Antimicrobial activity and SAR study of new gemini imidazolium-based chlorides

Article abstract of DOI:10.1111/cbdd.12236

A series of 70 new 3,3′(α,ω-dioxaalkyl)bis(1- alkylimidazolium) chlorides were synthesized. They were characterized with respect to surface active properties and antimicrobial activity against the following pathogens: Staphylococcus aureus, Enterococcus faecalis, Acinetobacter baumannii, Escherichia coli, Klebsiella pneumoniae, Enterobacter cloacae, Pseudomonas aeruginosa, Candida krusei, and Candida albicans. In this article, besides description of the synthesis, we characterize a set of features of these compounds, concerning their structure (described by the length of the dioxaalkan spacer and the length of the alkyl substituent in the aromatic ring) and surface active properties (critical micelle concentration, value of surface tension at critical micelle concentration, value of surface excess, molecular area of a single particle, and free energy of adsorption of molecule). Then, we present a SAR study for Staphylococcus aureus, as one of the most widespread pathogenic strains, conducted with the help of the Dominance-based Rough Set Approach (DRSA), that involves identification of relevant features and relevant combinations of features being in strong relationship with a high antimicrobial activity of the compounds. The SAR study shows, moreover, that the antimicrobial activity is dependent on the type of substituents and their position at the chloride moiety, as well as on the surface active properties of the compounds.

Full text of DOI:10.1111/cbdd.12236

Chem Biol Drug Des 2014; 83: 278–288
Research Article
Antimicrobial Activity and SAR Study of New Gemini
Imidazolium-Based Chlorides
1,
2
ꢀ
Key words: cheminformatics, drug design, structure-based
drug design
Łukasz Pałkowski *, Jerzy Błaszczynski ,
Andrzej Skrzypczak3, Jan Błaszczak3, Karolina
Kozakowska , Joanna Wroblewska , Sylwia
Kozuszko , Eugenia Gospodarek , Jerzy
Krysinski and Roman Słowinski
1
4
ꢀ
Received 11 March 2013, revised 6 September 2013 and
accepted for publication 9 September 2013
4
4
_
1
2,5
ꢀ
ꢀ
1Department of Pharmaceutical Technology, Nicolaus
Copernicus University, Jurasza 2, 85-094 Bydgoszcz,
Poland
Cationic surfactants such as ammonium, phosphonium,
pyridinium, and imidazolium salts have good biocidal activ-
ity against broad spectrum of bacteria, fungi, and algae
(1). Those compounds might also be used as anticorrosive
and anti-electrostatic agents in cosmetic, textile, and phar-
maceutical industries, as a biocides in wood protection
and in the agriculture as antifungal agents (2,3). The char-
acteristics of these compounds: the broad spectrum of
antimicrobial activity, low toxicity, good stability, and lack
of odor of usable solutions are the reasons for the com-
pounds’ wide usage as disinfectant, as cleaning agents in
food and pharmaceutical industries and in hospitals (4).
Bis-quaternary ammonium chlorides are a group of com-
pounds with good antimicrobial activity (5–7). Bis-imidazoli-
um derivatives are characterized by good surface
properties (8–10). Nevertheless, some microbial strains
might become resistant to those compounds over the
time. Those various properties of imidazolium salts form
the base for seeking and synthesizing new, active prod-
ucts. Numerous tests indicated that chemical and antimi-
crobial properties of gemini imidazolium-based chlorides
depend on their structure (e.g. length of alkyl group) (11).
The hydrophobic part of the molecule (alkyl chain) enables
penetration through the cell membrane, induce its damage
and exchange of extracellular with intracellular fluids, which
lead to cell lysis and its death (12).
2
ꢀ
Institute of Computing Science, Poznan University of
ꢀ
Technology, Piotrowo 2, 60-965 Poznan, Poland
Institute of Chemical Technology, Poznan University of
3
ꢀ
ꢀ
Technology, Skłodowskiej-Curie 2, 60-965 Poznan, Poland
4Department of Microbiology, Nicolaus Copernicus
University, Skłodowskiej-Curie 9, 85-094 Bydgoszcz,
Poland
5Systems Research Institute, Polish Academy of Sciences,
Newelska 6, 01-447 Warsaw, Poland
*Corresponding author: Łukasz Pałkowski,
lukaszpalkowski@cm.umk.pl
A series of 70 new 3,3´(a,x-dioxaalkyl)bis(1-alkylimi-
dazolium) chlorides were synthesized. They were
characterized with respect to surface active properties
and antimicrobial activity against the following patho-
gens: Staphylococcus aureus, Enterococcus faecalis,
Acinetobacter baumannii, Escherichia coli, Klebsiella
pneumoniae, Enterobacter cloacae, Pseudomonas
aeruginosa, Candida krusei, and Candida albicans. In
this article, besides description of the synthesis, we
characterize a set of features of these compounds,
concerning their structure (described by the length of
the dioxaalkan spacer and the length of the alkyl sub-
stituent in the aromatic ring) and surface active proper-
ties (critical micelle concentration, value of surface
tension at critical micelle concentration, value of sur-
face excess, molecular area of a single particle, and
free energy of adsorption of molecule). Then, we pres-
ent a SAR study for Staphylococcus aureus, as one of
the most widespread pathogenic strains, conducted
with the help of the Dominance-based Rough Set
Approach (DRSA), that involves identification of rele-
vant features and relevant combinations of features
being in strong relationship with a high antimicrobial
activity of the compounds. The SAR study shows,
moreover, that the antimicrobial activity is dependent
on the type of substituents and their position at the
chloride moiety, as well as on the surface active
properties of the compounds.
In the SAR analysis of the chemical compounds, different
chemometric statistical and nonstatistical methods (artificial
neural networks, principal component analysis, parametric
statistic: regression, correlation) have been used. In our
study, we used a method based on rough set theory. This
method is called Dominance-based Rough Set Approach
(DRSA) (13) and yields a description of structure-activity
relationships in terms of easily interpretable ‘if…, then…’
decision rules.
In this work, we synthesized a new group of 3,3′(a,x-
dioxaalkyl)bis(1-alkylimidazolium) chlorides, then determined
their surface properties and antimicrobial activity and finally
we performed SAR analysis using DRSA.
278
ª 2013 John Wiley & Sons A/S. doi: 10.1111/cbdd.12236

Antimicrobial Activity of New Gemini Chlorides
a capillary column HP-5 and a flame ionization detector to
check the purity of samples under study. All samples were
with the purity of 99.6–99.8%.
Methods and Materials
Materials
All reagents were obtained from Sigma-Aldrich or POCH
(Gliwice, Poland).
All a,x-di(chloromethoxy)alkanes readily hydrolyzed in the
presence of small amount of water to form HCl, which in
turn gives the 1-alkylimidazolium hydrochlorides. The sep-
aration of the quaternization product and hydrochloride is
practically impossible. For this reason, quaternization reac-
tions with a,x-di(chloromethoxy)alkanes were conducted
under strictly anhydrous conditions. The nucleophilic
substitution is a specific type of Menschutkin reaction that
involves a SN1 mechanism. The initial rate-determining
step involves the formation of the dimeric cation
+CH2O(CH2)nOCH2+. The obtained cation rapidly reacts
with 1-alkylimidazole giving a very high yield (92.7–98.5%)
of 70 gemini imidazolium based chlorides, which have
appropriate CH2O(CH2)nOCH2 spacer in the structure
(Figure 3).
Synthesis
1-Alkylimidazoles, besides 1-methyl- and 1-butylimidazole,
were synthesized according to the procedure described
earlier (14,15). Sodium (2.3 g/mol) was added to 75 mL of
anhydrous methanol and the imidazole (0.1 mol) was dis-
solved. The appropriate amount (0.11 mol) of one of the
bromoalkane: bromoethane, 1-bromopropane, 1-bromo-
pentane, 1-bromohexane, 1-bromoheptane, 1-bromooc-
tane, 1-bromononane, 1-bromodecane, 1-bromoundecane,
1-bromododecane, 1-bromotetradecane, or 1-bromohex-
adecane was added and the mixture was stirred for 6 h at
65 °C. Precipitated sodium bromide was filtered and the
methanol was removed. The final product, 1-ethyl-, 1-propyl-,
1-pentyl-, 1-hexyl-, 1-heptyl-, 1-octyl-, 1-nonyl-, 1-decyl-,
1-undecyl-, 1-dodecyl-, 1-tetradecyl-, 1-hexadecylimidazole,
was twice distilled under reduced pressure (1 Torr). The yields
were between 63% and 92% (Figure 1).
The synthesized compounds were determined on the
basis of NMR spectra (Bruker NMR spectrometer with tet-
ramethylsilane as an internal standard), and their purity
was confirmed by thin-layer chromatography (TLC Merck
silica gel 60 F254 plates) and elemental analysis (Element-
er Analyzensystem Vario EL III; see Table S1).
The degree of purity was determined using a Hewlett
Packard gas chromatograph 5890 Series II equipped with
a capillary column HP-5 and a flame ionization detector to
check the purity of samples under study. All samples were
with the purity of 99.9%.
Surface active properties evaluation
The surface tension was measured using Kruss K12 tensi-
ometer equipped with du Nouy Pt-Ir ring. All compounds
were diluted with triply distilled water. Measurements of
the surface tension were made under atmospheric pres-
sure, at the temperature of 20 ꢀ 0.01 °C, after the solu-
tion tested reached the temperature equilibrium using a
temperature-controlled bath. All measurements of surface
tension were repeated three times and each datum is an
average of three individual points, with an accuracy of
ꢀ0.2 mN/m.
a,x-Di(chloromethoxy)alkanes were obtained by passing
HCl-gas through a mixture of formaldehyde and appropri-
ate a,x-diol according to the procedure described earlier
(16). Of 0.2 mol of appropriate a,x-diol was dissolved in
toluene and 0.21 mol of formaldehyde was added. Then,
through the stirred mixture 5 mol of HCl-gas was passed
for 2 h at 17 °C (Figure 2).
When the reaction was finished, the water phase and
excess of formaldehyde was removed and the final
product,1,2-di(chloromethoxy)ethane,1,3-di(chloromethoxy)
propane,1,4-di(chloromethoxy)butane,1,5-di(chloromethoxy)
pentane, 1,6-di(chloromethoxy)hexane, was twice distilled
under reduced pressure (1 Torr). The yields were between
87% and 98%. The purity was controlled using a Hewlett
Packard gas chromatograph 5890 Series II equipped with
Surface active properties of synthesized chlorides were
described by the following parameters: critical micelle con-
centration (CMC; mol/L), value of surface tension at critical
micelle concentration (cCMC; mN/m with an accuracy of
ꢀ0.2 mN/m), value of surface excess (Γ 9 106; mol/m2
with an accuracy of 106 ꢀ 0.02 mol/m2;), molecular area
of a single particle (A 9 1020; m2 with an accuracy of
Figure 1: Synthesis of 1-
alkylimidazoles.
Figure 2: Synthesis of α,x-di(chloromethoxy)alkanes.
Chem Biol Drug Des 2014; 83: 278–288
279

Pałkowski et al.
Figure 3: Synthesis of gemini imidazolium based chlorides.
1020 ꢀ 0.08 m2), and free energy of adsorption of mole-
cule (ΔGads; kJ/mol with an accuracy of ꢀ 0.2kJ/mol).
Table 1: Numerical coding of the structure condition attributes
Condition attributes
Code
n
R
Strains and growth condition
1
2
3
4
5
6
7
8
9
10
11
12
14
16
CH3
C2H5
C3H7
C4H9
The antimicrobial activity of synthesized chlorides was
tested against: Staphylococcus aureus ATCC 25213,
Staphylococcus aureus MR-3 (methicillin resistant), Entero-
coccus faecalis ATCC 29212, Escherichia coli ATCC
25922, Klebsiella pneumoniae ATCC 700603, Pseudomo-
nas aeruginosa ATCC 27853, Acinetobacter baumanii
(multidrug resistant), Enterobacter cloacae, Candida albi-
cans ATCC 90028, and Candida krusei ATCC 6258. All
microbial strains were stored at ꢁ80 °C in Cryovial vials
C2H4
C3H6
C4H8
C5H10
C6H12
C5H11
C6H13
C7H15
C8H17
C9H19
C10H21
C11H23
C12H25
C14H29
C16H33
€
(Simport, Canada) and subcultured in Muller Hinton broth
before testing.
Minimum inhibitory concentration determination
Bacterial, fungal suspensions with a turbidity equivalent to
that of a 0.5 McFarland standard were prepared by sus-
pending growth from blood Trypticase Soy Agar (TSA,
BBL) plates in 2 mL of sterile saline. The antibacterial
effects of gemini imidazolium-based chlorides were evalu-
ated by determination of Minimal Inhibitory Concentration
(MIC, % w/V) values using the disk-diffusion method rang-
ing from 1.0 to 0.000125 and the control plate without
disinfectant. The test medium was Mueller–Hinton Agar
culture for the other bacteria and for Candida spp. Details
of the agar dilution procedure for testing microorganisms
have been published by Clinical and Laboratory Standards
Institute (17).
Rough set-based SAR study
In the course of the analysis, jRS and jMAF software
based on DRSA methodology were useda DRSA (13) was
chosen as the most suitable method to discover synthetic
rules that exhibit monotonic relationships between struc-
ture and surface active properties of the compounds on
ꢀ
(bio-Merieux). The MIC was determined as the lowest con-
centration of the tested agent at which no visible bacterial
growth could be detected. A barely visible haze of growth
was disregarded. MIC was read at 37 °C, after 24 or 48 h
280
Chem Biol Drug Des 2014; 83: 278–288

Antimicrobial Activity of New Gemini Chlorides
Table 2: Minimum inhibitory concentration for analyzed compounds
MIC (mM/L)
Compound
Sau MR-3
Efae
Aba
Eco
Kpn
Ecl
Pae
Ckr
Cal
1
2
33.875
28.468
13.181
6.590
5.741
2.697
0.326
0.077
0.037
0.017
0.017
0.032
0.116
0.108
14.826
27.375
25.423
5.933
1.112
2.618
0.158
0.009
0.069
0.008
0.008
0.008
0.056
0.211
28.468
26.363
26.360
2.871
0.345
0.079
0.010
0.009
0.009
0.008
0.008
0.061
0.056
0.052
27.375
25.423
5.933
2.781
0.168
0.158
0.019
0.142
0.008
0.008
0.015
0.030
0.055
0.102
13.181
0.817
0.712
33.875
28.468
13.181
6.590
11.483
10.788
2.543
0.308
0.037
0.017
0.017
0.032
0.029
0.108
29.652
27.375
25.423
5.933
22.249
5.235
0.158
0.036
0.069
0.008
0.008
0.124
0.056
0.106
28.468
26.363
26.360
22.966
2.697
0.079
0.308
0.018
0.009
0.008
0.008
0.061
0.056
0.207
27.375
25.423
23.730
22.249
5.235
4.945
0.300
2.226
0.034
0.008
0.031
0.030
0.110
0.102
6.591
13.180
22.966
33.875
28.468
13.181
6.590
5.741
10.788
2.543
0.597
0.146
0.278
0.265
1.978
1.817
13.441
29.652
27.375
6.356
11.865
5.562
0.649
0.613
0.150
0.069
0.526
0.502
0.960
7.123
13.192
28.468
26.363
3.295
11.483
2.697
1.261
0.308
0.146
0.278
0.066
0.981
0.470
0.056
12.952
27.375
25.423
11.865
22.249
10.471
19.780
1.162
4.451
0.068
0.129
0.062
0.119
0.110
1.590
6.591
26.360
22.966
33.875
28.468
13.181
6.590
5.741
1.338
0.631
0.308
0.073
0.017
0.017
0.063
0.451
1.680
29.652
27.375
12.711
5.933
5.562
2.618
1.226
0.300
0.142
0.008
0.259
0.124
0.883
1.649
28.468
26.363
26.360
0.712
10.788
2.543
1.193
0.292
0.139
0.265
0.032
0.470
0.112
0.052
27.375
25.423
23.730
22.249
5.235
4.945
0.581
4.451
0.034
0.016
0.062
0.059
0.110
0.204
6.591
13.180
2.871
33.875
28.468
13.181
6.590
5.741
5.394
2.543
0.597
0.566
0.538
0.513
0.981
1.817
1.680
29.652
27.375
25.423
5.933
11.125
5.235
0.613
0.581
0.552
0.271
0.502
0.480
7.123
6.596
28.468
26.363
26.360
22.966
5.394
1.261
0.308
0.146
0.069
0.133
0.063
0.940
0.433
0.052
27.375
25.423
23.730
2.781
20.942
4.945
18.740
0.285
2.120
0.259
0.062
0.119
0.425
1.590
3.295
26.360
22.966
33.875
28.468
13.181
13.180
11.483
5.394
1.261
0.308
0.146
0.139
0.265
0.491
0.901
1.680
29.652
27.375
25.423
5.933
11.125
2.618
0.613
0.581
0.069
0.136
0.502
0.480
7.123
1.649
28.468
26.363
26.360
22.966
10.788
1.261
0.597
0.566
0.278
0.265
0.063
0.940
0.433
0.052
27.375
25.423
23.730
2.781
20.942
4.945
18.740
0.285
2.120
0.129
0.062
0.059
0.110
0.789
3.295
26.360
22.966
33.875
28.468
13.181
13.180
11.483
10.788
5.086
1.193
1.132
0.538
0.513
0.491
1.817
1.680
29.652
27.375
25.423
5.933
22.249
10.471
4.945
1.162
4.451
0.526
0.502
15.484
7.123
13.192
28.468
26.363
26.360
11.483
10.788
5.086
1.193
2.283
0.278
0.066
33.875
14.234
13.181
1.634
0.712
0.669
0.020
0.010
0.002
0.002
0.002
0.002
0.002
0.027
0.237
13.687
1.576
2.966
1.379
0.649
0.077
0.150
0.009
0.032
0.008
0.008
0.007
0.013
0.114
6.591
3.295
2.871
1.338
0.326
0.010
0.009
0.009
0.008
0.008
0.008
0.028
0.006
0.109
6.356
2.966
0.779
0.008
9.890
0.075
0.142
0.085
0.065
0.004
0.004
0.003
0.006
0.026
0.105
11.483
16.937
3.558
3.295
1.634
0.712
0.086
0.020
0.005
0.002
0.017
0.017
0.016
0.116
0.108
29.652
27.375
6.356
5.933
1.379
0.082
0.158
0.009
0.034
0.008
0.081
0.008
0.014
0.106
7.117
3.295
3.295
0.712
0.173
0.010
0.010
0.009
0.009
0.008
0.008
0.030
0.007
0.104
6.844
3.178
0.736
0.690
0.168
0.158
0.094
0.071
0.008
0.004
0.004
0.030
0.027
0.102
1.634
1.634
0.712
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
3.957
0.940
0.223
12.952
27.375
25.423
23.730
11.125
20.942
2.472
18.740
2.226
0.136
0.065
0.248
0.460
0.849
12.721
26.363
26.360
22.966
Chem Biol Drug Des 2014; 83: 278–288
281

Pałkowski et al.
Table 2: continued
MIC (mM/L)
Compound
Sau MR-3
Efae
5.394
Aba
Eco
2.697
Kpn
Ecl
Pae
2.697
Ckr
0.669
Cal
0.345
60
61
62
63
64
65
66
67
68
69
70
0.669
0.041
0.077
0.018
0.009
0.017
0.004
0.015
0.029
0.108
0.100
21.576
5.086
2.406
0.566
0.139
0.066
0.063
0.015
0.233
3.360
1.562
21.576
10.172
0.597
0.292
0.278
0.133
0.253
0.030
0.233
0.417
3.125
21.576
20.344
0.308
0.292
0.139
0.133
0.127
0.030
0.058
0.108
6.249
0.163
0.154
0.037
0.009
0.033
0.004
0.015
0.029
0.108
0.050
0.326
0.308
0.073
0.009
0.017
0.004
0.030
0.116
1.680
0.387
5.086
1.193
0.566
1.077
0.033
0.016
0.470
3.634
0.417
1.562
0.631
2.406
0.009
0.035
0.004
0.002
0.002
0.002
0.002
0.006
0.010
0.038
0.005
0.002
0.008
0.002
0.008
0.015
0.054
0.100
Sau MR-3, Staphylococcus aureus MR-3; Efae, Enterococcus feacalis ATCC19212; Aba, Acinetobacter baumani; Eco, Escherichia coli
ATCC 25922; Kpn, Klebsiella pneumaniae ATCC 700603; Ecl, Enterobacter cloacae; Pae, Pseudomonas aeruginosa ATCC 27853;
Ckr, Candida krusei ATCC 6258; Cal, Candida albicans ATCC 90028.
the one hand, and their antimicrobial properties on the
other hand. It is well suited for this application because it
handles qualitative and quantitative attributes, without the
need of discretization of quantitative attributes or transfor-
mation of qualitative attributes into quantitative ones.
DRSA is also able to deal with possible inconsistencies in
the information table prior to induction of rules. Moreover,
it handles global or local monotonic relationships between
values of condition attributes and quality of the decision
class. These characteristics of DRSA perfectly fit the type
of data to be analyzed. Using this method, we obtain deci-
sion rules with ranges of values of condition attributes in
particular classes of activity. The rules are useful for syn-
thesis of new chemical entities with good antimicrobial
characteristics.
The bacteriostatic activity of imidazolium chlorides
depends on the species of the strains and the structure of
compounds. The highest values of MIC were obtained for
Pseudomonas aeruginosa ATCC 27853, Enterobacter
cloacae, and Klebsiella pneumaniae ATCC 700603. The
lowest values of MIC were obtained for staphylococci,
including the methicillin-resistant one. Low values of MIC
were also obtained for Candida yeasts. Chlorides with
short R-alkyl substituent (methyl to pentyl, compounds:
1–5, 15–19, 29–33, 43–47, 57–61) possess the weakest
antibacterial action. The most active are those with R-alkyl
chain from octyl to dodecyl one (compounds: 8–12,
22–26, 36–40, 50–54, 64–68). The activity of chlorides
with tetradecyl and hexadecyl substituents is decreasing.
In Table 3 surface active properties, structure of analyzed
compounds (according to Table 1 and Figure 2) and their
antimicrobial activity against Staphylcoccus aureus ATCC
25213 are presented. This table forms an information table
in SAR analysis.
Information table is the basis of SAR analysis of the
chemical compounds. It includes a set of objects (in
rows) described by a set of attributes (in columns). The
set of attributes is composed of condition and decision
attributes. In our case, condition attributes describe sur-
face active properties and structure of analyzed chlo-
rides. The decision attribute concerns antimicrobial
properties of bis-quaternary imidazolium chlorides repre-
sented by 3 classes of values of MIC for Staphylococcus
aureus ATCC 25213 in the following way: class of good
antimicrobial properties: MIC ≤ 0.02 mM/L, class of med-
ium antimicrobial properties: 0.02 < MIC < 1 mM/L, class
of weak antimicrobial properties: MIC ≥ 1 mM/L. Numeri-
cal coding of the structure condition attributes is pre-
sented in Table 1. The synthesized chlorides are listed in
Table 3.
Decision rules
Decision rules represent the most important cause–effect
relationships between values of condition attributes and
the class assignment. A decision rule induced from an
information table is denoted as E ? H, which reads as ‘if
E, then H ‘. A rule consists of a condition part (called also
premise or evidence) E, and a conclusion (called also deci-
sion part or prediction, or hypothesis) H. Considering a
finite set of condition attributes C = {q1, q2,… qn}, the con-
dition part of the rule is a conjunction of elementary condi-
tions on a particular subset of attributes:
Results and Discussion
E
=
ei1 ^ ei2 ^…^eip, where {i1, i2,… ip} ꢂ {1, 2,…, n},
p ≤ n, and eih is an elementary condition defined on the
value set of attribute qih, {i1, i2,… ip} (e.g.,
The antimicrobial activity of synthesized gemini imidazolium-
based chlorides is presented in Table 2.
h
2
eih ≡ qih > 0.5 or eih ≡ qih = 1 or eih ≡ qih 2 [0.5, 1]).
282
Chem Biol Drug Des 2014; 83: 278–288

Antimicrobial Activity of New Gemini Chlorides
Table 3: SAR table (Information table)
Condition attributes
No
n
R
ꢁlgCMC
cCMC
Γꢃ106
Aꢃ1020
ΔGads
MIC (mM/L)
Class
1
2
3
4
5
6
7
8
9
C2H5
C2H5
C2H5
C2H5
C2H5
C2H5
C2H5
C2H5
C2H5
C2H5
C2H5
C2H5
C2H5
C2H5
C3H7
C3H7
C3H7
C3H7
C3H7
C3H7
C3H7
C3H7
C3H7
C3H7
C3H7
C3H7
C3H7
C3H7
C4H9
C4H9
C4H9
C4H9
C4H9
C4H9
C4H9
C4H9
C4H9
C4H9
C4H9
C4H9
C4H9
C4H9
C5H11
C5H11
C5H11
C5H11
C5H11
C5H11
C5H11
C5H11
C5H11
C5H11
C5H11
CH3
C2H5
C3H7
C4H9
2.15
2.23
2.38
2.41
2.49
2.58
2.65
2.72
2.81
2.92
3.04
3.15
3.34
3.52
2.18
2.26
2.32
2.44
2.52
2.61
2.68
2.75
2.84
2.95
3.07
3.18
3.37
3.55
2.22
2.30
2.39
2.45
2.53
2.64
2.71
2.79
2.88
3.00
3.13
3.22
3.40
3.58
2.25
2.33
2.41
2.49
2.58
2.67
2.75
2.83
2.91
3.07
3.16
61.9
60.1
59.8
57.4
55.5
53.4
51.2
48.9
47.5
45.3
42.5
41.4
37.5
33.9
60.8
58.9
57.8
56.3
54.4
52.3
50.1
47.9
46.4
44.2
42.4
40.3
36.4
32.8
59.6
57.7
56.6
55.1
53.2
51.1
48.9
46.7
45.2
43.0
41.2
39.2
35.2
31.5
58.5
56.6
55.7
53.9
52.1
50.1
47.8
45.6
44.0
41.9
40.1
2.75
2.71
2.69
2.65
2.61
2.57
2.53
2.49
2.45
2.41
2.37
2.33
2.25
2.17
2.74
2.70
2.66
2.62
2.58
2.54
2.59
2.46
2.42
2.38
2.34
2.30
2.22
2.14
2.71
2.67
2.63
2.59
2.55
2.51
2.47
2.43
2.39
2.35
2.31
2.27
2.19
2.11
2.68
2.64
2.60
2.56
2.52
2.48
2.44
2.40
2.36
2.32
2.28
52
54
56
58
60
62
64
66
68
70
72
74
78
82
54
56
58
60
62
64
66
68
70
72
74
76
80
84
56
58
60
62
64
66
68
70
72
74
76
78
82
86
58
60
62
64
66
68
70
72
74
76
78
20.2
20.8
21.3
21.7
22.3
22.7
23.5
23.9
24.3
24.8
25.6
26.3
27.5
28.8
20.5
21.1
21.6
22.0
22.6
23.0
23.8
24.2
24.6
25.1
25.9
22.6
27.7
29.1
20.9
21.4
21.9
22.3
22.9
23.3
23.9
24.5
24.9
25.4
26.2
26.9
28.0
29.3
21.2
21.7
22.2
22.7
23.2
23.7
24.2
24.8
25.3
25.9
26.5
33.875
28.468
13.181
6.590
5.741
1.338
0.163
0.038
0.018
0.017
0.017
0.032
0.058
0.108
29.652
27.375
25.423
5.933
5.562
1.298
0.158
0.009
0.069
0.008
0.016
0.008
0.056
0.106
28.468
6.591
13.180
0.712
0.345
0.079
0.010
0.009
0.009
0.008
0.032
0.061
0.028
0.052
27.375
25.423
5.933
0.178
2.618
0.158
0.019
0.142
0.008
0.004
0.015
weak
weak
weak
weak
weak
weak
medium
medium
good
good
good
medium
medium
medium
weak
weak
weak
weak
weak
C5H11
C6H13
C7H15
C8H17
C9H19
C10H21
C11H23
C12H25
C14H29
C16H33
CH3
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
C2H5
C3H7
C4H9
C5H11
C6H13
C7H15
C8H17
C9H19
C10H21
C11H23
C12H25
C14H29
C16H33
CH3
weak
medium
good
medium
good
good
good
medium
medium
weak
weak
weak
medium
medium
medium
good
good
good
C2H5
C3H7
C4H9
C5H11
C6H13
C7H15
C8H17
C9H19
C10H21
C11H23
C12H25
C14H29
C16H33
CH3
C2H5
C3H7
C4H9
C5H11
C6H13
C7H15
C8H17
C9H19
C10H21
C11H23
good
medium
medium
medium
medium
weak
weak
weak
medium
weak
medium
good
medium
good
good
good
Chem Biol Drug Des 2014; 83: 278–288
283

Pałkowski et al.
Table 3: continued
Condition attributes
No
n
R
ꢁlgCMC
cCMC
Γꢃ106
Aꢃ1020
ΔGads
MIC (mM/L)
Class
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
C5H11
C5H11
C5H11
C6H13
C6H13
C6H13
C6H13
C6H13
C6H13
C6H13
C6H13
C6H13
C6H13
C6H13
C6H13
C6H13
C6H13
C12H25
C14H29
C16H33
CH3
C2H5
C3H7
3.25
3.43
3.61
2.29
2.37
2.45
2.54
2.62
2.71
2.79
2.87
2.95
3.07
3.20
3.29
3.47
3.65
38.1
33.9
29.6
47.1
44.8
42.6
39.4
37.6
36.3
35.1
34.2
33.3
32.2
31.2
30.3
28.3
26.4
2.24
2.16
2.08
2.73
2.66
2.59
2.53
2.47
2.41
2.34
2.28
2.22
2.15
2.09
2.03
1.90
1.77
80
84
88
56
59
62
65
68
71
74
77
80
83
86
89
95
101
27.1
28.3
29.6
29.3
30.2
30.9
31.5
32.1
32.6
33.1
33.6
34.0
34.4
34.8
35.2
36.1
37.1
0.030
0.027
0.204
6.591
0.817
0.712
0.669
0.041
0.077
0.018
0.009
0.017
0.004
0.015
0.029
0.027
0.100
medium
medium
medium
weak
medium
medium
medium
medium
medium
good
good
good
good
good
medium
medium
medium
C4H9
C5H11
C6H13
C7H15
C8H17
C9H19
C10H21
C11H23
C12H25
C14H29
C16H33
The rules induced from the information table structured
using DRSA represent all relevant patterns of cause–effect
relationships, which are free of inessential and redundant
information. The rules that are considered in DRSA are
mainly certain rules. This kind of rules covers consistent
objects only. To represent the most important cause–effect
dependencies between values of condition attributes and
value of decision attribute, decision rules are not only
involving important condition attributes but they also
include a minimal number of elementary conditions indis-
pensable for representation of the dependencies. The set
of decision rules may be understood as a representation
of cause–effect relationships existing in the information
table, from which all inessential and redundant information
was removed.
from Table 3. In particular, these decision rules involve spe-
cific values of condition attributes describing structure and
surface active properties. The rules are characterized by
various parameters, such as examples (i.e., number of
objects covering a given rule), strength (i.e. the proportion of
objects covered by premise that are also covered by conclu-
sion), or confirmation (i.e. measure that is quantifying the
degree to which premise provides evidence for conclusion;
see Appendix S1). Strong decision rules are supported by
majority of objects from the class that they refer to, while rel-
evant decision rules are characterized by high values of
Bayesian confirmation measures. The rules presented in
Table 4 cover 60% of chlorides that belong to class ‘good’
and 95% of chlorides that belong to class ‘weak’. Thus, one
can conclude that chlorides with good antimicrobial proper-
ties are more diverse and they are harder to describe by
strong (and relevant) rules. The decision rules provide guide-
lines for synthesis of new compounds with better antimicro-
bial properties. In decision rules covering chlorides with good
antimicrobial activity, the attribute n-spacer length is absent,
while attributes describing the surface active properties of
analyzed compounds are often present. Moreover, the value
of R-chain in these rules ranges from 6 to 11 (the alkyl chain
longer than hexyl and shorter than undecyl one), ꢁlogCMC
values are in the range (2.71; 3.13), values of surface tension
at critical micelle concentration are below 46.4, values of sur-
face excess are below 2.42, and values of free energy of
adsorption of molecule are >24.6. In decision rules covering
chlorides with weak antimicrobial activity, value of n-chain is
below 6 (the spacer shorter than [1,10-(2,9-dioxadecane)]),
value of ꢁlogCMC is below 2.68, value of surface tension at
Table 4 includes those from among all decision rules that
cover compounds from Table 3 belonging to ‘good’ or
‘weak’ class, which are strong and relevant. We do not
induce rules for class ‘medium’ since these rules are not
interesting from the view point of SAR analysis (it is more
important to know what are the features of chlorides with
definitely good or weak antimicrobial properties). Note, how-
ever, that the presence of chlorides from the ‘medium’ class
is important in the rule induction process. The rules with
conclusion ‘good’ discriminate chlorides with ‘good’ antimi-
crobial properties from those chlorides which have ‘med-
ium’ or ‘weak’ properties (analogously for rules with
conclusion ‘weak’). The decision rules from Table 4 can be
viewed as the most interesting certain patterns among all
patterns that can be found to describe features of chlorides
284
Chem Biol Drug Des 2014; 83: 278–288

Antimicrobial Activity of New Gemini Chlorides
Table 4: Decision rules
Condition attributes
No
n
R
ꢁlog CMC
cCMC
Γ 9 106
A 9 1020
ΔGads
Examples
Strength (%)
Confirmation measure c1
Decision class good
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
(2.71, 3.13)
<3.13
<3.13
<3.13
<3.13
<45.6
<45.6
<45.6
<45.6
12
12
11
11
11
11
11
11
11
11
11
11
10
10
10
10
60
60
55
55
55
55
55
55
55
55
55
55
50
50
50
50
0.7586
0.7586
0.7330
0.7330
0.7330
0.7330
0.7330
0.7330
0.7330
0.7330
0.7330
0.7330
0.7083
0.7083
0.7083
0.7083
>6
>6
<2.41
<2.41
>71
>24.8
>24.8
<3.13
(2.84, 3.13)
(2.83, 3.13)
(2.83, 3.13)
<3.13
(2.83, 3.13)
(2.71, 3.13)
<3.13
<46.4
<2.42
<2.40
>24.6
>24.8
>8
>8
(6, 11)
<11
<2.42
<3.13
>24.6
<45.6
<45.6
>2.71
Decision class weak
17
18
19
>44.8
>55.1
>44.8
>44.8
>2.59
>2.59
16
16
15
15
15
15
15
15
14
14
14
13
13
12
12
12
12
12
12
12
12
12
12
12
12
12
12
80
80
75
75
75
75
75
75
70
70
70
65
65
60
60
60
60
60
60
60
60
60
60
60
60
60
60
0.8703
0.8703
0.8409
0.8409
0.8409
0.8409
0.8409
0.8409
0.8125
0.8125
0.8125
0.7850
0.7850
0.7586
0.7586
0.7586
0.7586
0.7586
0.7586
0.7586
0.7586
0.7586
0.7586
0.7586
0.7586
0.7586
0.7330
<30.2
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
<6
>2.59
<2.45
<2.45
<30.2
<30.2
<62
<62
<62
<6
<6
<2.45
<22.3
<30.2
<4
<4
<7
>44.8
<4
<4
<4
<4
<4
<4
<4
<4
<4
<4
<4
<4
<6
<2.65
<2.68
>51.2
>50.1
>2.53
>2.53
>2.53
>2.53
<66
<23.8
<23.5
<23.5
<23.5
<23.5
<23.5
<12
<4
<3.18
>40.3
>2.30
<76
<59
CMC is over 44.8, value of surface excess is over 2.53, value
of molecular area of single particle is below 66, and value of
free energy of adsorption of molecule is below 30.2.
this analysis, consists of only 70 chlorides, we used 3-fold
stratified cross-validation procedure to assess the predic-
tive accuracy of rules. The cross-validation was repeated
100 times to obtain better reproducibility of results. The
predictive accuracy of DRSA decision rules was over 90%
for both classes, that is sufficient to consider these rules
as useful guidelines for synthesis of compounds with bet-
ter antimicrobial properties. The rules are further analyzed
to assess relevance of the condition attributes.
As we mentioned earlier, the decision rules presented so
far can be seen as a representation of the most important
relationships existing in the data table. Another aspect
of applying DRSA to SAR analysis is the predictive power of
the induced rules. Since the SAR table, that is the basis of
Chem Biol Drug Des 2014; 83: 278–288
285

Pałkowski et al.
Attribute relevance
substituent with regard to two opposite homologous
spacer type series of [1,6-(2,5-dioxahexane)] (n = 2) and
[1,10-(2,9-dioxadecane)] one (n = 6). In the first one, the
weakest bacteriostatic action was observed for com-
pounds 1–6 (R substituent from methyl to hexyl) and 12–14
(R substituent from dodecyl to hexadecyl). Compounds
with the strongest action are 9–11 (R substituent from no-
nyl to undecyl). It is worth mentioning that the chloride 1
with methyl substituent has the highest MIC value. In the
homologous series of [1,10-(2,9-dioxadecane)] (n = 6), the
best bacteriostatic action was observed for compounds
63–67 (R from heptyl to undecyl). The activity of com-
pounds with dodecyl to hexadecyl substituents was similar
to the activity of compounds from [1,6-(2,5-dioxahexane)]
(n = 2) homologous series. According to literature data,
compounds with high antimicrobial activity possess med-
ium length alkyl substituents and a short spacer (22). The
optimal activity was determined for the spacer with 4 car-
bon atoms (23,24). When the spacer is longer than hexyl
one, the CMC values are relatively high and the antimicro-
bial activity is decreasing (25).
We consider attribute relevance measures that satisfy the
property of Bayesian confirmation. These measures take
into account interactions between attributes present in the
decision rules (18,19). In this case, the property of confir-
mation is related to quantification of the degree to which
the presence of an attribute in the premise of a rule pro-
vides evidence for or against the conclusion of the rule.
The measure increases when more rules involving an attri-
bute suggest a correct decision, or when more rules that
do not involve the attribute suggest an incorrect decision,
otherwise it decreases (20,21).
The attributes with the highest values of confirmation mea-
sure c1 are the most relevant from the view point of cor-
rect class assignment of compounds to the MIC decision
classes. We can observe, in Figure 4, that attributes: ꢁlog
CMC (critical micelle concentration) and surface excess
have the highest impact on the correct assignment of the
analyzed compounds to the class good. On the other
hand, the assignment of the compounds to the class weak
is influenced the most by attributes ꢁlog CMC and surface
tension at CMC.
Monotonic dependencies between chemical structure and
surface active properties of analyzed chlorides were also
observed (Table 3). In each homologous series, the longer
was R-alkyl substituent, the smaller were values of surface
tension at CMC and surface excess, and the greater were
values of molecular area of single particle, and values of
free energy of adsorption of molecule. Similarly, the longer
was the spacer, the greater were values of molecular area
Surface active properties and antimicrobial activity of
obtained bis-imidazolium compounds depend on the
length of spacer and on the type of alkyl substituent in
imidazole ring. Clear dependence between bacteriostatic
action and structure was observed: in active compounds
the longer the spacer (in certain range) the shorter R-alkyl
Figure 4: Predictive attribute
confirmation calculated for MIC
decision classes: good (top) and
weak (bottom).
286
Chem Biol Drug Des 2014; 83: 278–288

Antimicrobial Activity of New Gemini Chlorides
of single particle and free energy of adsorption. The values
of surface tension at CMC and surface excess were
decreasing with elongation of the spacer.
through 4-aminobenzenesulfonic acid-chlorotriazine
adduct. Eur Polym J;42:3059–3067.
3. Shirai A., Sumitomo T., Kurimoto M., Maseda H., Kou-
rai H. (2009) The mode of the antifungal activity of
gemini-pirydinium salts against yeast. Biocontrol
Sci;14:13–20.
It might be concluded that surface active properties
influence strongly the antimicrobial activity. Structural
attributes are less significant. According to the literature
data, physicochemical properties of quaternary com-
pounds determine adsorption to the cell membrane and
their penetration to the bacterial cell, and therefore its lysis
as an effect of biocidal action (26–28). From the other
side, physicochemical properties of compounds are
dependent on their structure. In this work, only two struc-
tural and five surface active attributes were analyzed and
this could determine results of SAR analysis. From this
point of view, it might be interesting to study the influence
of spacer longer than [1,10-(2,9-dioxadecane)] (n = 6) on
surface active properties and antimicrobial activity of 3,3′-
[a,x-dioxaalkyl]bis(1-alkylimidazolium) derivatives.
€
4. Kummerera K., Eitela A., Brauna U., Hubnera P., Das-
chnera F., Mascartb G., Milandric M., Reinthalerd F.,
Verhoef J. (1997) Analysis of benzalkonium chloride in
the effluent from European hospitals by solid- phase
extraction and high-performance liquid chromatography
with post-column ion-pairing and fluorescence detec-
tion. J Chromatogr A;774:281–286.
5. Thebault P., Taffin de Givenchy E., Levy R., Vanden-
berghe Y., Guittard F., Geribaldi S. (2009) Preparation
and antimicrobial behaviour of quaternary ammonium
thiol derivatives able to be grafted on metal surfaces.
Eur J Med Chem;44:717–724.
6. Kourai H., Yabuhara T., Shirai A., Maeda T., Naga-
mune H. (2006) Syntheses and antimicrobial activity of
a series of new bis-quaternary ammonium compounds.
Eur J Med Chem;41:437–444.
Conclusions
7. Demberelnyamba D., Kim K.-S., Choi S., Park S.-Y.,
Lee H., Kim C.-J., Yoo I.-D. (2004) Synthesis and anti-
microbial properties of imidazolium and pyrrolidinonium
salts. Bioorg Med Chem;12:853–857.
8. Shirai A., Maeda T., Nagamune H., Matsuki H.,
Kaneshina S., Kourai H. (2005) Biological and physico-
chemical properties of gemini quaternary ammonium
compounds in which the position of a cross-linking sul-
fur in the spacer differ. Eur J Med Chem;40:113–123.
9. Avinash B., Sukhprint S. (2011) Synthesis and proper-
ties of thioether spacer containing gemini imidazolium
surfactants. Langmuir;27:14033–14044.
Antimicrobial activity of gemini imidazolium-based chlorides
with dioxaalkyl spacer depends mainly on the length of
alkyl substituents in the aromatic ring. Moreover, elonga-
tion of dioxaalkyl spacer might lead to active gemini imi-
dazolium salts with shorter substituent in imidazole ring
(heptyl, octyl). Obtained decision rules with their confirma-
tion measures indicate directions of synthesis of new
active bis-imidazolium compounds. Dominance-based
rough set approach (DRSA) is a valuable tool in SAR
analysis.
10. Sadula S., Paidimarla S., Rachapudi B., Sanjit K. (2011)
Synthesis and evaluation of new imidazolium-based
aromatic ether functionalized cationic mono and gemini
surfactants. Eur J Lipid Sci Technol;113:756–762.
11. Butorac R., Al-Deyab S., Cowley A. (2011) Syntheses, struc-
tures and antimicrobial activities of bis(imino)acenaphthene
(bian) imidazolium salts. Molecules;16:3168–3178.
12. Pang Z., Al-Mahrouki A., Berezovski M., Krylov S.
(2006) Selection of surfactants for cell lysis in chemical
cytometry to study protein-dna interactions. Electro-
phoresis;27:1489–1494.
Acknowledgment
The second and the tenth author wish to acknowledge
financial support from the Polish National Science Centre,
grant no. DEC-2011/01/B/ST6/07318. The third author
acknowledge financial support from the Polish National
Science Centre, grant no. NN 209437639.
Conflict of Interest
ꢀ
13. Greco S., Matarazzo B., Słowinski R. (2001) Rough
The authors have no conflict of interest to declare.
sets theory for multicriteria decision analysis. Eur J
Oper Res;129:1–47.
14. Emel’Yanenko V., Portnova S., Verevkin S., Skrzypczak
A., Schubert T. (2011) Building blocks for ionic
liquids: vapor pressures and vaporization enthalpies of
1-(n-alkyl)-imidazoles. J Chem Thermodyn;43:1500–1505.
15. Pernak J., Skrzypczak A., Lota G., Frackowiak E.
(2007) Synthesis and properties of trigeminal tricationic
ionic liquids. Chem Eur J;13:3106–3112.
References
1. Negma N.A., Zakia M.F., Salem M. (2010) Cationic
schiff base amphiphiles and their metal complexes:
surface and biocidal activities against bacteria and
fungi. Colloids Surf B Biointerfaces;77:96–103.
2. Young-A S., Byung-Soon K., Ravikumar K., Seung-
Goo L. (2006) Impacting durable antimicrobial proper-
ties to cotton fabrics using quaternary ammonium salts
16. Zabielska-Matejuk J., Urbanik E., Pernak J. (2004) New
bis-quaternary ammonium and bis-imidazolium chlo-
rides wood preservatives. Holzforschung;58:292–299.
Chem Biol Drug Des 2014; 83: 278–288
287

Pałkowski et al.
17. CLSI (2006) Performance Standards for Antimicrobial
Susceptibility Testing, Sixteenth informational supple-
ment. Pennsylvania: Clinical and Laboratory Standards
Institute, CLSI M100-S16 (M2-A9 and M7-A7).
surfactants: influence of the spacer length. Thermo-
chim Acta;379:255–260.
26. Zhang C., Su F.-Y., Zhang J.-F., Yan S.-T., Xing X.-H.
(2012) Luciferase and fluorescent protein as dual
reporters analyzing the effect of n-dodecyltrimethylam-
monium bromide on the physiology of Pseudomonas
putida. Appl Microbiol Biotechnol;93:393–400.
ꢀ
ꢀ
18. Błaszczynski J., Słowinski R., Susmaga R. (2011)
Rule-based estimation of attribute relevance. Lect
Notes Comput Sci;6954:36–44.
ꢀ
ꢀ
ꢀ
19. Pałkowski Ł., Błaszczynski J., Krysinski J., Słowinski
R., Skrzypczak A., Błaszczak J., Gospodarek E.,
27. Palermo E.F., Kuroda K. (2009) Chemical structure of
cationic groups in amphiphilic polymethacrylates mod-
ulates the antimicrobial and hemolytic activities. Bio-
macromolecules;10:1416–1428.
28. Huang L., Xiao Y.-H., Xing X.-D., Li F., Ma S., Qi L.-L.,
Chen J.-H. (2011) Antibacterial activity and cytotoxicity
of two novel cross-linking antibacterial monomers on
oral pathogens. Arch Oral Biol;56:367–373.
ꢀ
Wroblewska J. (2012) Application of rough set theory
to prediction of antimicrobial activity of bis-quaternary
ammonium chlorides. Lect Notes Comput Sci;7414:
108–116.
ꢀ
20. Greco S., Pawlak Z., Słowinski R. (2004) Can Bayesian
confirmation measures be useful for rough set decision
rules? Eng Appl Artif Intel;17:345–361.
ꢀ
21. Greco S., Słowinski R., Szczech I. (2012) Properties of
z
rule interestingness measures and alternative approaches
to normalization of measures. Inf Sci;216:1–16.
Note
ahttp://www.cs.put.poznan.pl/jblaszczynski/Site/jRS.html
22. Kuperkar K., Modi J., Patel K. (2012) Surface-active
properties and antimicrobial study of conventional cat-
ionic and synthesized symmetrical gemini surfactants.
J Surfactants Deterg;15:107–115.
Supporting Information
23. Li H., Yu C., Chen R., Li J. (2012) Novel ionic liquid-
type Gemini surfactants: synthesis, surface property
and antimicrobial activity. Colloid Surf A;395:116–124.
24. Bunio P., Chlebicki J. (2012) New sorbic-type quater-
nary ammonium single-chain and gemini polymerizable
surfactants: synthesis, interfacial properties and anti-
electrostatic activity. Colloid Surf A;413:119–124.
25. Grosmaire L., Chorro M., Chorro C., Partyka S., Lager-
ge S. (2001) Micellar structure of gemini cationic
Additional Supporting Information may be found in the
online version of this article:
Appendix S1. DRSA, confirmation measure, and chemis-
try details.
Table S1. Bis-imidazolium quaternary chlorides – elemental
analysis.
288
Chem Biol Drug Des 2014; 83: 278–288