Skip to main content
NCBI home page
MedChemComm logoLink to MedChemComm
. 2017 Apr 27;8(6):1143–1151. doi: 10.1039/c7md00101k

High throughput methods to measure the propensity of compounds to form intramolecular hydrogen bonding†

Giulia Caron a,✉, Maura Vallaro a, Giuseppe Ermondi a
PMCID: PMC6071820  PMID: 30108824

graphic file with name c7md00101k-ga.jpgImplementation of IMHB considerations in drug discovery needs robust and validated descriptors to experimentally verify the propensity of compounds to exhibit IMHBs.

Abstract

Implementation of IMHB considerations in drug discovery needs robust and validated descriptors to experimentally verify the propensity of compounds to exhibit IMHBs. The first part of the paper presents an overview of the most common techniques to measure the propensity of compounds to form IMHBs. Then we review and discuss recently proposed high throughput (HT) physicochemical descriptors (i.e. Δlog Poct–tol, EPSA and log k′80 PLRP-S) which provide the same information. Analysis of the available data enabled us to extract guidelines for the application of these descriptors in drug discovery programs.

Introduction

The formation of an intramolecular hydrogen bond (IMHB) requires the simultaneous presence on the same molecule of a hydrogen bond donor (HBD) group and a hydrogen bond acceptor (HBA) atom, the D–H pair and A in Fig. 1A, respectively. Any IMHB is governed by an equilibrium (Fig. 1B) described by a thermodynamic constant here named KIMHB which quantifies the relative amount of closed (IMHB) and open (no IMHB) forms (KIMHB = [open form]/[closed form]).

Fig. 1. Intramolecular hydrogen bond (IMHB). A) The two bond lengths (dDH and dHA) and the bond angle (Φ) that characterize the interaction are shown. B) The equilibrium governing the formation of an IMHB.

Fig. 1

Open in a new tab

The growing interest of medicinal chemists in IMHB is described in some recent literature reports, mostly provided by industrial researchers. In 2006 Rezai et al. supported that IMHB formation is critical for passive membrane permeability of cyclic peptides.1 In 2010 Kuhn et al. (Roche) proposed a rational use of IMHB in medicinal chemistry.2 In 2011 Alex et al. (Pfizer) described how IMHB could modulate lipophilicity in beyond rule of five (bRo5) chemical space.3 In 2012 Desai and coworkers (Eli Lilly) reported about the relevance of introducing IMHBs to increase membrane permeability and/or decrease P-gp efflux.4 Since then a significant number of papers published in medicinal chemistry journals used the formation of IMHBs as a strategy to obtain soluble, permeable and potent candidates. A few of these papers about the use of IMHB in drug discovery have been listed and discussed by some of us in a recent paper.5 Finally, some months ago, Giordanetto and coworkers remarked how the formation or disruption of IMHB could be one valuable strategy to transform molecules into drugs.6

IMHBs could improve permeability but not necessarily at the expense of solubility and binding affinity.5 In fact, a molecule could form IMHBs in non-polar environments (e.g. lipid membranes) and thus mask its polar properties. Meanwhile, the same molecule could expose HBD and HBA groups in polar media (e.g. aqueous environments). This is expected to be optimal for drug discovery purposes as it contributes to high aqueous solubility as well as high membrane permeability.3,7

Successful implementation of IMHB considerations in drug discovery programs requires experimental tools which assess the propensity of compounds to form IMHBs. In fact, molecular modeling could suggest structural modifications to transform a lead unable to form IMHB into an optimized lead with some propensity to form IMHB (and thus expected to have better ADME-Tox properties), but, once synthesized, the new molecule should be experimentally checked for its effective capacity for forming IMHBs.

Traditional methods to assess the propensity of compounds to form IMHBs

NMR techniques are the most common tools to measure the propensity of compounds to form IMHBs and the most appreciated by medicinal chemists.

One of the most convenient strategies to use NMR for the quantitative assessments of IMHB is described by Abraham and coworkers8 and Lessene et al.9 Briefly, the authors suggest measuring 1H NMR chemical shifts in chloroform (ε = 4.8) and DMSO (ε = 46.7). Then the difference in the chemical shifts of an OH or NH group in these two solvents can be converted into the hydrogen bond acidity (=hydrogen bonding donor (HBD) properties).

The formation of IMHBs also changes the temperature dependence of amide protons' chemical shifts (ΔσHN), and the temperature coefficients (TCs, i.e. ΔσHN/ΔT) can therefore be used to evaluate IMHBs.10,11 Amides involved in strong IMHBs tend to have less negative temperature coefficients than those that are not H-bonded. Although widely applied in the recent literature, this approach was developed for proteins and thus calibrations are required to extend its application to peptides (as reported by Alex and coworkers3) and small molecules.

A slow rate of deuterium exchange due to IMHB formation can also be used to indicate IMHB formation.12

Finally, 2D NMR techniques could also provide information about the propensity of compounds to form IMHB as for instance applied by Bockus and coworkers.13 Their review is beyond the scope of this paper.

NMR is a valuable technique to obtain structural information, but up to now it cannot be used efficiently in a high throughput (HT) mode, despite the progress in instrumentation increasing its performance in terms of the time and amount of compound required for the analysis.

In principle, direct experimental evidence of the presence of IMHB can be achieved by X-ray2,14 and neutron diffraction.15,16 The latter should be preferred since X-ray crystallography has severe limitations in determining the position of hydrogen atoms.15 However, the limited number of available diffraction measures collected in databases (i.e. the Cambridge Structural Database, CSD) does not permit an extensive use of these data.17

Crystallographic data are obviously not available during early drug discovery but they are very useful for retro prospective studies. This was chiefly shown by Kuhn and coworkers who, also supported by preliminary studies by Bilton et al.,14 performed exhaustive searches in crystal structure databases and provided the probability for IMHB formation of five- to eight-membered ring systems of relevance in drug discovery.18 More recently Doak and coworkers19 used crystallography to explore the potentialities of macrocycles as beyond Rule of 5 (bRo5) drugs.

Although perhaps biased by crystal packing effects and poorly refined geometries (mostly when PDB files are investigated),20 crystal structures are essential in retrospective studies since they are experimental data. Moreover, a comparative analysis of X-ray structures crystallised from solvents with different polarities could allow highlighting the impact of the environment (see below) on the propensity of compounds to form IMHBs.21 For instance this strategy was applied to investigate the unusual properties of cyclosporine A,3,22 a potent immunosuppressive drug to prevent graft rejection.

Infrared spectroscopy (IR) could also provide insight into IMHB existence since the formation of an IMHB causes a shift to lower wavenumbers in the IR stretching vibration band of the D–H bond (Fig. 1A).23 Unfortunately, the characterization of an IMHB by IR requires specific experimental conditions (non-hydrogen-bonding solvents, dilute conditions and solutes with a limited number of amidic NH groups9) and as such is poorly suited for drug discovery purposes.

Hydrogen deuterium exchange experiments coupled with mass spectrometry24 and circular dichroism (CD) are also methods which provide information about IMHB formation. Since they are applied to peptidic structures,25 they do not have an extensive application in medicinal chemistry.

As discussed above, the throughput of NMR and other traditional methods is still limited and this remains a challenge for their use in understanding structure–activity relationships (SAR) and driving rapid cycles of compound optimization. Therefore, there is a growing need for high throughput (HT) methods which require a smaller amount of substance.

HT physico-chemical descriptors to measure the propensity of compounds to form IMHBs

Since the pivotal paper by Lipinski et al.,26 physico-chemical descriptors are expected to be extensively used in drug discovery. Unfortunately, this is only partially true. For instance pKa is largely under-scrutinized27–29 and although a number of combined molecular descriptors could be used as sources of structural information,30,31 medicinal chemists often only use calculated log Poct and eventually experimentally determine log Doct in their daily work. However, to speed up the process of integrating IMHB considerations in drug discovery, ad hoc HT physico-chemical descriptors are needed.29 And in fact, a few HT descriptors have been recently proposed to measure the propensity of compounds to form IMHBs.

log Poct (or log Doct)

log Poct (or log Doct when ionisable compounds are considered) could in principle be used as a descriptor of the propensity of compounds to form IMHBs. Since log Poct is the most common lipophilicity descriptor, it would be fine if it could be routinely used for this purpose. Unfortunately, this is not often possible. In fact, in an octanol/water system the two phases (water and octanol saturated with water) provide a polar environment in which only strong IMHBs could be formed (see Discussion). This means that log Poct can only check if a compound has high but not intermediate or low propensity to form IMHBs. The failure of log Poct (or log Doct) in assessing intermediate capacity of compounds to form IMHB is a major limitation if one has interest in obtaining molecular chameleons, i.e. compounds that can arrange their physico-chemical profile and thus ADME properties according to the environment.32,33

Some papers reported about the use of log Poct to assess the (high) propensity of compounds to form IMHB. In these studies, pair analysis is recommended, namely, the comparison between a compound with a substructure prone to IMHB formation and a control compound incapable of forming that bond. For instance, Kuhn showed that 3a, the control, is unable to form any IMHB and thus it has a lower log Poct than 3b (0.68 vs. 1.39, respectively) which has high propensity to form IMHB2 (Fig. 2). This example reveals the difficulties of finding good controls for given samples (see Discussion). 3a in fact bears an additional methyl group which is expected to significantly increase molecular lipophilicity and thus making doubtful the comparison with 3b.

Fig. 2. log Poct detects compounds with high propensity to form IMHB: 3a, the control, is unable to form IMHB and thus it is more polar than 3b which has propensity to form IMHB.

Fig. 2

Open in a new tab

Another application of log Doct to highlight the propensity of compounds to form strong IMHBs was reported in 2015 by Over et al.34 who published about the physicochemical properties of a series of 8 stereoisomeric lactam inhibitors of T. Cruzi35 (Fig. 3).

Fig. 3. Chemical structures and lipophilicity data of the eight stereoisomeric lactam inhibitors of T. Cruzi described by Over and coworkers.34.

Fig. 3

Open in a new tab

In this elegant retrospective study, log Doct of compounds 1–4 (Fig. 3) is significantly higher than log Doct of compounds 5–8. This (and other experimental evidence) showed that four (1–4) out of eight stereoisomers have high propensity to form an IMHB. The remaining four stereoisomers (5–8) have lower propensity and do not form any IMHB in an apolar environment and thus exhibit lower lipophilicity.

Δlog Poct–alk

The most known physico-chemical descriptor for measuring the propensity of compounds to form IMHBs is Δlog Poct–alk36 which is the difference between log Poct (the logarithm of the partition coefficient P in an octanol/water system) and log Palk (the logarithm of the partition coefficient P in an alkane/water system). The interest in this descriptor mainly stems from the study by Young and coworkers who showed that Δlog Poct–alk could be used to describe brain penetration of drug candidates.37 Unfortunately, the experimental determination of this descriptor suffers from severe limitations mostly related to the poor solubility of compounds in alkanes.

To overcome this limitation, Δlog Poct–tol36 (i.e. the difference between log Poct and log Ptol, the logarithm of the partition coefficient P in a toluene/water system) was recently characterized and implemented in some drug discovery programs. Δlog Poct–tol is considerably more convenient than Δlog Poct–alk since candidates are more soluble in toluene, the information content (i.e. the solute exposed HBD capacity) being pretty similar.36 To obtain Δlog Poct–tol you have to measure both log Poct and log Ptol. Potentiometry as implemented using a Sirius T3 automatic titrator (; http://www.sirius-analytical.com/) and a miniaturized shake-flask can be used for this purpose, also starting from DMSO stock solutions. Also chromatographic tools (e.g. E log D (ref. 38)) could be used for the determination of the octanol–water distribution coefficients at pH 7.4.

Δlog Poct–tol is a clean descriptor of exposed HBD properties. This is shown (Fig. 4) by the Block Relevance (BR) analysis, a computational tool that allows the interpretation of 3D-QSAR/QSPR PLS models based on VolSurf+ (VS+) descriptors.39,40 In BR analysis, VS+ descriptors are aggregated into property-related groups (blocks, their list and significance are shown in Fig. 4B), thus providing a convenient framework for comparison and interpretation of molecular descriptors. The plot in Fig. 4A shows that Δlog Poct–tol is mostly governed by HBD solutes' properties (red block, positive sign) and it is poorly influenced by steric/hydrophobic descriptors (green and yellow blocks).41

Fig. 4. The BR analysis reveals that HBD properties (red block) govern (positive sign) Δlog Poct–tol.41 A) BR analysis graphical output and B) block significance.

Fig. 4

Open in a new tab

When a compound has a single HBD group, the interpretation of Δlog Poct–tol is straightforward. If Δlog Poct–tol is close to 0 then the compound has high propensity to form IMHBs. This also stems from Fig. 4A. In fact, only the red block is the major determinant of Δlog Poct–tol whereas the remaining blocks are less important (they also have small and comparable positive and negative contributions that tend to cancel one another). If the compound employs the HBD group in forming IMHB, the contribution of the red block decreases and thus Δlog Poct–tol is expected to be about 0.

Two examples are shown in Fig. 5.36 The large propensity to form IMHB of 1 is revealed by its value of Δlog Poct–tol which is about 0, and NMR experiments in the original paper36 supported this result. Conversely, quinine has a Δlog Poct–tol of about 2.3 and thus it does not form IMHBs. When Δlog Poct–tol is about 0, log Ptol is close to log Poct and this is possible only if the compound in toluene masks its polarity, i.e. it forms IMHBs. Conversely, the higher the Δlog Poct–tol, the lower the propensity of the molecule to form IMHBs.

Fig. 5. Δlog Poct–tol estimates the different propensities of forming IMHBs of compounds 1 (high propensity) and quinine (low propensity). The two compounds have a single HBD group.

Fig. 5

Open in a new tab

The interpretation of Δlog Poct–tol is complex when more HBD groups are present in the chemical structure and if only one of them could be involved in the formation of IMHB. In these situations, Δlog Poct–tol will be larger than zero since Δlog Poct–tol is a measure of the whole molecular HBD properties and some HBD could be exposed and not involved in the formation of IMHBs. In these cases, the interpretation of Δlog Poct–tol has to be performed as described by Shalaeva and coworkers.36 If Δlog Poct–tol of the sample is smaller than Δlog Poct–tol of the control, then the sample has a high propensity to form an IMHB. Conversely, if Δlog Poct–tol of the sample is larger than Δlog Poct–tol of the control, the sample has a low propensity to form IMHBs.

Summing up, Δlog Poct–tol is a powerful descriptor of the propensity of compounds to form IMHBs but solubility issues still persist and some expertise is required in its interpretation. For these reasons, chromatographic systems which can be easily set up with a variety of stationary and mobile phases to reproduce a few biological environments are preferred in industrial laboratories.42

EPSA

EPSA43 is an experimental descriptor of exposed polarity that can be obtained from supercritical fluid chromatography (SFC) retention.44 Briefly, a polar stationary phase (Chirex 3014) and a nonpolar mobile phase (supercritical CO2 with the addition of methanol as a modifier) create an apolar environment that favors folded conformations. Using a pairwise approach, EPSA allows one to distinguish samples with propensity to form IMHBs from controls with no propensity to form IMHBs. In fact, BR analysis showed that EPSA mostly describes HBD properties of compounds and is poorly influenced by the size of the sample.45

An application of EPSA is shown in Fig. 6 for a couple of regioisomers. The sample 1 has propensity to form IMHBs and thus it shows a lower polarity (EPSA = 108) than the control 2 (EPSA = 119).43

Fig. 6. EPSA estimates the different propensities of forming IMHBs of compounds 1 (the sample, high propensity to form IMHBs) and 2 (the control, no propensity). E log D values support the limitations of the octanol/water system in describing intermediate propensity of forming IMHBs (see text for more details).

Fig. 6

Open in a new tab

Also EPSA however suffers from some limitations. In particular, the HBD properties of the molecule can be underestimated by EPSA when the structure includes a number of HBA groups.45 Although not completely clarified, this evidence limits the domain of application of the method.

log k′80 PLRP-S

log k′80 PLRP-S is a molecular descriptor obtained using a reverse-phase (RP) chromatographic system with a polystyrene/divinylbenzene polymeric column as a stationary phase46 and a mixture of 80 : 20 = acetonitrile : buffer as a mobile phase.47 The stability of the PLRP-S polymeric column under different pH conditions enables measurements on the whole physiological pH range and thus compounds can also be analysed in their neutral form (the influence of ionization on the capacity for forming IMHB is discussed elsewhere48).

BR analysis showed that log k′80 PLRP-S and log Ptol are governed by a similar balance of intermolecular forces between the solutes and the system.47 In particular, given the relevant role played by HBD properties, the PLRP-S system is expected to detect the presence of IMHBs.47 In fact, samples with propensity to form IMHBs (less polar) are more retained than controls with no propensity to form IMHB (more polar). For instance, the two regioisomers reported in Fig. 7 can easily be distinguished using log k′80PLRP-S. Sample 5 has high propensity to form IMHBs and thus its log k′80 PLRP-S value is higher (–0.44) than that of control 6 (–1.04), which is more polar and thus less retained.

Fig. 7. log k′80 PLRP-S estimates the different propensities of forming IMHBs of compounds 5 (the sample, high propensity to form IMHBs) and 6 (the control, no propensity).

Fig. 7

Open in a new tab

Discussion

Most descriptors used to monitor the propensity of compounds to form IMHBs (also NMR chemical shift) include interferences (e.g. hydrophobic, not desired contributions). Therefore, a pairwise analysis using matched compound pairs is often needed to eliminate interferences. Matched pairs refer to a compound with a substructure prone to IMHB formation along with a control compound incapable of forming that bond, typically a regioisomer.43 Although the use of matched pairs does not require particular effort when working with a series of compounds, in the absence of regioisomers the identification of correct controls is not a trivial task.43

Measuring the propensity of compounds to form IMHBs mostly consists in measuring the variation in the exposed hydrogen bonding donor (HBD) properties of a sample in relation to its control. One could argue, why are HBD groups more important than hydrogen bonding acceptor (HBA) moieties? Although some authors do not share the same opinion,49 we believe that the impact of HBD properties on the formation of IMHBs is more critical than the HBA capacity since most HBD groups with few exceptions are also HBA but not all HBA are HBD. In practice, HBD capacity is more featuring than HBA and thus relevant in the formation of IMHBs.

Another major issue to be considered when designing compounds with propensity to form IMHBs is the crucial impact of the environment. In fact, as detailed elsewhere the propensity of compounds to form IMHBs depends on the hydrogen bond acceptor (HBA) and the hydrogen bond donor (HBD) capacity of the involved groups, the chemical structure (e.g. the distance between HBA and HBD and the angle Φ (Fig. 1A)) but also on the polarity of the environment.48,50 The latter is often underestimated but in the human body there is a variety of environments which could be roughly but efficiently characterized by their polarity. There are aqueous environments (ε = 80), apolar environments (e.g. the membrane/protein interior, ε = 2 (ref. 51 and 52)) and a number of intermediate situations. Fig. 8 schematizes the different environments created by a simplified model of cell membrane ideally composed of phospholipids (PPLs). Polar environments due to the polar heads of PPLs (Fig. 8A) shift the open/closed conformational equilibrium (Fig. 8B) towards the open (more polar) form. Conversely, apolar environments associated with the apolar core of PPLs (Fig. 8C) shift the same equilibrium towards the closed (less polar) form.

Fig. 8. The impact of the environment on the propensity of compounds to form IMHBs. A) A polar environment favors the formation of the open form (no IMHB); B) the open/closed conformational equilibrium; and C) an apolar environment favors the formation of IMHBs.

Fig. 8

Open in a new tab

Summing up, experimental conditions created by any method which assesses the propensity of compounds to form IMHBs are unique and specific (for instance, one method could reproduce the conditions shown in Fig. 8A and another those in Fig. 8C). Thus, two methods endowed with different experimental conditions could provide very different information about the capacity of samples for forming IMHBs.

Since the propensity of compounds to form IMHBs also depends on the environment, the polarity of the media created by any method determines the range of propensities for forming IMHBs that can be caught by the different descriptors. In practice, the more apolar the environment, the more sensible the descriptor to catch small propensities to form IMHBs, whereas in a polar environment only compounds with strong propensity to form IMHBs can be detected.

Although we could not experimentally measure the dielectric constant (ε) of the systems, we suggest ranking the HT physico-chemical descriptors previously described according to their polarity as follows:Δlog Poct–tol > log k′80 PLRP-S > EPSA ≫ log Poctwhere Δlog Poct–tol can catch modest and intermediate capacity of compounds to form IMHBs, followed by log k′80 PLRP-S, EPSA and finally log Poct which provides the most polar environment and thus can detect only compounds with high propensity to form IMHBs (roughly this scale corresponds to the sensibility of the system for the exposed HBD properties, as evidenced by BR analysis40,41,45,47). As a proof of this statement, let's consider compounds 1 (the sample) and 2 (the control) in Fig. 6 for which E log D and EPSA values are reported. 1 and 2 share very similar E log D (ref. 38) (0.9 and 0.8) but significantly different EPSA values (108 and 119). These data suggest that 1 has an intermediate/low (=not high) propensity to form IMHBs.

Is it possible to use Δlog Poct–tol, log k′80 PLRP-S and EPSA in any project? In principle, yes, since the three physicochemical descriptors could be applied to any chemical space. For instance, EPSA was determined for small molecules,43 but also for cyclic peptides,53 oxytocin and some of its macrocyclic derivatives.54 Besides the application to small molecules,36 we also used Δlog Poct–tol to investigate the IMHB profiles of macrocycles. In particular, we measured for rifampicin Δlog Poct–tol = 0.05 (unpublished value). This value supports (see above) the ability of the macrocycle to form IMHBs already experimentally revealed by the crystallographic structure (CSD code: MAPHIW). Also log k′80 PLRP-S was obtained for compounds belonging to different chemical spaces (unpublished data).

Δlog Poct–tol41 and EPSA45 measure the variation in the exposed hydrogen bonding donor (HBD) properties of compounds and thus they could be used to predict permeability. For instance, EPSA is relevant in describing the permeability of hydrophilic small molecules (clog P < 1)45 and cyclic peptides.53 Briefly it has been found that either a cyclic peptide or a hydrophilic small molecule with an EPSA value below 100 has a much greater chance of being permeable than a compound with a higher EPSA. These findings were applied to the design of cyclic peptides and in particular EPSA was used as a filter (100 was used as the cut-off) to not miss potentially permeable peptides.53,54 EPSA in fact was shown to respond to discrete modification, such as single amino acid sequence change and N-methylation.53N-Methylation of residues not involved in IMHB is a widely-known tactic to reduce exposed polarity of peptides. However, distinguishing the residues involved in IMHBs from the others is not an easy task. To do that, it is possible to evaluate the difference in EPSA (ΔEPSA) between the parent and the different methylated derivatives.54 If ΔEPSA is negative, a decrease in exposed polarity is found (i.e. N-methylation has been performed on a residue not involved in IMHB) and thus the permeability of the derivative is expected to be higher than that of the parent. The reverse is true if ΔEPSA is positive.54 Δlog Poct–tol could replace EPSA and be used to provide indications about structural modifications promoting the formation of IMHBs and to predict permeability. However, up to now, implementation of Δlog Poct–tol in industrial projects has not be reported and thus its usefulness in this field remains to be proven.

log k′80 PLRP-S alone cannot be used to predict permeability since it is not a pure descriptor of exposed HBD properties.47 To enable its application in this field, we need to limit/exclude the hydrophobic component. For instance, we could calculate Δlog k′ PLRP-S, i.e. the difference in log k’80 PLRP-S between the candidate expected to form IMHB and the parent unable to form IMHBs, and relate it to passive permeability. Work along these lines is in due course in our laboratories.

The strengths and limitations of HT physicochemical descriptors to experimentally determine the compounds' propensity to form IMHB are summarized in Table 1.

Table 1. Strengths (pros) and limitations (cons) of the physicochemical descriptors discussed in the text.

Descriptor Pros Cons
log Poct Often available for most candidates Not suited for the identification of molecular chameleons
EPSA Easy to obtain; already implemented in pharmaceutical companies for design purposes as a surrogate for permeability;53 Its interpretation could cause misunderstanding when multiple HBA groups are present in the molecule45
log k′80 PLRP-S Easy to obtain Contaminated by hydrophobicity
Δlog Poct–tol Pure descriptor of HBD properties Compound's solubility in toluene; two experimental determinations should be performed
Open in a new tab

The complexity of the topic highlighted here and in two more papers5,48 suggests that the prediction of the propensity of compounds to form IMHBs is not an easy task and a review of the in silico methods focusing on IMHBs goes beyond the aim of this paper.

However, there are two computational descriptors that deserve some comments in this minireview.

The first is COSMO-RS Δlog Poct–tol. This descriptor is provided by commercial software which is based on a quantum chemical approach combined with statistical thermodynamic treatment of surface interactions.55,56 COSMO-RS Δlog Poct–tol gives good predictions of Δlog Poct–tol for small molecules,36,47 but the high CPU resources required discourage its application to large and flexible molecules, such as cyclic peptides and macrocycles.

Conversely, the Polar Surface Area (PSA) is a very simple molecular descriptor widely used in medicinal chemistry also for discussing IMHB properties. However, PSA has severe limitations as a descriptor of HB properties since a) HBD and HBA strength and HB directionality are not considered and b) it does not distinguish HBA from HBD contributions. A fortiori information about the propensity of compounds to form IMHBs cannot be obtained from Topological Polar Surface Area (TPSA), the most common tool to calculate the polar surface area.57 This topic is discussed in more detail in a very recent paper.58

The advances of the field over the next 5 years mostly depend on persuading medicinal chemists that IMHB is a good opportunity to discover new drugs and log Doct is not the only available physicochemical descriptor to be used in drug discovery projects. Once obtained this “cultural” result of obtaining/sharing more data is desirable. In particular, it would be reasonable to find extensive application of these descriptors to cyclic peptides and macrocycles. In fact, numerical filters able to discriminate potential permeable compounds from the others in the very beginning of the drug discovery pipeline could provide practical guidelines for developing orally administered compounds in bRo5 space.

Conclusion

As proven by several papers and an entire session dedicated to IMHB at the 2016 EFMC International Symposium on Medicinal Chemistry (http://www.ldorganisation.com/v2/produits.php?cle_menus=1238915829), implementing IMHB considerations in drug discovery is coming out as a promising strategy for the optimization of drug candidates since the presence of IMHBs could positively impact the triad of solubility, permeability and potency.

Some reliable HT tools to measure the propensity of compounds to form IMHBs have been described and validated. Moreover, they could easily be implemented in property criteria profiles.29 However, for any method it is essential to characterize the provided information and related interferences and avoid overinterpretation of data.

The impact of the environment on the propensity of compounds to form IMHBs suggests that work be carried out using at least two different descriptors obtained in environments with different polarities. This remains true also when NMR studies are performed.

Although HT physico-chemical tools could be applied to both small and large molecules, few applications are reported in medicinal chemistry journals. For these reasons, some effort is still required to set up extensive datasets to subject to statistical analysis and extract more information about the domain of application of these descriptors.

Acknowledgments

We thank Marina Shalaeva from Pfizer, Groton laboratories, for E log D values shown in Fig. 6.

Biographies

graphic file with name c7md00101k-p1.jpg

Open in a new tab

Maura Vallaro, Giuseppe Ermondi and Giulia Caron

Giulia Caron is Associate Professor at the University of Torino and expert in molecular properties and physico-chemical descriptors. In particular, she designs and implements new lipophilicity and polarity determinants to help drug candidate ADME-Tox optimization. Her expertise covers experimental methods and computational tools. In the 2017 EFMC Symposium held in Manchester, she presented a talk about the need of implementing intramolecular hydrogen bonding considerations in drug discovery and how to do it.

Maura Vallaro is the laboratory technician of the CASSMedChem research group and critically provided most of the experimental data concerning the propensity of compounds to form IMHBs.

Giuseppe Ermondi is Associate Professor at the University of Torino and has a strong expertise in the application of in silico strategies to deconvolute information from inter- and intramolecular interactions. He is the main developer of the Block Relevance (BR) analysis, a tool that enables mechanistic interpretation of PLS-based QSAR/QSPR models.

Footnotes

†The authors declare no competing interests.

References

  1. Rezai T., Bock J. E., Zhou M. V., Kalyanaraman C., Lokey R. S., Jacobson M. P. J. Am. Chem. Soc. 2006;128:14073–14080. doi: 10.1021/ja063076p. [DOI] [PubMed] [Google Scholar]
  2. Kuhn B., Mohr P., Stahl M. J. Med. Chem. 2010;53:2601–2611. doi: 10.1021/jm100087s. [DOI] [PubMed] [Google Scholar]
  3. Alex A., Millan D. S., Perez M., Wakenhut F., Whitlock G. A. Med. Chem. Commun. 2011;2:669. [Google Scholar]
  4. Desai P. V., Raub T. J., Blanco M.-J. Bioorg. Med. Chem. Lett. 2012;22:6540–6548. doi: 10.1016/j.bmcl.2012.08.059. [DOI] [PubMed] [Google Scholar]
  5. Caron G., Ermondi G. Future Med. Chem. 2017;9:1–5. doi: 10.4155/fmc-2016-0195. [DOI] [PubMed] [Google Scholar]
  6. Giordanetto F., Tyrchan C., Ulander J. ACS Med. Chem. Lett. 2017;8:139–142. doi: 10.1021/acsmedchemlett.7b00002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Rafi S. B., Hearn B. R., Vedantham P., Jacobson M. P., Renslo A. R. J. Med. Chem. 2012;55:3163–3169. doi: 10.1021/jm201634q. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Abraham M. H., Abraham R. J., Acree W. E., Aliev A. E., Leo A. J., Whaley W. L. J. Organomet. Chem. 2014;79:11075–11083. doi: 10.1021/jo502080p. [DOI] [PubMed] [Google Scholar]
  9. Lessene G., Smith B. J., Gable R. W., Baell J. B. J. Organomet. Chem. 2009;74:6511–6525. doi: 10.1021/jo900871a. [DOI] [PubMed] [Google Scholar]
  10. Baxter N. J., Williamson M. P. J. Biomol. NMR. 1997;9:359–369. doi: 10.1023/a:1018334207887. [DOI] [PubMed] [Google Scholar]
  11. Cierpicki T., Otlewski J. J. Biomol. NMR. 2001;21:249–261. doi: 10.1023/a:1012911329730. [DOI] [PubMed] [Google Scholar]
  12. Haynes C. J. E., Busschaert N., Kirby I. L., Herniman J., Light M. E., Wells N. J., Marques I., Félix V., Gale P. A. Org. Biomol. Chem. 2014;12:62–72. doi: 10.1039/c3ob41522h. [DOI] [PubMed] [Google Scholar]
  13. Bockus A. T., Lexa K. W., Pye C. R., Kalgutkar A. S., Gardner J. W., Hund K. C. R., Hewitt W. M., Schwochert J. A., Glassey E., Price D. A., Mathiowetz A. M., Liras S., Jacobson M. P., Lokey R. S. J. Med. Chem. 2015;58:4581–4589. doi: 10.1021/acs.jmedchem.5b00128. [DOI] [PubMed] [Google Scholar]
  14. Bilton C., Allen F. H., Shields G. P., Howard J. A. K. Acta Crystallogr., Sect. B: Struct. Sci. 2000;56:849–856. doi: 10.1107/S0108768100003694. [DOI] [PubMed] [Google Scholar]
  15. Grabowski S. J. J. Phys. Org. Chem. 2004;17:18–31. [Google Scholar]
  16. Starikov E. B., Steiner T. Acta Crystallogr., Sect. B: Struct. Sci. 1998;54:94–96. [Google Scholar]
  17. Groom C. R., Allen F. H. Angew. Chem., Int. Ed. 2014;53:662–671. doi: 10.1002/anie.201306438. [DOI] [PubMed] [Google Scholar]
  18. Kuhn B., Mohr P., Stahl M. J. Med. Chem. 2010;53:2601–2611. doi: 10.1021/jm100087s. [DOI] [PubMed] [Google Scholar]
  19. Doak B. C., Zheng J., Dobritzsch D., Kihlberg J. J. Med. Chem. 2015;59:2312–2327. doi: 10.1021/acs.jmedchem.5b01286. [DOI] [PubMed] [Google Scholar]
  20. Davis A. M., Teague S. J., Kleywegt G. J. Angew. Chem., Int. Ed. 2003;42:2718–2736. doi: 10.1002/anie.200200539. [DOI] [PubMed] [Google Scholar]
  21. Weng Z. F., Motherwell W. D. S., Allen F. H., Cole J. M. Acta Crystallogr., Sect. B: Struct. Sci. 2008;64:348–362. doi: 10.1107/S0108768108005442. [DOI] [PubMed] [Google Scholar]
  22. Witek J., Keller B. G., Blatter M., Meissner A., Wagner T., Riniker S. J. Chem. Inf. Model. 2016;56:1547–1562. doi: 10.1021/acs.jcim.6b00251. [DOI] [PubMed] [Google Scholar]
  23. Takasuka M., Fujioka T., Iwata T. Vib. Spectrosc. 2005;37:11–20. [Google Scholar]
  24. Hyung S. J., Feng X., Che Y., Stroh J. G., Shapiro M. Anal. Bioanal. Chem. 2014;406:5785–5794. doi: 10.1007/s00216-014-8023-1. [DOI] [PubMed] [Google Scholar]
  25. Perczel A., Hollosi M., Foxman B. M., Fasman G. D. J. Am. Chem. Soc. 1991;113:9772–9784. [Google Scholar]
  26. Lipinski C. A., Lombardo F., Dominy B. W., Feeney P. J. Adv. Drug Delivery Rev. 1997;23:3–26. doi: 10.1016/s0169-409x(00)00129-0. [DOI] [PubMed] [Google Scholar]
  27. Manallack D. T., Prankerd R. J., Yuriev E., Oprea T. I., Chalmers D. K. Chem. Soc. Rev. 2013;42:485–496. doi: 10.1039/c2cs35348b. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Manallack D. T., Prankerd R. J., Nassta G. C., Ursu O., Oprea T. I., Chalmers D. K. ChemMedChem. 2013;8:242–255. doi: 10.1002/cmdc.201200507. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Caron G., Ermondi G., Drug Discovery Today10.1016/j.drudis.2016.11.017 , (in press) . [Google Scholar]
  30. Caron G., Reymond F., Carrupt P., Girault H., Testa B. Pharm. Sci. Technol. Today. 1999;2:327–335. doi: 10.1016/s1461-5347(99)00180-7. [DOI] [PubMed] [Google Scholar]
  31. Caron G., Ermondi G. Mini-Rev. Med. Chem. 2003;3:821–830. doi: 10.2174/1389557033487665. [DOI] [PubMed] [Google Scholar]
  32. Carrupt P. A., Testa B., Bechalany A., El Tayar N., Descas P., Perrissoud D. J. Med. Chem. 1991;34:1272–1275. doi: 10.1021/jm00108a005. [DOI] [PubMed] [Google Scholar]
  33. Whitty A., Zhong M., Viarengo L., Beglov D., Hall D. R., Vajda S. Drug Discovery Today. 2016;21:712–717. doi: 10.1016/j.drudis.2016.02.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Over B., McCarren P., Artursson P., Foley M., Giordanetto F., Grönberg G., Hilgendorf C., Lee M. D., Matsson P., Muncipinto G., Pellisson M., Perry M. W. D., Svensson R., Duvall J. R., Kihlberg J. J. Med. Chem. 2014;57:2746–2754. doi: 10.1021/jm500059t. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Dandapani S., Germain A. R., Jewett I., Le Quement S., Marie J. C., Muncipinto G., Duvall J. R., Carmody L. C., Perez J. R., Engel J. C., Gut J., Kellar D., Siqueira-Neto J. L., McKerrow J. H., Kaiser M., Rodriguez A., Palmer M. A., Foley M., Schreiber S. L., Munoz B. ACS Med. Chem. Lett. 2014;5:149–153. doi: 10.1021/ml400403u. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Shalaeva M., Caron G., Abramov Y. A., Connell T. N. O., Plummer M. S., Yalamanchi G., Farley K. A., Goetz G. H., Philippe L., Shapiro M. J. J. Med. Chem. 2013;56:4870–4879. doi: 10.1021/jm301850m. [DOI] [PubMed] [Google Scholar]
  37. Young R. C., Mitchell R. C., Brown T. H., Ganellin C. R., Griffiths R., Jones M., Rana K. K., Saunders D., Smith I. R., Sore N. E., Wilks T. J. J. Med. Chem. 1988;31:656–671. doi: 10.1021/jm00398a028. [DOI] [PubMed] [Google Scholar]
  38. Lombardo F., Shalaeva M. Y., Tupper K. A., Gao F. J. Med. Chem. 2001;44:2490–2497. doi: 10.1021/jm0100990. [DOI] [PubMed] [Google Scholar]
  39. Ermondi G., Caron G. J. Chromatogr. A. 2012;1252:84–89. doi: 10.1016/j.chroma.2012.06.069. [DOI] [PubMed] [Google Scholar]
  40. Caron G., Vallaro M., Ermondi G. Med. Chem. Commun. 2013;4:1376–1381. [Google Scholar]
  41. Ermondi G., Visconti A., Esposito R., Caron G. Eur. J. Pharm. Sci. 2014;53:50–54. doi: 10.1016/j.ejps.2013.12.007. [DOI] [PubMed] [Google Scholar]
  42. Valkó K. J. Chromatogr. A. 2004;1037:299–310. doi: 10.1016/j.chroma.2003.10.084. [DOI] [PubMed] [Google Scholar]
  43. Goetz G. H., Farrell W., Shalaeva M., Sciabola S., Anderson D., Yan J., Philippe L., Shapiro M. J. J. Med. Chem. 2014;57:2920–2929. doi: 10.1021/jm401859b. [DOI] [PubMed] [Google Scholar]
  44. Desfontaine V., Guillarme D., Francotte E., Nováková L. J. Pharm. Biomed. Anal. 2015;113:56–71. doi: 10.1016/j.jpba.2015.03.007. [DOI] [PubMed] [Google Scholar]
  45. Goetz G. H., Shalaeva M., Caron G., Ermondi G., Philippe L. Mol. Pharmaceutics. 2017;14:386–393. doi: 10.1021/acs.molpharmaceut.6b00724. [DOI] [PubMed] [Google Scholar]
  46. Abraham M. H., Chadha H. S., Leitao R. A., Mitchell R. C., Lambert W. J., Kaliszan R., Nasal A., Haber P. J. Chromatogr. A. 1997;766:35–47. [Google Scholar]
  47. Caron G., Vallaro M., Ermondi G., Goetz G. H., Abramov Y. A., Philippe L., Shalaeva M. Mol. Pharmaceutics. 2016;13:1100–1110. doi: 10.1021/acs.molpharmaceut.5b00910. [DOI] [PubMed] [Google Scholar]
  48. Caron G. and Ermondi G., manuscript submitted for pubblication.
  49. Kenny P. W., Montanari C. A., Prokopczyk I. M., Ribeiro J. F. R., Sartori G. R. J. Med. Chem. 2016;59:4278–4288. doi: 10.1021/acs.jmedchem.5b01946. [DOI] [PubMed] [Google Scholar]
  50. Desiraju G. R. and Steiner T., The Weak Hydrogen Bond in Structural Chemistry and Biology, Oxford University Press, Oxford, UK, 1999. [Google Scholar]
  51. Gramse G., Dols-Perez A., Edwards M. A., Fumagalli L., Gomila G. Biophys. J. 2013;104:1257–1262. doi: 10.1016/j.bpj.2013.02.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Li L., Li C., Zhang Z., Alexov E. J. Chem. Theory Comput. 2013;9:2126–2136. doi: 10.1021/ct400065j. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Goetz G. H., Philippe L., Shapiro M. J. ACS Med. Chem. Lett. 2014;5:1167–1172. doi: 10.1021/ml500239m. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Sciabola S., Goetz G. H., Bai G., Rogers B. N., Gray D. L., Duplantier A., Fonseca K. R., Vanase-Frawley M. A., Kablaoui N. M. Bioorg. Med. Chem. 2016;24:3513–3520. doi: 10.1016/j.bmc.2016.05.062. [DOI] [PubMed] [Google Scholar]
  55. Wittekindt C., Klamt A. QSAR Comb. Sci. 2009;28:874–877. [Google Scholar]
  56. Klamt A., Eckert F., Reinisch J., Wichmann K. J. Comput.-Aided Mol. Des. 2016;30:959–967. doi: 10.1007/s10822-016-9927-y. [DOI] [PubMed] [Google Scholar]
  57. Ertl P., Rohde B., Selzer P. J. Med. Chem. 2000;43:3714–3717. doi: 10.1021/jm000942e. [DOI] [PubMed] [Google Scholar]
  58. Caron G., Ermondi G. Future Med. Chem. 2016;8:2013–2016. doi: 10.4155/fmc-2016-0165. [DOI] [PubMed] [Google Scholar]

Articles from MedChemComm are provided here courtesy of Royal Society of Chemistry

RESOURCES