quantitative structure–activity and structure–toxicity relationships of 4-aminodiphenyl sulphone...
TRANSCRIPT
Quantitative Structure±Activity and Structure±ToxicityRelationships of 4-Aminodiphenyl Sulphone Derivatives withAntiin¯ammatory Activity
Joachim K. Seydel1*, Hanna BuÈrger1, Anil K. Saxena1**, Michael D. Coleman2, Stephen N. Smith2 and Alan D. Perris2
1Borstel Research Center, Center for Medicine and Biosciences, D-23845 Borstel, Germany2Department of Pharmaceutical Sciences, Aston University, Aston Triangle, Birmingham B4 7ET, UK
Dedicated to Prof. Corwin Hansch on the occasion of his 80th birthday
Abstract
It has been shown that dapsone and its derivatives are not
only potential inhibitors of bacterial and plasmodial folate
synthesis but also possess antiin¯ammatory activity. The
application of dapsone is, however, limited by its toxic side
effects at higher dose manifested especially by its potential
to produce methaemoglobin. We have analyzed the
structural dependence of this property on a set of 29
derivatives. A highly signi®cant nonlinear dependence on
the lipophilicity could be derived. Two exceptions to the
general relation were found and the mechanism for these
could be derived. It was also shown that the bilinear
dependence is not due to lipophilicity mediated diffusion
into erythrocytes. It was found that the diffusion is linearly
dependent on lipophilicity. Principal Component (PC)-
Analysis revealed that the antiin¯ammatory activities,
determined as inhibition of zymosane stimulated cell burst
and cell adhesion are orthogonal to methaemoglobin
formation and cell death potential of the studied derivatives.
The antiin¯ammatory activity seems to depend on elec-
tronic and steric effects of the 2 0-substituents. The ortho-
gonality of substituent in¯uences on the toxic and wanted
effects opens up the possibility to further optimize the
selectivity of aminodiphenylsulphones.
Quant. Struct.-Act. Relat., 18 (1999) # WILEY-VCH Verlag GmbH, D-69469 Weinheim 0931-8771/99/0510-0043 $17.50+.50/0 43
Quantitative Structure±Activity and Structure±Toxicity Relationships QSAR
1 Introduction
4-Aminodiphenyl sulphones and especially 4,4 0-Diamino-
diphenylsulphone (Dapsone) are known and used in the
therapy of bacterial infections, malaria, pneumocystis
carinii and toxoplasmosis. They act as inhibitors of
dihydropteroic acid synthase in the de novo synthesis of
folate [1±3] and combined with dihydrofolate reductase
inhibitors such as trimethoprim (TMP) lead to synergistic
effects in the inhibition of parasitic growth [4]. Dapsone is
the drug of choice for the treatment of dermatitis
herpetiformis and related in¯ammatory conditions which
are characterized by tissue neutrophil in®ltration [5, 6].
Dapsone inhibits neutrophil adhesion, chemotaxis, lipoxy-
genase activity and the ability of cells to generate oxidative
species [7±11].
Unfortunately the applications of dapsone are dose-limited
due to its hepatic phase I activation to haemotologically
toxic hydroxylamines [12, 13].
Clinically, dermatitis herpetiformis patient responses vary
and dosages can range from 50±300 mg dayÿ1 [14]. At the
higher dose, hydroxylamine metabolites of dapsone can
cause signi®cant methaemoglobinaemia and haemolysis
[14, 15]. Short-term efforts to reduce dapsone toxicity and
improve patients' tolerance involving co-administration of
cimetidine have been moderately successful [12, 14].
* To receive all correspondence
** A. v. Humboldt-Fellow from the Central Drug Research Institute,
Lucknow, India
Key words: 2 04 0-substituted 4-aminodiphenyl sulphones, anti-
in¯ammatory activity, cell burst, cell adhesion, methaemoglobin
formation, diffusion into erythrocytes, PC-and multiple regression
analysis.
Abbreviations: TMP, trimethoprim; DDS, dapsone; RBC, red
blood cell; PBS, phosphate buffer salin; QDDS, effect of drug at
10 mM drug/effect of DDS at 10 mM; Derythr., partitioning of drug
into erythrocytes; cellbr, inhibition of neutrophil zymosan-
mediated respiratory burst; celladh, inhibition of neutrophil
interleukin-1 stimulated cell adhesion.
However, a number of potentially less toxic sulphones
which display increased antibacterial activity have been
synthesized in our laboratories [3, 16].
QSAR-studies to explain the observed variation in anti-
bacterial activity have led to the conclusion that electronic
and steric substituent effects are decisive for this activity
and that sulphones and sulfonamides bind to the same
receptor site [2, 3, 17]. Preliminary animal and in vitro
studies [16, 18] have indicated that a number of these
analogues are far less toxic than dapsone. In consequence a
series of these dapsone analogues have been toxicologically
tested using human and rat tissues and their inhibitory
effects on neutrophil function determined in vitro [19, 20].
The aim of this study was to derive quantitative structure±
toxicity and structure±activity relationships which could
allow further optimization of the antiin¯ammatory activity
and selectivity of sulphones.
2 Material and Methods
Chemicals
The synthesis of the studied dapsone analogues has been
described in detail [16]. Dapsone was obtained from
Aldrich (Poole, UK). The compounds structures are listed
in Table 1.
Experimental
Methaemoglobin Formation
a) 1-Compartment System
The ®nal reaction mixture contained: 1 ml erythrocyte
suspension, 0.25 ml NADPH; 0.2 ml 0.3 M nicotinamide;
0.3 ml 0.1 M MgCl2; 0.1 ml 5% glucose; 2.05 ml phosphate
buffer (pH 7.4); 0.1 ml ethanolyH2O solution of the
sulphone derivative; 1 ml of rat liver microsome prepara-
tion. The reaction mixture was incubated for 2 hrs at 37�C,
then placed in ice water and methaemoglobin readings were
taken according to Havemann and BuÈrger [18, 21]. Read-
ings were taken with an Eppendorf photometer at 650 nm.
As a measure of methaemoglobin formation the percent
methaemoglobin produced by 10 mM of the drug is de®ned
[18]. In a later paper in another laboratory the following
2-compartment system has been used [19].
b) 2-Compartment System
Rat (male Sprague-Dawley) microsomal preparations were
prepared according to the method of [22]. The assay
involved a two-compartment in vitro system which consists
of two Te¯on compartments separated by a cellulose
(molecular weight cut-off 5000 Da) membrane [23]. Com-
partment A contained 2 mg rat liver microsomes, 1 mM
NADPH and 5 ml DMSO solution of a test analogue.(®nal
concentration 100 mM). This concentration was applied in
previous analogue studies [24]. Vehicle controls contained
1% DMSO and buffer and microsomal controls contained
all the above except for NADPH. Compartment B contained
500 mL washed (50% haematocrit) human erythrocytes.
Incubations were carried out in triplicate. In all cases, no
methaemoglobin was generated in NADPH-free incubations
prior to determination of methaemoglobin formation using
an IL-482 CO-oximeter.
Partitioning into Erythrocytes
The partition of the compounds in the system erythrocyte
(RBC's)ybuffer was determined in RBC suspensions in pH
7.4 phosphate buffer at room temperature (21�C) essentially
according to our previously reported method [25] except
that the concentration of free drug before and after
equilibration was determined in the buffer phase by HPLC
instead of using the Bratton and Marshall photometric
method. The partition coef®cient D is de®ned:
D � ��AT�y�AB�� � �1yH� ÿ 1yH� 1
H� hematocrit; VB true� true volume of buffer; VT true �true volume of whole blood; VRBC true� true volume of red
Table 1. List of compounds used in derivation of structure±toxicity and structure±activity relationships of 2 0,4 0 substituted 4-aminodiphenylsulphones
Nr Substitution Nr. Substitution
1 (1±7) 4 0-amino (dapsone) 24 (2±4,6) 2 0-OCH3; 4 0-NHC2H5
2 (1,4,5) 4 0-H 25 (4) 2 0-OCH3; 4 0-NHC3H7
3 (1,4,5) 4 0-NHCH3 26 (2±4,7) 3 0-OCH3; 4 0-NHC2H5
4 (4) 4 0-NHC2H5 27 (4) 3 0-OCH3; 4 0-NHC3H7
5 (2,3) 4 0-N(CH3)2 28 (4) 2 0-CH3; 4 0-NH2
6 (1±3) 4 0-NHCH2CH2OH 29 (2,3,7) 2 0-CH3; 4 0-NHCH3
7 (1±3) 4 0-NHCH2COOH 30 (2±4,7) 2 0-CH3; 4 0-NHC2H5
8 (1,4,5) 4 0-NHCH2CONHNH2 31 (3, 4) 2 0-CH3; 4 0-NHC3H7
9 (1,4,5) 4 0-OH 32 (1±3,6) 2 0-CH3, 4 0-NHC5H9
10 (1,4,5) 4 0-OCH3 33 (2,3,6) 2 0-CH3; 4 0-NHC6H13
11 (1,4,5) 4-CH3 34 (2,3,7) 2 0-CH3; 4 0-N(CH3)2
12 (1,5) 4 0-COOH 35 (1±3,6) 2 0-CH2OH; 4 0-NHC2H5
13 4 0-COOCH3 36 (1±3,6) 2 0-CF3; 4 0-NH2
13a 4 0-COOC3H7
14 (1±5) 4 0-CONHNH2 37 (1±3,6,7) 2 0-CF3; 4 0-NHC2H5
15 (1) 4 0-NO2 38 (2,3,7) 2 0-COCH3;4 0-NH2
16 (1) 4 0-CL 39 (2,3,7) 2 0-CONH2; 4 0-NH2
17 (1,4,5) 4 0-Br 40 (2,3,7) 2 0-COOCH3; 4 0-NH2
18 (2,5,7) 2 0-NH2 4 0-H 41 (2,3,7) 2 0-CL; 4 0-NH2
19 (6) 2 0-NH2 ; 4 0-NH2 42 (2,3,7) 2 0-Br; 4 0-NH2
20 (2,3,7) 2 0-NHCH3; 4 0-NHC3H7 43 (2,3) 2 0-NO2; 4 0-NH2
21 (1±3,6) 2 0-OH; 4 0-NH2 44 (2,3,7) 2 0-CN; 4 0-NHC2H5
22 (2±4,7) 2 0-OH; 4 0-NHC2H5 45 4,4 0-CH3NH-
23 (2±4,6) 2 0-OH; 4 0-NHC3H7
The numbers in parentheses indicate the regression Eqs. where this
derivative is included.
QSAR Joachim K. Seydel, Hanna BuÈrger, Anil K. Saxena, Michael D. Coleman, Stephen N. Smith and Alan D. Perris
44 Quant. Struct.-Act. Relat., 18 (1999)
blood cells. H or VB trueyVT true� 1-H; AT� total amount in
the synthetic drug; AB� amount of drug in buffer.
Inhibition of Zymosan-mediated Human Neutrophil
Respiratory Burst
Whole blood was obtained by venepuncture from healthy
volunteers. The ability of neutrophils to generate a
respiratory burst was measured using whole blood, whereas
adhesion to endothelial cells was measured using puri®ed
neutrophil preparations. Neutrophil respiratory burst was
elicited by means of an opsonized zymosan preparation.
The respiratory burst was assessed by means of lucigenin-
enhanced chemiluminescence in a Bio-Orbit 1523 lumin-
ometer (Labtech International, Sussex, UK). Chemilumi-
nescence readings were taken in intervals. Before analysis
samples of puri®ed neutrophiles or whole blood were
incubated with the compounds (0.5, 1.0, or 1.5 mM) for
30 min at 37�C. For details see [20].
Inhibition of Interleukin-1-stimulated neutrophil adhesion
Isolation of neutrophils before measurement of adhesion
were performed according to Afford [26] with some
modi®cations. Neutrophil adherence studies were per-
formed using transformed human umbilical vein cells
(ECV 304) grown to con¯uence in 96-well plates. The
cells were then incubated in a 95 : 5% air-CO2 mixture at
37�C for 4 h during which time up-regulation of adhesion
molecules occurs on the endothelial cell surface. The
medium was then removed and puri®ed isolated neutrophils
suspension pretreated with dapsone or analogues; (100 mL)
were added. The neutrophils were left for 30 min to adhere,
then unadhered neutrophils were removed by washing with
PBS. The adherent neutrophils were then assessed by means
of the method of Junger [27]. The neutrophils were lyzed
with Triton X-100 in PBS and the liberated myeloperox-
idase measured using hydrogen peroxide (0.02%) as
substrate and O-dianisidine dihydrochloride (0.34 mM) as
chromogen at pH 5. Absorbance was measured at 405 nM.
The concentration range 0.5±1.5 mM was chosen on the
basis of in vitro pilot studies which showed complete
inhibition of both neutrophil responses by dapsone at 1 mM.
The twelve candidate analogues were compared with the
parent dapsone for their ability to inhibit adhesion and
respiratory burst completely in this concentration range.
All statistical analysis was performed by Student's t-test.
Where more than one comparison was made with data, the
Bonferroni correction was employed where the acceptable
level of signi®cance was reduced to 0.05yk (where k is the
number of tests) to compensate for the increased testing
likelihood of reaching P< 0.05 during multiple testing [28].
Physicochemical Parameters
HPLC-capacitor factors, in form of log k 0, determined on a
octanol coated column have been used as descriptors of
lipophilic substituent effects [3]. They are signi®cantly
correlated with p-values (r2 � 0:91; n � 25; F � 227;
Q2 � 0:89).
The van der Waals volume, VW, the parameters for molar
polarizability, MR, and s-Hammett values and Swain-
Lupton parameters used in this paper for substituents in
o- and p-position have been taken from compilations
[29, 30].
Correlation Analysis
Linear and nonlinear regression analysis as well as Principal
Component-Analysis were carried out using programmes
locally written.
3 Results and Discussion
Methaemoglobin formation in microsomal preparations of
rat liver
It is highly likely that the structural analogues of dapsone
under study in this report are also oxidized as the parent
drug is, to hydroxylamines, which in turn may be cytotoxic
as well as generators of methaemoglobin [12, 24]
Methaemoglobin formation is the most common side-effect
which limits dapsone's effectiveness in patients [15], and is
thus valuable in the study of sulphone analogue toxicity
estimations [23]. In general, sulphone-derived hydroxyl-
amines react with oxyhaemoglobin, which is thought to
result in the formation of both methaemoglobin as well as of
short-lived nitroso derivative [31].
From previous studies in vitro using a simple 1-compart-
ment test system and almost solely 4 0-substituted 4-
aminodiphenylsulphones, a bilinear correlation has been
observed between the logarithm of the % methaemoglobin
formation at a constant drug concentration of 10 mM divided
by the amount produced by dapsone (QDDS) and log k 0 ([18]
Eq. 1). The methaemoglobin production increases with
increasing lipophilicity up to an optimum log k 0 of about 1.6
and then levels of or decreases.
log QDDS�0:74��0:10� log k 0ÿ0:94��0:31� log�0:08 k 0�1�ÿ 0:48��0:13� �1�
n � 18 r2 � 0:92 s � 0:23 log k 0max � 1:63 F � 27:8
Three derivatives did not follow this general relationship
and were excluded from the regression (Eq. 1). These are
the two carbonic esters (Table 1, #13, 13a) and the 4,4 0
Quantitative Structure±Activity and Structure±Toxicity Relationships QSAR
Quant. Struct.-Act. Relat., 18 (1999) 45
methylamino derivative (Table 1, # 45), the latter possesses
no primary amino group. It was found that the two carbonic
acid esters are rapidly hydrolyzed by the microsomal liver
preparation to the corresponding acids. After 15 minutes of
incubation no ester can be detected. In the methaemoglobin
assay the readings for methaemoglobin formation are
however, taken after 2 hrs. It is therefore not surprising
that a very low amount of methaemoglobin is produced by
these derivatives corresponding to the low amount formed
by the hydrophilic acid (#12). In experiments using
haemolyzed erythrocytes it was checked if permeation
problems are responsible for the low amount of methae-
moglobin formed by the carbonic acids. It was found that
there was no difference in methaemoglobin formation using
haemolyzed or intact erythrocytes so that permeation
problems are not responsible but the lipophilicity was
found to be low in accordance with the general correlation.
Detailed studies were also performed to ®nd the reason for
the low production of methaemoglobin by the 4,4 0-methylamino derivative. It was known that N-alkylated
derivatives can lead to methaemoglobin formation or can be
N-oxidized, respectively, [32, 33]. Also in our experiments
methaemoglobin formation did occur. At higher substrate
concentration, however, a reduction in the formation of
methaemoglobin was observed. (Figure 1). This could have
been due to the low solubility of the compound. Another
explanation could be substrate or product inhibition. To
evaluate this possibility three series of concentrations with
three different microsome dilutions, i.e. with different
substrateyreceptor ratios were incubated simmultaneously.
In case of substrate- or product autoinhibition a shift of the
maximum of methaemoglobin formation to lower substrate
concentration at reduced enzymeysubstrate ratio should be
observable. This was indeed seen (maximum at approxi-
mately 36, 20, 10, respectively, Figure 2). In a further
experiment we wanted to evaluate if the inhibition was
restricted to the metabolism of the 4,4 0-methylamino
derivative or if the presence of this derivative also inhibitedthe metabolism of other derivatives. Figure 3 shows the
dose-response curves for the metabolism (methaemoglobin
formation for DDS and 4,4 0-methylamino diphenylsulphone
alone and for the mixture of both. At higher dapsone
concentrations the methaemoglobin formation is inhibited
by the presence of the 4,4 0-methylamino derivative. This
could indicate that the metabolism of the primary amino
group of DDS and the metabolism of the secondary amino
group of the 4,4 0-methylamino derivative are catalyzed by
the same enzyme.
The two examples underline the power of QSAR to not only
detect structure activity relationships but also to ®nd
signi®cant exceptions to these relationships and to shed
light on the mechanism of action of the exceptions.
We have extended these studies on methaemoglobin
formation to a larger group with additional substitution in
Figure 1. Methaemoglobin production by DDS and 4,4 0-methyl-amino-diphenylsulphone at increasing drug concentrations (stan-dardized microsome test, 1-compartment system (BuÈrger 1990)).
Figure 2. Methaemoglobin production by 4,4 0-methylamino-diphenylsulphone at different microsome dilutions. Drug concen-trations as indicated on the ordinate. The microsome dilutions 1 : 2;1 : 6 and 1 : 12 correspond to a ®nal dilution of the microsomes inthe reaction mixture (rat liver weight) of 1 : 60; 1 : 30 and 1 : 10,respectively.
Figure 3. Concentration dependent methaemoglobin productionby DDS and 4,4 0-methylamino-diphenylsulphone alone and incombination. Ratio in the mixture of DDS plus 4,4 0-amino-diphenylsulphone is 1 : 1.
QSAR Joachim K. Seydel, Hanna BuÈrger, Anil K. Saxena, Michael D. Coleman, Stephen N. Smith and Alan D. Perris
46 Quant. Struct.-Act. Relat., 18 (1999)
2-position. These derivatives are of interest with respect to
their toxic properties because of their increased antibacterial
and antiin¯ammatory activities [3, 19, 20]. The two
compartment system as described in the experimental
section was used to determine the % methaemoglobin
produced by 100 mM of the test substances.
A highly signi®cant correlation could be derived showing
again a bilinear dependence of the methaemoglobin
formation on log k 0 (Figure 4). The predictive power was
tested by two derivatives not included in the test set (Eq. 2),
derivatives #8 and # 31 (Table 1). The calculated and found
formation of methaemoglobin were 22.2% calc. 22.9% obs.
and 67.9% calc., 64.6% obs., respectively. The physico-
chemical descriptor used to derive quantitative structure-
toxicity relationships and the observed and calculated % of
methaemoglobin formation using the most signi®cant
regression equation derived are given in Table 2 (DDS
effect� 100%).
log %methaemoglobin � 0:528��0:094� log k 0
ÿ 0:608��0:088� log�0:391��0:29�k 0 � 1�� 1:765��0:117�
n � 27 r2 � 0:85 s � 0:12 F � 31:6 �2�
including #8 and #31:
log %methaemoglobin �0:528��0:089� log k 0 ÿ 0:608��0:084�
log�0:387��0:26�k 0 � 1� � 1:763��0:105�
n � 29 r2 � 0:86 s � 0:11 F � 37:4 �3�
The results are in excellent agreement with the previously
derived structure toxicity relationships of a different set of
sulphones using a simpler test system and lower drug
concentrations (Eq. 1) [18].
The k 0-values have been determined at the pH of the test
system so that the in¯uence of ionizable substituents like
22OH and 22COOH is considered.
In contrast, studies performed on some derivatives in vivo in
cats demonstrated that more lipophilic derivatives showed
less formation of methaemoglobin (Table 3). The same was
observed when serum albumin was added to the in vitro
experiments, indicating a competitive protecting function of
the albumin (Figure 5). The more lipophilic compounds are
more tightly bound to serum albumin.
To produce methaemoglobin in the erythrocytes, the
sulphones must partition into the erythrocytes; this is also
a necessary precondition for their antimalarial activity. To
evaluate if the partitioning might be the rate limiting step
we have studied the partitioning of a series of 20 sulphones
into erythrocytes at pH 7.4. The results are listed in Table 2
(log Derythrocytes). Regression analysis leads to a highly
signi®cant linear correlation with log k 0 (Eq.4), despite the
fact that the log k 0 values are exceeding the value of 1.6
found as optimum value for methaemoglobin production for
a similar data set. It supports the argument that competition
with protein binding is reponsible for the nonlinear relation
found for methaemoglobin formation and the observed
lower toxicity of the highly lipophilic derivatives in
experimental animal studies (Table 3) and in in vitro
studies where serum albumin has been added to the reaction
mixture (Fig. 5).
One referee suggested to ®t the data to a nonlinear function,
log Derythr.� log(A6k 0B� 1)�C which describes better
the highly hydrophilic derivative 12, but then derivatives 8,
26, 27 are deviating more from the general regression than
12 before in the linear ®t (Eq. 4, deviation was less than
2 sd).
Figure 4. Bilinear dependence of methaemoglobin production byvarious substituted 4-amino-diphenylsulphones (Table 1, 2) as afunction of lipophilicity (log k 0).
Figure 5. Methaemoglobin production by DDS and 4-aminoÿ4 0-ethylamino-diphenylsulphone in the absence -d- and in thepresence -s- of serum albumin (4gy100 mL). Time of incubation2.5 h, dilution of microsome preparation 1 : 50 (BuÈrger 1991).
Quantitative Structure±Activity and Structure±Toxicity Relationships QSAR
Quant. Struct.-Act. Relat., 18 (1999) 47
It can also be shown that the methaemoglobin formation for
10 derivatives studied in both systems correlates with the
partitioning into erythrocytes (Eq. 5, Table 2).
log Derythr: � 0:164��0:015� log k 0 � 0:511��0:024� �4�n � 20 r2 � 0:87 s � 0:07 F � 126 Q2 � 0:78
log QDDS � 10:17��3:65��log D�2erythrocytes
ÿ 6:34��3:19� log Derythrocytes
ÿ 3:94��0:96� �5�n � 10 r2 � 0:95 s � 0:20 F � 66 Q2 � 0:90
PC-analysis
The antiin¯ammatory activities, expressed as inhibition of
the neutrophil zymosan-mediated respiratory burst (cellbr),
the inhibition of neutrophil interleukin-1 stimulated cell
adhesion (celladh) and the methaemoglobin formation of
sulphones have been studied on a smaller set of 11 deri-
vatives [20]. First a PC-analysis has been performed to
analyze for possible multiple intercorrelations amongst the
three test systems.The result shows that with 2 PC's 89% of
the total information can be extracted. The ®rst PC after
varimax rotation is loaded by the cellbr- and celladh-data, the
second with the methaemoglobin formation ability (Table 4).
The total orthogonality of the methaemoglobin formation
ability, the cell toxicity and the properties of the compounds
to exert antiin¯ammatory activities suggests great promise
for the process of further optimization of the antiin¯amma-
tory effect of the sulphones.
More than 12 physicochemical parameters describing
electronic steric and lipophilic substituent effects in o-
and p-position have thus been analyzed for their ability to
explain the observed variance in the biological activity data.
Ortho and related proximate substituent effects can be
separated according to Charton [34] into inductive,
resonance and steric contributions.
log k � asI � bsR �CrV � h
Table 2. Methaemoglobin formation and partitioning into ery-throcytes by 4-aminodiphenylsulphones Structure±toxicity rela-tionships
log QDDS log % Methaemgl. log DErythrocytes
1-compartment 2-compartment
obs. calc. obs. calc. obs. calc.
Nr log k 0 eq. 1 eq. 2 eq. 4
1 0.339 0.00 ÿ0.29 2.00 1.78 0.59 0.57
2 1.375 0.21 0.09 Ð Ð 0.67 0.74
3 1.20 0.02 0.06 Ð Ð 0.66 0.71
4 1.646 Ð Ð Ð Ð 0.71 0.78
5 1.770 Ð Ð 1.74 1.86 Ð Ð
6 0.155 ÿ0.14 ÿ0.41 1.83 1.83 Ð Ð
7 ÿ1.716 ÿ2.00 ÿ1.75 1.04 0.86 Ð Ð
8 ÿ0.705 ÿ1.30 ÿ1.01 Ð Ð 0.29 0.39
9 0.965 0.08 0.00 Ð Ð 0.72 0.67
10 1.506 0.06 0.10 Ð Ð 0.68 0.76
11 1.890 ÿ0.04 0.09 Ð Ð 0.81 0.82
12 ÿ2.471 ÿ2.00 ÿ2.31 Ð Ð 0.23 0.11
14 ÿ0.197 ÿ1.09 ÿ0.64 1.49 1.60 0.42 0.48
15 1.589 ÿ0.03 0.10 Ð Ð Ð Ð
16 2.27 0.08 0.05 Ð Ð Ð Ð
17 2.38 0.07 0.03 Ð Ð 0.94 0.90
18 1.08 Ð Ð 1.97 1.88 Ð Ð
20 1.86 Ð Ð 1.75 1.86 Ð Ð
21 ÿ0.009 ÿ0.37 ÿ0.51 1.56 1.68 Ð Ð
22 1.258 Ð Ð 1.82 1.88 0.75 0.72
23 1.932 Ð Ð 1.72 1.85 0.82 0.81
24 1.277 Ð Ð 2.01 1.88 0.69 0.72
25 1.850 Ð Ð Ð Ð 0.86 0.82
26 2.03 Ð Ð 1.94 1.85 0.71 0.84
27 2.604 Ð Ð Ð Ð 1.09 0.94
28 0.833 Ð Ð Ð Ð 0.67 0.65
29 1.659 Ð Ð 1.96 1.87 Ð Ð
30 2.15 Ð Ð 1.95 1.84 0.88 0.86
31 2.22 Ð Ð Ð Ð 0.95 0.88
32 3.75 ÿ0.03 ÿ0.06 1.74 1.71 Ð Ð
33 4.25 Ð Ð 1.72 1.67 Ð Ð
34 3.03 Ð Ð 1.77 1.77 Ð Ð
35 0.89 Ð Ð 1.96 1.87 Ð Ð
36 1.41 0.13 0.09 1.91 1.88 Ð Ð
37 2.45 0.11 0.02 1.79 1.82 Ð Ð
38 0.192 Ð Ð 1.76 1.74 Ð Ð
39 ÿ1.534 Ð Ð 0.78 0.95 Ð Ð
40 1.625 Ð Ð 1.75 1.87 Ð Ð
41 1.058 Ð Ð 1.90 1.88 Ð Ð
42 1.32 Ð Ð 1.78 1.88 Ð Ð
43 1.31 Ð Ð 1.92 1.88 Ð Ð
44 1.86 Ð Ð 1.70 1.85 Ð Ð
45 2.05 ÿ0.25 Ð Ð Ð Ð Ð
QDDS� effect of drugyeffect of DDS at 10mM.
log Derythrocytes�Partitioning into Erythrocytes.
Table 3. Methaemoglobin-formation (%) in cats 6 hrs and 24 hrsafter iv administration of the indicated sulphones (200 mgykg) andthe corresponding log k 0-values
Substance
% methaemo-
globin
6 hrs after
administration
24 hrs after
administration
dapsone (1) 6.09 4.4 0.34
28 0.7 1.2 0.83
31 0.0 0.0 0.00
Table 4. Loadings of the original scaled activity variables onprincipal components after Varimax rotation
PC 1 PC 2
I log % Methaem. formation 0.013 0.992
II log Cell burst inhibition 0.902 0.178
III log Cell adhesion inhibition 0.909 0.147
n� 10, derivative 35 omitted, no effects in cell adhesion test, 2PCs contain
89% of the information of the 4 original variables.
QSAR Joachim K. Seydel, Hanna BuÈrger, Anil K. Saxena, Michael D. Coleman, Stephen N. Smith and Alan D. Perris
48 Quant. Struct.-Act. Relat., 18 (1999)
Similarly a treatment of o-substituent effects based on
Swain-Lupton treatment [35] has been proposed by Fujita
and Nishioka [36].
log k � ff� rR � dEos � h
A highly signi®cant correlation was derived for the
inhibition of zymosan-mediated respiratory burst. The most
signi®cant correlation was found with sI The activity
increases with increasing sI-effect of the substituents in o-
position.(log Cellbr.� log effectDDSylog effect drug at
1 mM).
log cellbr: � 0:860��0:276�sI � 0:043��0:084� �6a�n � 11 r2 � 0:85 s � 0:081 F � 50 Q2 � 0:80
Alternatively the ®eld effect f has been used (Eq. 6b). The
intercorrelation sIyf in this data set is r2� 0.80.
log cellbr: � 0:762��0:286�f� 0:086��0:085� �6b�n � 11 r2 � 0:80 s � 0:092 F � 36 Q2 � 0:73
The inclusion of the resonance effect, R, in Eqs. 6a,b led to
a slight increase in r2 (0.86, 0.81 respectively) was,
however, not signi®cant. The inclusion of a steric parameter
(MR2) led also to an increase in r2 (0.85) and was signi®cant
on the 82% level.
Later this series was extended using a smaller drug
concentration of 0.1 mM instead of 1 mM.
Again a highly signi®cant correlation with sI is found but in
addition the MR2-value of the o-substituents plays a role in
explaining the observed variance. In this extended data set
substituents with larger steric demands are included (see
Tables 1, 5).
log 1y%cellbr � 0:133��0:071�sI ÿ 0:0117��0:004�MR2
ÿ 1:701��0:032� �7a�n � 15 r2 � 0:77 s � 0:002 F � 21 Q2 � 0:67
log 1y%cellbr � 0:094��0:066�fÿ 0:0123�0:005�MR2
� 1:697�0:036� �7b�n � 15 r2 � 0:75 s � 0:025 F � 18 Q2 � 0:64
Table 5. Observed and calculated antiin¯ammatory activities and descriptors used in the derivation of the regression equations
log cellbra log 1y%cellbrb
Eq. 6a Eq.7a log
cell
Nr. obs. calc. obs. calc. adhcc sI f MR2 R logk 0
1 0.00 0.04 ÿ1.72 ÿ1.71 0.00 0.00 0.00 1.03 0.00 0.34
18 Ð Ð ÿ1.73 ÿ1.74 Ð 0.17 0.02 5.42 ÿ0.68 1.08
19 0.176 0.189 Ð Ð 0.129 0.17 0.02 5.42 ÿ0.68 ÿ0.096
20 Ð Ð ÿ1.82 ÿ1.80 Ð 0.13 ÿ0.11 10.33 ÿ0.74 Ð
21 0.163 0.249 Ð Ð 0.00 0.24 0.29 2.85 ÿ0.64 ÿ0.009
22 Ð Ð ÿ1.71 ÿ1.70 Ð 0.24 0.29 2.85 ÿ0.64 1.26
23 0.419 0.249 Ð Ð 0.280 0.24 0.29 2.85 ÿ0.64 1.93
24 0.335 0.307 ÿ1.72 ÿ1.75 0.501 0.30 0.26 7.87 ÿ0.51 1.77
26 Ð Ð 0.00 0.00 1.03 0.00 2.03
29 Ð Ð ÿ1.73 ÿ1.77 Ð ÿ0.01 ÿ0.04 5.65 ÿ0.13 1.66
30 Ð Ð ÿ1.76 ÿ1.77 Ð ÿ0.01 ÿ0.04 5.65 ÿ0.13 2.15
32 0.016 0.034 Ð Ð 0.155 ÿ0.01 ÿ0.04 5.65 ÿ0.13 3.75
33 0.034 0.035 Ð Ð 0.083 ÿ0.01 ÿ0.04 5.65 ÿ0.13 4.25d
34 Ð Ð ÿ1.79 ÿ1.77 Ð ÿ0.01 ÿ0.04 5.65 ÿ0.13 3.03
35 0.187 0.138 Ð Ð Ð 0.11 0.00 7.19 0.00 0.89
36 0.415 0.387 Ð Ð 0.063 0.40 0.38 5.02 0.19 1.41
37 0.260 0.387 ÿ1.71 ÿ1.71 0.063 0.40 0.38 5.02 0.19 2.45
38 Ð Ð ÿ1.79 ÿ1.79 ± 0.30 0.32 11.18 0.20 0.192
39 Ð Ð ÿ1.79 ÿ1.78 ± 0.28 0.24 9.81 0.14 ÿ1.53
40 Ð Ð ÿ1.82 ÿ1.81 ± 0.32 0.33 12.87 0.15 1.63
41 Ð Ð ÿ1.73 ÿ1.71 ± 0.47 0.41 6.03 ÿ0.15 1.06
42 Ð Ð ÿ1.71 ÿ1.74 ± 0.47 0.44 8.88 ÿ0.17 1.32
43 0.629 0.619 Ð Ð 0.266 0.67 0.67 7.63 0.16 1.31
44 Ð Ð ÿ1.70 ÿ1.70 ± 0.63 0.51 6.33 0.19 1.86
(a) log cellbr� log (effect DDSyeffect drug) at 1 mM [20].
(b) log %cellbr at 0.1 mM of the drugs.
(c) log celladh� log effect DDSyeffect drug at 1 mM [20].
(d) calc.
Quantitative Structure±Activity and Structure±Toxicity Relationships QSAR
Quant. Struct.-Act. Relat., 18 (1999) 49
As the descriptors used in Eqs. 7a and 7b have very
different values (size), it is dif®cult to judge their relative
importance. All parameters have therefore been scaled:
1ylog %cellbr � 0:592��0:316�sI ÿ 0:867��0:316�MR2
n � 15 r2 � 0:77 s � 0:53 F � 21 Q2 � 0:67 �8a�
1ylog %cellbr: � 0:556��0:310�fÿ 0:835��0:323�MR2
n � 15 r2 � 0:75 s � 0:56 F � 18 Q2 � 0:64 �8b�
Intercorrelation MR2ysI, r2 � 0:10
Inclusion of the resonance effect in Eqs. 7ay7b did slightly
improve r2 but the contribution was again not signi®cant.
The QSAR results on inhibition of respiratory cell burst
using drug concentrations of only 0.1 mM have, however, to
be considered with caution because of the small range in
biological activities (Eqs. 7ayb). This also may be the
reason for the lack of signi®cant correlations for the data on
inhibition of cell adhesion. The best equation obtained was
with sI in accordance with the results of the PC-analysis.
The scores of PC 2 which contain the information extracted
from the original variables cellbr and celladh correlate with
sI:
PC2 � 3:88��2:36�sI ÿ 0:931��0:75�n � 10 r2 � 0:64 s � 0:67 F � 14:0 Q2 � 0:53 �9�
The two most deviating derivatives are #23 and 24 (Table
1). A reason for this cannot be given.
4 Conclusion
The toxic effect (methaemoglobin formation) of 2-sub-
stituted 4-aminodiphenylsulphones depends solely on log k 0
in a nonlinear function.In contrast, the antiin¯ammatory
activity, expressed as inhibition of cell burst and cell
adhesion, depends on electronic and steric substituent
effects of the substituent in 2 0-position as found previously
also for their antibacterial effect. No in¯uence on changes
in lipophilicity occurred with the exception that under in
vivo conditions a further decrease in methaemoglobin
formation with increasing lipophilicity was observed. This
can also be demonstrated in vitro upon addition of serum
albumin. The orthogenality in the in¯uence of physico-
chemical properties on the toxic effects compared to the
wanted antiin¯ammatory and antibacterial effects can lead
to an optimization in the selectivity of the sulphones.
In summary, we have used a wide range of biological data
including cytotoxicity, methaemoglobin formation, cell
burst and cell adhesion assays and physicochemical
properties in a mathematical analysis to derive QSAR
relationships aimed at optimizing the selectivity and
minimizing the toxicity of antibacterial and antiin¯amma-
tory sulphone drugs.
References
[1] Seydel, J.K., Coats, A.E., Cordes, H-P., Richter, M., Wiese,M., and Kulkarni,V.M., Application of QSAR in thedevelopment of model biological test systems. Diphenylsul-phones versus cell-free and whole cell systems. in:Quantitative Approaches in Drug Design, J.C. Dearden ed.,Elsevier, Amsterdam, 1983, pp. 253±258.
[2] Coats, A.E., Cordes, H.-P., Kulkarni, V.M., Richter, M.,Schaper, K.-J., Wiese, M., and Seydel, J.K. Multipleregression and principal component analysis of antibacterialactivities of sulfones and sulfonamides in whole cell and cell-free systems of various DDS sensitive and resistant strains.Quant. Struct.-Act. Relat. 4, 99±109 (1985).
[3] Wiese, M., Seydel, J.K., Pieper, H., KruÈger, G., Noll, K.R.,and Keck, J., Multiple regression analysis of sulfones ofantimalarial activities in cell-free systems and principalcomponent analysis to compare with antibacterial activities.Quant.Struct.-Act. Relat. 6, 164±172 (1987).
[4] Seydel, J.K., Wempe, E.G., Rosenfeld,M., Jaganathan, R.,Mahadevan, P.R., Dhople,A.M. In vitro and in vivoexperiments with the new inhibitor of Mycobacterium leprae,Brodimoprim, alone and in combination with dapsone.Arzneim-Forsch. 40, 69±75 (1990).
[5] Zone, J.J., Dermatitis herpetiformis. Curr. Probl. Derm. 3,4±42 (1991).
[6] Uetrecht, J.P., The role of leukocyte-generated reactivemetabolites in the pathogenesis of idiosyncratic drugreactions. Drug Metab. Rev. 24, 299±366 (1992).
[7] Kettle, A.J., Winterbourne, C.C., Mechanism of the inhibi-tion of myeloperoxidase by anti-in¯ammatory drugs. Bio-chem. Pharmacol. 41, 1485±1492 (1991).
[8] van Zyl, J.M., Basso, K., Kriegler, A., van der Walt, B.J.,Mechanisms by which clofazimine inhibits the myeloperox-idase system. Biochem. Pharmacol. 42, 599±608 (1991).
[9] Booth, S.A., Mody, C.E., Dahl, M.V., Herron, M.J., Nelson,R.D., Dapsone suppresses integrin-mediated neutrophiladherence function. J. Invest. Pharmacol. 98, 135±140(1992).
[10] Thuong-Nguyen, V., Kadunce, D.P., Hendrix, J.D., Gammon,W.R., Zone, J.J., Inhibition of neutrophil adherence toantibody by dapsone: a possible therapeutic mechanism ofdapsone in the treatment of IgA dermatoses. J. Invest.Pharmacol. 100, 349±355 (1993).
[11] Wozel, G., Lehmann, B., Dapsone inhibits generation of5-lipoxygenase products in human polymorphonuclearleukocytes. Skin Pharmacol. 8, 196±202 (1995).
[12] Coleman, M.D., Breckenridge, A.M., Park, B.K., Bioactiva-tion of dapsone to a cytotoxic metabolite by human hepaticmicrosomal enzymes. Br. J. Clin. Pharmacol. 28, 389±395(1989).
[13] Coleman, M.D., Scott, A.K., Breckenridge, A.M., Park, B.K.,The use of cimetidine as a selective inhibitor of dapsone N-hydroxylation in man. Br. J. Clin. Pharmacol. 30, 761±767(1990)
[14] Coleman, M.D., Rhodes, L.A., Scott, A.K., Breckenridge,A.M., Park, B.K., The use of cimetidine to reduce dapsone-dependent methaemoglobinaemia in dermatitis herpetiformispatients. Br. J. Clin. Pharmacol. 34, 244±249 (1992).
QSAR Joachim K. Seydel, Hanna BuÈrger, Anil K. Saxena, Michael D. Coleman, Stephen N. Smith and Alan D. Perris
50 Quant. Struct.-Act. Relat., 18 (1999)
[15] Manfredi, G., De Pau®lis, G., Zampetti, M., Allegra, F.,Studies on dapsone-induced haemolytic anaemia: methae-moglobin production and glucose-6-phosphate activity incorrelation with dapsone dosage. Br. J. Dermatol. 100, 427±432 (1979).
[16] Pieper, H., Seydel, J.K., KruÈger, G., Noll, K., Keck, J.,Wiese, M., Preparation and biological activity of newsubstituted antimalarial diaminodiphenylsulfones. Arzneim.-Forsch. 39, 1073±1080 (1989).
[17] Kulkarni, V.M., Seydel, J.K., Inhibitory activity and mode ofaction of DDS in cell-free folate synthesizing systemsprepared from M. lufu and M. lepraeÐA comparison,Chemotherapy 29, 58±67 (1983).
[18] BuÈrger, H., Entwicklung eines in vitro Versuchssystems zurErmittlung methaemoglobinbildender Eigenschaften vonDiaminodiphenylsulfon und Ableitung von quantitativenStruktur-Wirkungsbeziehungen. Dissertation, UniversitaÈtKiel, 1991.
[19] Coleman, M.D., Smith, St., Kelly, D.E., Kelly, St.L., andSeydel, J.K., Studies on the toxicity of novel analogues ofdapsone in-vitro using rat, human and heterologouslyexpressed metabolizing systems. J. Pharm. Pharmacol. 48,945±950 (1996).
[20] Coleman, M.D., Smith, J.K., Perris, A.D., Buck, N.S., andSeydel, J.K., Studies on the inhibitory effects of analogues ofdapsone on neutrophil function in-vitro. J. Pharm. Pharma-col. 23, 53±57 (1997).
[21] Havemann,R., Jung, R., v.Issekutz, B., Die Bestimmung vonMethaemoglobin im Blut mit dem lichtelektrischen Kolori-meter. Biochem. Z. 301, 116±124 (1939).
[22] Purba, H.S., Maggs, J.L., Orme, M.L.E., Back, D.J., andPark, B.K., The metabolism of 17aethinyloestradiol byhuman liver microsomes: formation of catechol and chemi-cally reactive metabolites. Br. J. Clin. Pharmacol. 23, 447±453 (1987).
[23] Tingle, M.D., Coleman, M.D., and Park, B.K., Investigationinto the role of metabolism in dapsone-induced methaemo-globinaemia using a two-compartment in-vitro test system.J. Clin. Pharmacol. 30, 829±838 (1990).
[24] Coleman, M.D., Tingle, M.D., Hussain, F., Storr, R.C., Park,B.K., An investigation into the haematological toxicity ofstructural analogues of dapsone in-vitro and in-vivo. Br. J.Clin.Pharmacol. 43, 779±784 (1991).
[25] Saxena, A.K., and Seydel, J.K., QSAR studies of potentialantimalarial sulfonamide partitioning into erythrocytes andbinding to erythrocytes. Eur. J. Med. Chem. 15, 241±246(1980).
[26] Afford, S.C., Pongracz,J., Stockley, R.A., Croker, J., Burnett,D., The induction by human interleukin-6 of apoptosis in thepromonocytix cell line U937 and human neutrophils. J. Biol.Chem. 267, 21612±21616 (1992).
[27] Junger, W.G., Cardoza, T.A., Liu, F.C., Hoyt, D.B., Good-win, R., Improved rapid photometric assay for quantitativemeasurement of PMN migration. J. Immunol. Methods 160,73±79 (1993).
[28] Elashoff, J.D., Down with multiple `t' tests. Gastroenter-ology 80, 615±619 (1981).
[29] Seydel, J.K., and Schaper K-.J., Chemische Struktur undBiologische AktivitaÈt von Wirkstoffen, Verlag Chemie,Weinheim 1979.
[30] Hansch, C., and Leo, A., Substituent Constants for Correla-tion Analysis in Chemistry and Biology, Wiley Interscience,New York 1979.
[31] Kiese, M., Reinwein, D., Waller, H.D., Kinetik derHaemoglobinbildung. IV. Mitteilung. Die Haemoglobinbil-dung durch Phenylhydroxylamin und Nitrobenzol in rotenZellen in vitro. Naunyn-Schmiedebergs Archiv. Ex. Path.Pharm. 210, 393±398 (1950).
[32] Willi, P., Bickel, M.H., Liver metabolic reactions: Tertiaryamine N-dealkylation, tertiary amine N-oxidation, N-oxidereduction, and N-oxide N-dealkylation, II. N,N-Dimethylani-line. Arch. Biochem. Biophys. 156, 772±779 (1973).
[33] Houdi, A.A., Damani, L.A., N-demethylation and N-oxida-tion of N-dimethylaniline by rat liver microsomes. in: GorrodJ.W., Damani L.A., eds. Biological Oxydation of Nitrogen inOrganic Molecules, Chapt. 8, pp. 96±100, Verlag Chemie,Weinheim 1985.
[34] Charton, M., The quantitative treatment of the ortho-effect.Prog. Phys. Org. Chem. 8, 235±317 (1971).
[35] Swain, C.G., and Lupton, E.C., Field and resonancecomponents of substituent effects. J. Am. Chem. Soc. 90,4328±4337 (1968).
[36] Fujita, T., and Nishioka, T., The analysis of the ortho-effect.Prog. Org. Phys. Chem. 12, 49±89 (1976).
Received on May 25, 1998; accepted on August 31, 1998
Future Articles ISSUE 2/99
D. Douguet and G. Grassy: A Quantitative Structure±Activity Relationship Studies of RAR a, b, g,Retinoid Agonists
Q. Chen, C. Wu, D. Maxwell, G. A. Krudy, R. A. F. Dixon and T. J. You: A 3D QSAR Analysis of in vitroBinding Af®nity and Selectivity of 3-Isoxazolylsulfonylaminothiophenes as Endothelin ReceptorAntagonists
M. Fjalali-Heravi and F. Parastaar: Computer Modeling of the Rate of Glycine Conjugation of SomeBenzoic Acid Derivatives: A QSAR Study
M. Seel, D. B. Turner and P. Willett: Effect of Parameter Variations on the Effectiveness of HQSARAnalyses
O. Mekenyan, N. Nikolova, S. Karabunarliev, S. P. Bradbury, G. T. Ankley and B. Hansen: NewDevelopments in a Hazard Identi®cation Algorithm for Hormone Receptor Ligands
Quantitative Structure±Activity and Structure±Toxicity Relationships QSAR
Quant. Struct.-Act. Relat., 18 (1999) 51