search or to direct students to specific therapeutic targets. It is
important to note that rational drug design relies on availability
of target structures. Structures for many targets are available in
the PDB, but for others only homologous structures may be available (e.g., the mouse protein rather than human, or a different
receptor type). Homologs can be used for the simplest purpose of
lessons 3 and 4 (to illustrate drug design as the process of designing a ligand that has the highest affinity to the target), but students
engaged in an inquiry project should understand the caveats. Most
importantly, the active site of a nonhuman target may differ in
amino acid sequence and conformation, which may then affect
ligand binding. In principle, students could try to build a homology model of a human protein based on the structure of a homolog, but that would be a long and complex extension of the
lesson we are proposing here. We refer readers interested in
extending the project and including homology modeling to
Schmidt et al. (2014). At the end of this part of the lesson, students
should write a brief report on what structures are available in the
PDB and whether the structure available is a homolog.
Once a target structure has been identified, students proceed to
calculating all potential druggable cavities in the protein surface
using the DogSiteScorer server (Table 2). To input the target structure to the server, students can (1) specify the PDB ID, (2) download
the selected protein structure prior to using DogSiteScorer and then
upload it from disk, or (3) upload any PDB protein file that has been
user-modified. The server offers multiple different programs; for this
lesson, the DogSiteScorer module should be selected after uploading
the protein file. Students should write a report in which they record
the number and locations of cavities that may be potential binding
sites as well as their physical and chemical properties (e.g., volume
and hydrophobicity) directly from the DogSiteScorer output. If the
target PDB ID corresponds to a protein–ligand complex, they identify and analyze the binding site they have already seen in the previous lesson and evaluate its “druggability” compared to other cavities
in the protein surface. Students then use these data to design their
own ligands, namely small organic molecules that are to be tested
as the drug candidates. For example, if the pocket is highly hydrophobic and roughly spherical, students can design a drug ligand
whose structure corresponds to these characteristics.
The second goal of lesson 3 is to create 3D input files for the
candidate drugs. Students are familiar with drawing two-dimensional
chemical formulae of molecules, but they need to use a program to
convert each such formula to spatial coordinates for all atoms that
are then saved in a text file. Many programs are available to do so.
We recommend that students design their own novel ligand molecules
and create the 3D input files using ProDrg or Avogadro (Table 2). One-Click Docking is also very convenient in that it allows the user to
streamline the ligand structure drawing, choice of docking target,
and docking simulations. When saving the ligand’s 3D coordinates,
we recommend choosing mol2 format (for more information on file
formats, see Supplemental Material, section II).
The best form of this exercise is for students to identify at least
one ligand that is known to bind to their target protein and use this
ligand structure as the basis for chemical modifications to match the
amino acid structure of the binding pocket and/or the shape and
physical properties of the binding cavity. This would correspond
to drug optimization (e.g., improving binding affinity). Alternatively,
novel structures can be designed following Lipinski’s rule of five to
discover novel binders, compounds with higher affinity, and/or
potentially better pharmacokinetic parameters (Lipinski et al.,
1997). Lipinski’s rule of five is an empirically based predictor of
absorption or permeation of a drug and includes, for example, the
number of hydrogen bond donors and acceptors in the ligand mole-
cule. Students should write a report in which they show the ligands
they have designed and explain (1) how the new structure relates to
other known ligands/drugs and (2) why they think the ligand will
“fit” one of the previously identified pockets.
We recommend that students submit their input files (target pro-
tein and ligand structure files) for docking simulations (Table 2) as
homework after lesson 3. This will ensure that the simulation
results are returned by servers in time for analysis during lesson
4. Alternatively, part of lesson 4 could be devoted to starting additional simulations or data analysis, and the report could be completed as homework after lesson 4.
Lesson 4: Docking Simulations
There are multiple free and licensed programs to carry out protein–
ligand simulations, but for this activity we have chosen to use two
online-based programs that are the easiest to learn and fastest to execute. We suggest SwissDock or One-Click Docking servers (Table 2)
to carry out the docking simulations (for additional comments and
tips, see Supplemental Material, section II). Both servers will also display the results (docked ligand poses and estimated binding energies)
at the end of the simulation for students to analyze their results. We
recommend that in their written reports, students should (1) record
the estimated binding energies for different poses or different ligands,
(2) paste in and/or describe the location of the binding pocket within
the protein structure, and (3) identify the most favorable orientation
of the ligand in the pocket. The students should try to predict the
most promising ligand and suggest further modifications that can be
tested. More advanced students can use VMD to list all the amino
acids forming contacts with the ligand and determine the type of contacts ( i.e., hydrogen bonding, hydrophobic, charge-charge, etc.). Several possible versions of this exercise can be implemented:
• Students design a few chemical modifications to the known
drug molecule and dock them to a single target protein (drug
• Students dock several known drugs to proteins other than their
designed targets (identifying potential off-targets).
• Students design a set of completely novel molecules and dock
them to a selected single target to explore possible regulatory
sites (looking for binding to non-orthosteric cavities).
• Students conduct a series of consecutive docking simulations,
each with a ligand progressively modified according to the
properties of the binding pocket.
• Students can evaluate the accuracy of the docking method by
“re-docking” a known ligand to reproduce the crystal conformation of the complex (several different docking methods could be
used for comparison if this variant is selected).
Questions that students should answer after the exercise include:
• Which ligand binds with greater affinity?
• Is this ligand likely to be selective?