Introduction

The incidence of invasive microbial infections caused by opportunistic pathogens, often lead to a large number of people to death every year. Over the past few years, patients that become severely immunocompromised, because of underlying diseases such as leukemia or recently acquired immunodeficiency syndrome or patients who undergo cancer chemotherapy or organ transplantation are particularly susceptible to opportunistic fungal infections.

The number of cases of multidrug resistant bacterial infections is increasing at an alarming rate. Clinicians have to become reliant on few antimicrobial drugs available in the market but that is not sufficient as microbial species are getting resistant very fastly. In order to meet these challenges there is need for the development of novel antimicrobial drugs to which the microbes have never been presented before (Emami et al., 2008).

It has been well documented that 2-amino benzoic acid derivatives exhibit broad spectrum of activity including antiviral (Selvam et al., 2008), anticancer (Cocco et al., 2004), anti-alzheimer (Simons et al., 2009), antiallergic (Inglis et al., 2007), diuretic (Dambrosio et al., 1965) and insecticidal (Raman et al., 2008) activities.

Quantitative structure activity relationship (QSAR) is a methodology mostly used to correlate properties (such as biological activities) with chemical structures. It is a statistically validated mathematical model of correlation between the chemical structures and their activity profiles (Sperandio et al., 2004).

In light of above facts and in continuation of our research efforts in development of novel antimicrobial drugs (Minu et al., 2008; Narasimhan et al., 2003, 2004, 2006a, b, 2007a, b, c), we hereby report the synthesis, antimicrobial evaluation and QSAR studies of a series of 2-amino benzoic acid derivatives.

Experimental

Starting materials were obtained from commercial sources (HiMedia Chemicals, Mumbai and LOBA Chemie, Mumbai) and were used without further purification. Reaction progress was observed by thin layer chromatography (TLC) making use of commercial silica gel plates (Merck). Melting points were determined in open capillary tubes on a Sonar melting point apparatus and are uncorrected. 1H nuclear magnetic resonance (1H NMR) spectra were determined by Bruker Avance II 400 NMR spectrometer in appropriate deuterated solvents and are expressed in parts per million (δ, ppm) downfield from tetramethylsilane (internal standard). NMR data are given as multiplicity (s singlet, d doublet, t triplet, m multiplet) and number of protons. Infrared (IR) spectra were recorded on a Perkin Elmer FTIR spectrometer.

General procedure for synthesis of amides/anilides derivatives of 2-amino benzoic acid (1–20)

The solution of corresponding amine/aniline (0.1 mol) in ether (50 ml) was added drop wise to a solution of 2-amino benzoyl chloride (0.1 mol) in ether (50 ml) maintained at 0–10°C temperature. The solution was stirred for 30 min and the precipitated amide was separated by filtration. The crude amide was recrystallized with alcohol. In case of anilides, the precipitated crude anilide was treated with 5% hydrochloric acid, 4% sodium carbonate and water to remove residual aniline and the resultant anilide was recrystallized with alcohol.

Procedure for synthesis of 8-hydroxy quinoline ester of 2-amino benzoic acid (21)

For the preparation of 2-amino benzoyl chloride, thionyl chloride (0.3 mol) was added gradually to 2-amino benzoic acid (0.25 mol) in a round bottom flask. After addition of thionyl chloride, the mixture was stirred for 4 h and heated to 80°C for 30 min in water bath. The excess of thionyl chloride was removed by distillation. A solution of 8-hydroxy quinoline (0.05 mol) in ether (50 ml) was added to a solution of 2-amino benzoyl chloride (0.05 mol) in ether (50 ml). The mixture was heated on a water bath until no further evolution of hydrogen chloride was observed and completion of reaction was checked by single spot TLC. The mixture was cooled to room temperature and evaporation of solvent yielded the crude product which was purified by recrystallization with alcohol.

General procedure for synthesis of ester derivatives of 2-amino benzoic acid (22–28)

A mixture of 2-amino benzoic acid (0.08 mol) and appropriate alcohol (0.74 mol) was heated under reflux in presence of sulphuric acid (Scheme 2) till the completion of reaction which was checked by single spot TLC. Then the reaction mixture was poured in 200 ml ice cold water, neutralized with sodium bicarbonate solution followed by the extraction of ester with ether (50 ml). The ether layer was separated, which on evaporation yielded the ester derivatives of 2-amino benzoic acid.

Compound 2: M.P. (°C) 258–260; Yield 34.45%; 1H NMR (DMSO): 3.67 (s, 3H, OCH3), 7.17–8.05 (m, 8H of ArH); IR (KBr pellets, cm−1): 1609.5 (C=O str., secondary amide), 3464.56 (N–H sym. str., amine), 1772.92 (C–H out of plane bending of Ar), 1290.77 (C–N str., N–H bending of sec., amide), 1388.83 (C–N str., aryl primary amine), 1511.41 (N–H in plane bending of sec., amide), 1021.08 (CH in plane bending of phenyl ring), 775.92 (C–H out of plane bending of 1,2 disubstituted benzene ring); Compound 10: M.P. (°C) 248–250; Yield 48.46%; 1H NMR (DMSO): 5.21 (s, 2H, NH2 of ArNH2), 8.51 (s, 1H, H of CONH), 7.06–8.02 (m, 7H, H of ArH); IR (KBr pellets, cm−1): 1768.64 (C–H out of plane bending of Ar), 1613.57 (C=O str., secondary amide), 1499.16 (N–H in plane bending of secondary amide), 1213.16 (C–F str.); Compound 12: M.P. (°C) 178–180; Yield 37.09%; 1H NMR (DMSO): 7.17–8.63 (m, 4H, H of ArH), 3.07–4.70 (m, 8H of morpholine); IR (KBr pellets, cm−1): 1772.92 (C–H out of plane bending of Ar), 1609.48 (C=O str. of tertiary amide), 1033.34 (ring str., of morpholine); 771.83 (C–H out of plane bending of 1,2 disubstituted benzene ring); Compound 16: M.P. (°C) 183–185; Yield 37.59%; 1H NMR (DMSO): 3.40 (s, 2H, H of ArNH2), 6.17–8.98 (m, 4H of ArH); IR (KBr pellets, cm−1): 3460.47 (N–H asym. of primary amine), 3354.24 (N–H sym. str., of primary amine), 1768.84 (C–H out of plane bending of Ar), 1650.34(C=O str., primary amide), 1605.39 (N–H in plane bending of primary amine), 1572.71 (N–H in plane bending primary amide), 1507.33 (C=C str., skeletal of Ar), 1155.93 (C–H in plane bending of phenyl ring); Compound 28: B.P. (°C) 105–107; Yield 82.7%; 1H NMR (DMSO): 3.46 (s, 2H, H of ArNH2), 7.31–7.79 (m, 4H of ArH); IR (KBr pellets, cm−1): 1654.43 (C=O str., ester), 1209.04 (C–N str., aryl primary amine), 1115.06 (C–O–C str., of aromatic ester), 1037.43 (sym. str., of aryl ester).

Antimicrobial evaluation

Antibacterial assay

A 24 h fresh culture was obtained by inoculation of respective bacteria in double strength nutrient broth-I.P. followed by incubation at 37 ± 1°C. The stock solution of synthesized 2-amino benzoic acid derivatives was serially diluted in tube containing 1 ml of sterile double strength nutrient broth-I.P. to get a concentration of 50–1.56 μg/ml and then inoculated with 100 μl of suspension (with a count of 105 cfu/ml) of respective microorganisms (Gram-positive Staphylococcus aureus, Bacillus subtilis, Gram negative Escherichia coli) in sterile saline. The inoculated tubes were incubated at 37 ± 1°C for 24 h and minimum inhibitory concentration (MIC) was determined.

Antifungal assay

The antifungal activity of synthesized 2-amino benzoic acid derivatives against the fungal species Candida albicans and Aspergillus niger was determined by serial dilution method similar to antibacterial assay using Sabouraud dextrose broth-I.P. The inoculated tubes were incubated at 37 ± 1°C and 25 ± 1°C for a period of 2 and 7 days in case of C. albicans and A. niger, respectively, and minimum inhibitory concentration (MIC) was determined.

Determination of MBC/MFC

The minimum bactericidal concentration (MBC) and fungicidal concentration (MFC) were determined by subculturing on fresh medium 100 μl of culture from each tube that remained clear in the MIC determination. MBC and MFC values represent the lowest concentration of compound that produces a 99.9% end point reduction (Rodriguez-Arguelles et al., 2005).

QSAR studies

The calculations of molecular descriptors as well as regression analysis were carried by using molecular package TSAR 3D version 3.3 (2000). The description of these descriptors is available in the literature (Hansch et al., 1973; Kier and Hall, 1976; Randic, 1975, 1993; Balaban, 1982; Wiener, 1947).

The predictive powers of the equations were validated by leave one out (LOO) cross-validation method (Agrawal et al., 2006), where a model is built with N − 1 compounds and Nth compound is predicted. Each compound is left out of the model derivation and predicted in turn. An indication of the performance is obtained from cross-validated r 2 method which is defined as

$$ q^{ 2} = 1-\Upsigma \left( {Y_{\text{predicted}} -Y_{\text{actual}} } \right)^{ 2} /\Upsigma \left( {Y_{\text{actual}} -Y_{\text{mean}} } \right)^{ 2} $$

where, Y predicted, Y actual and Y mean are predicted, actual and mean values of target property (pMIC), respectively. Σ(Y predicted − Y actual)2 is predictive residual error sum of squares.

Results and discussion

A series of 2-amino benzoic acid derivatives (1–28) were synthesized by following the general pathway depicted in Schemes 1 and 2. The amide/anilide derivatives (1–20) of 2-amino benzoic acid were prepared by addition of amines/anilines to a solution of 2-amino benzoyl chloride. The ester derivative (21) having the quinoline nucleus was obtained by the reaction of 2-amino benzoyl chloride with 8-hydroxy quinoline. The esters of 2-amino benzoic acid (22–28) were prepared by reaction of 2-amino benzoic acid with appropriate alcohol in presence of sulphuric acid. The physiochemical characteristics of synthesized 2-amino benzoic acid derivatives (1–28) are presented in Table 1.

Scheme 1
scheme 1

Scheme for synthesis of amide/anilide derivatives of 2-aminobenzoic acid

Full size image
Scheme 2
scheme 2

Scheme for synthesis of ester derivatives of 2-aminobenzoic acid

Full size image
Table 1 Physicochemical characteristics of 2-amino benzoic acid derivatives
Full size table

The synthesized compounds were characterized by their consistent IR and NMR spectral characteristics. The appearance of IR bands around 1630–1610 cm−1 indicates the formation of secondary amides and anilides. The appearance of aromatic ring is indicated by the IR bands at 775.92, 1021.08, 1772.92 cm−1 in case of compound 2. The formation of anilides by the reaction of 3-chloro, 4-fluoro aniline is confirmed by the appearance of IR bands corresponding to C–F stretch around 1213.13 cm−1 apart from the appearance of C–H stretch of aromatic ring at 1768.84 cm−1 in case of compound 10. The formation of tertiary amide containing morpholine (12) was confirmed by the appearance of IR peak at 1033.34 cm−1 which corresponds to morpholine ring in addition to IR peaks for tertiary amide and primary amino group of anthranilic acid. The formation of primary amide (16) was demonstrated by the shifting of IR band towards higher wave number 1650.34 cm−1 from 1630–1610 cm−1 appeared in case of secondary amide. Further, the formation of compound 16 was indicated by the appearance of IR peak at 1605.39 cm−1 and 3460.47 cm−1 which corresponds to NH in plane bending of primary amine and NH stretch of primary amide, respectively. The formation of aromatic esters is confirmed by the appearance of C–O–C stretch of aromatic esters at 1115.06 cm−1 along with the IR C=O stretch for esters.

The 1H NMR spectra of synthesized compounds showed a multiplet signal at δ 7.02–8.98 ppm corresponding to the presence of aromatic nucleus. The singlet signal for the methoxy group in compound 2 was found at δ 3.67 ppm. The presence of aromatic amino group in the synthesized compounds was evidenced by appearance of singlet signal around δ 3.40 ppm. A multiplet signal at δ 3.07–4.70 ppm demonstrated the presence of morpholine nucleus in compound 12.

The synthesized 2-amino benzoic acid derivatives were screened in vitro for their antibacterial activity against Gram positive S. aureus, B. subtilis and Gram negative E. coli and in vitro antifungal activity against C. albicans and A. niger by tube dilution method (Cappucino and Sherman, 1999). Double strength Nutrient broth-I.P. and Sabouraud dextrose broth-I.P (Pharmacopoeia of India, 2007) have been employed as media for growth of bacterial and fungal species, respectively. The results of antimicrobial activity are presented in Table 2.

Table 2 Antimicrobial activity of 2-amino benzoic acid derivatives
Full size table

In case of S. aureus compound 21 was found to be active having pMIC value of 1.93 (Table 2). Against B. subtilis compounds 10, 14 and 20 emerged as most active ones with pMIC values of 2.23, 2.27 and 2.23 (Table 2), respectively. Compounds 24 and 25 demonstrated high antibacterial activity against the Gram negative bacteria, E. coli with pMIC values 1.85 and 1.88 (Table 2), respectively.

For antifungal activity against C. albicans, compounds 21 and 25 exhibited their antifungal potential at pMIC values 1.63 and 1.58, respectively, which is high in comparison to the antifungal activity of other synthesized 2-amino benzoic acid derivatives. In case of antifungal activity against A. niger compounds 23 and 24 were found to be active ones with pMIC values 1.72 and 1.85, respectively. In general, the compound 21, which has 8-hydroxyquinoline nucleus, has shown appreciable antimicrobial activity against all the microorganisms under test. The idea of coupling heterocyclic moiety to anthranilic acid gave appreciable antimicrobial potential to the anthranilic acid.

The minimum bactericidal concentration/minimum fungicidal concentration (MBC/MFC) (Table 3) determination results revealed that, in general, the synthesized 2-amino benzoic acid derivatives were bacteriostatic/fungistatic in action except in case of B. subtilis where most of the synthesized derivatives were found to be bactericidal in action (a drug is considered to be bacteriosatic/fungistatic when its MBC and MFC values are 3-fold higher than its MIC values) (Emami et al., 2004).

Table 3 MBC/MFC values of 2-amino benzoic acid derivatives
Full size table

From the aforementioned antimicrobial activity results following structure–activity relationship (SAR) of 2-amino benzoic acid derivatives can be deduced:

  1. 1.

    Compound 21 (2-amino-benzoic acid quinolin-8-yl ester) was found to be active against S. aureus. This indicates that heterocyclic nucleus quinoline is essential for an anthranilic acid derivative to be effective against S. aureus.

  2. 2.

    In contrast to S. aureus, the esters of saturated alcohols, 2-amino-benzoic acid hexyl ester (24) and 2-amino-benzoic acid heptyl ester (25) are effective against E. coli.

  3. 3.

    In case of antifungal activity against C. albicans and A. niger, the anthranilic acid requires esterification.

  4. 4.

    The bulky aromatic substitution in the anilide portion is essential for antibacterial activity of anthranilic acid derivatives against B. subtilis.

  5. 5.

    It is important to note here that compound 10 that has a single phenyl group attached to amide nitrogen is as equally active as compound 14 and 20 with bulky aromatic group against B. subtilis. This higher activity of compound 10 may be attributed to the presence of electron withdrawing group, i.e. F and Cl in compound 10. The role of electron withdrawing groups improving antibacterial activity is supported by the findings of Sharma et al. (2004).

  6. 6.

    In general the esters are more active than the amides and anilides against the tested panel of microorganisms. The anilides were active against B. subtilis.

  7. 7.

    The amide formation does not improve the antimicrobial profile of 2-amino benzoic acid as none of the synthesized amides were found to be active.

  8. 8.

    The aforementioned facts demonstrated that different compounds are active against different microorganisms. This clearly indicates that different structural requirements are essential for a compound to be effective against different microorganisms which is similar to the results of Sortino et al. (2007).

The above findings are summarized in Fig. 1.

Fig. 1
figure 1

SAR for antimicrobial activity of 2-amino benzoic acid derivatives

Full size image

QSAR studies

Development of one-target QSAR model

In order to understand the experimental antimicrobial data on theoretical basis, we established a quantitative structure activity relationship (QSAR) between the in vitro antimicrobial activity of 2-amino benzoic acid derivatives and descriptors coding for lipophilic, electronic, steric and topological parameters of the molecules under consideration using the linear free energy relationship model (LFER) described by Hansch and Fujita (1964). Biological activity data determined as MIC values was first transformed into pMIC values on molar basis, which was used as dependent variable in QSAR study. The different molecular descriptors (independent variables) calculated as: log of octanol − water partition coefficient (log P), molar refractivity (MR), Kier’s molecular connectivity (0χ, 0χv, 1χ, 1χv, 2χ, 2χv) and shape (κ1, κα 1) topological indices, Randic topological index (R), Balaban topological index (J), Wiener topological index (W), Total energy (T e), energies of highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO), dipole moment (μ), electronic energy (Ele.E), nuclear energy (Nu.E) and molecular surface area (SA) (Hansch et al., 1973; Kier and Hall, 1976; Randic, 1975, 1993; Balaban, 1982; Wiener, 1947). The structures of 2-aminobenzoic acid derivatives are first pre-optimized with the Eigen-vector following procedure included in TSAR for Windows, and the resulting geometries are further refined by means of the semiempirical method AM-1 (Hamiltonian type). We chose a time limit of 3600 s for the geometry optimization and the electronic parameters T e, HOMO, LUMO, dipole moment, electronic energy (Ele.E) and nuclear energy (Nu.E) were calculated. The values of selected molecular descriptors used in the QSAR study are presented in Table 4.

Table 4 Value of QSAR descriptors of 2-amino benzoic acid derivatives used in the present study
Full size table

In the present study, a data set of 28 synthesized compounds was subjected to linear free energy regression analysis for model generation. Preliminary analysis was carried out in terms of correlation analysis. A correlation matrix constructed for antibacterial activity against S. aureus is presented in Table 5. The high interrelationship was observed between 1χ and W (r = 0.989) and low interrelationship was observed between LUMO and μ as well as κα3 and 3χv (r = 0.008).

Table 5 Correlation matrix for antibacterial activity of 2-amino benzoic acid derivatives against S. aureus
Full size table

The correlations of different molecular descriptors with antimicrobial activity are presented in Table 6. For S. aureus the QSAR model (Eq. 1) was developed with topological parameter, first order molecular connectivity index, 1χ (r = 0.723, Tables 5, 6).

Table 6 Correlation of molecular descriptors with antimicrobial activity of 2-amino benzoic acid derivatives
Full size table

ot-QSAR model for antibacterial activity against S. aureus

$$ \begin{aligned} {\text{pMIC}}_{\text{sa}} = 0.0 9 4{}\,^{1} \chi + 0. 5 1 2 \\ n = 2 8\quad r = 0. 7 2 3\quad q^{ 2} = 0. 3 9 3\quad s = 0. 1 2 9\quad F = 2 8. 5 3 \end{aligned} $$
(1)

Here and thereafter, n is the number of data points, r is the correlation coefficient, q 2 is the cross validated r 2 obtained by LOO method, s is the standard error of the estimate and F is the Fischer statistics.

Being the coefficient of 1χ in Eq. 1 is positive, the antibacterial activity against S. aureus will increase with increase in value of 1χ. This is clearly evident that compounds 14 and 21 having high 1χ values 10.77 and 9.75, respectively (Table 4) have high pMIC values 1.36 and 1.93, respectively (Table 2). Similarly, compound 19 having low 1χ value (Table 4) has shown minimum antibacterial activity against S. aureus.

In order to improve value of regression coefficient (r), we coupled topological parameter, first order molecular connectivity index, 1χ with electronic parameter, dipole moment, based on their low inter-relationship between them (r = 0.365, Table 5), resulted in marginal improvement in r-value from 0.723 to 0.767 in the developed QSAR model (Eq. 2).

ot-QSAR model for antibacterial activity against S. aureus

$$ \begin{gathered} {\text{pMIC}}_{\text{sa}} = 0. 10 7{}\,^{1} \chi - 0.0 50\mu + 0. 5 2 2\hfill \\ n = 2 8\quad r = 0. 7 6 7\quad q^{ 2} = 0. 3 2 9\quad s = 0. 1 2 3\quad F = 1 7. 8 2\hfill \\ \end{gathered} $$
(2)

Further in search of a better QSAR model, we coupled topological parameter, first order molecular connectivity index, 1χ with electronic parameter, LUMO and the change resulted in Eq. 3 having value of regression coefficient 0.794.

ot-QSAR model for antibacterial activity against S. aureus

$$ \begin{gathered} {\text{pMIC}}_{\text{sa}} = 0.0 7 1{}\,^{1} \chi - 0. 4 3 7 {\text{ LUMO}} + 0. 6 4 8\hfill \\ n = 2 8\quad r = 0. 7 9 4\quad q^{ 2} = 0. 3 3 9\quad s = 0. 1 1 6\quad F = 2 1. 2 7\hfill \\ \end{gathered} $$
(3)

As coupling of topological parameter, first order molecular connectivity index, 1χ with electronic parameters, μ and LUMO resulted in improvement in value of regression coefficient (Eqs. 2, 3), we finally coupled both μ and LUMO with 1χ in order to achieve the best QSAR model with highest value of regression coefficient (Eq. 4).

ot-QSAR model for antibacterial activity against S. aureus

$$ \begin{gathered} {\text{pMIC}}_{\text{sa}} = 0.0 8 3{}\,^{1} \chi - 0. 3 8 3 {\text{ LUMO}} - 0.0 3 8\mu + 0. 6 3 9\hfill \\ n = 2 8\quad r = 0. 8 1 6\quad q^{ 2} = 0. 3 9 3\quad s = 0. 1 1 3\quad F = 1 5. 9 6\hfill \\ \end{gathered} $$
(4)

The developed QSAR model (Eq. 4) was cross validated by q 2 value (q 2 = 0.393) obtained by LOO method. The value of q 2 less than 0.5 indicated that the developed model is an invalid one. But one should not forget the recommendations of Golbraikh and Tropsha who have recently reported that the only way to estimate the true predictive power of a model is to test their ability to predict accurately the biological activities of compounds. As the observed and predicted values are close to each other (Table 7), the QSAR model for S. aureus (Eq. 4) is a valid one (Golbraikh and Tropsha, 2002). The comparison of observed and predicted antibacterial activities is presented in Table 7. It can be seen from the results that the observed and predicted antimicrobial activities lie close to each other as evidenced by their low residual values. The plot of predicted pMICsa against observed pMICsa (Fig. 2) also favours the developed model expressed by Eq. 4. Further, the plot of observed pMICsa versus residual pMICsa (Fig. 3) indicated that there was no systemic error in model development as the propagation of error was observed on both sides of zero (Kumar et al., 2007).

Table 7 Observed and predicted antibacterial activity of 2-amino benzoic acid derivatives against S. aureus, B. subtilis, E. coli and C. albicans using the best ot-QSAR models
Full size table
Fig. 2
figure 2

Plot of predicted pMICsa against the observed pMICsa for the regression model developed by Eq. 4

Full size image
Fig. 3
figure 3

Plot of residual pMICsa values against the experimental pMICsa values

Full size image

Equations 5–7 were developed to predict the antimicrobial activity of synthesized 2-amino benzoic acid derivatives against the B. subtilis, E. coli and C. albicans, respectively.

ot-QSAR model for antibacterial activity against B. subtilis

$$ \begin{gathered} {\text{pMIC}}_{\text{bs}} = 0. 2 1 4 {\text{ logP}}-0. 1 2 5\kappa \alpha_{3} + 1. 9 2 7\hfill \\ n = 2 8\quad r = 0. 6 1 7\quad q^{ 2} = 0. 1 8 7\quad s = 0. 20 2\quad F = 7. 7 1\hfill \\ \end{gathered} $$
(5)

ot-QSAR model for antibacterial activity against E. coli

$$ \begin{gathered} {\text{pMIC}}_{\text{ec}} = 0.0 1 2 {\text{ Te}} + 0.0 7 3 {\text{ logP}}-0. 4 9 6 {\text{ LUMO}} + 1. 4 2 5\hfill \\ n = 2 8\quad r = 0. 7 5 2\quad q^{ 2} = 0. 30 1\quad s = 0. 1 6 9\quad F = 10. 4 4\hfill \\ \end{gathered} $$
(6)

ot-QSAR model for antifungal activity against C. albicans

$$ \begin{gathered} {\text{pMIC}}_{\text{ca}} = 0. 1 1 8{}\,^{1} \chi - 0. 5 7 1{}^{3} \chi^{\text{v}} + 0. 5 3 2\hfill \\ n = 2 8\quad r = 0. 7 5 7\quad q^{ 2} = 0. 3 8 5\quad s = 0. 1 3 8\quad F = 1 6. 8 2\hfill \\ \end{gathered} $$
(7)

In case of B. subtilis, the developed QSAR model (Eq. 5) indicated the predominance of lipophilic parameter, log P and topological parameter Kier’s alpha third order shape index, κα3 in describing the antibacterial activity. The coefficient of log P is positive which shows that the antibacterial activity will increase with the increase in log P value of the synthesized compounds, which is clearly evident from the results of antibacterial activity against B. subtilis (Table 2) and values of log P presented in Table 4.

For antibacterial activity against E. coli, the developed QSAR model (Eq. 6) describes the importance of total energy (T e), log P and LUMO. In this case a positive correlation was observed between T e and antibacterial activity against E. coli which reveals that decrease in T e value (Table 4) will decrease the antibacterial activity against E. coli.

The model described by Eq. 7 depicted the importance of first order molecular connectivity index, 1χ and valence third order molecular connectivity index, 3χv in describing the antifungal activity against C. albicans. It is important to note that no significant correlation was found between molecular descriptors of the synthesized compounds and their antifungal activity against A. niger.

The validity and predictability of the QSAR models, i.e. Eqs. 5–7 is indicated by high values of their correlation coefficient (r) as well as the low residual values (Table 7).

Topological indices are numerical quantifier of molecular topology and are sensitive to bonding pattern, symmetry, content of heteroatom as well as degree of complexity of atomic neighbourhood (Lather and Madan, 2005). The molecular connectivity topological index, χ signifies the degree of branching, connectivity of atoms and the unsaturation in the molecule which accounts for variation in activity (Gupta et al., 2003).

In the matter of QSAR, the total energy plays important role. Total energy of a molecular system is the sum of the total electronic energy, and the energy of internuclear repulsion (Pasha et al., 2006).

The electronic parameter LUMO, which denotes the energy of lowest unoccupied molecular orbital directly relates to the electron affinity and characterizes the sensibility of the molecule towards an attack by nucleophile. The contribution of LUMO in describing antifungal activity may be attributed to the interaction of 2-amino benzoic acid derivatives with nucleophilic amino acid residue like cysteine of fungi (Kumar et al., 2008).

The importance of dipole moment in modulating antibacterial activity against S. aureus may be due to the presence of carbonyl group (C+–O−) where permanent polarization is seen due to electro negativity difference between the atoms. The carbonyl oxygen of substituted benzimidazoles may involve in making fruitful binding interactions with amino acid present at the target site, through hydrogen bonding. The molecular property dipole moment plays a critical role in modulating antibacterial profile of this class of compounds (Pillai et al., 2005).

It is important to note that Eqs. 2–7 were derived using the entire data set as there were no outliers in the data set. In multivariate statistics, it is common to define three types of outliers (Furusjo et al., 2006).

  1. 1.

    X/Y relation outliers are substances for which the relationship between the descriptors (X variables) and the dependent variables (Y variables) is not the same as in the (rest of the) training data.

  2. 2.

    X outliers are substances whose molecular descriptors do not lie in the same range as the (rest of the) training data.

  3. 3.

    Y outliers are only defined for training or test samples. They are substances for which the reference value of response is invalid.

The sample size and the ‘rule of thumb’ allowed us to go for development of tetra-parametric model in multiple linear regression analysis. The ‘rule of thumb’ gives information about the number of parameters to be selected for regression analysis in QSAR based on the number of compounds (Narasimhan et al., 2007c).

Generally for QSAR studies, the biological activities of compounds should span 2–3 orders of magnitude. But in the present study the range of antimicrobial activities of the synthesized compounds is within one order of magnitude. This is in accordance with results suggested by the Bajaj et al., who stated that the reliability of the QSAR model lies in its predictive ability even though the activity data are in the narrow range (Bajaj et al., 2005). Further, recent literature reveals that the QSAR have been applied to describe the relationship between narrow range of biological activity and physicochemical properties of the molecules (Narasimhan et al., 2007c; Sharma et al., 2006; Hatya et al., 2006; Kumar et al., 2006). When biological activity data lies in the narrow range, the presence of minimum standard deviation of the biological activity justifies its use in QSAR studies (Kumar et al., 2007; Narasimhan et al., 2007c). The minimum standard deviation (Table 2) observed in the antimicrobial activity data justifies its use in QSAR studies.

The value of q 2 less than 0.5 indicated that some of the developed models are invalid ones. But it is important to note that the predictability of the QSAR models developed in the present study is highly evidenced by the low residual values. This is in accordance with the recommendations of Golbraikh and Tropsha, who have recently reported that the only way to estimate the true predictive power of a model is to test their ability to predict accurately the biological activities of compounds. As the observed and predicted values are close to each other (Tables 7, 8), the developed QSAR models are valid ones (Golbraikh and Tropsha, 2002).

Table 8 Observed and predicted antibacterial, antifungal and antimicrobial activities of 2-amino benzoic acid derivatives using the best mt-QSAR models
Full size table

Even though the sample size and the ‘rule of thumb’ allowed us to go for development of penta-parametric model in multiple linear regression analysis, the high colinearity among the parameters restricted us to go for mono-parametric model in most cases. The multi-colinearity occurs when two independent variables are correlated with each other. One should note that the change in signs of the coefficients, a change in the values of previous coefficient, change of significant variable into insignificant one or an increase in standard error of the estimate on addition of an additional parameter to the model are indications of high interrelationship among descriptors (Kumar et al., 2007).

Development of multi-target QSAR model

According to ot-QSAR models one should use five different equations with different errors to predict the activity of a new compound against the five microbial species. However, very recently the interest has been increased in development of multi-target QSAR (mt-QSAR) models. In opposition to ot-QSAR, the mt-QSAR model is a single equation that considers the nature of molecular descriptors which are common and essential for describing the antimicrobial activity (Prado-Prado et al., 2008; Gonzalez-Diaz et al., 2007, 2008; Cruz-Monteagudo et al., 2007).

In the present study, we have attempted to develop three different types of mt-QSAR models viz. mt-QSAR model for describing antibacterial activity of synthesized compounds against S. aureus, B. subtilis and E. coli, mt-QSAR model for describing antifungal activity of synthesized compounds against C. albicans and A. niger as well a common mt-QSAR model for describing the antimicrobial activity of 2-amino benzoic acid derivatives against all the aforementioned microorganisms.

In order to develop mt-QSAR models, initially we have calculated the average antibacterial activity, antifungal activity and antimicrobial activity of substituted 2-amino benzoic acid derivatives which are presented in Table 2. These average activity values were correlated with the molecular descriptors of synthesized compounds (Table 4).

The mt-QSAR model of antibacterial activity displayed the importance of log P and LUMO (Table 6), in describing the antibacterial activity of 2-amino benzoic acid derivatives.

mt-QSAR model for antibacterial activity

$$ \begin{gathered} {\text{pMIC}}_{\text{ab}} = 0.0 8 1 {\text{ logP}}-0. 3 6 2 {\text{ LUMO}} + 1. 3 70 \hfill \\ n = 2 8\quad r = 0. 7 4 6\quad q^{ 2} = 0. 3 3 6\quad s = 0. 10 2\quad F = 1 5. 6 9\hfill \\ \end{gathered} $$
(8)

The mt-QSAR model for antifungal activity reveals the importance of Wiener topological index (W) and valence third order molecular connectivity index, 3χv in describing antifungal activity.

mt-QSAR model for antifungal activity

$$ \begin{gathered} {\text{pMIC}}_{\text{af}} = - 0. 6 5 6{}\,^{3} \chi^{\text{v}} + 0.000 6 {\text{ W}} + 1. 2 5 1\hfill \\ n = 2 8\quad r = 0. 7 3 3\quad q^{ 2} = 0. 40 1\quad s = 0. 1 2 2\quad F = 1 4. 5 5\hfill \\ \end{gathered} $$
(9)

The mt-QSAR model of antimicrobial activity (Eq. 10) depicted the importance of Wiener topological index (W), valence third order molecular connectivity index, 3χv and LUMO in describing the antimicrobial activity of synthesized 2-amino benzoic acid derivatives.

mt-QSAR model for antimicrobial activity

$$ \begin{gathered} {\text{pMIC}}_{\text{am}} = - 0. 3 6 1{}\,^{3} \chi^{\text{v}} + 0.000 4 {\text{W}}-0. 2 4 8 {\text{ LUMO}} + 1. 3 7 2\hfill \\ n = 2 8\quad r = 0. 7 5 6\quad q^{ 2} = 0. 3 5 9\quad s = 0.0 9 6\quad F = 10. 7 1\hfill \\ \end{gathered} $$
(10)

The Wiener index (W) was introduced by Wiener to demonstrate correlations between physicochemical properties of organic compounds and the topological structure of their molecular graphs in terms of sum of distances between any two carbon atoms in the molecules, in terms of carbon–carbon bonds (Wiener, 1947).

The data presented in Table 8 and the plot of observed versus predicted antimicrobial activity (Fig. 4) obtained by the mt-QSAR model (Eq. 10) indicated their predictability as both were close to each other. Further the plot of observed activity and residual antimicrobial activity (Fig. 5) indicated that no systemic error was involved in the development of mt-QSAR model as the propagation of residuals was found on both the sides of zero.

Fig. 4
figure 4

Plot of predicted pMICam against the observed pMICam for the regression model developed by Eq. 10

Full size image
Fig. 5
figure 5

Plot of residual pMICam values against the experimental pMICam values

Full size image

It was observed from mt-QSAR models (Eqs. 8–10) that the antibacterial activity, antifungal activity and overall antimicrobial activity of synthesized 2-amino benzoic acid derivatives is governed by the structural parameters especially the 3χv, LUMO and W. The other ot-QSAR and mt-QSAR models derived for antimicrobial activity of 2-amino benzoic acid derivatives are presented in Table 9.

Table 9 Regression analysis and quality of correlation for modeling antibacterial and antifungal activity of synthesized 2-amino benzoic acid derivatives
Full size table

Conclusion

In the present study, 2-amino benzoic acid derivatives (1–28) were synthesized and evaluated for their in vitro antimicrobial activity against S. aureus, B. subtilis, E. coli, C. albicans and A. niger. The results of antimicrobial studies indicated that, in general, synthesized compounds were found to be bacteriostatic and fungistatic in action. Further, the antimicrobial results of synthesized 2-amino benzoic acid derivatives indicated that different structural requirements are essential for a compound to be selected as antibacterial or antifungal agent. Further, QSAR investigation was performed by development of one target and multi target models. The multi-target model was found to be effective in describing the antimicrobial activity of 2-amino benzoic acid derivatives in comparison to the one target models and indicated the importance of the 3χv, LUMO and W in describing the antimicrobial activity.