Design, synthesis, biological evaluation, NMR and DFT studies of structurally simplified trimethoxy benzamides as selective P-glycoprotein inhibitors: the role of molecular flatness

Received Date : 22-Jan-2016

Revised Date : 31-Mar-2016

Accepted Date : 18-Jun-2016

Article type : Research Article

Design, synthesis, biological evaluation,

NMR and DFT studies of structurally-

simplified trimethoxy benzamides as P-

glycoprotein selective inhibitors: the role of

molecular flatness

‡

1

‡1

‡

§

Angela Stefanachi , Giuseppe Felice Mangiatordi , Piero Tardia , Domenico Alberga ,

‡

§

Francesco Leonetti , Mauro Niso , Nicola Antonio Colabufo , Carlo Adamo Orazio

,

‡

*

‡*

Nicolotti and Saverio Cellamare

‡

Dipartimento di Farmacia – Scienze del Farmaco, Università di Bari Aldo Moro, Via

Orabona 4, 70125, Bari, Italy.

#

Dipartimento di Fisica, Università di Bari Aldo Moro, INFN & TIRES, Via Amendola 173,

0126 Bari, Italy

7

This article has been accepted for publication and undergone full peer review but has not

been through the copyediting, typesetting, pagination and proofreading process, which may

lead to differences between this version and the Version of Record. Please cite this article as

doi: 10.1111/cbdd.12811

§

PSL Research University, Chimie ParisTech-CNRS, Institut de Recherche de Chimie Paris,

Paris, France and Institut Universitaire de France, 103 Bd Saint-Michel, F-75005 Paris,

France

1

these authors contributed equally to this study

Abstract

In a recent investigation carried out on a panel of trimethoxybenzanilides, we showed that the

formation of an intramolecular hydrogen bond is a key element for tuning P-gp inhibitory

activity. In the present study, we designed new structurally-simplified trimethoxy benzamides

(5-17, Table 1) with the aim to uncover the minimal molecular requirements needed for P-gp

inhibition. The new prepared smaller-sized compounds exhibited IC in the low micromolar

5

0

range. The combined use of NMR and DFT studies suggested that molecular flatness is

causatively related to the P-gp inhibition. Our results clearly pointed out that concerted

theoretical and experimental approaches herein presented might be very helpful in addressing

the design of structurally-simplified and highly efficient compounds biasing P-gp protein.

1

. Introduction

It is widely acknowledged that molecular size, basicity and lipophilicity are determinant

1

,2

molecular properties to obtain potent and selective P-gp ligands. Less attention has instead

been paid to other molecular features, such as the occurrence of intramolecular hydrogen

bonds (IMHBs) and, more importantly, the propensity to result ring planar conformations. As

well known, five- to eight-membered ring systems, and primarily the six-membered motifs,

can benefit of π-delocalization. In this respect, an important effect of IMHB is that of

increasing the lipophilic content since the formation of planar ring conformations reduces the

level of exposure of polar groups. Of course, the conformation-dependent gain of

3

,4

lipophilicity is relevant also in terms of membrane permeability. This issue has been the

5

,6

focus of recent important studies in medicinal chemistry.

ABCB1 (also named P-glicoprotein, P-gp or MDR1) and ABCC1 (Multidrug Resistance

Protein 1 or MRP1) are efflux pumps members of the ATP binding cassette (ABC)

transporters superfamily playing a crucial role in the active transport across plasma

membrane of metabolites and detoxification of xenobiotics and drugs in numerous human

7

-9

tissues (such as blood-brain barrier, intestinal barrier and blood-cerebrospinal fluid barrier).

The protective role of the ABC transporters is enhanced by the low substrate specificity that

allows the active efflux of numerous structurally unrelated compounds. Chemotherapeutic

pressure induces overexpression of these transporters in a wide range of cancer cells (breast

cancer, colon cancer, leukemia), which reduces sensitivity to multiple anticancer agents

1

0-12

(

Multidrug Resistance, MDR).

P-gp is the best-characterized member of the ABC family and evidences collected suggest

that the protein is able to catch its substrates from the inner membrane leaflet and release

1

3

them to either the extracellular medium or the outer membrane leaflet. Furthermore, the

recently solved X-ray crystal structure of Caenorhabditis elegans P-gp showed the presence

of an opening, easily accessible from the inner membrane bilayer, thus suggesting that drugs

1

4

can enter the binding pocket from the plasma membrane. Consequently, it is easy to

postulate that high lipophilicity is a key feature for high P-gp inhibitory activity. This is also

supported by the linear correlation between this physicochemical property and the inhibitory

1

5,16

activity found for several molecular scaffolds.

In a recent work carried out on a series of trimethoxybenzanilides, we reported that the

1

7

establishment of an IMHB allows the fine-tuning of the P-gp inhibitory activity. This was

likely due to the increase of the conformation-dependent lipophilicity that enabled to reach

sub-micromolar IC₅₀values in terms of P-gp inhibitory activity. Interestingly, quantum

mechanics calculations and molecular dynamics simulation studies performed in the same

previous work demonstrated that the IMHB was able to hold the most active compounds in a

flat conformation and envisaged a certain trend between the strength of the HB and the P-gp

inhibitory activity (1-4 Figure 1).

These intriguing results prompted us to prepare a new set of structurally-simplified

trimethoxybenzanilides, designed with two aims: a) elucidating the impact of the IMHB at

conformational level and its effects on the P-gp inhibitory activity; b) identifying the minimal

molecular requirements needed to gain P-gp modulation. Thus, we designed and synthesized

a set of trimethoxybenzanilides whose general scaffold was obtained by structural

simplification of the previously reported compounds, a strategy widely applied in drug design

1

8-20

to shorten synthetic routes and to reduce costs while keeping a certain biological activity.

Indeed, designing potent smaller-sized ligands is of utmost importance in medicinal

chemistry to better elucidate, at molecular level, the nature of the driving forces behind the

experimentally observed interactions and to minimize ADME attrition in late stage

2

1

development. Herein, the molecular scaffold is simply made of a trimethoxybenzoyl moiety

bound to an aniline ring (see Table 1). As far as the trimethoxybenzoyl moiety is concerned,

all the possible regioisomers were designed and prepared to unveil the role of the ortho

methoxy substituent in determining high P-gp inhibition and its importance in the formation

of a stable IMBH (hereafter referred to as IMBH ). On the other side, the aniline ring was

1

decorated by using two different substituents that are a methoxy group and a bromine atom.

The rationale behind this strategy of molecular design was that the ortho methoxy group

1

(

position R of the aniline ring, see Scheme 1) might favor the occurrence of a second IMHB

hereafter referred to as IMHB ) while the bromine atom would support the lipophilicity

2

content. Satisfactorily, this approach allowed obtaining ligands with P-gp inhibitory activity

in the low micromolar range despite a lower molecular weight compared to previous

17,22,23

studies.

At the best of our knowledge, levels of P-gp inhibitory activity as good as those

herein presented have never been observed so far for equally sized compounds. Finally, the

results obtained employing Density Functional Theory (DFT) and NMR unequivocally

suggest that molecular flatness could be a key property to enhance P-gp inhibition.

2. Chemistry

The synthesis of our derivatives 5-17 was accomplished very easily by a condensation

between the corresponding acid, activated with thionyl chloride, and the aniline (Scheme 1).

2

4

25

26

27

28

The synthesis of compounds 5, 6, 9, 10, and 13 was already described and their

reported physical-chemical properties were coincident with that observed by us.

3

. Results and discussion

3

.1 Biochemical and lipophilicity studies

P-gp and MRP1 modulating activities of tested compounds were determined by fluorescence

measurements, using calcein-AM fluorescent probes, in MDCK-MDR1 and MDCK-MRP1

2

9

cell lines. These cells overexpress only P-gp or MRP1 transporters and, thus, the observed

biological effects can be ascribed to the specific inhibition of these pumps. Calcein-AM is a

lipophilic substrate of both P-gp and MRP1, able to cross the cell membrane. Inside the cell

compartment, it is hydrolyzed by endogenous cytoplasmic esterases, yielding highly

fluorescent calcein. The hydrolyzed compound is not a P-gp or MRP1 substrate, and it cannot

cross the cell membrane via passive diffusion because it is too hydrophilic. Thus, a rapid

increase in the fluorescence of cytoplasmic calcein can be monitored. P-gp and MRP1

transporters, expressed in the cell membrane, rapidly efflux the calcein-AM before its

entrance into the cytosol, giving a reduction of the fluorescent signal due to a decrease in the

accumulation of calcein. Evaluation of P-gp or MRP1 activity in the presence of pump

inhibitors can be performed in a competitive manner. Compounds that block P-gp and MRP1

pumps inhibit calcein-AM efflux, increasing intracellular accumulation of fluorescent

calcein. Values of IC₅₀for calcein-AM uptake (Table 1) were determined by measuring

relative fluorescence values obtained after 30 min of incubation at 37 °C. Verapamil and

30

MC18 were used as reference compounds for MRP1 and P-gp inhibition, respectively. The

accuracy and reproducibility of the estimated IC values are in accordance with the most

5

0

3

1

17

recent guidelines. We have recently reported that there is a good relationship between the

lipophilicity and the P-gp inhibitory activity of trimethoxybenzanilides. Actually, such a

trend can be appreciated by measuring the value of log k’ (the interested reader is referred to

the “Experimental section” for details), an experimental lipophilicity index determined by

HPLC using a reversed stationary phase in isocratic conditions. Building on this evidence, we

have measured log k’ values for the herein presented new set of structurally-simplified

1

compounds. Inhibition data of P-gp and MRP1, lipophilicity measures and the H NMR

chemical shifts of the amidic proton (δNH) for all the investigated derivatives 5-17 are

reported in Table 2. As shown in Scheme 1, all the investigated compounds are characterized

by a common molecular scaffold consisting of an aniline and a trimethoxybenzoyl ring,

respectively, joined through an amide bond. Importantly, a good activity is obtained (see for

instance compounds 10, 14 and 17) despite the performed structural simplification of

compounds (i.e., 1, 2, 3 and 4), already discussed in our recently published work. Notably,

such structural changes did not affect the high selectivity towards P-gp with respect to MRP1.

Building on these observations, we attempted to identify the minimal molecular requirements

needed for both efficient and selective P-gp modulation. In this respect, data reported in

Table 1 enclose a wealth of information and enable us to draw important clues. First and

foremost, the majority of the compounds whose NMR data indicated the presence of IMHB₁

show both higher activity and lipophilicity with respect to regioisomers lacking of IMHB₁.

For instance, this becomes clear by comparing 5 (IC = 70.7 µM) with 6 (IC = 21.9 µM)

5

0

50

or, alternatively, 9 (IC = 27.0 µM) with 10 (IC = 9.3 µM). We also noticed that, despite

5

0

50

2

the presence of a methoxy group in position R (potentially IMHB acceptor from the amidic

nitrogen atom), appreciable increases of lipophylicity and activity are not observed in case of

2,4,6-trimethoxy regioisomers with respect to the corresponding 3,4,5-trimethoxy ones. To

this end, the interested reader can compare 8 (log k’= -0.16, IC >100 µM), 12 (log k’= 0.16,

5

0

IC = 76.2 µM) and 16 (log k’= 0.13, IC = 47.6 µM) with 5 (log k’= 0.09, IC = 70.7 µM),

5

0

50

9

(log k’= 0.52, IC₅₀= 27 µM) and 13 (log k’= 0.15, IC₅₀= 26.2 µM). Furthermore,

comparing experimental data of 14 (log k’= 0.51, IC = 3.52 µM) and 17 (log k’= 0.87, IC

5

0

50

=

1.54 µM) with those of 6 (log k’= 0.26, IC = 21.9 µM) and 10 (log k’= 0.65, IC = 9.3

5

0

50

1

µM) respectively, it is evident that the presence of a methoxy group in position R is

responsible for both higher activity and lipophylicity. In particular, a close look at the δNH

data (see Table 2) would suggest the likely formation of IMHB , whose main effect, at

2

molecular level, would be that of favoring the occurrence of flat tetracyclic-like

conformations. Surprisingly, 2,3,4-trimethoxy regioisomers, namely compounds 7 (IC >100

5

0

µM) and 15 (IC₅₀= 30.4 µM), do not display higher activities with respect to the

corresponding 3,4,5-trimethoxy regioisomers, namely compounds 5 (IC = 70.7 µM) and 13

5

0

(

IC = 26.2 µM). The only exception is represented by compound 11, whose activity (IC =

5

0

50

19.4 µM) is slightly higher than the corresponding 3,4,5-trimethoxy regioisomer, namely

compound 9 (IC = 27 µM). This is observed despite their ability to establish IMHB , as

5

0

1

denoted by an increase of lipophylicity and of δNH (see Table 2). Such unexpected behavior

of 2,3,4-trimethoxy regioisomers is also supported by their ability to drop down the linear

relationship obtained when plotting log k’70 vs pIC . The removal of regioisomers 7, 11 and

50

2

1

5 has the effect of improving correlation (from r = 0.563 to r = 0.797, see Figure 2). Next,

the obtained data clearly indicate that the bromine atom at the para position of the aniline ring

strongly increases both lipholicity, according to its tabulated positive Hansch-Fujita π

3

2

substituent constant, and P-gp activity, as evident by comparing compounds 9 (log k’= 0.52,

IC = 27 µM) with 5 (log k’= 0.09, IC = 70.7 µM); 10 (log k’= 0.65, IC = 9.3 µM) with 6

5

0

50

(

log k’= 0.26, IC = 21.9 µM); 11 (log k’= 0.75, IC50 = 19.4 µM) with 7 (log k’= 0.35, IC50

5

0

>

100 µM); 12 (log k’= 0.16, IC = 76.2 µM) with 8 (log k’= -0.16, IC >100 µM) and

5

0

50

finally 17 (log k’= 0.87, IC₅₀= 1.54 µM) with 14 (log k’= 0.51, IC₅₀= 3.52 µM). In

summary, the above-discussed data unequivocally indicate some specific structural features

2

4

5

for P-gp inhibition: 1) methoxy groups in positions R , R and R , 2) bromine in position R,

1

3

) a methoxy substituent in position R . The sole compound fulfilling all these molecular

requisites is the 17, which thus, not surprisingly, shows the highest activity (IC = 1.54 µM)

5

0

in the panel of the tested derivatives. Finally, although a good correlation between

lipophilicity and activity was confirmed, three compounds were provided with IC values

5

0

not complying the experimental lipophilicity data. Taken together these observations

prompted us to seek further and more convincing explanations affecting the P-gp inhibitory

activity.

3.2 H-NMR

Although new methods based on physicochemical parameters such as log P have been

3

proposed to evaluate the presence of IMHB, NMR spectroscopy still represents the most

useful and accurate technique. In order to demonstrate the presence of IMHB, Δδ chemical

shift for the amide NH was evaluated and 2D-NOESY experiments were performed.

Amide proton hydrogen bonded generally present higher δ chemical shift and it is reasonable

to accept the same trend for amide NH intramolecular hydrogen bonded. Consequently, the

1

difference of H-NMR chemical shift (Δδ) between NH of isomers differing for the ability of

forming IMHB (i.e. 3,4,5 trimethoxy regioisomers) represents a feasible tool for obtaining

1

evidence of this interaction. Thus the NH H-NMR chemical shifts were measured and the Δδ

were evaluated.

As expected, the Δδ observed for compounds bearing the 2-methoxy group on the benzoyl

moiety are generally positive, clearly indicating the IMHB with NH (Tables 2a and 2b).

1

Interestingly, despite the presence of two methoxy groups on position 2 and 6, compounds 8,

1

2 and 16 show negative Δδ. It is reasonable to accept that the steric clash and the

electrostatic repulsion between the carbonyl group and the 6-methoxy group force the amide

linker and the aromatic ring in different planes disrupting the proximity required for the

IMHB , as confirmed by DFT analysis.

1

Furthermore, compounds 13-17 present a 2-methoxy group on the aniline moiety matching

the proximity required for establishing IMHB . Δδ was evaluated comparing analogs

2

differing for the presence of the methoxy group on the aniline ring (Δδ’). The Δδ’ observed

are positive in all cases, indicating the formation of the second IMHB (Figure 3b).

In order to confirm the results obtained from the evaluation of Δδ and Δδ’, 2D-NOESY

assessments on 9, 10 and 17 as model compounds were performed. The hydrogen bonds

freeze the molecular conformation and force the methoxy groups to lie in proximity to the

amidic proton. Consequently, we focused on the interaction between the amide NH and both

the ortho methoxy groups on the aniline and benzoyl moieties. Compound 9 does not present

the features required for the intramolecular hydrogen bonding. Thus, the coupling between

a

b

the amidic proton (H , s, 8.01 ppm) and both the in ortho position protons (H , s, 7.02 ppm

c

and H , m, 7.52 ppm) is detectable (Figure 3a).

a’

As expected, the cross-peack relative to the interaction between the amide NH (H , s, 9.86

d

ppm) and the ortho methoxy protons (H , s, 4.04 ppm) occurs in 10, clearly indicating the

presence of the IMHB. This assumption is also confirmed by the lack of interaction with the

b’

H proton (Figure 3b). Compound 17 presents both the methoxy groups in proximity to the

a’’

amide linker. The NOE effect of the amidic proton (H , s, 10.58 ppm) with the protons of

d’

e

the methoxy goups (H , s, 4.06 ppm and H , s, 3.97 ppm) and the lack of interaction with

c’’

b’’

H and H confirm both the establishment of the IMHB with the ortho methoxy group on

the aniline moiety and the presence of the three-center hydrogen bond (Figure 3c).

3.3 DFT

Based on a DFT-based investigation, we recently provided a solid rationale for the different

1

7

P-gp activities (at the submicromolar level) observed for compounds 1-4. Notably, the

strength of the IMHB occurring in such compounds was proved to be a sensitive parameter

for tuning, at molecular level, P-gp inhibition. Building on this evidence, we extended such

approach to all the compounds herein investigated.

Preliminary studies on the optimized geometries

As a first step, we conducted an in-depth analysis of the obtained DFT optimized structures

(see Figure 4).

As expected, one IMHB can be detected in compounds 6, 7, 8, 10, 11, 12 and 13 while the

1

presence of methoxy substituent in position R (see scheme 1) allows the formation of a

second IMHB in compounds 14, 15, 16 and 17. The reliability of the obtained DFT

1

geometries is supported by the high correlation observed between the H NMR chemical shift

2

of the amidic proton and the distance computed from the proton to the nitrogen atom (r =

0.873, see Figure 5).

As shown in Figure 4, all the computed minima disclose an almost planar conformation,

except for the trimethoxy phenyl moiety, whose coplanarity with respect to the remaining

molecule depends on the presence of an IMHB involving such moiety and the amidine group.

However, the picture emerged from this preliminary investigation suggests that, albeit

necessary, the presence of one sole IMHB could not be a sufficient requisite to constrain flat

molecular conformations. Indeed, it is certainly true that the computed minima for

compounds where an IMHB cannot be established (5 and 9) prefer other than planar

structural poses. Nevertheless, the computed minima of 8, 12, and 16 were far from flatness

although an IMHB could be envisaged for the presence of a methoxy substituent at the ortho

position. However, in this case, the second methoxy substituent at the other ortho position

(2,4,6 trimethoxy regioisomers) would prevent hypothetic planar rearrangement to avoid the

likely repulsive interaction between the amidic oxygen atom and the methoxy oxygen atom at

6

position R (see lateral views in Figure 4). Importantly, slightly not-planar minima are

obtained optimizing all the 2,3,4 trimethoxy regioisomers, despite the presence of an IMHB,

as indicated also by the above-discussed lipophilicity and δNH values (see Table 2). Such

geometries give a glimpse of a possible rational explanation of the unaccountably lower

values of activity observed for all the 2,3,4 trimethoxy regioisomers with respect to the 2,4,5

trimethoxy regioisomers. It is indeed worthy to mention that molecular flatness was

3

4

hypothesized to be a key requisite for obtaining good P-gp inhibitory as observed for 6, 10,

1

7, and 14 whose energetically favored conformations are planar.

Geometrical analysis

In order to assess in more detail the putative relationship between molecular flatness and P-gp

response, we performed an in-depth analysis of the optimized structures. As shown in Figure

6

, the molecular flatness of the tested derivatives could be evaluated through different

dihedral angles:

-

φ, measuring the rotation around the bond established between the amidic carbon atom and

the benzoic ring;

-

ψ, measuring the rotation around the amidic bond;

ω, measuring the rotation around the bond engaged between the nitrogen atom and the

aniline ring. In other words, we assume that a structure can be considered totally planar if φ,

ψ and ω are at the same time equal to 180°. This held certainly true if we limit our definition

to the molecular scaffold irrespective of all the substituents decorating the two aromatic rings.

Importantly, relationships between the IC₅₀values (µM) and the deviation from the total

planarity (180-|x| where x is the value of the dihedral resulting from the optimized geometry)

computed for dihedrals φ, ψ and ω and considering all compounds having at least one IMHB

can be found. Needless to say that compounds 7 and 8 were not considered since a specific

value of IC is not available for such molecules (see Table 1).

5

0

Together with the dihedral angle φ, already proved to be important for tuning P-gp

1

7

inhibition, a certain trend can be detected also for dihedrals ψ and ω (see Figure S1 in the

Supporting Information). Notably, the closer is the dihedral angle to the complete flatness,

the higher is the P-gp activity. This would suggest that behind the formation of IMHB₁

2

between the methoxy substituent in position R , a complete flatness of the entire molecular

scaffold is highly desirable to increase P-gp inhibitory activity. This is also consistent with

the evidence that the presence of two IMHB, potentially able to constrain the molecular

scaffold into a tetracyclic conformation, favours P-gp inhibition, as reported in Table 1.

Beyond the performed geometrical analysis allowing us to speculate about a putative

relationship between P-gp inhibitory activity and the DFT optimized geometries, we moved

forward studying the energetic aspects that already proved to be crucial in our previous work.

This was necessary to properly investigate the behaviour also of those compounds missing of

IMHB and thus having a higher conformational freedom.

Energetic analysis

The picture emerged from the above discussed geometrical analysis indicates that the

extended flatness to the entire molecular scaffold, shared by all the investigated derivatives,

could be a key property to improve the P-gp inhibitory activity. As reported, all the

investigated compounds preferred to stay in planar conformation during geometrical

optimization, until a steric hindrance takes place. In other words, the optimized geometry is,

in the end, that having the highest flatness. Building on this evidence, we estimated the

energy required to reach a new stable although less planar conformation (local minimum)

starting from the obtained minima. In doing that, we calculated the Potential Energy Profile

(PEP) for the torsion around the dihedrals angle φ, ψ and ω, through relaxed SCAN

calculations. Such profiles allowed us to derive an energetic parameter (E ), computed

p

following the equation (1):

ꢀ_ꢁ

=ꢂmin_ꢃ_ꢄ_ꢅ_,_ꢆ_,_ꢇꢈꢀ_ꢃ^ꢉ^ꢊ^ꢋꢌ

1)

(

bar

Where E_icorresponds to the energy barrier required reaching a local (less planar) minimum

starting from the optimized geometry, as resulting from the PEP of the dihedral i torsion (see

bar

Figure S2 in the supporting information for a three meaningful examples of the E_i

computation). Briefly, E indicates the minimum value of energy barrier required to move far

p

away from point intercepting the highest flatness, displayed by the optimized structure,

whatever dihedral (φ, ψ and ω) is considered. For each considered compound both E value

p

and dihedral angle providing the lowest E_b_a_rvalue are reported in Table S1 (see Supporting

bar

Information). Note that dihedral ψ returns always E_ψmuch higher than the others since it

describes the torsion around a partial double bound. For instance, considering compound 6,

bar

E

φ

was equal to the energy required to break the IMHB occurring between the trimethoxy

moiety and the amide group. Such value was obviously higher with respect to that required

reaching a new local minimum having a less-planar ω angle, thus we considered for

bar

ω

2

compound 6 E = E . Importantly, the relationship (r = 0.835) between E and the IC50

p

values (µM) is shown in Figure 7. Note that herein all compounds for which a value of

inhibitory activity is available were considered. Satisfactory, a very good correlation was

detected confirming the importance of flatness for P-gp activity. More importantly, such

relationship comprised also compounds 11 and 15, whose activities were not predictable on

the basis of lipophylicity and NMR studies.

The robustness of the herein relationship was further challenged performing an intensive y-

3

5

randomization analysis to avoid the risk of chance correlations. In this respect, we

performed 1 million of y-scrambling runs to assess that the reliability and goodness of the

2

correlation shown in Figure 7. Satisfactorily, all the scrambled r obtained values were far

from that reported in Figure 7. For the sake of completeness, we reported the distribution of

2

scrambled r in Figure 8. Finally, the robustness of the found correlation is also supported by

2

the computed r value equal to 0.730, which largely exceeds the recommended threshold of

p

3

6,37

0

.500

(the interested reader is referred to the Experimental section for methodological

details).

Conclusions

The present paper reports how the harmonized use of experimental and theoretical

approaches can be helpful to address the design of small-sized and efficient molecules

selectively inhibiting P-gp. Our effort to streamline the molecular structures of our recently

1

7

published compounds allowed us to detect the essential requirements needed to gain P-gp

modulation, being the obtained low micromolar activity never observed before for equally

sized compounds. Furthermore, the results obtained from an in-depth DFT based

investigation suggest a solid and causative relationship between the energy required reaching

a local (less planar) minimum starting from an optimized geometry and the observed activity.

We do believe that our achievements might constitute a feasible molecular strategy to address

the design of new, potent and selective P-gp inhibitors.

Experimental section

High analytical grade chemicals and solvents were purchased from commercial suppliers.

When necessary, solvents were dried by standard techniques and distilled. Thin layer

chromatography (TLC) was performed on aluminum sheets precoated with silica gel 60 F254

(0.2 mm) (E. Merck). Chromatographic spots were visualized by UV light. Purification of

crude compounds was carried out by flash column chromatography on silica gel 60

(Kieselgel 0.040−0.063 mm, E. Merck) or by preparative TLC on silica gel 60 F254 plates or

1

crystallization. H NMR spectra were recorded in DMSO-d or CDCl at 300 MHz on a

6

3

Varian Mercury 300 instrument. Chemical shifts (δ scale) are reported in parts per million

ppm) relative to the central peak of the solvent. Coupling constant (J values) are given in

(

hertz (Hz). Spin multiplicities are given as s (singlet), br s (broad singlet), d (doublet), t

triplet), dd (double doublet), dt (double triplet), or m (multiplet). LRMS (ESI) was

(

performed with an electrospray interface ion trap mass spectrometer (1100 series LC/MSD

trap system Agilent, Palo Alto, CA). In all cases, spectroscopic data are in agreement with

known compounds and assigned structures. Combustion analyses were performed by

Eurovector Euro EA 3000 analyzer (Milan, Italy) and gave satisfactory results (C, H, N

within 0.4% of calculated values).

Chromatographic settings. Logk’ values and purity determinations were carried out using a

Zorbax 300SB-C18 4.6 x 250mm, with 5 µm size particles, built on a Waters double pump

HPLC system in isocratic conditions. Injection volumes were 5 μL, flow rate was 1 mL/min,

detection was performed with UV (λ=230 nm and 254 nm). Samples were prepared by

dissolving 0.1 mg/mL of the solute in 10% (v/v) DMSO and 90% (v/v) methanol. Retention

times (tr) were measured at least from three separate injections, dead time (t ) was the

0

retention time of deuterated water. The mobile phase was filtered through a Nylon 66

membrane 0.45μm (Supelco, USA) before use. For each reference compound, the average

retention time (tr) of three consecutive injections of 5 μL of sample was used to calculate the

log k’ values (log k’ = log[(tr - t )/t ]). The eluent consisted of three different mixtures of

0

methanol and PBS buffer at pH 7.4: the first measurement was 80% (v/v) methanol and 20%

(v/v) of buffer; 70% (v/v) methanol and 30% (v/v) of buffer; 60% (v/v) methanol and 40%

(v/v) of buffer.

2

6

The synthesis of N-(4-Bromo-phenyl)-3,4,5-trimethoxy-benzamide 9 was already reported.

General procedure for the preparation of benzamides 5-17

The starting benzoic acid (0.75 mmol) was suspended under argon in thionyl chloride (1.0

mL) and stirred for 3h at room temperature. The unreacted excess of thionyl chloride was

removed under nitrogen flow to afford the corresponding benzoyl chloride. The solid

obtained and the appropriate aniline (0.50 mmol) were dissolved in the minimum volume of

THF dry and stirred at room temperature for 6h. The suspension obtained was filtered and the

residue was purified by crystallization or column chromatography affording the desired

product.

2

4

3

,4,5-Trimethoxy-N-phenyl-benzamide 5

1

Purified by chromatography (hexane/ethyl acetate, 60:40). Yield:37 %; mp:141-143 °C; H

NMR (CDCl ) δ: 3.90 (s, 6H), 3.93 (s, 3H), 7.07 (s, 2H), 7.16 (tt, J=1.2 , 7.8, 1H), 7.38 (t,

3

-

J=7.8 Hz, 2H), 7.63 (dd, J=1.2 ,7.8 Hz, 2H), 7.72 (s, 1H); LRMS (ESI) m/z: = 287 [M-H] .

2

5

2

,4,5-Trimethoxy-N-phenyl-benzamide 6

1

Purified by chromatography (hexane/ethyl acetate, 60:40). Yield:43%; m.p.=119-121 °C; H

NMR CDCl δ 3.93 (s, 3H), 3.96 (s, 3H), 4.05 (s, 3H), 6.57 (s, 1H), 7.14-7.08 (m,1H),7.30-

3

7.41 (m, 2H), 7.67 (dd, J=1.1, 8.5 Hz 2H), 7.82 (s, 1H), 9.84 (s, 1H); LRMS (ESI) m/z: =

-

2

87[M-H] .

2

,3,4-Trimethoxy-N-phenyl-benzamide 7

1

Purified by chromatography (hexane/ethyl acetate, 70:30). Yield:53 %; m.p. =107-109 °C; H

NMR CDCl δ 3.91 (s, 3H), 3.93 (s, 3H), 4.06 (s, 3H), 6,85 (d, J=8.9 Hz, 1H), 7.09-7.15

3

(m,1H), 7.34-7.39(m, 2H), 7.68 (dd, J=0.96, 8.7 Hz, 2H), 7.99 (d, J= 8.9 Hz, 1H), 9.97 (s,

-

1

H); LRMS (ESI) m/z: = 287 [M-H] .

2

,4,6-Trimethoxy-N-phenyl-benzamide 8

1

Purified by chromatography (hexane/ethyl acetate, 70:30).Yield:17 %; p.f.:155-157 °C; H

NMR (CDCl ) δ: 3.82 (s, 6H), 3.84 (s, 3H), 6.14 (s, 2H), 7.1 (t. J=7.4 Hz,1H), 7.35-7.31 ( m

3

-

2

H), 7.49 (s, 1H), 7.64 (d, J=7.9 Hz, 2H), LRMS (ESI) m/z : = 287[M-H] .

27

N-(4-Bromo-phenyl)-2,4,5-trimethoxy-benzamide 10

1

Purified by chromatography (hexane/ethyl acetate, 60:40).Yield: 15 %; mp:82-84 °C; H

NMR (CDCl ) δ: 3.92 (s, 3H), 3.96 (s, 3H), 4.05 (s, 3H), 6.56 (s, 1H), 7.45 (dd, J=1.9, 6.9

3

Hz, 2H),7.57 (dd, J=1.92 , 6.87, 2H), 7,79 (s, 1H), 9.86 (s, 1H); LRMS (ESI) m/z: = 367 [M-

-

H] .

N-(4-Bromo-phenyl)-2,3,4-trimethoxy-benzamide 11

1

Purified by chromatography (hexane/ethyl acetate, 70:30).Yield: 61%; mp:145-147 °C; H

NMR (CDCl ) δ: 3.90 (s, 3H), 3.93 (s, 3H), 4.06 (s, 3H), 6.83 (d J=9.1 Hz, 1H), 7.46 (dd,

3

J=2.2, 6.8 Hz , 2H), 7.58 (dd, J=2.2, 6.8 Hz, 2H), 7.97 (d, J=9.1 Hz, 1H), 10.00 (s, 1H);

-

LRMS (ESI) m/z: = 367 [M-H] .

N-(4-Bromo-phenyl)-2,4,6-trimethoxy-benzamide 12

1

Purified by chromatography (hexane/ethyl acetate, 70:30).Yield: 22 %; mp:218-220 °C; H

NMR CDCl δ 3.82 (s, 6H), 3.84 (s, 3H), 6.14 (s, 2H), 7.43 (d. J=8.8, 2H), 7.52 (s, 1H).7,54

3

-

(

d, J=8.79, 2H); LRMS (ESI) m/z: = 367 [M-H] .

2

8

3,4,5-Trimethoxy-N-(2-methoxy-phenyl)-benzamide 13

1

Purified by chromatography (hexane/ethyl acetate, 60:40).Yield: 63%; mp:118-120°C; H

NMR CDCl δ 3.90 (s, 3H), 3.92 (s, 3H), 3.94 (s, 6H), 6.92 (dd, J=1.7, 7.9 Hz, 1H), 7.12-

3

-

6

.99 (m, 4H), 8.43 (s, 1H), 8.47 (dd, J=1.7, 7.9 Hz, 1H); LRMS (ESI) m/z: [M-H] = 317.

2,4,5-Trimethoxy-N-(2-methoxy-phenyl)-benzamide 14

1

Purified by crystallizationEtOH/Et O. Yield: 34%; mp:180-182; °C, H NMR CDCl δ 3.93

2

3

(s, 3H), 3.95 (s, 3H), 3.96 (s, 3H), 4.06 (s, 3H), 6.57 (s, 1H), 6.93 (dd, J=1.8, 7.6 Hz, 1H),

7.00-7.05 (m, 2H), 7.83 (s,1H), 8.63 (dd, J=1.8, 7.6 Hz, 1H), 10.62 (s, 1H); LRMS (ESI) m/z:

-

=

317 [M-H] .

2,3,4-Trimethoxy-N-(2-methoxy-phenyl)-benzamide 15 Purified by chromatography

1

(

hexane/ethyl acetate, 70:30). Yield: 63%; mp:118-120 °C; H NMR (CDCl ) δ: 3.91: (s,

3

3H), 3.93 (s, 3H), 3.96 (s, 3H), 4.09 (s, 3H), 6.82 (d, J=9.1 Hz, 1H), 6.92 (dd, J=1.5, 7.8 Hz,

1H), 7.00-7.08 (m, 2H), 8.03 (d, J=9.1 Hz, 1H), 8.62 (dd, J=1.9, 7.8 Hz, 1H), 10.72 (s, 1H);

-

LRMS (ESI) m/z: = 317 [M-H] .

2,4,6-Trimethoxy-N-(2-methoxy-phenyl)-benzamide 16

1

Purified by crystallization EtOH/Et O. Yield: 15% ; mp:168.5-170.5 °C; H NMR (CDCl ) δ

2

3

3.81 (s, 9H), 3.84 (s, 3H), 6.14 (s, 2H), 6.87 (dd, J=2.0; 7.6 Hz, 1H), 6.96-7.06 (m, 2H), 8.21

-

(

s,1H), 8,63 (dd, J=2.0, 7.6 Hz, 1H). LRMS (ESI) m/z: = 317[M-H] .

4-bromo-2-methoxy-N-(2,4,5-trimethoxyphenyl)benzamide 17

Purified by chromatography (hexane/ethyl acetate, 70:30). Yield:15 %; m.p.:168,5-170,5 °C;

1

H NMR (CDCl ) δ: 3.94 (s, 3H), 3.96 (s, 3H), 3.98 (s, 3H), 4.07 (s, 3H), 6.56 (s, 1H), 7.03

3

(d, J=1.5 Hz, 1H), 7.14 (dd, J=1.5, 8.3 Hz, 1H), 7.81 (s,1H), 8.55 (d, J=8.3 Hz, 1H), 10.58 (s,

-

1

H); LRMS (ESI) m/z: = 395 [M-H] .

Biology

Materials. CulturePlate 96/wells plates was purchased from PerkinElmer Life and Analytical

Sciences (Boston, MA, USA). Cell culture reagents were obtained from EuroClone (Milan,

Italy). Calcein-AM, was obtained from Sigma-Aldrich (Milan, Italy).

Cell Culture. MDCK-MDR1 and MDCK-MRP1 cells was a gift of Prof. P. Borst, NKI-AVL

Institute, Amsterdam, Nederland. MDCK-MDR1 and MDCK-MRP1 cells were grown in

DMEM high glucose supplemented with 10% fetal bovine serum, 2 mM glutamine, 100

U/mL penicillin, 100 μg/mL streptomycin, in a humidified incubator at 37 °C with a 5 %

CO2 atmosphere.

Calcein-AM experiment. These experiments were carried out as described by Colabufo et

2

9

al. with minor modifications. Calcein-AM is a profluorescent probe and is a P-gp substrate.

In cells overexpressing P-gp or MRP1, Calcein-AM is not able to permeate cell membrane

whereas when the efflux pump is not present or is inhibited, the probe enters living cells and

is converted to fluorescent Calcein by intracellular esterases. Calcein is not able to diffuse

through the membrane since it is hydrophilic and is not a P-gp or MRP1 substrate; thus,

Calcein accumulates in cells when the pump is blocked. Therefore, the fluorescent signal is

directly correlated to the amount of P-gp or MRP1 interaction. MDCK-MDR1 or MDCK-

MRP1 cells (50,000 cells per well) were seeded into black CulturePlate 96/wells plate with

1

00 μL medium and allowed to become confluent overnight. 100 μL of test compounds were

solubilized in culture medium and added to monolayers. 96/Wells plate was incubated at 37

°C for 30min. Calcein-AM was added in 100 μL of Phosphate Buffered Saline (PBS) to yield

a final concentration of 2.5 μM and plate was incubated for 30 min. Each well was washed 3

times with ice cold PBS. Saline buffer was added to each well and the plate was read to

Victor3 (PerkinElmer) at excitation and emission wavelengths of 485 nm and 535 nm,

respectively. In these experimental conditions Calcein cell accumulation in the absence and

in the presence of tested compounds was evaluated and fluorescence basal level was

estimated by untreated cells. In treated wells the increase of fluorescence with respect to

basal level was measured. IC values were determined by fitting the fluorescence increase

5

0

percentage versus log[dose].

Computational details

3

8

All DFT calculations were carried out using the Gaussian 09 package. Geometry

3

9

optimizations were performed at the B3LYP /6-311G(d,p) level of theory and in gas phase.

Harmonic vibrational frequency analyses were applied in order to confirm the nature of the

localized stationary states. Relaxed scan calculations were performed using the same level of

theory and a step size equal to 10 degrees (i.e. total number of steps equal to 36 for each

considered dihedral).

Model validation

The robustness of the found correlation was challenged by performing a randomization

analysis. In order to assess the risk of chance correlation 1 million of y-scrambled models

2

were generated and for each of them a value of r was computed. Notably, the found

correlation can be considered statistically significant if for all the randomized models a value

2

of r lower that that observed for the non-randomized model is detected. Furthermore we also

2

37

computed the r parameter following the equation:

p

ꢎ

ꢍ = ꢍ ∗ ꢏꢐꢍ − ꢍ

ꢔ

ꢁ

ꢑꢒꢓ

(2)

2

Where r is the squared correlation coefficient of the non-randomized model and r

is the

SCR

2

squared correlation coefficient averaged over all the randomized models. A value of r_p

3

7

higher than 0.5 ensures that the obtained correlation is not obtained by chance.

Acknowledgements

The authors thank the program FIRB (Futuro in Ricerca 2012, RBFR12SJA8_003) and the

Programma IDEA 2011. We acknowledge the CINECA awards no. HP10CWS1PD under the

ISCRA initiative for the availability of high-performance computing resources and support.

Conflict of interest

The authors declare no conflict of interest

AUTHOR INFORMATION Corresponding Authors: phone, +39-080-5442551/+39-080-

5442090;

fax,

+39-080-5442230;

E-mails,

orazio.nicolotti@uniba.it/saverio.cellamare@uniba.it.

Abbreviations: MDR. multidrug resistance; P-gp, P-glycoprotein; MRP1, human multidrug

resistance associate-protein 1; intramolecular hydrogen bond , IMHB ; Density Functional

Theory DFT

Scheme captions

Scheme 1. Synthesis of compounds 5-17.

Figure captions

Figure 1. Chemical structure and P-gp inhibitory activity of trimethoxy benzamides

investigated in the previous study.

Figure 2. Relationships between P-gp inhibitory potency (pIC ) of compounds listed in

5

0

Table 1 and experimental log k’₇₀_%_M_e_O_Hvalues.

Figure 3. a) 2D NOESY of compound 9; b) 2D NOESY of compound 10; c) 2D NOESY of

compound 17.

Figure 4. Structures of energy minima resulting from DFT optimization. All calculations

were performed at the B3LYP/6-311G(d, p) level of theory. IMHBs are depicted as dotted

lines. Computed distances between the hydrogen and the acceptor oxygen involves in all the

detected IMHB (d

) and IMHB (d₂₍_O_-_H₎) are reported in Å.

1

1(O-H)

2

1

Figure 5. Relationship between the experimental H NMR chemical shift of the

exchangeable NH proton and the computed H-N distance (Å) of the optimized DFT

structures.

Figure 6. Sketch of the dihedrals angles considered as representative of the molecular

flatness in the investigated compounds.

Figure 7. Relationship between the IC values and E values computed according to the

5

0

p

equation (1).

2

Figure 8. Histogram of r

vs. number of randomized models over 1 million y-scrambling

SCR

runs.

Table Captions

Table 1. Chemical structure, biological activity and experimental measures of lipophilicity of

the compounds 5-17.

1

Table 2. H NMR chemical shifts of the amidic proton (δNH) of the considered compounds

and a) its variation with respect to the corresponding 3,4,5, trimethoxy regioisomers (ΔδNH);

b) its variation with respect to the analog differing for the presence of the methoxy group on

the aniline ring (Δδ’NH).

Scheme 1

Reagents and conditions: i) neat Thionyl chloride 3h; reflux; ii) DRY THF, r. t., 6h.

Figure 1

Figure 2

a)

b)

Figure 3

a)

b)

c)

Figure 4

TOP VIEW

LATERAL VIEW

TOP VIEW

LATERAL VIEW

5

6

7

8

9

1

12

d₁₍_O_-_H₎= 2.32 Å

13

d₁₍_O_-_H₎= 1.86 Å

d₁₍_O_-_H₎= 2.10 Å

^d1(O-H)⁼¹^.⁸⁹^Å

14

^d1(O-H)= 1.87 Å

^d2(O-H)= 2.15 Å

d₁₍_O_-_H₎= 2.39 Å

1

5

6

d₁₍_O_-_H₎= 1.91 Å

d₂₍_O_-_H₎= 2.15 Å

1

0

d₁₍_O_-_H₎= 1.85 Å

d₁₍_O_-_H₎= 1.88 Å

d₁₍_O_-_H₎= 2.56 Å

d₂₍_O_-_H₎= 2.13 Å

11

17

d₁₍_O_-_H₎= 1.87 Å

d₂₍_O_-_H₎= 2.17 Å

Figure 5

Article Doi

Design, synthesis, biological evaluation, NMR and DFT studies of structurally simplified trimethoxy benzamides as selective P-glycoprotein inhibitors: the role of molecular flatness

DOI: 10.1111/cbdd.12811

Source and publish data:

Authors:

Article abstract of DOI:10.1111/cbdd.12811

Full text of DOI:10.1111/cbdd.12811

Products guided by the article

R&D Labs maybe for 299950-93-9

Relevant to this article

Hot Product