key parameter in HINT calculations is the hydrophobic atom constant which must be assigned for each atom in each molecule. This step is termed partitioning as the hydrophobic atom constants are derived from experimental solvent partition experiments. There are two approaches to partitioning employed by HINT. The first method, Calculate, directly utilizes the SYBYL atom types and connection matrix of the molecule to assign hydrophobic atom constants and to calculate a LogP (partition coefficient for water/octanol) which is the sum of the hydrophobic atom constants. This is the method of choice for small molecules. For macromolecules composed of distinct monomer or substructure units (especially proteins where the monomers are amino acid residues) the Dictionary method is recommended for consistency and reproducibility. Here, HINT assigns the hydrophobic atom constants based on the substructure and atom names. The LogP for the macromolecule is reported by HINT, but this result is only of tangential interest or validity. In either case, Hydrogen Treatment must be specified. This parameter describes how the Partitioning will treat hydrogens; there are three options: United Atom (using an approach that treats all hydrogens implicitly as part of their parent heavy atom), Essential (using an approach that treats (only) polar hydrogens explicitly), and Retain All. Note that the selection of this parameter has implications in following interaction calculations. For example, hydrogen bonding will be incorrectly modeled by HINT if the United Atom option is chosen. In general the Essential option is best for most applications. The Retain All option appears to "dilute" the hydrophobic density, but treats aromatic hydrogens as potential hydrogen bond donors.
- Calculate Method for Partitioning
  If the Calculate method is chosen, one other parameter must be specified: the Polar Proximity. When two or more polar groups are in "in proximity" within a molecule, their effects are cumulatively diminished -- this is termed the polar proximity effect and is dealt with by the Leo method of LogP estimation and by HINT as corrective factors to the hydrophobic atom constants. HINT offers three options for Polar Proximity: Off, Via Bond, and Through Space. The Via Bond method is most compatible with the Leo system, and is the option recommended. Through Space may be appropriate for some larger molecules that have significant intramolecular non-covalent interactions between polar groups. It uses a through-space distance function to estimate the Polar Proximity effect.
- Dictionary Method for Partitioning
  The Dictionary method of partitioning requires the selection of an additional parameter to specify the Solvent Condition for the molecule. The HINT partition dictionary cross- references each substructure type and its included atom by three solvent conditions: Acid, Neutral, and Base. For amino acids these conditions refer to fully protonated (both acids and bases), protonated bases and ionized acids, and fully ionized (both acids and bases). A fourth option (Inferred) uses the actual protonation state of each amino acid residue to select the appropriate solvent condition and HINT parameters for the residue. This option is to be preferred if the hydrogens of the input molecule (protein) have been very accurately modeled. Missing or superfluous hydrogens will be reflected in an incorrect assignment of the HINT parameters for the monomer, however.
Step 3: Setting up the Calculation Parameters
- Defining the HINT Distance Function
  The HINT distance function represents the mathematical equation of the distance behavior for interacting atoms in the HINT model. Two terms can be controlled representing the hydrophobic distance effect and the steric distance effect. The former term is not experimentally well understood and is coded in the HINT model as either an exponential or 1/rⁿ. In general the exponential function is more empirically realistic and is recommended for most calculations. The steric term in the overall HINT distance function is simply the Lennard-Jones function representing the dispersion attraction/repulsion between atoms at or near van der Waals distance from each other. Use of the steric term is optional, and for some grid calculations (e.g., HintMap Molecule and HintMap Complement) the steric term should normally be off. The Steric/Hydro Scaler parameter balances the contributions of the hydrophobic and steric terms. If using the exponential hydrophobic term, a scaling factor of 50.0 generates an overall interaction that is about 2/3 hydropathic and 1/3 steric between two attractive atoms that are at optimum distance for interaction. The primary contribution of the steric term to the HINT model, however, is to penalize atom-atom contacts that are too close. Another option of the distance function is Directionality vectors. These let you control the shape of lone pairs on polar atoms by defining a preferred interaction direction either along the bond axis or lone pair/pi orbital axes. The associated vector focus controls the "tightness" of the direction.
  Other parameters that indirectly affect the distance function are Cut Off Radius and Van der Waals Limit. The Cut Off Radius is a range cut-off that serves to reduce computation time by restricting the HINT calculation to grid points or interactions within range of the atom of interest. Values in the 6-8 è range are typically used. The Van der Waals Limit parameter is a multiplier for the atomic van der Waals radii to "tighten" or "loosen" the spectrum of acceptable contacts. In theoretical terms, adjustment of the Van der Waals Limit shifts the bottom of the van der Waals interaction potential well. By increasing Van der Waals Limit to 1.10, more atom-atom contacts will be recorded by HINT as being repulsive by virtue of the HINT steric term. Likewise, by reducing Van der Waals Limit to 0.90, some interactions that would have been sterically repulsive would be "allowed". This parameter is included for an additional flexibility in balancing the contributions of the hydropathic and steric terms of the HINT calculation.
- Defining the Grid Region
  An important part of most HINT calculations involves mapping the molecule or interacting molecules onto a three-dimensional grid. The Region Defintion Dialog Box defines this grid. There are three basic quantities that must be specified: the Grid Center, Grid Size, and Grid Resolution.
  The most common method used to define the grid involves specifying the Molecule Extents to be placed in the center of the grid, along with a Margin to surround this region. A special case arises for InterMolecular interaction maps where the best description of the region may be defined in terms of the Interfacial Region (or overlap) and Margin between two molecules. Alternatively the Grid Center may be specified as any arbitrary cartesian coordinate value.
  In contrast to electrostatic mapping programs, there are no "boundary effects" in the HINT calculation. There is no contribution from bulk dielectric. Thus, the selected margin must only satisfy your desire for a complete map. As the exponential distance function most often used in HINT decays much more rapidly than a 1/r2 function, a margin in the vicinity of 4-5 Angstrom is usually sufficient.
  The only quantity left to specify is the Grid Resolution. This is a single value described as the spacing between grid points in Angstroms. It is recommended that a grid resolution of 1.0 Angstroms or less be chosen for the construction of HINT maps due to the rapid decay of the HINT distance function.
- Specifying Map Sub-Types
  There are Grid Map Sub-Types that may be specified during the setup of a HintMap calculation. For the HintMap Molecule and HintMap Complement commands the Grid Types are Hydrophobic/Polar and Acid/Base. The former map describes hydrophobic portions of the hydropathic field as positive grid points and polar (hydrophilic) portions of the field as negative grid points. The second map, Acid/Base, plots acidic polar regions as positive and basic polar regions as negative (hydrophobic regions are zero). Note that HINT provides three levels of Acid Base Definition models to aid in description of the three-dimensional hydropathic environment of a Molecule or its Complement. The first level, Coulombic, uses only the formal charge of the atom to categorize acid (positive) or base (negative) properties. The second and third levels are based on the Br nsted and Lewis models, respectively. This option provides additional flexibility in HINT calculations. The default and recommended model is the Lewis acid/base definition as it is the most comprehensive because it recognizes some acid/base interactions not relevant to the Bronsted definition.
  For setting up HintMap Interaction (both InterMolecular and IntraMolecular) maps the Interactions parameter provides an option to split interactions involving Hydrophobic groups/atoms from those that are purely Polar in nature. The All option includes both. The most informative maps are obtained by performing dual calculations (Hydrophobic and Polar) on the same InterMolecular or IntraMolecular interaction.
- Defining Table Extents
  For HintTable calculations the Table Extents must be set. The HINT interaction calculation includes all atom-atom pairs, but it is usually desirable to limit the scope of the interactions written to the Interaction Table file. You can control the Table Extents with two parameters, Output Radius and Output Value. Output Radius limits the table to atom-atom interactions that are within the specified value. Output Value further refines the Table Extents by restricting the table to interactions having |bij| larger than the specified value. Clearly, an InterMolecular HintTable calculation involving two large molecules (e.g., proteins) can have more than 10^6 specific atom-atom interactions -- an uninterpretable quantity of data. Output Radius of 5-7 and Output Value of 2-10 are customarily chosen and typically yield 10^2 - 10^3 tabulated interactions. A further refinement of the Table content is offered by the Proton Suppress parameter. If this parameter is on, protons are not explicitly listed in the HintTable. However the interaction information from the protons is combined with their associated heavy atoms in the table entries. If the Proton Suppress parameter is off, protons in the molecule(s) will be listed as distinct atoms in the table. In either case, the user can control the definition of hydrogen bonds with H-Bnd Distance X-Y if Proton Suppress is on and H-Bnd Distance X-H if Proton Suppress is off.
- Output File Specification
  Three types of files are generated by HINT, depending on the specific calculation being performed. Any Partition operation can produce a Partition File (*.par). These are of some interest for archival purposes, but it is generally not necessary to retain these files. HintMap calculations produce contour files (*.cnt) and HintTable calculations produce table files (*.tab).
  The hydropathic maps from HintMap calculations are generally the most important and useful HINT output files. These contain either the value of the hydropathic density (for Molecule and Complement grids) or the HINT Interaction density at every grid point. Interpretation of the Interaction maps can often be made easier by producing a HintTable hardcopy for the same interaction calculation. This is an ASCII output file that, referenced by atom, lists the specific implicated interactions.
Step 4: Performing HINT Calculations
After all the parameters are specified, you are ready to run the HINT calculations. All of the HINT calculations can be run either in Interactive mode or as background jobs through the SYBYL NetBatch utility. The choice of run mode is dependent on the complexity of the calculation, in particular the size of the molecule(s). For small molecules or regions HintMap Molecule calculations take on the order of a few seconds, such that choosing to calculate the map interactively is not an inconvenience. In contrast, for HintMap InterMolecular calculations between, for example, two protein subunits, the calculation may take several hours.
Step 5: Analyzing the Results
HINT calculation results are analyzed in two ways: graphical and analytical. The graphical analysis is accomplished through contour maps. By far, the best method for analysis of the analytical HintTable results is by obtaining a printout of the file.
- HintTable Output (*.tab)
  The HintTable output file contains an itemized atom-by-atom listing of the specific interactions involved in the calculation, including the relevant parameters, the distance between the interacting atoms, and the HINT MicroInteraction constant or score. All implicated atoms are fully identified and their hydrophobic atom constants and solvent accessible surface areas are listed. The distance between each pair of interacting atoms is tabulated in two ways -- an Angstrom scale, and percentage of van der Waals radii sum. This latter scale permits an obvious scan for problematic close contacts (i.e., those less than 100.0). The HINT MicroInteraction constant is listed in the next column of the HintTable file. This value represents the HINT quantitation of the relative significance of the atom-atom interaction. Positive values represent interactions that are "good", while negative values represent interactions that are "bad" for the substrate binding, protein folding, or quaternary subunit interaction being modeled. The final column lists a preliminary characterization of the interaction type, i.e., whether hydrophobic, acid-base, acid-acid, base-base, hydrophobic-polar, or hydrogen bonding.
- Molecule and Complement HintMaps
  At completion of an Interactive HintMap calculation a dialog box will appear to assist in contouring the HintMaps. (For batch calculations this dialog box can be invoked by selecting the Contour option on the HintMap submenu.) Typical contour values for Molecule and Complement HintMaps are on the order of +10 to -40. For a HintMap calculated with the Grid Type set as Hydrophobic/Polar the positive contours represent hydrophobic regions of the molecule or assembly and the negative contours represent the polar or hydrophilic regions of the molecule or assembly. (For Complement HintMaps these are regions of the implied interacting species.) Maps calculated as Acid/Base give further information on the Polar portion of the HintMap by categorizing regions as Acidic or Basic. If both Grid Types have been calculated for a molecule, you can get the best combination of information by contouring the positive portion of the Hydrophobic/Polar map and both the negative and positive portions of the Acid/Base map.
- IntraMolecular and InterMolecular HintMaps
  IntraMolecular and InterMolecular HintMaps provide unique tools to visualize non-covalent interactions within or between molecules. You can contour a map by using the Contour dialog under the HintMap submenu. Typical contoured values for interaction maps are on the order of -400 to +400. The most information can be obtained by calculating (and mapping) the Polar interactions separate from the Hydrophobic interactions. Then, positive contours of the Hydrophobic map represent hydrophobic-hydrophobic interactions; negative contours represent polar-hydrophobic interactions which are generally unfavorable. It is advisable to not overevaluate the importance of this contribution as this type of interaction is virtually unavoidable in the biological environment. Positive contours of the Polar map represent polar interactions such as acid-base, hydrogen bonding and Coulombic effects; negative contours represent unfavorable polar interactions such as acid-acid, base-base, etc. These latter negative effects may often be reduced by the contribution of water molecules acting as hydrogen bond donors/acceptors or by the protonation or deprotonation of acids and/or bases at the interface.

Typical HINT Scenarios

The preceding section described how to setup and run some HINT calculations without particular regard to the problem being solved. This section describes several common types of interaction calculations for which HINT is appropriate and the strategies used to obtain the desired results. This is not intended as an exhaustive list of the uses of HINT, but rather as an introduction to some underlying principles that can be extended to many other types of applications.

Examination of Hydropathic Field around Molecule

One of the most common uses of HINT is to calculate the hydropathic potential field around a small molecule or macromolecule, and display this result as three-dimensional contours. This provides qualitative information that may be used to explain the interaction of the molecule with other molecules, or suggest a rationale for the observed structure of the molecule (i.e., intramolecular non-covalent forces). The most detailed information would be obtained by parallel runs using the Hydrophobic/Polar and Acid/Base Grid Types, which can be contoured together to display the hydrophobic, acidic, and basic regions of the molecule. The preferred HINT Distance_ Function for Molecule maps is one using the exponential hydropathic term and no steric term.

Prediction of Hydropathic Field for Complementary Species

HINT can be used to predict the Hydropathic Field for a Complementary species. When the structure of a receptor is known it is useful to qualitatively identify the hydropathic structural features of the ideal Complementary molecule (termed the "Key"). With this information it may be possible to construct a molecule with those features, that would be presumed to be the "best" substrate for the receptor site. Likewise, with a known drug, a "Lock" map can be constructed that may reveal the hydropathic structural features of an unknown receptor.

A special parameter in the HintMap Complement command, Asymmetry Factor, controls the directionality of the complementary hydropathic density. With no directionality the density indicates little preference for frontside vs. backside interactions. Using an Asymmetry Factor of 1.0 directs the Lock or Key density along bond axes to be consistent with frontside non-covalent interaction.)

Tabulation/Visualization of InterMolecular Interactions

The best visualization of InterMolecular interactions is obtained by using two HINT runs. One to calculate the Hydrophobic interactions and the second to calculate the Polar interactions. Both runs will be calculated over the same grid space, and can be contoured and displayed simultaneously. Interpretation of the maps is aided by the preparation of a HintTable for the same InterMolecular interaction. This, in essence, will give you a legend to the maps.

The preferred HINT Distance_Function for interaction maps is one which includes both the exponential Hydropathic term and the Steric term. We have found a Steric/Hydro Scaler of 50.0 to be satisfactory in its balance of the two terms. Note that the majority of quantitatively large negative interactions are due to steric violations where the two interacting atoms are too close to each other.

Analysis of a Series of Substrates

If you have a series of molecules known or suspected to bind at the same receptor HINT provides tools for analysis of these data. If the receptor structure is known, you can perform InterMolecular (Binding) HintTable calculations for each substrate with the receptor. The resulting HINT interaction constants may be then correlated with known binding constants for the substrate/receptor binding to construct a calibration curve for the system. HINT has been shown in several cases to order predicted vs. actual binding constants. It is particularly convenient to use the HINTScore option in the molecular spreadsheet if you organized your ligands that way in Sybyl. The next obvious step, once a calibration curve is available, is to predict the binding constants for new molecules. It is important to note the large number of potential uncertainties in this approach. Each atom-atom interaction contributes to the total interaction constant; and each atom-atom microbinding constant is significantly dependent on the distance between the atoms, and by inference, the quality of the molecular model construction.

Secondly, you can examine a series of substrates using the CoMFA technique in the SYBYL QSAR module. HINT adds functionality to CoMFA by providing a hydropathic field to supplement the basic steric and electrostatic CoMFA fields. A second entry point to HINT has been provided in the SYBYL Molecular Spreadsheet to use the HINT calculation of LogP for the QSAR.

Hydropathic Studies of Site-Directed Mutations

Both the Molecule and Interaction HintMaps can be useful tools for investigating the effects of site-directed protein mutations. The Difference command in the HintMap submenu will take the difference (subtract or add) between two contour maps. One strategy would be to compare the Hydropathic maps for "native" and "mutant" Molecules to determine the effect of the mutation on Hydropathic structure. This could reveal, for instance, whether there is an increase in hydrophobicity at the active site due to the mutation.

If the mutation involves structural features implicated in substrate binding, at a subunit interface or in protein folding, difference maps from InterMolecular or IntraMolecular interaction calculations may be useful tools to gain a qualitative understanding of the effect of the mutation on the non-covalent forces at the molecular interface.