Intramolecular hydrogen bonding in medicinal chemistry

Article abstract of DOI:10.1021/jm100087s

The formation of intramolecular hydrogen bonds has a very pronounced effect on molecular structure and properties. We study both aspects in detail with the aim of enabling a more rational use of this class of interactions in medicinal chemistry. On the basis of exhaustive searches in crystal structure databases, we derive propensities for intramolecular hydrogen bond formation of five- to eight-membered ring systems of relevance in drug discovery. A number of motifs, several of which are clearly underutilized in drug discovery, are analyzed in more detail by comparing small molecule and protein-ligand X-ray structures. To investigate effects on physicochemical properties, sets of closely related structures with and without the ability to form intramolecular hydrogen bonds were designed, synthesized, and characterized with respect to membrane permeability, water solubility, and lipophilicity. We find that changes in these properties depend on a subtle balance between the strength of the hydrogen bond interaction, geometry of the newly formed ring system, and the relative energies of the open and closed conformations in polar and unpolar environments. A number of general guidelines for medicinal chemists emerge from this study

Full text of DOI:10.1021/jm100087s

J. Med. Chem. 2010, 53, 2601–2611 2601
DOI: 10.1021/jm100087s
Intramolecular Hydrogen Bonding in Medicinal Chemistry
Bernd Kuhn,* Peter Mohr, and Martin Stahl
Discovery Chemistry, F. Hoffmann-La Roche AG, CH-4070 Basel, Switzerland
Received January 20, 2010
The formation of intramolecular hydrogen bonds has a very pronounced effect on molecular structure
and properties. We study both aspects in detail with the aim of enabling a more rational use of this class
of interactions in medicinal chemistry. On the basis of exhaustive searches in crystal structure databases,
we derive propensities for intramolecular hydrogen bond formation of five- to eight-membered ring
systems of relevance in drug discovery. A number of motifs, several of which are clearly underutilized in
drug discovery, are analyzed in more detail by comparing small molecule and protein-ligand X-ray
structures. To investigate effects on physicochemical properties, sets of closely related structures with
and without the ability to form intramolecular hydrogen bonds were designed, synthesized, and
characterized with respect to membrane permeability, water solubility, and lipophilicity. We find that
changes in these properties depend on a subtle balance between the strength of the hydrogen bond
interaction, geometry of the newly formed ring system, and the relative energies of the open and closed
conformations in polar and unpolar environments. A number of general guidelines for medicinal
chemists emerge from this study.
Introduction
Druglike organic molecules almost invariably contain a
number of functional groups capable of forming hydrogen
bonds, rendering them soluble and giving them the ability to
form specific interactions with their biomolecular targets.
When a donor and an acceptor are in proximity on the same
molecule, an equilibrium may exist between closed conforma-
tions in which an intramolecular hydrogen bond is formed,
creating a temporary ring system, and open conformations in
which the polar groups are exposed to solvent (Figure 1).
These sets of conformations are not only structurally distinct.
It is intuitively clear that the closed forms, hiding polarity
from the environment, should be more lipophilic and might
display a higher membrane permeability, whereas the open
forms should be more water-soluble. In drug discovery, it is
therefore important to recognize the potential for intramole-
cular hydrogen bond formation and to be aware of its con-
sequences. Here we perform a systematic study of topologies
prone to intramolecular hydrogen bond formation to enable
the rational use of such interactions in medicinal chemistry.
To this end, we combine data mining in crystal structure
databases with the detailed analysis of measured and calcu-
lated molecular properties in illustrative model systems.
Several publications have reported beneficial effects of
intramolecular hydrogen bonding on ligand-receptor bind-
ing and rationalized this with conformational restriction in
which the small molecule substituents are favorably aligned
with the protein pockets.1-3 In these cases, the intramolecular
hydrogen bond is crucial and removal of either hydrogen
bond acceptor or donor results in a drastic loss of binding
affinity. Rational design of internal hydrogen bonds for
Figure 1. Molecules with hydrogen bond donor (D-H) and accep-
tor (A) functionalities in proximity often occur in a thermodynamic
equilibrium between closed (left) and open (right) conformations.
conformational preorganization was pursued in scaffold re-
placements for a diverse set of targets.4-8 In addition to its
effects on receptor binding, an improved property profile has
been associated with molecules containing intramolecular
hydrogen bonds. Increased brain penetration and pharmaco-
logical activity was observed for NK1a receptor antagonists
and attributed to the higher apparent lipophilicity due to
intramolecular hydrogen bonding.9 A team at Takeda has
reported improved oral absorption and an excellent pharma-
cokinetic profile for luteinizing hormone-releasing hormone
receptor antagonists when an intramolecular hydrogen bond
was established.10 Studies on cyclic peptides support the
notion that the ability to form internal hydrogen bonds is
critical for passive membrane permeability.11
Databases of crystal structures are rich sources of informa-
tion on molecular interactions and conformations. Infantes
et al. have investigated the competition between inter-
and intramolecular hydrogen bonding and concluded that
a Abbreviations: NK1, neurokinin 1; CSD, Cambridge Structural
Database; PDB, Protein Data Bank; B3LYP, Becke three-parameter,
Lee, Yang, and Parr; MMFF94x, Merck molecular force field 94x;
DMF, dimethylformamide; THF, tetrahydrofuran; LYSA, lyophilized
solubility assay; PAMPA, parallel artificial membrane permeability
assay.
*To whom correspondence should be addressed. Phone: þ41
616889773. Fax: þ41 616886459. E-mail: bernd.kuhn@roche.com.
r
2010 American Chemical Society
Published on Web 02/22/2010
pubs.acs.org/jmc

2602 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 6
Kuhn et al.
Figure 2. CSD search queries for NH hydrogen bond donor vs different acceptors: (a) CdO; (b) SdO; (c) disubstituted nitrogen acceptor in a
heterocycle; (d) alkoxy group. X stands for any atom, Y for carbon or nitrogen. Subscripts denote coordination numbers. Semicircles represent
˚
cyclic bonds. Hydrogen bonds are defined by the geometric constaints d e 2.35 A and 100° e R e 180°.
intramolecular hydrogen bonds are preferred when five- or
six-membered conjugated rings are formed.12 A pioneering
statistical analysis of intramolecular hydrogen bonds in the
Cambridge Structural Database (CSD)13 was performed by
Bilton et al.14 This analysis resulted in 50 intramolecular
hydrogen bond topologies and their propensity of formation
within small molecule crystal structures. However, a consider-
able number of these motifs are not relevant to medicinal
chemistry because of undesired substructure liabilities.
Our search focuses exclusively on motifs of interest to phar-
maceutical drug design, and by profiting from the more
than 2-fold higher number of entries in the current database,
we are able to identify several additional motifs. Moreover,
we illustrate the frequently occurring motifs with selected
crystal structures from both CSD and Protein Data Bank
(PDB).15,16
cutoffs were used to classify interactions as hydrogen bonds:
˚
˚
upper limits d = 2.35 A for N-H N/O and d = 2.30 A for
3 3 3
O-H N/O contacts14 were used, and the angle R at the
3 3 3
hydrogen atom was confined to 100° e R e 180° (Figure 2).
For the PDB database, there is more ambiguity in identifying
structures with intramolecular hydrogen bond, as hydrogen
atoms in protein-ligand structures are typically not resolved.
Searches were performed only for a subset of interactions,
namely, six- and seven-membered hydrogen bond motifs of
CdO acceptor with NH and OH donors. Slightly modified
search queries had to be used: distance donor acceptor d e
˚
3 3 3
3.0 A, angle base-donor acceptor 70° e R1 e 180°, and angle
3 3 3
donor acceptor-base 70° e R2 e 180°. NH and OH donors
3 3 3
were defined as nitrogen and oxygen atoms that are connected
by single bonds to two and one neighbor atoms, respectively.
Apart from these different geometric and donor type defini-
tions, queries were run in identical manner in the CSD and the
PDB ligand database. The propensity of a particular substruc-
ture to form an intramolecular hydrogen bond is expressed as
the ratio between the number of entries that fulfill these geo-
metric criteria and the total number of entries in the database
containing this substructure. Certain topologies cannot possibly
adopt a conformation required for intramolecular hydrogen
bonding. These were eliminated by requiring that the torsion
angles about any cyclic or acyclic double bond be within the
interval -90° to 90°. Furthermore, we removed trivial solutions
in which the number of consecutive cyclic bonds between donor
and acceptor is g(total number of linker bonds - 1), as these
have only 1 degree of freedom at best.
Experimental Section
Database Searches. All statistical data described in this article
were derived with the ConQuest 1.11 program. For small
molecule crystal structures the CSD, version 5.30 (November
2008), was searched. Unless noted otherwise, the following
general search flags were set: R e 0.10, “3D coordinates
determined”, “not disordered”, “no ions”, “no errors”, “not
polymeric”, and “only organic”. This results in a search space of
124 791 structures. Conformations of protein-bound ligands
were analyzed by extracting ligands in SD format from Proa-
sis2,17 a curated version of the PDB, and by generating a
separate database in CSD format with the program PreQuest.
The ligand structures were taken from all HET entries, except
metals and commonly found small ions, in the PDB as of
December 8, 2008. Ligands with less than 5 or more than 100
atoms were removed, excluding in particular large peptidic
We report intramolecular hydrogen bond motifs that show
more than five hits in the CSD database and contain a sub-
structure that is represented more than five times in Prous. We
distinguish between Csp2 (C3), Csp3 (C4), N, and O atom types
as well as cyclic (c) and acyclic (a) bond types. The linker
between donor and acceptor is abbreviated by the sequence of
bond and atom types; i.e., cC3aC3a stands for a linker that
connects hydrogen bond acceptor and donor by cyclic-Csp2-
acyclic-Csp2-acyclic bond and atom types.
˚
ligands. A resolution limit of 3.0 A was applied, resulting in a
database of 73 158 ligands derived from 23 291 PDB entries.
CSD search queries are depicted in Figure 2.
The CSD database contains a number of entries that have
substructures typically not found in drugs. To focus on the
chemical space relevant for medicinal chemistry, we report only
those intramolecular hydrogen bond motifs that are present to a
larger extent (>5 hits) in a filtered Prous Integrity database.18
This filtered database contains all compounds that have pro-
gressed into phase I clinical trials or beyond, including currently
marketed drugs. The following keywords were used as search
criteria: “phase I”, “phase I/II”, “phase II”, “phase II/III”,
“phase III”, “pre-registered”, “recommended approval”, “re-
gistered”, or “launched”. Molecular weight was limited between
250 and 800. Salts and entries with elements other than H, C, N,
O, F, S, P, Cl, Br, and I were removed. SD files of the resulting
4555 compounds were converted into a Conquest-searchable
database using Prequest to be able to apply consistent search
queries for all databases.
Conformational Energy Calculations. For model compounds
1d, 2d, 3b, and 4b, a systematic conformational search was
performed with MOE19 (version 2008.10) using the MMFF94x
force field and without nonbonded interaction cutoff. Also cis-
amides were included in the sampling and the resulting con-
formations minimized with default settings prior to analysis.
Conformational searches were performed both in vacuo and in
water (generalized Born solvation model). The conformations
with lowest molecular mechanical energy of open and closed
form each were selected for subsequent quantum mechanical
optimizations with Jaguar20 using the density functional B3LYP
and a 6-31G** basis set. Both conformational energy differences
were calculated in the gas phase and in the Minnesota water
solvation model SM8.
Synthesis of Model Compounds. Amides 1a-d were prepared
from the commercially available precursor acids by transform-
ing them into the corresponding acid chlorides with oxalyl
chloride/cat. DMF in toluene at ambient temperature followed
by reaction with either methylamine in THF (2 M) or dimethyl-
amine in THF (2 M). 1-Methyl-3-phenylurea 2c was synthe-
sized by reacting aniline with methyl isocyanate in DMF at
room temperature overnight, whereas 1,1-dimethyl-3-phenylurea
We generated CSD statistics for all five- to eight-membered
intramolecular hydrogen bond motifs of CdO, SdO, disubsti-
tuted nitrogen acceptor in a heterocycle, and alkoxy groups as
hydrogen bond acceptors with NH donors. For OH donors,
additional searches were performed for five- to seven-membered
rings with carbonyl and nitrogen as acceptors. Geometric

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 6 2603
2a was obtained by stirring aniline with dimethylcarbamoyl
chloride in the presence of triethylamine in CH2Cl2.21 1,1-
Dimethyl-3-pyridin-2-ylurea 2b was synthesized pursuing the
characterization of all model compounds is provided in the
Supporting Information (pp S2-S48).
Molecular Property Measurements and Calculations. All com-
pounds were tested with respect to their physicochemical beha-
vior. The log D values were determined by the new CAMDIS
method.24 Kinetic solubilities were measured according to the
LYSA protocol and are reported in μg/mL.25 Permeation con-
stants were obtained from the PAMPA assay26 and are ex-
pressed in 10-6 cm/s. Details of each experimental setup are
given in the Supporting Information (pp S49-S50). Lipophili-
city was calculated with the program ClogP, version 4.94.27
€
same method (alternatively with Hunig’s base instead of
triethylamine); however, because of the ambidentic nature and
the low reactivity the yield was extremely low (8%), and
scrupulous purification was necessary to get the product in pure
form. 1,3-Dipyridin-2-ylurea was formed as typical side pro-
duct. 1-Methyl-3-pyridin-2-ylurea 2d, finally, was prepared in
decent yield by reacting 2-aminopyridine with methyl iso-
cyanate again in DMF at ambient temperature. N-(1-Methyl-
1H-benzoimidazol-2-yl)acetamide 3a was produced by acetyla-
tion of commercially available 1-methyl-1H-benzoimidazol-2-
ylamine with acetyl chloride/pyridine in CH2Cl2.22 Extensive
NMR spectroscopy was necessary to prove that the acetyl group
is attached to the endocyclic nitrogen; reaction with acetic
anhydride (neat) led to the identical product. We note in passing
that 3a exists as tautomeric mixture; as a consequence thereof, it
was not possible to record a well resolved 13C NMR. 3b, on the
other hand, was similarly synthesized from 1H-benzoimidazol-
2-ylamine with neat acetic anhydride.23 N-(1H-Benzoimidazol-
2-ylmethyl)acetamide 4b was bought from a commercial
supplier (Princeton), whereas 4a, finally, was obtained from
C-(1H-Benzoimidazol-2-yl)methylamine by acetylation with
acetyl chloride/pyridine in CH2Cl2. All compounds were pur-
ified by chromatography and/or crystallization to a purity of
at least 95% as determined by HPLC. Full spectroscopic
Results and Discussion
In the following, we first discuss the statistical trends
observed for intramolecular hydrogen bonds forming five-
to eight-membered rings as observed in crystallographic
databases. In the second part of the paper, we analyze
molecular property differences of specifically designed sets
of model compounds in which one molecule can form an
intramolecular hydrogen bond while one or more closely
related structures cannot.
1. Analysis of Crystallographic Data. The probability of
intramolecular hydrogen bond formation is far greater for
Table 1. CSD Hit Statistics of Intramolecular Hydrogen Bond Forma-
tion of Ring Sizes 5-7 for Different Combinations of Hydrogen Bond
Acceptors (Carbonyl, Disubstituted N in Heterocycle) and Donors
(NH, OH)a
NH
OH
Five-Membered
CdO
N
13.7
46.1
16.8
11.5
Six-Membered
64.7
56.0
CdO
N
48.6
42.4
Seven-Membered
CdO
N
7.4
6.7
9.0
10.8
Figure 3. CSD-derived probabilities for the formation of intramo-
lecular hydrogen bonds of ring sizes 5-8 with NH as hydrogen bond
donor and different hydrogen bond acceptors: CdO (red), hetero-
cyclic N acceptor (blue), SdO (yellow), alkoxy (cyan).
a Numbers indicate the percentage of entries with intramolecular
hydrogen bond (% hb).
Figure 4. Plot of hydrogen bond distance vs angle for NH N and NH OdC. Ring sizes 5 (magenta), 6 (green), and 7 (red) are plotted
along with the distribution for unconstrained intermolecular hydrogen bonds (gray). The geometry constraints used in this study are plotted as
thick lines.
3 3 3
3 3 3

2604 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 6
Kuhn et al.
Figure 5. Selected topologies, CSD hit statistics, and crystal structure examples of intramolecular NH OdC and NH N interactions in
3 3 3
3 3 3
six-membered rings. The topology abbreviation lists the sequence of bond and atom types in the linker connecting hydrogen bond acceptor with
donor. C3 and C4 are sp2 and sp3 carbon atoms, and a and c denote acyclic and cyclic bond types, respectively. Statistics for other six-membered
ring topologies are supplied in the Supporting Information (pp S51-S55).
six-membered rings than for any other system (Figure 3).
Here, the probability of intramolecular ring formation cor-
relates with the acceptor strength in the order CdO >
heterocyclic N acceptor > SdO > alkoxy.28 In all other
cases, specific substructures with a particularly high propen-
sity of hydrogen bond formation dominate the statistics.
These will be discussed in the individual paragraphs below.
Geometric parameters of all small-ring intramolecular
hydrogen bonds lie outside the typically observed ranges
Intermolecular hydrogen bonds are close to linear, prefer-
entially with angles greater than 150°. Only seven-membered
ring intramolecular hydrogen bonds come close to these
values. Six-membered rings have angles between 130° and
140° but the same distances as intermolecular hydrogen
bonds. Five-membered rings have longer distances and
smaller angles just within the cutoffs applied here
(decreasing the lower cutoff to 75° changes the statistics only
marginally by ∼1%). On these grounds it may be argued that
five-membered rings should not be regarded as classical
hydrogen bonds but, more general, as favorable electro-
static interactions. We will still cover them here, since the
for unconstrained systems. Figure
4 illustrates this
for carbonyl and nitrogen acceptors. A linear correlation
between bond length and angle is observed in all cases.

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 6 2605
Figure 6. Left: Plot of torsion angle vs hydrogen bond distance for arylether oxygen atoms interacting with an NH group separated by three
linker atoms. Intramolecular hydrogen bonds are not uncommon in drugs. Top right: Small-molecule crystal structure of the citrate salt of
sildenafil revealing an intramolecular hydrogen bond (CSD entry FEDTEO). Bottom right: Sildenafil in complex with phosphodiesterase 5a
with intermolecular hydrogen bonds only (PDB code 1tbf).
conformation-property relationships for larger rings
described in this article are applicable to such systems as
well.
Table 1 allows a comparison between the two donor types
(NH and OH) and the two most important acceptor types
(CdO and cyclic N). As observed by Bilton et al.,14 the
overall propensities are quite similar for the two donors. An
in the hydrogen-bonded form in 81% of the entries and with
CdO as acceptor in 85% of the entries.
Queries on the PDB confirm the preference for intramo-
lecular hydrogen bond formation in the case of sp2-hybri-
dized linkers, albeit with lower propensity (Supporting
Information p S56). For nonplanar rings the quantitative
agreement between CSD and PDB results is higher, indicat-
ing that the increased probability for planar hydrogen
bonded systems in small molecule crystals is caused by
packing effects.
Slightly less pronounced preferences for resonance-
assisted hydrogen bonds are found with weaker acceptors
such as alkoxy groups bound to aromatic rings (Supporting
Information p S55). CSD statistics show that the planar
hydrogen bonded conformations prevail in the solid state,
but a significant number of open forms exist as well (Figure
6). The vasodilator sildenafil30 is an example of such a
system. When complexed with its target, phosphodiesterase
5a, the lactame unit, is engaged in an intermolecular hydro-
gen bond with a Gln side chain.31 When crystallized as its
citrate salt, sildenafil adopts a planar conformation with the
pyrimidinone NH interacting intramolecularly with the
ethoxy group (Figure 6).32
exception is unusually frequent NH N five-membered
3 3 3
ring systems, dominated by amide-substituted heterocycles
adopting a high propensity planar conformation. There
is also no general difference between nitrogen and carbonyl
as an acceptor, indicating that formal lone pair directionality
is not correlated with the likelihood of hydrogen bond
formation.
Six-Membered Rings. Individual topologies, hit statistics,
and illustrative examples of selected interactions are sum-
marized in Figure 5. Detailed statistics for all other systems is
supplied in the Supporting Information (pp S51-S55). All
substructures with a high probability of hydrogen bond
formation are variations on a common theme: They all share
two sp2-hybridized linker atoms (amides, heteroaromatic
rings) leading to planar, conjugated systems. The high
stability of this type of intramolecular hydrogen bond has
been termed resonance-assisted hydrogen bonding and can
be rationalized by enhanced π-delocalization.29 Substruc-
tures containing sp3 atoms in the linker do not profit from
resonance assistance and are conformationally more flexible
resulting in a preference for the open form (e.g., aC4aC3a).
Further constraints in the form of cyclic linker bonds slightly
increase the propensity for internal hydrogen bond forma-
tion, as illustrated by the comparison of the topologies
aNaC3c and aNaC3a. The general observation that
carbonyl and nitrogen acceptors have a similar propensity
for hydrogen bond formation also holds for specific linkers.
For example, the aNaC3a linker with the N acceptor occurs
The general observation that the propensity for intra-
molecular hydrogen bonding correlates with increasing hydro-
gen bond acceptor strength is nicely illustrated by the
preferred conformations of ester groups interacting with
NH donors (Figure 7). Because of its higher acceptor
strength, the CdO group of esters is found considerably
more often in the syn than in the anti arrangement (CSD
ratio of ∼20:1).
The high propensity to form resonance-assisted hydrogen
bonds has been used in the design of new kinase inhibitors.6
On the basis of a known bicyclic kinase inhibitor
scaffold, pyrimidin-4-ylureas were suggested as bioisosteres

2606 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 6
Kuhn et al.
Figure 7. CSD-derived relative preferences of alkoxy and carbonyl
functionalities of ester groups to engage in six-membered reso-
nance-assisted intramolecular hydrogen bonds with NH donors.
(Figure 8a). Molecules containing this substructure indeed
turned out to be inhibitors of multiple kinases, and the
binding mode was confirmed by a cocrystal structure. The
strong preference for intramolecular hydrogen bond forma-
tion of urea-substituted pyri(mi)dines has also been con-
firmed by NMR spectroscopy33 and by model calculations
on adenosine receptor agonists.34 Hydrogen bonds of this
type were shown to persist in water.35 In our studies, we find
a probability of 81% that intramolecular hydrogen bond
Figure 8. Examples of designed scaffold replacements comprising
an intramolecular hydrogen bond: (a) pyrimidin-4-ylurea kinase
inhibitors;6 (b) anthranilamide kinase inhibitors;5 (c) 4-aminoquin-
azoline scytalone dehydratase inhibitors;4 (d) alkoxybenzamide
poly(ADP-ribose) polymerase inhibitors.7 Intramolecular hydro-
gen bonds are highlighted in red.
formation occurs for such a topology (Figure 5, N HN,
3 3 3
aNaC3a). The literature contains a number of further
examples where six-membered hydrogen bonded pseudo-
rings effectively mimic an aromatic ring, offering a similar
extent of π-delocalization and pharmacophore features
(Figure 8b-d), but overall this possibility of “scaffold hop-
ping” still seems underexplored.4,5,7,8,36
Seven-Membered Rings. In contrast to six-membered
rings, planar, resonance-assisted hydrogen bonds are con-
siderably less likely for seven-membered rings because of
steric repulsion. In fact, we could identify only one substruc-
ture that fulfills these criteria (and it is rare in druglike
compounds). As shown in Figure 9, two amide substituents
in the ortho position off a five-membered ring heterocycle
are in plane with each other with a propensity of 60% (11 hits
in 18 CSD structures). No analogous examples with intra-
molecular hydrogen bonds are known for six-membered
aromatic rings in the linker (23 CSD entries), most likely
because of higher angle strain.
Figure 9. (a) Planar intramolecular seven-membered ring system
involving five-membered heterocycles (BEZYAG). (b) No intra-
molecular hydrogen bonds are observed for systems involving six-
membered rings (LICKIS).
form an intramolecular hydrogen bond. In the PDB, 75 of
1786 ligands (4%) form γ-turns. It might be argued that the
more lipophilic protein environment promotes the forma-
tion of the less polar hydrogen bonded form (“hydro-
philic collapse”). Many PDB structures are not pure γ-turns
but intermediate structures between β sheet and cyclic struc-
ture, with bifurcated inter- and intramolecular hydrogen
bonds.
We are not aware of examples where seven-membered ring
intramolecular hydrogen bonds have been designed in med-
icinal chemistry. Where they were observed, they had not
been predicted by modeling.39
In the seven-membered ring CdO HN structures
3 3 3
(Figure 10), donor and acceptor are not in plane; the
˚
acceptor O atom can be as far as 1.8 A above the donor
plane. Specific topologies occur quite frequently in the ring-
closed form. Strikingly, the aC4aC3cC3a ring system (first
entry in Figure 10) is homologous to cyclohepta-1,4-diene in
terms of electronic structure and geometry. One may regard
the carbonyl group and the aromatic bond participating in
the seven-membered ring as double bond equivalents. The
conformation of this ring as exemplified by MISBEW is a
“folded” structure, consisting of two sets of five and four
coplanar atoms, corresponding to the lowest energy confor-
mer of cyclohepta-1,4-diene.37
Eight-Membered Rings. The overall frequency of eight-
membered ring intramolecular hydrogen bonds declines
further; the number of alternative conformations is even
larger here. Two substructures still stand out: intramolecular
The statistics for N-Csp3-Csp2 linkers connecting CdO
acceptor and NH donor is dominated by peptide structures.
The cyclic conformation, which corresponds to the γ-turn,38
is rare: In the CSD, only 5 out of 1138 CSD entries (0.4%)
SdO NH hydrogen bonds are more frequent in eight-
3 3 3
than in seven-membered rings. This is in particular true for
Csp2-Csp2-Csp3-Csp3 linkers, even if there are few examples
(6 CSD hits in 10 entries, Figure 11a). Analogous systems

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 6 2607
Figure 10. Topologies, CSD hit statistics, and crystal structure examples for the CdO NH interaction in seven-membered rings. Statistics
for other seven-membered ring topologies are supplied in the Supporting Information (pp S51-S53).
3 3 3
Figure 11. Examples for eight-membered rings with relatively high
propensity: (a) SdO NH interaction with topology aC3cC3a-
3 3 3
C4aC4a (DIGNAJ); (b) CdO NH interaction with topology
aNaC4aC4aC4a (WUGJIR).
3 3 3
with CdO acceptors have a lower propensity for this ar-
rangement and are characterized by longer hydrogen bond
distances. Apparently, the special geometry of the SdO
group is well suited to engage in this type of intramolecular
hydrogen bond. The second substructure is CdO NH
3 3 3
contacts connected by a N-Csp3-Csp3-Csp3 linker (6 CSD
hits in 21 entries, Figure 11b).
Five-Membered Rings. For substructures where a donor is
separated by three bonds from an acceptor, we often find a
strong preference for planar geometries that is not observed
when the acceptor is replaced by carbon. This difference in
conformation is caused by a combination of reduced steric
repulsion and favorable electrostatic interaction between
H-bond donor and acceptor. This effect is illustrated
with torsion histograms for two frequently occurring
motifs, pyridine-2-amides and alkoxyacetamides, and their
Figure 12. Torsion histograms for two analogue systems in which
the planar motif (black) contains a hydrogen bond acceptor that can
corresponding carbon analogues (Figure 12). These exam-
ples are representative of the many planar NH
interact with the NH donor, while the corresponding carbon con-
gener is bent (gray).
N
3 3 3

2608 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 6
Kuhn et al.
Figure 13. Experimental and calculated molecular properties of four different model systems. Molecules 1d (2-methoxy-N-methylbenzamide),
2d (1-methyl-3-pyridin-2-ylurea), 3b (N-(1H-benzoimidazol-2-yl)acetamide), and 4b (N-(1H-benzoimidazol-2-ylmethyl)acetamide) are able to
form intramolecular hydrogen bonds and are highlighted in red. Experimental properties are PAMPA permeation constant Pe (black), LYSA
water solubility (green), and octanol/water partition coefficient log D (blue). clogP (gray) is calculated log P.
and NH alkoxy arrangements observed in the CSD (see
not readily explained on the basis of the two-dimensional
structures alone. Most likely the pronounced out-of-plane
rotation of the tertiary amide substituent affects distribution
behavior.
3 3 3
Supporting Information p S57 for crystal structure exam-
ples). Such motifs occur very frequently in development
compounds.
2. Analysis of Molecular Properties. The formation of an
intramolecular hydrogen bond should effectively remove
one donor and one acceptor function from the surface
of a molecule and thus result in increased lipophilicity
and membrane permeability and in reduced aqueous solu-
bility. To study this hypothesis in detail, we synthesized
and characterized four sets of closely related chemical struc-
tures. In each set, only one structure is able to form an
intramolecular hydrogen bond; the others lack the necessary
hydrogen bond donor or acceptor. Figure 13 gives an
overview of the structures and their respective proper-
ties. We measured the passive permeability (PAMPA) per-
meation constant Pe, the octanol/water partition coefficient
log D, and kinetic solubility. In addition, the calculated
lipophilicity clogP is contrasted with the experimental log D
values.
Molecular property changes of the benzamides 1a-d
follow the expected trends. Starting from 1a, introduction
of a weak acceptor (1b) or a donor (1c) decreases perme-
ability and increases solubility. The presence of both func-
tional groups in 1d has the opposite effect. The log D value
also decreases from 1a to 1b. The log D values of 1a and 1c
are identical in spite of the introduction of the donor. This is
The benzimidazoles 3a and 3b also follow the above
trends. Removal of a methyl group in 3a leads to 3b, which
has a higher log D value, slightly higher permeability, and
decreased solubility. Although we do observe the expected
trends here, it is noted that 3a is in equilibrium with a second
tautomer, which results from a hydrogen shift of the amide
NH to the nonmethylated benzimidazole nitrogen. This
tautomer can form a resonance-assisted six-membered
intramolecular hydrogen bond (CdO HN) of topology
3 3 3
aNaC3c. The balance between the two tautomers very likely
depends on the nature of the solvent. We have not performed
further analyses here.
It is instructive to compare clogP values and experimental
log D values for the four model systems (Figure 13). Not
unexpectedly, clogP calculations underestimate the lipophi-
licity of molecules with intramolecular hydrogen bonds
because the underlying fragment-based approach does not
take into account internal pairing of polar groups. This is
most pronounced in the pair 3a and 3b. The two molecules
are predicted to have equal lipophilicity (clogP of 1.46 vs
1.41), whereas the measured log D value of the hydrogen
bonded analogue 3b is 0.71 units higher than that of 3a. On
average, it seems that clogP predictions should be increased

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 6 2609
permeability, and only a modest increase in lipophilicity. The
prerequisite for such an effect seems to be that the polar,
open conformation should be a low-energy conformation in
water. The benzamide 1d also shows attractive overall
properties and suffers only from slightly reduced solubility.
In this case, crystallographic data (Figure 6) indicate that the
intramolecular hydrogen bond is rather weak, leaving the
amide torsion rather flexible. This leaves us with two alter-
native strategies for the introduction of intramolecular
hydrogen bonds with a beneficial overall profile: choosing
a strong hydrogen bond with a distinct alternative open
conformation accessible in water or choosing a relatively
flexible and weak hydrogen bond with a continuum of open
and closed states.
Conclusions
By means of database searches in the CSD, we have
identified a considerable number of substructures with the
propensitytoformintramolecularhydrogenbonds. We found
several indicators that intramolecular hydrogen bonding is
correlated with acceptor strength. The highest frequency of
intramolecular hydrogen bonds have planar, six-membered
rings stabilized by conjugation with a π-system. Such sub-
structures have been actively applied in drug design. A variety
of further, far less explored topologies have been identified:
Weaker six-membered ring hydrogen bonds containing one
sp3 center and, in particular, a number of nonplanar seven-
membered and eight-membered ring topologies. Five-mem-
bered ring intramolecular hydrogen bonds have the smallest
angles and the longest distances and are just within the
geometric limits of a hydrogen bond as defined here. Never-
theless, because of their planarity, these topologies can be
useful modules in conformation design.
Figure 14. Quantum mechanically (B3LYP/6-31G**) calculated
energy differences between open and closed conformations. Differ-
ences are given in kcal/mol for gas phase (dielectric constant ε = 1)
and water environment (ε = 80). Hydrogen bonded 3D conforma-
tions are shown: 1d (analogue of Figure 6, top right), 2d (Figure 5,
aNaC3a), 3b (Figure 5, aNaC3c), 4b (analogue of Figure 10,
aNaC4aC3a).
by a correction factor of 0.4 for each intramolecular hydro-
gen bond. Note that polar surface area calculations of
molecules with intramolecular hydrogen bonds are over-
estimates for the same reason.
In contrast to the structures 1 and 3, the molecular
property profiles of the urea-substituted aryls 2 and the
benzimidazoles 4 deviate from the above simple trends. In
both cases, the structures with a potential intramolecular
hydrogen bond exhibit unexpectedly high aqueous solubility
and, for the urea structures, a less consistent trend in log D
values.
The molecular properties measured for four specifically
designed model systems did not reveal a uniform behavior
with respect to membrane permeability, solubility, and lipo-
philicity. To understand the trends within these systems, it is
important to consider the equilibrium between the polar open
conformations and the unpolar closed conformation both in
aqueous solution. Simultaneous high solubility and perme-
ability can be achieved when the open conformation is of
sufficiently low energy to be significantly populated in water.
This can be predicted to some extent by quantum mechanical
calculations of conformational energies in a high dielectric
environment. The targeted design of intramolecular hydrogen
bonds also requires an assessment of hydrogen bond interac-
tion strength. Weaker intramolecular hydrogen bonds avoid
an extreme shift of the equilibrium to the closed form. In that
respect, potential design strategies are to reduce the hydrogen
bond acceptor or donor strengths28 or to capitalize on non-
planar seven- instead of planar six-membered rings. Finally,
introductionofstericstraininthe closedconformationmay be
a useful strategy to affect the equilibrium. In seven-membered
ring topologies similar to Figure 9a (pyrazoles instead of
imidazoles), we have found drastically improved aqueous
solubility when the amide nitrogen not involved in intramo-
lecular hydrogen bonding is fully substituted. X-raystructures
showed an intramolecular hydrogen bond for both primary
and tertiary amides but very different torsional angles of the
two amide planes of 30° and 70°, respectively (results not
shown).
The unexpectedly high solubility of 2d and 4b might be due
to a different equilibrium between the open (polar) and
hydrogen-bonded (unpolar) conformations in aqueous solu-
tion than in the cases of 1d and 3b. To investigate this
hypothesis, we performed high-level quantum mechanical
calculations of conformational energy differences of the
lowest open and closed forms of the respective hydrogen
bonded representatives 1d, 2d, 3b, and 4b in both a low- and
high-dielectric environment (Figure 14). In the low dielectric
environment, the energy difference between the open and
closed forms is prohibitively high in all cases. In the high
dielectric, however, the energy difference stays high for 1d
and 3b but decreases to about 1.5 kcal/mol for 2d and 4b.
Taken at face value, this energy difference corresponds to a
population of approximately 10% of the higher energy
conformer in aqueous solution. We interpret the energy
differences as a trend only, indicative of a significant popula-
tion of the open conformation in aqueous solution positively
affecting solubility.
The lesson to be learned from systems 2 and 4 is that the
introduction of a structural feature facilitating intramolecu-
lar hydrogen bond formation can be beneficial for the overall
property profile, leading to increased solubility, enhanced
Intramolecular hydrogen bonds are abundant in medicinal
chemistry; one will find examples in any recent issue of this

2610 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 6
Kuhn et al.
journal.40,41 We are confident that our analysis of topologies
and properties of this interaction type provides a useful basis
for its more directed and rational use.
(12) Infantes, L.; Motherwell, W. D. S. Hydrogen bond competition
between chemical groups: new methodology and the Cambridge
Structural Database. Z. Kristallogr. 2005, 220, 333–339.
(13) Allen, F. The Cambridge Structural Database: a quarter of a
million crystal structures and rising. Acta Crystallogr. B 2002, 58,
380–388.
(14) Bilton, C.; Allen, F. H.; Shields, G. P.; Howard, J. A. K. Intramo-
lecular hydrogen bonds: common motifs, probabilities of forma-
tion and implications for supramolecular organization. Acta
Crystallogr. B 2000, 56, 849–856.
Acknowledgment. Synthesis of model compounds by
€
Stefan Burli, exhaustive spectroscopic characterization by
Christian Bartelmus, Markus Burkler, Dr. Joseph Schneider,
€
and Dr. Monira Siam, and physicochemical measurements by
€
Isabelle Parrilla, Frank Senner, and Bjorn Wagner are grate-
fully acknowledged.
(15) http://www.rcsb.org/pdb/.
(16) Bernstein, F. C.; Koetzle, T. F.; Williams, G. J. B.; Meyer, E. F., Jr.;
Brice, M. D.; Rodgers, J. R.; Kennard, O.; Shimanouchi, T.;
Tasumi, M. The protein data bank: A computer-based archival
file for macromolecular structures. J. Mol. Biol. 1977, 112,
535–542.
Supporting Information Available: Synthesis and character-
ization data for model compounds (pp S2-S48); methods for
measurement of physicochemical properties (pp S49-S50);
(17) http://www.desertsci.com/.
(18) Prous Science, S. A. Prous Science Integrity. http://integrity.prous.
com
(19) http://www.chemcomp.com/.
(20) http://www.schrodinger.com/.
(21) Woods, W. G.; Crawford, R. F. 1,1-Dialkyl-3-(pyridyl)uea
Hercides. U.S. Patent 3330641, 1967.
(22) Pilyugin, V. S.; Sapozhnikov, Y. E.; Davydov, A. M.; Chikisheva,
G. E.; Vorob’eva, T. P.; Klimakova, E. V.; Kiseleva, G. V.;
Kuznetsova, S. L.; Davletov, R. D.; Sapozhnikova, N. A.;
Yumadilov, R. K. 13C NMR spectra and biological activity of
N-(1H-benzimidazol-2-yl)benzamides. Russ. J. Gen. Chem. 2006,
76, 1653–1659.
(23) Terzioglu, N.; van Rijn, R. M.; Bakker, R. A.; De Esch, I. J. P.;
Leurs, R. Synthesis and structure-activity relationships of indole
and benzimidazole piperazines as histamine H4 receptor antago-
nists. Bioorg. Med. Chem. Lett. 2004, 14, 5251–5256.
(24) Fischer, H.; Kansy, M.; Wagner, B. Determination of High
Lipophilicity Values. U.S. Patent 2006211121, European Patent
1705474, 2006.
(25) Drug Bioavailability: Estimation of Solubility, Permeability, Ab-
sorption and Bioavailability; Van de Waterbeemd, H., Lennernas, H.,
Artursson, P., Eds.; Methods and Principles in Medicinal Chemistry,
Vol. 18; Wiley-VCH: Weinheim, Germany, 2003; 579 pp.
(26) Kansy, M.; Avdeef, A.; Fischer, H. Advances in screening
for membrane permeability: high-resolution PAMPA for medi-
cinal chemists. Drug Discovery Today: Technol. 2004, 1, 349–355.
(27) BioByte Corp. http://www.biobyte.com.
(28) Laurence, C.; Brameld, K. A.; Graton, J.; Le Questel, J.-Y.;
Renault, E. The pKBHX database: toward a better understanding
of hydrogen-bond basicity for medicinal chemists. J. Med. Chem.
2009, 52, 4073–4086.
(29) Gilli, G.; Bellucci, F.; Ferretti, V.; Bertolasi, V. Evidence for
resonance-assisted hydrogen bonding from crystal-structure cor-
relations on the enol form of the b-diketone fragment. J. Am.
Chem. Soc. 1989, 111, 1023–8.
(30) Terrett, N. K.; Bell, A. S.; Brown, D.; Ellis, P. Sildenafil (Viagra), a
potent and selective inhibitor of type 5 cGMP phosphodiesterase
with utility for the treatment of male erectile dysfunction. Bioorg.
Med. Chem. Lett. 1996, 6, 1819–1824.
detailed CSD statistics for CdO H-N, CdO H-O,
3 3 3 3 3 3
N
H-N, N H-O, and alkoxy H-N interactions (pp
3 3 3 3 3 3
3 3 3
S51-S55); a comparison between CSD and PDB statistics of
CdO H-N and CdO H-O interactions (p S56); crystal
3 3 3 3 3 3
structure examples of N H-N and alkoxy H-N interac-
3 3 3
3 3 3
tions in five-membered rings with particularly high propensities
(p S57). This material is available free of charge via the Internet
at http://pubs.acs.org.
References
(1) Nomura, M.; Kinoshita, S.; Satoh, H.; Maeda, T.; Murakami, K.;
Tsunoda, M.; Miyachi, H.; Awano, K. (3-Substituted benzyl)thia-
zolidine-2,4-diones as structurally new antihyperglycemic agents.
Bioorg. Med. Chem. Lett. 1999, 9, 533–538.
(2) Harter, W. G.; Albrect, H.; Brady, K.; Caprathe, B.; Dunbar, J.;
Gilmore, J.; Hays, S.; Kostlan, C. R.; Lunneya, B.; Walker, N. The
design and synthesis of sulfonamides as caspase-1 inhibitors.
Bioorg. Med. Chem. Lett. 2004, 14, 809–812.
(3) Van Zandt, M. C.; Sibley, E. O.; McCann, E. E.; Combs, K. J.;
Flam, B.; Sawicki, D. R.; Sabetta, A.; Carrington, A.; Sredy, J.;
Howard, E.; Mitschler, A.; Podjarny, A. D. Design and synthesis
of highly potent and selective (2-arylcarbamoyl-phenoxy)-
acetic acid inhibitors of aldose reductase for treatment of chronic
diabetic complications. Bioorg. Med. Chem. 2004, 12, 5661–
5675.
(4) Hodge, C. N.; Pierce, J. A diazine heterocycle replaces a six-
membered hydrogen-bonded array in the active site of scytalone
dehydratase. Bioorg. Med. Chem. Lett. 1993, 3, 1605–1608.
(5) Furet, P.; Bold, G.; Hofmann, F.; Manley, P.; Meyer, T.; Altmann,
K.-H. Identification of a new chemical class of potent angiogenesis
inhibitors based on conformational considerations and database
searching. Bioorg. Med. Chem. Lett. 2003, 13, 2967–2971.
(6) Furet, P.; Caravatti, G.; Guagnano, V.; Lang, M.; Meyer, T.;
Schoepfer, J. Entry into a new class of protein kinase inhibitors
by pseudo ring design. Bioorg. Med. Chem. Lett. 2008, 18,
897–900.
(31) Zhang, K. Y. J.; Card, G. L.; Suzuki, Y.; Artis, D. R.; Fong, D.;
Gillette, S.; Hsieh, D.; Neiman, J.; West, B. L.; Zhang, C.; Milburn,
M. V.; Kim, S.-H.; Schlessinger, J.; Bollag, G. A glutamine switch
mechanism for nucleotide selectivity by phosphodiesterases. Mol.
Cell 2004, 15, 279–286.
(7) Menear, K. A.; Adcock, C.; Alonso, F. C.; Blackburn, K.; Copsey,
L.; Drzewiecki, J.; Fundo, A.; Le Gall, A.; Gomez, S.; Javaid, H.;
Lence, C. F.; Martin, N. M. B.; Mydlowski, C.; Smith, G. C. M.
Novel alkoxybenzamide inhibitors of poly(ADP-ribose) polymer-
ase. Bioorg. Med. Chem. Lett. 2008, 18, 3942–3945.
(32) Yathirajan, H. S.; Nagaraj, B.; Nagaraja, P.; Bolte, M. Sildenafil
citrate monohydrate. Acta Crystallogr. E 2005, 61, 489–491.
(33) Sudha, L. V.; Sathyanarayana, D. N. Infrared and proton NMR
study of molecular conformation of some N,N0-arylalkylureas.
J. Mol. Struct. 1984, 125, 89–96.
(8) Lord, A.-M.; Mahon, M. F.; Lloyd, M. D.; Threadgill, M. D.
Design, synthesis, and evaluation in vitro of quinoline-8-carbox-
amides,
a
new class of poly(adenosine-diphosphate-ribo-
se)polymerase-1 (PARP-1) inhibitor. J. Med. Chem. 2009, 52,
868–877.
(9) Ashwood, V. A.; Field, M. J.; Horwell, D. C.; Julien-Larose, C.;
Lewthwaite, R. A.; McCleary, S.; Pritchard, M. C.; Raphy, J.;
Singh, L. Utilization of an intramolecular hydrogen bond to
increase the CNS penetration of an NK1 receptor antagonist.
J. Med. Chem. 2001, 44, 2276–2285.
(10) Sasaki, S.; Cho, N.; Nara, Y.; Harada, M.; Endo, S.; Suzuki, N.;
Furuya, S.; Fujino, M. Discovery of a thieno[2,3-d]pyrimidine-2,
4-dione bearing a p-methoxyureidophenyl moiety at the 6-position:
a highly potent and orally bioavailable non-peptide antagonist for
the human luteinizing hormone-releasing hormone receptor.
J. Med. Chem. 2003, 46, 113–124.
(11) Rezai, T.; Bock, J. E.; Zhou, M. V.; Kalyanaraman, C.; Lokey,
R. S.; Jacobson, M. P. Conformational flexibility, internal hydro-
gen bonding, and passive membrane permeability: successful in
silico prediction of the relative permeabilities of cyclic peptides.
J. Am. Chem. Soc. 2006, 128, 14073–14080.
(34) Ferretti, V.; Pretto, L.; Tabrizi, M. A.; Gilli, P. Role of strong
intramolecular N-H N hydrogen bonds in determining the
3 3 3
conformation of adenosine-receptor ligands. Acta Crystallogr. B
2006, 62, 634–641.
(35) Jansma, A.; Zhang, Q.; Li, B.; Ding, Q.; Uno, T.; Bursulaya, B.;
Liu, Y.; Furet, P.; Gray, N. S.; Geierstanger, B. H. Verification of a
designed intramolecular hydrogen bond in a drug scaffold by
nuclear magnetic resonance spectroscopy. J. Med. Chem. 2007,
50, 5875–5877.
(36) Zhu, G.-D.; Gandhi, V. B.; Gong, J.; Thomas, S.; Luo, Y.; Liu, X.;
Shi, Y.; Klinghofer, V.; Johnson, E. F.; Frost, D.; Donawho, C.;
Jarvis, K.; Bouska, J.; Marsh, K. C.; Rosenberg, S. H.; Giranda,
V. L.; Penning, T. D. Synthesis and SAR of novel, potent and orally
bioavailable benzimidazole inhibitors of poly(ADP-ribose) poly-
merase (PARP) with a quaternary methylene-amino substituent.
Bioorg. Med. Chem. Lett. 2008, 18, 3955–3958.

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 6 2611
(37) Allen, F. H.; Garner, S. E. Symmetry-modified conformational
mapping and classification of the medium rings from crystallo-
graphic data. III. endo-Unsaturated seven-membered rings. Acta
Crystallogr. B 1994, 50, 395–404.
(38) Nemethy, G.; Printz, M. P. The gamma-turns, a possible folded
conformation of the polypeptide chain. Comparison with the beta-
turn. Macromolecules 1972, 5, 755–758.
(39) Foloppe, N.; Fisher, L. M.; Howes, R.; Kierstan, P.; Potter, A.;
Robertson, A. G. S.; Surgenor, A. E. Structure-based design of
novel Chk1 inhibitors: insights into hydrogen bonding and pro-
tein-ligand affinity. J. Med. Chem. 2005, 48, 4332–4345.
(40) Bikker, J. A.; Brooijmans, N.; Wissner, A.; Mansour, T. S.
Kinase domain mutations in cancer: implications for small
molecule drug design strategies. J. Med. Chem. 2009, 52, 1493–
1509.
(41) Schroeder, G. M.; An, Y.; Cai, Z.-W.; Chen, X.-T.; Clark, C.;
Cornelius, L. A. M.; Dai, J.; Gullo-Brown, J.; Gupta, A.; Henley,
B.; Hunt, J. T.; Jeyaseelan, R.; Kamath, A.; Kim, K.; Lippy, J.;
Lombardo, L. J.; Manne, V.; Oppenheimer, S.; Sack, J. S.; Schmidt,
R. J.;Shen, G.; Stefanski, K.; Tokarski, J. S.; Trainor, G. L.;Wautlet,
B. S.; Wei, D.; Williams, D. K.; Zhang, Y.; Zhang, Y.; Fargnoli, J.;
Borzilleri, R. M. Discovery of N-(4-(2-amino-3-chloropyridin-4-
yloxy)-3-fluorophenyl)-4-ethoxy-1-(4-fluorophenyl)-2-oxo-1,2-dihy-
dropyridine-3-carboxamide (BMS-777607), a selective and orally
efficacious inhibitor of the Met kinase superfamily. J. Med. Chem.
2009, 52, 1251–1254.