Benzimidazole-based derivatives as privileged scaffold developed for the treatment of the RSV infection: a computational study exploring the potency and cytotoxicity profiles

Abstract Respiratory syncytial virus (RSV) has been identified as a main cause of hospitalisation in infants and children. To date, the current therapeutic arsenal is limited to ribavirin and palivizumab with variable efficacy. In this work, starting from a number of in-house series of previously described anti-RSV agents based on the benzimidazole scaffold, with the aim at gaining a better understanding of the related chemical features involved in potency and safety profiles, we applied a computational study including two focussed comparative molecular fields analysis (CoMFA) and comparative molecular similarity indices analysis (CoMSIA). The results allowed us to derive useful suggestions for the design of derivatives and also to set up statistical models predicting the potency and selectivity index (SI = CC50/EC50) of any new analogue prior to synthesis. Accordingly, here, we discuss preliminary results obtained through the applied exhaustive QSAR analyses, leading to design and synthesise more effective anti-RSV agents.


Introduction
Respiratory syncytial virus (RSV) is the aetiological agent of serious and widespread respiratory tract infections that represent the major cause of hospitalisation of infants and children. The virus is highly contagious and frequent re-infections may cause morbidity in elderly people affected by chronic illnesses and in immunocompromised individuals. Epidemiologic studies proved RSV as a ubiquitous contributor to pulmonary exacerbation, particularly in cystic fibrosis (CF) patients, causing severe clinical symptoms and promoting comorbidities with bacterial pathogens 1 .
Despite the huge economic impact and the medical needs associated with severe RSV infection, therapy is restricted to ribavirin, whose effectiveness, however, is highly questionable 2 . Humanised monoclonal antibody palivizumab (SynagisV R ) is used for the prevention in high-risk infants, with partial efficacy in reducing RSV hospitalisation rates 3 . Insufficient clinical observation exist to concretely support whether this drug is cost-effective and safe 4,5 . Thus, there is a clear need for new drugs to prevent and treat RSV infection.
Member of Paramyxoviridae family, RSV has a negativestranded RNA genome that encodes 11 viral proteins 6 . The main antigenic determinants, F (fusion) and G (attachment) glycoproteins, are expressed on the virion surface and play a key role in viral entry into host cells; therefore, they represent interesting targets of small molecule inhibitors.
Case in point as a future RSV therapeutic is the orally administered GS-5806 7 that was found as a potent inhibitor of a broad range of RSV clinical isolates by blocking the virus-cell fusion process 8 . GS-5806 needs to be improved for its low efficacy against the emergent mutant strains, an issue that elicits concern for efficacy of the current antiviral therapies, in general 9 .
Benzimidazole and its derivatives are important bioactive molecules class in drugs and pharmaceuticals fields 10 : they exhibit significant activity against several viruses and are also endowed with antimicrobial, anti-inflammatory, antitumour, antiparasitic, antiprotozoal properties.
The activity against RSV is peculiar for several benzimidazolebased chemotypes 11 , which were demonstrated to target viral replication machinery by blocking fusion and entry processes 12,13 . Continue efforts are devoted to design new RSV inhibitors, by decorating the benzimidazole core structure, from which promising leads reached clinical trials 14 . The benzimidazole derivative, TMC353121, has been shown to inhibit the viral fusion process in an African green monkeys model 15 : administered under continuous intravenous infusion, it achieved a complete inhibition of RSV entry into host cells in a dose-dependent manner, in contrast to previously described fusion inhibitors that exhibited only a reduction in viral load.
Recently, we identified three principle classes of anti-RSV benzimidazole derivatives, such as 2-benzylbenzimidazoles 16 , 2-phenylbenzimidazoles 17 and [(benzotriazol-1/2-yl)methyl] benzimidazoles 18 ; among them, the most active compounds showed even submicromolar potency associated with a high safety profile (SI ¼ CC 50 /EC 50 ). In fact, the 75% of the compound library displayed a SI value from >14.3 up to >3333, comparing favourably with the drug ribavirin (SI >14.3), used as positive control.
In this work, starting from these previously studied series of benzimidazole-based anti-RSV agents [16][17][18] including 156 compounds ( Figure 1), aiming to gain a better understanding of the related chemical features involved in potency and safety profiles, we applied a computational study including exhaustive quantitative structure-activity relationship (3D-QSAR) analyses. Indeed, the ligand-based approach represents a useful tool guiding for rational drug design, as shown in the literature [19][20][21] .
In detail, we developed two comparative molecular fields analysis (CoMFA) and comparative molecular similarity indices analysis (CoMSIA). The first one was performed around the ligands potency values (pEC 50 ), the second one using the pCC 50 parameter (cytotoxicity) against human MT-4 cell line. The results allowed us to derive useful suggestions for the design of new derivatives and also to set up statistical models predicting the potency and selectivity index (SI ¼ CC 50 /EC 50 ) of any new analogue prior to synthesis.
Basing on CoMFA and CoMSIA analyses, we deemed interesting to design and synthesise two new compounds (157, 158), that bear in position 1 of the (benzotriazol-1-yl)methyl]benzimidazole scaffold the 2-(N,N-dimethylamino)ethyl or homo-lupinyl chains, which were previously found as the best suited substitutions, responsible for a high potency profile. These chemical features have been now associated with the new investigated apolar CH 3 group in position 5 of the benzimidazole ring, with the aim at improving the activity and even more at reducing the cytotoxicity. Compound were evaluated against RSV in Vero-76 cell cultures in order to validate the computational analyses results.

Data set
All the compounds here evaluated were built in silico and energy minimised within MOE using MMFF94 force field 22 . Therefore, all the compounds were also parameterised by means of the Gasteiger-H€ uckel method. All calculations were carried out using a PC running the Windows XP operating system.

3D-QSAR analyses
The benzimidazoles 1-156 have been aligned and submitted to 3D-QSAR studies through CoMFA and CoMSIA analyses, by means of Sybyl-X1.0 23 . Model A and model B CoMFA and CoMSIA analyses were performed to analyse the impact played by steric, electrostatic, hydrophobic, H-bond acceptor and H-bond donor features around the potency as anti-RSV agents and the cytotoxicity trend of these series of benzimidazoles.

Training set and test set
Starting from all compounds, a number of benzimidazoles were grouped into a training set, for model generation, and a test set, for model validation, containing: (i) 50 and 11 compounds for model A; (ii) 123 and 32 derivatives for model B, respectively. In any model, the molecules selected for the training and the test set pools were chosen manually, based on representative criteria of the overall biological activity trend and structural variations.
For model A and model B analyses, the RSV EC 50 values and the human MT-4 cell line CC 50 ones have been transformed into pEC 50 and pCC 50 values, respectively, and then used as response variables.

CoMFA and CoMSIA models and statistical evaluation
CoMFA 24 and CoMSIA 25 methods are widely used 3D-QSAR techniques being useful to relate any variation of an experimentally determined parameter (dependent variables), related to a set of molecules, with respect to specific descriptors which are considered as independent variables. In particular, the steric and electrostatic fields and especially the hydrophobic, H-bond donor, H-bond acceptor ones were calculated by CoMFA and CoMSIA analysis, respectively. Starting from a proper molecule alignment within a 3D cubic lattice (with a 2 Å grid spacing), any descriptor was calculated, using the standard Tripos force field method. Successively, the reliability of the derived models can be evaluated using specific statistical tools, such as partial least square (PLS) analysis and cross-validation methods.
Finally, the predictive ability about those compounds included in the test set (r 2 pred ) was also calculated, by means of the following equation: being SD and PRESS the sum of the squared deviations between the biological activities of the test set molecules and the mean activity of the training set compounds, and the squared deviation between the observed and the predicted activities of the test set compounds, respectively. Any further detail concerning the (standard) CoMFA and CoMSIA procedures and the statistical and predictive evaluation we applied, were previously elucidated in a consistent number of our works [26][27][28][29] .

Chemistry
General synthetic methods Melting points were taken in open glass capillaries on a B€ uchi apparatus and were uncorrected. Elemental analyses were performed on a Carlo Erba EA-1110 CHNS-O instrument in the Microanalysis Laboratory of the Department of Pharmaceutical Sciences of Genoa University. The analytical results are within ±0.4% of calculated values. 1 H NMR and 13 C NMR spectra (reported as supplemental material S1-S2) were recorded in CDCl 3 or DMSO-d6 on Varian Gemini-200 spectrometer; d in ppm rel. to Me4Si as internal standard. J in Hz. Q ¼ quinolizidine ring. Results of elemental analyses, thin-layer chromatography (TLC) and nuclear magnetic resonance (NMR) spectra indicated that the purity of all compounds was !95%. Chemicals, solvents and commercially available intermediates were purchased from Aldrich (Milan). The non-commercially available intermediates were prepared according to the literature, or as follows, when not previously known. Intermediates. N-substituted 1,2-phenylendiamines General synthetic method. To a stirred solution of the above nitroderivative (1.3 mmol) in EtOH (7 mL), a solution of SnCl 2 .2H 2 O (3.9 mmol, 0.88 g) in conc. HCl (10 mL) was slowly added. The mixture was refluxed for 6 h and then concentrated in vacuo. The residue was taken up in H 2 O, alkalinised with a solution of 6N NaOH and then extracted with Et 2 O. The organic layer was dried (Na 2 SO 4 ), filtered and then evaporated, affording a yellow oil. A small amount was converted into dihydrochloride salt with a 1N ethanolic solution of HCl for performing elemental analysis.  Cytotoxicity assays a. Exponentially growing MT-4 cells were seeded at an initial density of 1 Â 10 5 cells/mL in 96-well plates in RPMI-1640 medium supplemented with 10% foetal bovine serum (FBS), 100 units/mL penicillin G and 100 mg/mL streptomycin. Cell cultures were then incubated at 37 C in a humidified, 5% CO 2 atmosphere in the absence or presence of serial dilutions of test compounds. Cell viability was determined after 96 h at 37 C by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide (MTT) method 30 . b. Vero-76 cells were seeded at an initial density of 4 Â 10 5 cells/ mL in 24-well plates, in culture medium (Dulbecco's modified Eagle medium (D-MEM) with L-glutamine, supplemented with foetal bovine serum (FBS), 0.025 g/L kanamycin). Cell cultures were then incubated at 37 C in a humidified, 5% CO 2 atmosphere in the absence or presence of serial dilutions of test compounds. Cell viability was determined after 48-96 h at 37 C by the crystal violet staining method. c. The results are expressed as CC 50 , which is the concentration of compound necessary to inhibit cell growth by 50%. Each CC 50 value is the mean and standard deviation of at least three separate experiments performed in duplicate.

Antiviral assays
Antiviral activity against RSV was determined by plaque reduction assays in infected cell monolayers. To this end, Vero 76-cells were seeded in 24-well plates at a density of 2 Â 10 5 cells/well and were allowed to form confluent monolayers by incubating overnight in growth medium (Dulbecco's modified Eagle medium (D-MEM) with L-glutamine and 4500 mg/L D-glucose and 0.025 g/L kanamycin, supplemented with 10% FBS) at 37 C in a humidified CO 2 (5%) atmosphere. Then, monolayers were infected for 2 h with 250 mL of proper viral dilutions to give 50 to 100 PFU/well. Following removal of unadsorbed virus, 500 mL of maintenance medium [D-MEM with L-glutamine and 4500 mg/L D-glucose, supplemented with 1% inactivated FBS] containing 0.75% methylcellulose, without or with serial dilutions of test compounds, were added. Cultures were incubated at 37 C for 5 days and then fixed with PBS containing 50% ethanol and 0.8% crystal violet, washed and air-dried. Plaques were then counted.

Linear regression analysis
The extent of cell growth/viability and viral multiplication, at each drug concentration tested, were expressed as percentage of untreated controls. Concentrations resulting in 50% inhibition (CC 50 or EC 50 ) were determined by linear regression analysis.

Chemistry
Preparations of compounds 157 and 158 were accomplished by the general synthetic steps in Scheme 1.

3D-QSAR analyses
Starting from the in-house compounds 1-156 (Table 1), CoMFA and CoMSIA analyses here reported were used to explore, through quantitative methods, the main features responsible for the anti-RSV activity (model A) of benzimidazole-based derivatives and also for the related cytotoxicity profile (model B).
For both models, calculations were developed using CoMFA steric and electrostatic fields, and CoMSIA steric, electrostatic, hydrophobic, H-bond acceptor and H-bond donor parameters, as independent variables. On the other hand, the RSV pEC 50 and the MT-4 pCC 50 were employed as dependent variables for model A and B, respectively (experimental section).
Concerning model A, CoMFA and CoMSIA analyses were performed choosing among compounds 1-156 a training set pool molecules (2, 9, 11, 12, 14, 15, 21,  All statistical parameters supporting the two series of 3D-QSAR analyses are reported in Table 2, while final models A and B experimental and predicted pEC 50 (pCC 50 ) values are listed in Tables 3-6. Any detail is described as follows.
The final model A CoMFA was generated by employing noncross-validated PLS analysis with the optimum number of components (ONC ¼ 5) to give a non-cross validated r 2 (r 2 ncv ) ¼ 0.92, a test set r 2 (r 2 pred ) ¼ 0.88, standard error of estimate (SEE) ¼ 0.279, steric contribution ¼ 0.574 and electrostatic contribution ¼ 0.426.
The related CoMSIA analysis was derived using a statistical PLS analysis leading to the following results: ONC ¼ 5, a non-cross validated r 2 (r 2 ncv ) ¼ 0.89, a test set r 2 (r 2 pred ) ¼ 0.88, standard error of An overall overview of the predictive ability of model A study can be obtained from graphical distributions of the predicted

CoMFA and CoMSIA contour maps
In order to deeply discuss the contribution of any feature displayed within the series of benzimidazoles here investigated with respect to their potency as anti-RSV agents, and also to their cytotoxicity profile, the description of the related 3D-QSAR models (A and B, respectively) was organised in two sections separately.
According to the CoMFA steric map descriptors, green polyhedra represent those areas that should be decorated with bulky groups, while yellow maps highlight those regions related to an unfavourable or slightly allowed presence of substituents. On the other hand, the CoMFA electrostatic descriptors are shown as blue areas around those regions predicted to be beneficial for electropositive or electron donor moieties, while red polyhedra occupy any area recommended for much more electronegative or electron withdrawing. Concerning CoMSIA analysis, the hydrophobic map reveals through yellow and white polyhedra those ligand features predicted to be favoured for lipophilic and polar groups, respectively.
The introduction of H-bond acceptor and H-bond donor moieties results to be encouraged or discouraged by the presence of magenta and cyan and of red and purple areas, respectively.

Model a CoMFA and CoMSIA analyses (benzimidazoles anti-RSV profile)
On the basis of model A, for all the data set, the steric contour map predicts favourable substitutions at the benzimidazole position 5 (green polyhedral), while the presence of any bulky group at the position 6 results to be detrimental for the anti-RSV activity (yellow polyhedral). The reliability of these information is supported by the inactivity of the 5,6-dichloro-benzimidazoles 44-49 and by the poor anti-RSV potency values of the 6-substituted benzimidazoles 102 (pEC 50 5 4.20) and 103 (pEC 50 5 4.35).
Notably, the presence of a (even small) substituent linked to the benzimidazole position 2 is favoured. Accordingly, compound 2 (R2 ¼ trifluoromethyl; pEC 50 ¼ 4.66) display a better potency profile if compared with the analogue 21 (R2 ¼ 4-bromobenzyl; pEC 50 ¼ 4.60), suggesting that a proper decoration on the benzimidazole position 5 and at the R1 group could be successfully accompanied by a small moiety in R2.
On the other hand, for those 2-substituted benzimidazoles bearing a rigid phenyl group in R2 (such as 27, 29-31 and 102, 103), any further substitution at the ortho positions proves to be encouraged, falling in a green area, while any other at the meta and para ones are disfavoured, being surrounded by yellow polyhedra (Figure 2(a)).
These results are confirmed by the inactivity of compounds 40-43 (pEC 50 < 4.00), which are poly-substituted at the phenyl ring in R2. Moreover, for this series of compounds, any group eventually placed in R1 results to be disfavoured. Indeed, also compounds 97 (R1 ¼ methyl; pEC 50 < 4.00) and 98 (R1 ¼ cyclohexyl; pEC 50 < 4.00) are not interesting as anti-RSV agents.
Notably, this information is also in accordance with the promising potency profile displayed within the series of 2-benzotriazolylmethyl-benzimidazoles, such as the analogues 120, 122 and 125 (pEC 50 ¼ 7. 22-7.52).
Finally, it should be noticed that the presence of the N(1)-substituted benzotriazole ring rather than the N(2)-substituted benzotriazole promote a favourable positioning of the R1 group, such as that shown by the homolupinyl chain of 126 and 148, depicted in Figure 3. Indeed, the R1 of 126 much more effectively fall in a beneficial green steric map.
On all these basis, the 3D-QSAR study confirms that the introduction of an N(1)-substituted benzotriazole ring in R2 represents the better choice at the benzimidazole position 2, in comparison with the other chemical moieties explored within the dataset.
For model A, the electrostatic CoMFA map revealed for all the molecules here investigated a blue area (beneficial for electropositive moieties) in proximity of the R3 substituent, while two other polyhedral favourable for electropositive functions would involve a (bulky) substituent eventually present in R1. Notably, these results suggest the introduction of an electron-donor group onto the benzimidazole position 5 (R3) rather than an electron-withdrawing one, and also strongly promote the presence of a basic chain or substituent properly connected with the benzimidazole N1 nitrogen atom.
In addition, a consistent blue region is placed near one of the ortho positions of the R2 phenyl-and benzyl-substituted benzimidazoles, as shown for compounds 44 and 11 in Figure 4. The reliability of this information is confirmed by the poor activity as anti-RSV agents of compounds 25 (R2 ¼ 2-nitrophenyl-; pEC 50 ¼ 4.40) and 49 (R2 ¼ 2,6-difluorophenyl-; pEC 50 <4.00), bearing highly electron-withdrawing functions at the R2 phenyl ortho positions.
On the other hand, the introduction of much more electronegative groups is recommended in proximity of the benzimidazole core and of the R2 phenyl and benzyl ring in R2 of 44 and 11. Consequently, the presence of electron-donor groups properly decorating the aromatic R2 substituents results to be the most beneficial. Indeed, among those compounds being unsubstituted in R1, compound 11 (R2 ¼ 4-methoxybenzyl-; pEC 50 ¼ 5.30) results to be more effective than 48 (R2 ¼ 4-nitrophenyl-; pEC 50 < 4.00) and 49 (R2 ¼ 2,6-difluorophenyl-; pEC 50 < 4.00).
In addition, contour maps predicted to be favoured for electronegative moieties are also shown around the electropositive core eventually involving R1, and in the area placed between the benzimidazole positions 6 and 7, even opening the possibility to design new derivatives based on a tricyclic heteroaromatic ring in place of the benzimidazole scaffold.
The most effective benzotriazole-based benzimidazoles follow the aforementioned preferred electrostatic profile, being in particular the N(1)-substituted benzotriazoles, rather than the N(2)-substituted ones, more able to fulfil the electronegative area occupied by the triazole ring, as depicted in Figure 5 around 126 and 148.
Moreover, the N(1)-substituted benzotriazoles better arrange the basic core included in R1 towards the electropositive blue area, being in accordance with the behaviour previously discussed about the steric map. Interestingly, within the R1 basic moieties here explored, the bulkier homolupinyl and the more flexible dialkylaminoalkyl chains prove to be the most effective, optimising any contacts with the CoMFA contour maps. Accordingly, compound 126 is the most active of the whole series (R1 ¼ homolupinyl-; R2 ¼ N(1)-benzotriazolyl; pEC 50 ¼ 7.70).  The information derived by the CoMSIA hydrophobic map highlights the relevance of lipophilic group in R3 (yellow regions), while a small yellow polyhedral is also located in proximity of the substituent eventually placed in R1. Conversely, more polar function (white regions) are predicted to be favoured at R4 and at the R2 group, as shown around the phenyl and benzyl functions of 44 and 11 (Supplemental material S7).
The related hydrophobic map depicted around the benzotriazole-based benzimidazoles 126 and 148 points out the beneficial role played by a high lipophilic substituent falling in R1, such as the homolupinyl and N,N-diethylamino-ethyl ones ( Figure 6).
The anti-RSV activity of benzimidazoles is enhanced by H-bond acceptor functions (magenta polyhedral) near the ortho positions of those analogues decorated with a phenyl group in R1, and at the para position of those compounds bearing a benzyl moiety (Supplemental material S8). On the contrary, it results to be disfavoured by any H-bond acceptor function eventually placed in R3 or R4 (green areas).
Concerning the most promising benzotriazole-based benzimidazoles, they also follow the aforementioned H-bond acceptor preferences, being in particular the N(1)-substituted benzotriazoles, rather than the N(2)-substituted ones, more able to arrange near the favoured magenta area involving R2, as depicted in Figure 7.
Finally, the H-donor map discloses the relevance of favoured Hbond donor groups (cyan polyhedral) only at the central core of the area occupied by R1, being disfavoured at the surrounding region (see Figure 8 depicted around 126 and 148 and Supplemental material S9 concerning the phenyl and benzyl  derivatives 44 and 11). Notably, this information is probably related to the presence of the key basic chain, as we previously highlight through the CoMFA electrostatic map. In particular, the presence in R2 of the quinolizidine ring (lupinyl, epilupinyl and honolupinyl) let the topology of the benzotriazole nucleus be arranged in allowed conformations.

Model B CoMFA and CoMSIA analyses (benzimidazoles cytotoxicity profile)
Model B CoMFA and CoMSIA analyses allowed us to highlight any key feature eventually increasing the cytotoxicity against the human MT-4 cell line.
In detail, the CoMFA steric map reveals, for all the compounds, a probable higher toxicity profile related to the presence of substituents near R3 or in the area placed between the positions 6 and 7 of the benzimidazole ring (green area; Figure 9), and also in the case of even small groups linked in R2 (compare compounds 2 (R3 ¼ trifluoromethyl-; pCC 50 ¼ 4.66) and 4 (R2 ¼ cyclopentylmethyl; pCC 50 ¼ 4.96) with 19 (R2 ¼ benzyl; pCC 50 ¼ 4.00)).
In the case of bulkier decorations in R2, the most flexible ones, such as the benzyl-substituted benzimidazoles and the N(1)-substituted benzotriazoles-based derivatives, are projected towards yellow disfavoured regions, being therefore characterised by an adequate safety profile. Conversely, the more rigid R2 phenyl-or N(2)-substituted benzotriazoles-based analogues much more overlap a large green area, thus confirming their related possible cytotoxic role. Interestingly, the reliability of these results is supported by the lower MT-4 pCC50 value of 11 (R2 ¼ 4-methoxybenzyl-; pCC 50 ¼ 4.54) if compared with the analogue 45 (R2 ¼ 4-methoxyphenyl-; pCC 50 ¼ 6.00). On the other hand, based on the  experimental data about the pattern of cytotoxicity investigated within the benzotriazole analogues, they proved to display a quite comparable profile.
The CoMFA electrostatic contour map depicted in Figure 10 describes an overall incisive effect played by electropositive functions included in a bulky R1 group, at the central core of the R2 substituent and in the area near R3. On the other hand, electronegative functions or electron-withdrawing moieties are favoured as small R1 group, at the latter portion of R2 and in proximity of the N2 nitrogen atom of the benzimidazole ring. In particular, the R2 phenyl or benzyl analogues, further decorated with electrowithdrawing or electron-rich functions especially at the para position, represent those compounds which better fit the electrostatic map, while the N(1)-substituted benzotriazoles, followed by the N(2)-substituted benzotriazoles ones (lacking any further group on the heterocyclic system) are projected elsewhere. Notably, these information allow to retain the series including benzotriazolebased decoration as the most safety ones with respect to the others shown within the data set.
Concerning the effect played by hydrophobic substitutions around the benzimidazole scaffold with respect to the related cytotoxicity profile, a critical role proves to be determined by lipophilic groups at the latter part of R1 and in R3 and R4 (Figure 11).  As a consequence, most of the compounds bearing a trifluoromethyl in R3 or a dichloro substitution at R3 and R4 led to quite toxic compounds [see 44-49 (pCC 50 ¼ 4.77-6.00)].
As we previously discussed for the electrostatic map, even in this case the R2 phenyl-or benzyl-group analogues, rather than the N(1)-substituted benzotriazoles-and N(2)-substituted benzotriazoles-based ones, are those that fully occupy the hydrophobic map. In details, the R1 aromatic ring falls in a disfavoured hydrophobic region (white polyhedral), while any substituent eventually placed at the phenyl-or benzyl-para position is also in contact with a favourable yellow area. Interestingly, these data underline that the benzimidazole cytotoxicity profile could be enhanced by halogens in R3 and/or R4 and also by the presence of para-substituted phenyl-or benzyl-carboxamide groups in R2, falling this moiety in a white area.
As shown in Figure 12, H-bond acceptor functions result detrimental for the benzimidazole selectivity index by increasing cytotoxicity especially when they are placed around the (benzotriazol-2-yl) scaffold (magenta areas) and at the R2 phenyl-or benzyl-para positions and, confirming the critical role of the carboxamide function at this position.
Accordingly, also the H-donor contour map points out the presence of favoured (cyan areas) and disfavoured (purple areas) donor functions in R2, representing the aforementioned carboxamide moiety ( Figure 13).
Based on an overall analysis of all the information coming from model A and model B, all the three much more consistent series of benzimidazoles, being substituted in R2 with a phenyl, benzyl or benzotriazole ring, could be optimised through the introduction of small group in R3 endowed with electropositive profile and a less hydrophobic one than that displayed by the trifluoromethyl group or chlorine atom. Concerning R1 it should be noticed that model A underlines a key role played by H-bond acceptor moieties included in bulky and hydrophobic appendages. On the contrary, the corresponding map developed around the MT-4 pCC 50 values does not reveal any effect of H-bond acceptor functions at this level. Consequently, all these series of derivatives should be decorated with proper basic chain, such as the effective homolupinyl and dialkylaminoalkyl ones (see the model A CoMFA electrostatic results). In particular, new 2-phenyl-substituted  benzimidazoles could particularly take advantage from such a decoration in R1. Finally, the most promising substituent in R2 explored within the dataset proved to be the (benzotriazolyl)methyl one, opening the possibility to be optimised through a series of isostere and/or by a proper selection of other R3 substituents.
Concerning this issue, it should be noticed that the anti-RSV clinical candidate BMS-433771 was designed starting from a benzimidazolone-based compound, being bioisostere of the here discussed benzotriazol-1-yl moiety (Figure 14) 34 .
Notably, BMS-433771 qualitatively fulfils any key features recommended by the 3D-QSAR maps here proposed, also about the substituent linked at the benzimidazole position 1, giving a further validation of the applied computational protocol.
In this context, in order to gain a preliminary evaluation of the reliability of the applied 3D-QSAR study, we synthesised two new compounds exploring the effectiveness of the apolar and electrodonor CH 3 group in position 5 of the 2-[(benzotriazol-1-yl)methyl]benzimidazole framework.
The presence of (quinolizidinyl)alkyl residue associated with the 5-Cl group of benzimidazole derivative 126 ring (pEC 50 ¼ 7.70, MT-4 pCC 50 ¼ 4.35) contributed to the best activity against RSV, also leading to an increased cytotoxicity. Notably, based on these data and on the information coming from 3D-QSAR studies, the proposed 5-methyl analogue 158 proved to be endowed with the same potency, but lower toxicity against both the cell lines; underlying that 5-methyl substitution is fruitful for modulating the safety profile (CC 50 /EC 50 ).

Conclusions
The computational studies here presented highlight and discuss the role played by the steric and electrostatic features and also by hydrophobic, H-bond acceptor and donor moieties, in terms of anti-RSV activity and cytotoxicity around the benzimidazole scaffold.
Through focussed CoMFA and CoMSIA analyses, any pattern of requirements able to specifically discriminate much more effective antiviral agents endowed with an increased selectivity index was addressed and deeply discussed. Based on information coming from CoMFA and CoMSIA studies, we designed and synthesised two new benzimidazole derivatives (157, 158) in order to verify the reliability of computational studies. These compounds, being actually endowed with the same high potency but improved safety profiles, allowed us to pave the way for a more focussed rational design process. A much more consistent number of more drug-like anti-RSV agents will be disclosed in due course.