Quantitative structure-antifungal activity relationships of some benzohydrazides against Botrytis cinerea

Article abstract of DOI:10.1021/jf0704211

Fourteen benzohydrazides have been synthesized and evaluated for their in vitro antifungal activity against the phytopathogenic fungus Botrytis cinerea. The best antifungal activity was observed for the N′,N′- dibenzylbenzohydrazides 3b-d and for the N-am

Full text of DOI:10.1021/jf0704211

J. Agric. Food Chem. 2007, 55, 5171−5179 5171
Quantitative Structure−Antifungal Activity Relationships of
Some Benzohydrazides against Botrytis cinerea
JOSEÄ L. REINO,† LIANE SAIZ-URRA,‡ ROSARIO HERNAÄ NDEZ-GALAÄ N,†
VICENTE J. ARAÄ N,§ PETER B. HITCHCOCK,| JAMES R. HANSON,|
,#
MAYKEL PEREZ GONZALEZ,‡,
AND ISIDRO G. COLLADO*,†
Departamento de Qu´ımica Orga´nica, Facultad de Ciencias, Universidad de Ca´diz, Apartado 40,
11510 Puerto Real, Ca´diz, Spain, Chemical Bioactive Center, Central University of Las Villas,
Santa Clara, Villa Clara, C.P. 54830, Cuba, Instituto de Qu´ımica Me´dica, Consejo Superior de
Investigaciones Cient´ıficas, c/ Juan de la Cierva, 3, 28006 Madrid, Spain, Department of Chemistry,
University of Sussex, Brighton, Sussex, BN1 9QJ, United Kingdom, and Service Unit, Experimental
Sugar Cane Station “Villa Clara-Cienfuegos”, Ranchuelo, Villa Clara, C.P. 53100, Cuba
Fourteen benzohydrazides have been synthesized and evaluated for their in vitro antifungal activity
against the phytopathogenic fungus Botrytis cinerea. The best antifungal activity was observed for
the N′,N′-dibenzylbenzohydrazides 3b-d and for the N-aminoisoindoline-derived benzohydrazide 5.
A quantitative structure-activity relationship (QSAR) study has been developed using a topological
substructural molecular design (TOPS-MODE) approach to interpret the antifungal activity of these
synthetic compounds. The model described 98.3% of the experimental variance, with a standard
deviation of 4.02. The influence of an ortho substituent on the conformation of the benzohydrazides
was investigated by X-ray crystallography and supported by QSAR study. Several aspects of the
structure-activity relationships are discussed in terms of the contribution of different bonds to the
antifungal activity, thereby making the relationships between structure and biological activity more
transparent.
KEYWORDS: Benzohydrazides; crop protection agents; Botrytis cinerea; antifungal activity; QSAR;
spectral moments
INTRODUCTION
Hydrazides, which are the subject of this study, are known
to exhibit various biological activities including tuberculostatic
(5, 6), antibacterial (7), anticonvulsant (8), and antifungal
activities (9).
Botrytis species include some serious fungal plant pathogens,
which are implicated in many diseases affecting flowers, fruits,
cereals, legumes, and other vegetables. In particular, Botrytis
cinerea Pers. ex Fr. attacks economically important crops such
as carrots, grapes, lettuce, strawberries, and tobacco, producing
various leaf spot diseases and powdery gray mildews (1).
Quantitative structure-activity relationships (QSAR) are
widely used in the search for new and better compounds with
a specific biological activity. This methodology uses molecular
descriptors reflecting the structure of the molecules, which can
then be used in the development of predictive models. This
methodology gives rise to cost savings by reducing laboratory
resources used and research time required. Another characteristic
of this technique is its potential to analyze theoretical aspects
of the role of specific molecules or molecule fragments in a
certain biological activity.
In spite of the fact that many QSAR studies have been re-
ported about antifungal activity, a small number of limited
studies has been conducted aimed at exploring the use of QSAR
techniques in the rational search for new fungicides against B.
cinerea (10-13). It is important to highlight that the largest
QSAR unified model for antifungal compounds reported up to
date does not include B. cinerea among the strains taken into
account (14).
Following the discovery of the systemic antifungal properties
of carboxamides, several compounds, including fenfuram,
carboxin, and oxycarboxin, were synthesized and introduced
as agricultural fungicides (2). Within this group, the benzanilide
fungicides such as benodanil (1), flutolanil, mebenil, mepronil,
salicylanilide, and tecloftalam were found to exhibit a wide
spectrum of activity against pathogens such as Rhizoctonia
solani and various Typhula, Corticium, and Gymnosporangium
species (3, 4).
* To whom correspondence should be addressed. Tel: 34 956 016371.
Fax: 34 956 016193. E-mail: isidro.gonzalez@uca.es.
† Universidad de Ca´diz.
‡ Central University of Las Villas.
§ Consejo Superior de Investigaciones Cient´ıficas.
| University of Sussex.
Two- and three-dimensional molecular descriptors have been
widely used in QSAR research. However, these descriptors have
Experimental Sugar Cane Station.
# Universidad Santiago de Compostela.
10.1021/jf0704211 CCC: $37.00
© 2007 American Chemical Society
Published on Web 06/02/2007

5172 J. Agric. Food Chem., Vol. 55, No. 13, 2007
Reino et al.
triethylamine or sodium hydrogencarbonate as bases (previously
reported methods A1 and A2) (15).
not previously been used in the development of models to
predict fungicidal activity against B. cinerea.
2-Iodo-N′,N′-dimethylbenzohydrazide (2d). Yield, 75% (method
A1); mp 155-157 °C (toluene). IR (KBr) νmax: 3217 (NH), 1650 (CO)
cm-1. 1H NMR (CDCl3): δ ) 7.86-6.98 (m, 4 H, aromatic H, Z and
E rot.), 6.59 (br s, E rot.) and 6.48 (br s, Z rot.) (1 H, NH), 2.71 (s, Z
rot.) and 2.50 (s, E rot.) (6 H, CH3) (Z rot./E rot. ratio: 78/22). 13C
NMR (CDCl3): δ ) 172.5 (E rot.) and 167.0 (Z rot.) (CO), 142.3 (E
rot.) and 140.7 (Z rot.) (C-1), 139.4 (Z rot.) and 138.9 (E rot.) (C-3),
131.0 (Z rot.) and 129.9 (E rot.) (C-4), 128.3 (Z rot.), 128.0 (Z rot.),
127.4 (E rot.) and 126.3 (E rot.) (C-5, -6), 92.6 (Z rot.) and 91.4 (E
rot.) (C-2), 48.4 (E rot.) and 47.1 (Z rot.) (CH3). EI MS: m/z (%) )
290 (M+, 10), 248 (89), 247 (79), 231 (100), 203 (45), 105 (14), 76
(40), 59 (77). C9H11IN2O (290.10) calcd: C, 37.26; H, 3.82; N, 9.66;
I, 43.74. Found: C, 36.98; H, 4.01; N, 9.82; I, 43.50%.
Additionally, our ready access to compounds containing a
benzohydrazide skeleton analogous to the structure of the
commercial fungicide benodanil (1) and our interest in the
control of B. cinerea has prompted us to undertake the synthesis
of several N′,N′-disubstituted benzohydrazides.
N′,N′-Dibenzylbenzohydrazide (3a). Yield, 89% (method A2); mp
168-170 °C (EtOH); lit. (18), 169 °C.
N′,N′-Dibenzyl-2-chlorobenzohydrazide (3b). Yield, 89% (method
A2); mp 125-127 °C (2-PrOH). IR (KBr) νmax: 3216 (NH), 1658 (CO)
1
cm-1. H NMR (CDCl3): δ ) 7.50-6.28 (m, 15 H, aromatic H and
NH, Z and E rot.), 4.30 (s, Z rot.) and 3.78 (s, E rot.) (4 H, CH2) (Z
rot./E rot. ratio: 78/22). 13C NMR (CDCl3): δ ) 171.2 (E rot.) and
166.0 (Z rot.) (CO), 137.2 (Z rot.) and 135.3 (E rot.) (C-1′’), 135.6 (E
rot.) and 134.3 (Z rot.) (C-1), 131.0 (Z rot.) and 129.6 (E rot.) (C-4),
130.9 (Z rot.) and 129.92 (E rot.) (C-2), 129.90 (Z rot.) and 128.9 (E
rot.) (C-3), 129.8 (E rot.), 129.3 (Z rot.), 128.4 (E rot.) and 128.3 (Z
rot.) (C-2′′, -3′′, -5′′, -6′′), 129.2 (Z rot.) and 128.0 (E rot.) (C-6), 127.7
(E rot.) and 127.4 (Z rot.) (C-4′’), 126.7 (Z rot.) and 126.1 (E rot.)
(C-5), 61.5 (E rot.) and 59.4 (Z rot.) [C-1′ (CH2)]. EI MS: m/z (%) )
350 (M+, 0.4), 259 (14), 211 (28), 196 (53), 195 (66), 194 (56), 156
(41), 139 (55), 111 (22), 91 (100). C21H19ClN2O (350.85) calcd: C,
71.89; H, 5.46; N, 7.98; Cl, 10.10. Found: C, 72.02; H, 5.70; N, 8.08;
Cl, 10.21%.
N′,N′-Dibenzyl-2-iodobenzohydrazide (3c). Yield, 90% (method
A2); mp 139-141 °C (EtOH). IR (KBr) νmax: 3218 (NH), 1654 (CO)
In this paper, we describe both the synthesis and the results
of our tests on the fungistatic activity of these benzohydrazides.
Several aspects of the stereochemical relationship between their
structure and biological activity are discussed. Another aim is
to develop a predictive model of the antifungal activity for this
group of compounds and to determine the contribution made
by several structural fragments to the relationship, between
structure and biological activity.
1
cm-1. H NMR (CDCl3): δ ) 7.86-6.24 (m, 15 H, aromatic H and
NH, Z and E rot.), 4.31 (s, Z rot.) and 3.81 (s, E rot.) (4 H, CH2) (Z
rot./E rot. ratio: 80/20). 13C NMR (CDCl3): δ ) 173.2 (E rot.) and
168.3 (Z rot.) (CO), 141.7 (E rot.) and 140.4 (Z rot.) (C-1), 139.7 (Z
rot.) and 138.6 (E rot.) (C-3), 137.2 (Z rot.) and 135.1 (E rot.) (C-1′′),
131.0 (Z rot.) and 129.8 (E rot.) (C-4), 129.8 (E rot.), 129.3 (Z rot.),
128.4 (E rot.) and 128.3 (Z rot.) (C-2′′, -3′′, -5′′, -6′′), 127.9 (Z rot.),
127.7 (Z rot.), 127.5 (E rot.) and 127.2 (E rot.) (C-5, -6), 127.8 (E
rot.) and 127.4 (Z rot.) (C-4′′), 92.6 (Z rot.) and 91.5 (E rot.) (C-2),
61.3 (E rot.) and 59.4 (Z rot.) [C-1′ (CH2)]. EI MS: m/z (%) ) 442
(M+, 0.4), 351 (14), 248 (54), 231 (55), 211 (31), 203 (24), 196 (58),
195 (71), 194 (47), 118 (10), 105 (15), 91 (100). C21H19IN2O (442.30)
calcd: C, 57.03; H, 4.33; N, 6.33; I, 28.69. Found: C, 57.21; H, 4.22;
N, 6.51; I, 28.91%.
MATERIALS AND METHODS
We prepared a series of substituted benzohydrazides 2a-d, 3a-d,
4a-e, and 5 by acylating the hydrazines with the corresponding acid
chlorides (15). These compounds were then screened for antifungal
activity at 50, 100, and 200 µg/mL against the phytopathogen B.
cinerea. The reference substance, benodanil (1) (50 µg/mL), was
included in all tests as a positive control. The X-ray crystal structures
of compounds 3a,b were determined.
General Experimental Procedures. Melting points were determined
with a Reichert-Jung Thermovar hot-stage microscope. 1H (300 MHz)
and 13C (75, 100, or 125 MHz) NMR spectra were recorded on Varian
Inova 300, Varian Inova 400, or Varian Unity 600 spectrometers.
Chemical shifts are reported in ppm from TMS (δ scale) but were
measured against the solvent signal (CDCl3: δH ) 7.24; δC ) 77.0).
Z/E rotamer ratios of hydrazides were determined using the relative
integrals for the N[CHn]2 signals corresponding to each isomer in the
1H NMR spectra. δC Values reported in the literature for 2-substituted
benzoic acids (16) and esters (17) were consulted for the assignment
of 13C NMR spectra. Electron impact (EI) mass spectra were obtained
at 70 eV on a Hewlett-Packard 5973 MSD spectrometer. Microanalyses
were performed by the Departamento de Ana´lisis, Centro de Qu´ımica
Orga´nica “Manuel Lora Tamayo”, CSIC (Madrid, Spain).
N′,N′-Dibenzyl-2-methylbenzohydrazide (3d). Yield, 91% (method
A2); mp 126-128 °C (2-PrOH). IR (KBr) νmax: 3226 (NH), 1665 (CO)
1
cm-1. H NMR (CDCl3): δ ) 7.50-6.30 (m, 15 H, aromatic H and
NH, Z and E rot.), 4.25 (s, Z rot.) and 3.69 (s, E rot.) (4 H, CH2), 2.06
(s, Z rot.) and 1.89 (s, E rot.) (3 H, CH3) (Z rot./E rot. ratio: 91/9). 13
C
NMR (CDCl3): δ ) 173.7 (E rot.) and 169.2 (Z rot.) (CO), 137.3 (Z
rot.) and 135.2 (E rot.) (C-1′′), 136.2 (Z rot.) and 134.8 (E rot.) (C-2),
135.5 (E rot.) and 134.9 (Z rot.) (C-1), 130.7 (Z rot.) and 130.0 (E
rot.) (C-4), 129.7 (Z rot.) and 128.9 (E rot.) (C-3), 129.5 (E rot.), 129.2
(Z rot.), 128.5 (E rot.) and 128.3 (Z rot.) (C-2′′, -3′′, -5′′, -6′′), 127.8
(E rot.) and 127.5 (Z rot.) (C-4′′), 126.7 (E rot.) and 126.5 (Z rot.)
(C-6), 125.4 (Z rot.) and 124.5 (E rot.) (C-5), 61.8 (E rot.) and 59.8 (Z
rot.) [C-1′ (CH2)], 19.04 (Z rot.) and 18.98 (E rot.) (CH3). EI MS: m/z
(%) ) 330 (M+, 1), 239 (14), 211 (37), 196 (73), 195 (62), 194 (46),
136 (50), 119 (63), 104 (6), 91 (100). C22H22N2O (330.43) calcd: C,
79.97; H, 6.71; N, 8.48. Found: C, 80.12; H, 6.43; N, 8.28%.
N-Piperidinobenzamide (4a). Yield, 82% (method A1); mp 197-
199 °C (toluene); lit. (19), 193-195 °C; lit. (20), 198-199 °C.
2-Iodo-N-piperidinobenzamide (4d). Yield, 88% (method A1); mp
Preparation of Substituted Benzohydrazides. The preparation of
hydrazides 2a-c, 4b,c, and 5 has been previously described (15).
Hydrazides 2d, 4a,d,e, and 3a-d were prepared by acylation of the
corresponding hydrazines with acid chlorides using, respectively,
147-149 °C (toluene). IR (KBr) νmax: 3212 (NH), 1663 (CO) cm-1
.
1H NMR (CDCl3): δ ) 7.84-6.98 (m, 4 H, aromatic H, Z and E rot.),
6.57 (br s, E rot.) and 6.40 (br s, Z rot.) (1 H, NH), 2.88 (m, Z rot.)

Benzohydrazides against Botrytis cinerea
J. Agric. Food Chem., Vol. 55, No. 13, 2007 5173
Table 2. Crystal and Structure Refinement for Compound 3a
Table 1. Antifungal Bioassays of Compounds 2a
−d, 3a−d, 4a−e, and
5 against B. cinerea
empirical formula
formula weight
temperature
wavelength
crystal system
space group
C21H20N2O
316.39
173(2) K
0.71073
monoclinic
P21/c (no. 14)
% inhibitiona
100 g/mL
45
compounds
200
µ
g/mL
µ
50
µ
g/mL
2a
2b
2c
2d
3a
3b
3c
3d
4a
4b
4c
4d
4e
5
51
±
2
1
2
1
3
2
1
2
1
1
3
2
1
1
±
3
1
0
2
2
2
2
2
1
6
2
1
1
1
24
12
0
±
±
±
±
±
±
±
±
±
±
±
±
±
±
4
0
0
1
0
1
1
4
3
1
4
1
1
1
32
33
33
44
86
82
88
45
51
57
46
51
91
±
±
±
±
±
±
±
±
±
±
±
±
±
22
0
±
±
±
±
±
±
±
±
±
±
±
±
±
unit cell dimensions
a
b
c
)
)
)
9.9014(3) Å
17.0048(6) Å
10.0709(3) Å
R ) 90°
â ) 96.307(2)
°
28
42
42
50
55
43
51
40
42
47
50
19
10
31
27
48
37
46
36
33
38
42
γ ) 90
°
volume
Z
density (calculated)
absorption coefficient
F(000)
1685.39(9) Å3
4
-
3
1.25 mg m
0.08 mm
672
-1
crystal size
0.2
×
0.2
×
0.1 mm3
°
11,
20,
θ
range for data colletion
index ranges
3.89−25.02
−11
−20
−11
e
e
e
h
k
e
e
l
e 11
reflections collected
independent reflections
11875
2943 [R(int)
2389
99.0%
full-matrix least-
squares on F2
2943/0/298
0.989
a Results are the means of at least three independent experiments conducted
in triplicate.
) 0.048]
reflections with I > 2σ(I)
completeness to θ ) 25.02
°
and 2.70 (br s, E rot.) (4 H, 2′-, 6′-H), 1.75 (m), 1.42 (m) and 1.25 (m)
(6 H, 3′-, 4′-, 5′-H, Z and E rot.) (Z rot./E rot. ratio: 82/18). 13C NMR
(CDCl3): δ ) 172.6 (E rot.) and 166.7 (Z rot.) (CO), 142.3 (E rot.)
and 140.9 (Z rot.) (C-1), 139.4 (Z rot.) and 138.7 (E rot.) (C-3), 131.0
(Z rot.) and 129.7 (E rot.) (C-4), 128.4 (Z rot.), 128.0 (Z rot.), 127.2 (E
rot.) and 126.5 (E rot.) (C-5, -6), 92.7 (Z rot.) and 91.6 (E rot.) (C-2),
57.7 (E rot.) and 56.7 (Z rot.) (C-2′, -6′), 25.1 (Z and E rot., C-3′,
-5′), 23.1 (Z rot.) and 22.7 (E rot.) (C-4′). EI MS: m/z (%) ) 330
(M+, 3), 325 (7), 296 (4), 286 (11), 248 (45), 247 (30), 231 (43), 210
(11), 203 (30), 186 (19), 105 (11), 99 (100), 83 (58), 76 (31). C12H15-
IN2O (330.17) calcd: C, 43.65; H, 4.58; N, 8.48; I, 38.44. Found: C,
43.75; H, 4.52; N, 8.28; I, 38.21%.
2-Methoxy-N-piperidinobenzamide (4e). Yield, 77% (method A1);
mp 92-94 °C (cyclohexane). IR (KBr) νmax: 3325 (NH), 1663 (CO)
cm-1. 1H NMR (CDCl3): δ ) 8.50 (br s, 1 H, NH), 8.15 (dd, Jo ) 7.7,
Jm ) 1.6, 1 H, 6-H), 7.40 (m, 1 H, 4-H), 7.04 (m, 1 H, 5-H), 6.93 (d,
Jo ) 8.1, 1 H, 3-H), 3.93 (s, 3 H, CH3), 2.89 (br t, J ) 5.3, 4 H, 2′-,
6′-H), 1.72 (m, 4 H, 3′-, 5′-H), 1.43 (m, 2 H, 4′-H) (100% of Z rot.).
13C NMR (CDCl3): δ ) 163.0 (CO), 156.9 (C-2), 132.5 (C-4), 132.2
(C-6), 121.3 (C-1, -5), 111.2 (C-3), 56.7 (C-2′, -6′), 55.8 (CH3), 25.2
(C-3′, -5′), 23.4 (C-4′) (100% of Z rot.). EI MS: m/z (%) ) 234 (M+,
3), 217 (3), 203 (2), 190 (10), 152 (58), 135 (100), 105 (12), 99 (51),
92 (20), 84 (56), 77 (38). C13H18N2O2 (234.30) calcd: C, 66.64; H,
7.74; N, 11.96. Found: C, 66.86; H, 7.70; N, 12.18%.
Microorganism and Antifungal Assays. The isolate used in this
work, B. cinerea 2100, was obtained from the “Centro Espan˜ol de
Cultivos Tipos” (CECT), Facultad de Biolog´ıa, Universidad de Valencia
(Spain), where this strain is on deposit. Bioassays were performed by
measuring radial growth inhibition on an agar medium in a Petri dish.
The test compound was dissolved in acetone to give final compound
concentrations ranging from 50 to 200 µg/mL. Solutions of the test
compound were added to a glucose-malt-peptone-agar medium (61
g of glucose-malt-peptone-agar per liter, pH 6.5-7.0). The final
acetone concentration was identical in both the control and the test
cultures. The medium was poured into sterile plastic Petri dishes
measuring 9 cm in diameter, and a mycelial disc of B. cinerea cut
from an actively growing culture and measuring 5 mm in diameter
was placed at the center of the agar plate. The radial growth inhibition
was measured for 6 days. Percentages of growth inhibition were calcu-
lated by comparing the mean value of the diameters of the mycelia in
the test plates with that of the untreated control plates. Each determi-
nation assay was conducted in triplicate and repeated at least three times.
X-ray Crystal Data and Structure Determination of N′,N′-
Dibenzylbenzohydrazide (3a). Crystallographic data (Table 2) have
been deposited at the Cambridge Crystallographic Data Centre, CCDC
No. 251521.
refinement method
data/restraints/parameters
goodness-of-fit on F2
final R indices [I > 2
R indices (all data)
σ
(I)]
R1
R1
0.102(8)
0.39 and
)
0.040, wR2
0.056, wR2
)
)
0.099
0.119
)
extinction coefficient
largest diff peak and hole
-3
−
0.36 e Å
Table 3. Crystal and Structure Refinement for Compound 3b
empirical formula
formula weight
temperature
wavelength
crystal system
space group
C21H19ClN2O
350.83
173(2) K
0.71073
triclinic
P (no. 2)
unit cell dimensions
a
b
c
)
)
)
9.6251(3) Å
10.9307(4) Å
17.9455(6) Å
R ) 74.513(2)
°
°
°
â ) 85.429(2)
γ ) 89.801(2)
volume
Z
density (calculated)
absorption coefficient
F(000)
1813.38(11) Å3
4
-
3
1.29 mg m
0.22 mm
736
-1
crystal size
0.30
3.73
×
0.15
×
0.10 mm3
θ
range for data colletion
index ranges
−25.11
°
11,
13,
e 21
−11
−12
−21
e
e
e
h
k
l
e
e
reflections collected
independent reflections
23862
6370 [R(int)
4563
98.5%
full-matrix least-
squares on F2
6370/0/603
1.030
) 0.081]
reflections with I > 2σ(I)
completeness to θ ) 25.02
°
refinement method
data/restraints/parameters
goodness-of-fit on F2
final R indices [I > 2
R indices (all data)
σ
(I)]
R1
R1
0.20 and
)
0.051, wR2
0.085, wR2
)
)
0.108
0.122
)
-3
largest diff peak and hole
−
0.37 e Å
Molecular Descriptors. Spectral Moments. Spectral moments are
based on the calculation of the bond matrix, whose theoretical basis
has been widely described in previous reports (21, 22). Essentially,
the bond matrix is the square and symmetric matrix whose entries are
ones or zeros according to whether the corresponding bonds are adjacent
or not. The order of this matrix (m) is the number of bonds in the
molecular graph, two bonds being adjacent if they have one common
atom. The spectral moments of the edge adjacency matrix are defined
N′,N′-Dibenzyl-2-chlorobenzohydrazide (3b). Crystallographic data
(Table 3) have been deposited at the Cambridge Crystallographic Data
Centre, CCDC No. 251522.

5174 J. Agric. Food Chem., Vol. 55, No. 13, 2007
Reino et al.
as the traces, that is, the sum of the main diagonal of the different
powers of such a matrix.
underestimating the utility of the regression coefficient model. The
Randic´ method of orthogonalization has been described in detail in
several publications (29-33).
Validation of the Models. The models obtained were validated by
calculating the cross-validated squared regression coefficient (q2) values.
The q2 values are calculated from the LOO test and from the LGO
test, also known as cross-validation.
The following steps were followed in applying the above approach
to the development of a model for predicting antifungal activity. (i)
An adequate training set of chemicals was selected; (ii) the molecular
graphs for each of the training set were drawn; (iii) molecular bonds
with appropriate weights were differentiated; (iv) the spectral moments
of the bond matrix for each molecule in the data set were computed;
(v) a QSAR was derived by using multiple regression analysis (MRA);
(vi) the predictive performance of the model was assessed employing
the leave one out (LOO) and leave group out (LGO) cross-validation
methodologies; and (vii) the contribution of the different fragments
was ascertained to determine their quantitative contribution to the
antifungal activity of the molecules studied.
In the case of LOO, a data point is removed from the set, and the
regression is recalculated; the predicted value for that point is then
compared to its actual value. This is repeated until each piece of data
has been omitted once; the sum of squares of these deletion residuals
can then be used to calculate q2, an equivalent statistic to R2. In the
LGO method, 25% of the data was eliminated every time in 100
different forms. In this way, we ensured that the same group of
compounds was not used for every validation. The q2 values can be
considered a measure of the predictive power of a regression equation,
whereas R2 can always be increased artificially by adding more
parameters (descriptors), q2 decreases if a model is overparameterized
(34), and is therefore a more meaningful summary statistic for QSAR
models.
Computation of Bond Contributions. Calculation of the fragment
contribution has been described previously (35, 36); we will now
provide an overview of the methodology used. Bond contributions to
antifungal activity were calculated on the basis of the local spectral
moments, which were defined as the diagonal entries of the different
powers of the weighted bond matrix
For the development of the MRA, we used the following expression
P ) a0µ0 + a1µ1 + a2µ2 + ... + akµk + b
(1)
where P is the activity under study (the percentage of growth inhibition),
µk is the kth spectral moment, and the ak values are the coefficients
obtained by the regression equation.
Selection of Bond Weights and Calculation of Molecular De-
scriptors. Hydrophobicity (Hyd), molar refractivity (Mol), and Gasteiger-
Marsilli (GM) charge were used as bond weights. The latter is an atomic
contribution, and it was transformed into bond contributions, according
to eq 2. This transformation has been previously described (23)
µTk(i) ) bii(T)k
(3)
wi wj
w(i, j) )
+
(2)
δi δj
where µT(i) is the kth local moment of the bond i, bii(T) are the
k
where wi and δi are the atomic weight and vertex degree of atom i,
respectively. The calculation of the spectral moment descriptors was
carried out with Modeslab 1.0 software (24). The input of the chemical
structures into the Modeslab software was carried out using the familiar
SMILES notation. We calculated the first 15 spectral moments
(µ1-µ15) for each bond weight and the number of bonds in the mole-
cules (µ0). Because of the nonlinearity of the biological process under
study (fungicidal activity), the interactions between µ0 and µ1 and with
all variables (descriptors) were evaluated. Spectral moments were
calculated considering the molecules in the absence of hydrogen atoms.
Computational Strategies. The mathematical model was obtained
by means of the MRA as implemented in the STATISTICA software
version 6.0 (25). The data for the compounds which had been
synthesized were that for their biological activities reported for 200
µg/mL and 3 days of observation The most significant parameters were
identified from the data set using genetic algorithm (GA) analysis as a
variable selection strategy of the descriptors obtained by TOPS-MODE
computer software. The GA is a class of methods based on biological
evolution rules (26-28). The first step was to create a population of
linear regression models. These regression models mate with one
another, mutate, cross over, reproduce, and then evolve through
successive generations toward an optimal solution. The particular GA
simulation conditions applied here are as follows: 1000 generations,
0.5 for the tradeoff between cross over and mutation parameter, 1
smoothness factor, and 300 model populations.
Analysis of residuals and deleted residuals from the regression
equations was used to identify outliers. The statistical significance of
the models was determined by examining the squared regression
coefficient (R2), the standard deviation (S), the number of variables,
and the Fisher ratio (F).
Orthogonalization of Descriptors. The orthogonalization process
of molecular descriptors was introduced by Randic´ several years ago
as a way of improving the statistical interpretation of the model, which
had been built by using inter-related indices (29-33). The main
philosophy of this approach is to avoid the exclusion of descriptors on
the basis of their collinearity with other variables previously included
in the model. The acceptable level of collinearity to avoid is a more
subjective issue. In our view, the collinearity of the variables should
be as low as possible because inter-relatedness among the different
descriptors can result in a highly unstable regression coefficient, making
it impossible to know the relative importance of an index and
diagonal entries of the weighted bond matrix, and T is the type of the
bond weight: Std, Dip, and GM.
It is plain to see that the total moments are the sum of the local
moments. Consequently, we can substitute eq 3 into the QSAR (eq 4)
in such a way that the total contribution of the different bonds in a
specific molecule was obtained as follows
P ) bo +
ak ‚ µTk
(4)
∑
k
These contributions represent the additive features of the property
modeled, and they can be expressed as fragment contributions, the sum
of the contributions of different bonds that are inside the substructure
whose contribution is under examination.
RESULTS AND DISCUSSION
The results of the antifungal activity screening test on B.
cinerea show that in general the benzohydrazides displayed
antifungal activity (see Table 1) with compounds 3b-d and 5
being the most active against the fungus. It is worth noting that
the most active compounds bear an additional aromatic ring on
the nitrogen, N′, of the hydrazide group, a fact that lends
credence to the idea that a second aromatic nucleus on the
benzohydrazide skeleton is necessary for a high degree of
activity. Thus, while compound 3b at 200 µg/mL inhibited
growth by 86% after 6 days, compound 3d inhibited growth by
88, 55, and 48% after 6 days at 200, 100, and 50 µg/mL,
respectively. Compound 5 was the most active compound, with
a 91% inhibition rate after 6 days at 200 µg/mL. These results
also indicate that an ortho substituent on the aromatic ring of
the benzoic acid contributes to an increase in the activity, as
can be seen from a comparison of the activity of compounds
3a-d.
Previous studies (37-39) have noted that many hydrazides
appear in solution (NMR) as mixtures of Z and E rotamers
because of their restricted rotation around the N-CO bond
(Figure 1), and this may have an impact on their biological
activity. Consequently, we investigated several previously

Benzohydrazides against Botrytis cinerea
J. Agric. Food Chem., Vol. 55, No. 13, 2007 5175
Figure 1. Rotamers Z and E of benzohydrazides.
Figure 2. X-ray molecular structure of compound 3a (ORTEP drawing).
described (15) benzohydrazides as well as some reported here
for the first time and established the Z/E rotamer ratios in CDCl3
at room temperature. These ratios were for the 2-fluoro- (2b,
4b, and 5), 2-chloro- (2c, 3b, and 4c), 2-iodo- (2d, 3c, and 4d),
2-methyl- (3d), and 2-methoxybenzoic acid (4e) derivatives
being ca. 93:7, 77:23, 80:20, 91:9, and 100:0, respectively. The
signals corresponding to each rotamer were assigned by
comparing the NMR spectra of 2-substituted hydrazides with
those of unsubstituted benzoic acid derivatives [2a (15), 3a (18),
and 4a (19)] in which only one species, claimed to be the Z
rotamer in the case of compound 3a (39), can be detected.
Figure 3. X-ray molecular structure of compound 3b (ORTEP drawing):
molecule A and molecule B.
which makes the amide N-H more readily available for
hydrogen bonding, could account for the increase in biological
activity.
A comparison of the X-ray crystal structures (Figures 2 and
3) of compounds 3a,b reveals a possible role for the ortho chloro
substituent. There are two competing influences on the confor-
mation of the aromatic carboxyhydrazide. On the one hand, there
is the tendency for the carbonyl group to become coplanar with
the aromatic ring influenced by the conjugation between the
two groups. On the other hand, in the crystalline state, there is
hydrogen bonding between the amide N-H of one molecule
and the carbonyl oxygen of the adjacent molecule. This is
favored by twisting the plane of the amide out of the plane of
the aromatic ring. In the unsubstituted case, this leads to an
angle between the plane of the aromatic ring and the plane of
the amide [C(1):C(7):N(1):O(1)] of 35.3° and a hydrogen bond
length [d(H‚‚‚A)] of 2.07 Å. The ortho chloro compound 3b
crystallizes with two independent molecules. However, there
is an angle between the planes of 81.5° (60.0° in the second
molecule) and a shorter hydrogen bond length (2.01 and 2.00
Å). The angle of the hydrogen bond in 3b is more near linearity
(173 and 174°) as compared to 3a (161°), indicating the potential
for a stronger hydrogen bond in 3b. Thus, the increased rotation
of the carboxyhydrazide brought about by the ortho chlorine
facilitates hydrogen bonding between the N-H and another
molecule. The biological activity is likely to involve a hydrogen
bond interaction between the hydrazide and the fungus. This
extra twisting, which is apparent in the crystalline state and
A QSAR study based on the synthesized compounds was
carried out to go deeper into the causes of the increase of the
antifungal activity in regard to the chemical structure, emphasiz-
ing the bond contributions of the different molecules from the
data.
First, three different models were developed taking into
account the percentage of the fungal growth inhibition reported
at three different concentrations, 200, 100, and 50 µg/mL, to
find the one that best describes the antifungal activity of these
benzohydrazides through the variables from the TOPS-MODE
approach by using GA to select the most significant variables.
Model selection was subjected to the principle of parsimony
(34). Then, to select the model, a high statistical significance
was taken into account in searching for a function with as few
parameters as possible. The results of this study are contained
in Table 4 where N is the number of compounds included in
the model, R2 is the square of the correlation coefficient, S is
the standard deviation of the regression, F is the Fisher ratio, p
2
is the significance of the model, and qCV-LOO is the cross-
validated squared regression coefficient. The parameter q2 is
used as a criterion of both the robustness and the predictive
ability of the model. Many authors consider high q2 (for instance,
q2 > 0.5) as an indicator that the model is predictive (40).
It is worth noting that all models are suitable for the prediction
of antifungal activity according to the statistical parameters;

5176 J. Agric. Food Chem., Vol. 55, No. 13, 2007
Reino et al.
Table 4. Models Developed for the Percentage of the Fungal Growth
Table 5. Models Developed for the Percentage of the Fungal Growth
Inhibition at 200, 100, and 50
µ
g/mL
Inhibition at 200, 100, and 50
µ
g/mL Keeping the Same Variables of
the Model at 200
µ
g/mL
model 1
model 2
model 3
model 1
model 4
model 5
GM
GM
Dip
Pol
Hyd
Pol
variables
µ0µ15
,
µ11
,
µ0µ6
,
µ0µ1
,
µ0µ6
,
µ1 ,
GM
Dip
Dip
GM
GM
GM
GM
GM
GM
µ0µ1
µ6
µ15
variables
µ0µ15
,
µ11
,
µ0µ15
,
µ11
,
µ0µ15 , µ11 ,
GM
GM
GM
concentration
200
14
0.983
4.02
µ
g/mL
100
14
µ
g/mL
50
14
µg/mL
µ0µ1
µ0µ1
µ0µ1
N
R2
S
concentration
200
µ
g/mL
100
µ
g/mL
50 µg/mL
R2
S
0.983
4.02
0.394
9.974
2.2
0.146
2.5
0.526
9.248
3.7
0.900
4.056
29.93
0.817
5.749
14.88
F
190.355
F
190.355
-
-
5
-4
-
5
-4
p
qCV
<10
2.6
×
10
5.0
×
10
p
qCV
<10
0.970
5.0
×
10
2
2
0.970
0.785
0.709
-
LOO
9.01
-LOO
however, the model related with the percentage of the fungal
growth inhibition at 200 µg/mL concentration is the best
since it shows the best statistical values by describing 98.3%
of the variance and with the greatest predictive capability of
97.0%.
Once the optimum condition of the concentration was selected
to continue the QSAR study, we developed several models for
the percentage of fungal growth inhibition at 200 µg/mL by
changing the number of variables in every step of the analysis.
In other words, once we developed a model, we calculated its
statistical parameters and determined whether the introduction
of the new variable into the model was justified. If it was, we
compared the results with the previous models. We then repeated
this analysis whenever a new variable was included. The first
model contained two variables and was described with the
following equation and the statistical parameters of the regres-
sion presented
as can be seen in the statistical parameters related to the
following equation
GM
%GI ) -71.54 ‚ µ12GM + 126.02 ‚ µ0µ12
+
12.56 ‚ µ14Hyd - 55.03 ‚ µ0µ1Mol + 46.17 (7)
where N ) 14, R2 ) 0.986, S ) 3.745, F ) 165.116, p < 10-5
,
2
2
qCV-LOO ) 0.959, SCV-LOO ) 6.548, qCV-LGO ) 0.929, and
SCV-LGO ) 7.801.
Despite the relative improvement that certain parameters
underwent, such as R2, S, and F, the value of qLOO worsened
2
(decrease from 0.970 to 0.959), along with qLGO2 (decrease from
0.947 to 0.929) as well as SLOO (increase from 5.267 to 6.548)
and SLGO (increase from 6.971 to 7.801).
All things considered, we selected eq 6 as the best model
because it features better statistical parameters than the model
containing two variables and the greatest predictive capability
with lower variability than the model described the eq 7.
In addition, a comparison between the model finally selected
and another two models was developed with the same three
variables for the inhibition growth percentage at the different
concentrations 100 and 50 µg/mL. The results are given next.
As can be seen in Table 5, the variables considered as the
best in the modeling at 200 µg/mL are not suitable to model
the antifungal activity against B. cinerea at the other two
concentrations. These variables are not well-correlated with the
antifungal activity displayed by these compounds at 100 and
50 µg/mL concentrations, and it is worth noting that the
concentration of the compounds tested might be related to their
mechanism of action and the best weight for the variables is
not necessarily the same for these concentrations.
The presence of outliers in QSAR models could lead to a
degree of bias such that the model is unable to predict the “real”
biological activity. Consequently, we looked for the presence
of outliers in eq 6 but none were found because all of the
compounds included in the training set had a standard residual
value under 2 ‚ δ where δ is equivalent to the standard deviation.
However, another test was reserved for the analysis of deleted
residual. In other words, the analysis of the residual value for
every compound considered in this case was not included in
the regression analysis. In this connection, we found that the
deleted residual value of the compounds did not differ signifi-
cantly from the respective standardized residual values, from
which we deduced that there was not any outlier in the data
set.
%GI ) -52.38 ‚ µ0µ15GM + 76.01 ‚ µ7GM + 46.17 (5)
where N ) 14, R2 ) 0.955, S ) 6.193, F ) 116.93, p < 10-5
,
2
2
qCV-LOO ) 0.927, SCV-LOO ) 7.877, qCV-LGO ) 0.831, and
SCV-LGO ) 9.214 and where %GI is the percentage of the fungal
growth inhibition and the parameters N, R2, S, F, p, and
2
qCV-LOO have been defined above. Also, we calculated the
validation parameters shown previously, that is, cross-validated
squared regression coefficient q2 and the standard deviation Scv
of the LOO and LGO procedures.
In this case, the statistic parameters quality was adequate.
Nonetheless, we looked into the possibility of obtaining a new
model with better statistical results and therefore introduced a
new variable. The resulting model is shown below with its
statistical parameters.
GM
%GI ) 94.33 ‚ µ0µ15GM - 58.71 ‚ µ11
+
21.43 ‚ µ0µ1GM + 46.17 (6)
where N ) 14, R2 ) 0.983, S ) 4.02, F ) 190.355, p < 10-5
qCV-LOO ) 0.970, SCV-LOO ) 5.267, qCV-LGO ) 0. 947, and
SCV-LGO ) 6.971.
,
2
2
As we had expected, the quality of statistic parameters
improved by increasing the value of the R2 (from 0.955 to
0.983) and of F (from 116.93 to 190.355) and by decreasing
the value of S (from 6.193 to 4.02). The predictive capability
of the model likewise was enhanced. This can be seen by the
rise in cross-validation parameters, qLOO2 (from 0.927 to 0.970)
In addition, the analysis of the model domain of application
(MDA) was carried out to find the applicability of the model
and emphasize the absence of outliers. For this aim, the leverage
values were calculated for every compound and plotted vs
standard residuals (Y-axes). Then, the domain of applicability
of the model is defined as a squared area within the (2 band
2
and qLGO (from 0.0.831 to 0.947), and the fall in SLOO (from
7.877 to 5.267) and for both procedures SLGO (from 9.214 to
6.971).
However, in the specific case of the model containing four
variables, the introduction of the new variable was not justified,

Benzohydrazides against Botrytis cinerea
J. Agric. Food Chem., Vol. 55, No. 13, 2007 5177
Figure 4. Model domain of application (MDA).
for residuals and a leverage threshold of h ) 0.857 (41). The
result of this study can be seen in the next table.
All compounds used in the training set lie within this area,
and any outliers were detected as the previous analysis showed.
Collinearity among variables was avoided by making an
orthogonalization of molecular descriptors because interrelated-
ness among different descriptors can result in highly unstable
models.
The QSAR model obtained with the spectral moments (eq
8) after orthogonalization and standardization is given below,
together with the statistical parameters of regression analysis.
%GI ) 24.50 ‚ Ω1µ0µ15GM + 7.65 ‚ Ω2µ0µ1
-
GM
7.16 ‚ Ω3µ11GM + 46.17 (8)
where N ) 14, R2 ) 0.983, S ) 4.02, F ) 190.355, p < 10-5
qCV-LOO ) 0.970, SCV-LOO ) 5.267, qCV-LGO ) 0.947, and
SCV-LGO ) 6.971.
,
Figure 5. Structures of group number 1. Compounds 2a−d. Fragments
in blue have a negative contribution, and the red ones have a positive
contribution.
2
2
Once we obtained the definitive model, we developed an
analysis of the bond contribution taking into account several
fragments of the molecules and dividing the compounds into
three groups to discover tendencies in the contributions made
by these fragments to antifungal activity with the aim of
extending the structure-activity relationship.
explained above, as the number of aromatic rings rises, not all
of these compounds affect it in a significant way (see compound
3a). However, when a substituent in the ortho position is
introduced in the benzoic acid, the contribution of this ring rises
significantly whereas the others practically double their con-
tributions. The reason for this could be that the inclusion of a
substituent in this ring at the ortho position does not permit the
rotation of the σ bond that exists between the ring and the
carbonyl group as easily as in the absence of the substituent.
A postoptimization conformational analysis of the structures
by the semiempiric PM3 method shows that the ring in the 3a
compound may rotate 60° to one side or the other. It must cross
an energy barrier with a maximum of 35 kcal/mol, until it comes
up against barriers of higher energy content. These are difficult
to overcome at the physiological temperature of any living
organism, to be able to rotate bigger angles. Nevertheless,
compounds 3b, 3d, and 3c with chloride, methyl, and iodide in
the ortho position must overcome a higher energy barrier for a
rotation of only 25° in either direction. This rotation becomes
increasingly difficult as the size of the substituent increases in
this position (Figure 6). This could be because this planar
system interacts more readily with certain proteins from a
An analysis of bond contributions for the first compound set
containing 2a-d shows that the aromatic ring makes a positive
contribution to biological activity and does not undergo
significant modification by adding one electron-withdrawing
substituent. This fact leads us to believe that low electron density
at the aromatic ring does not tend to strengthen activity in this
case. Another interesting fact is that while the molecular size
of the substituent increases, the contribution of the ring
decreases, possibly due to increasing difficulty in remaining on
the same plane as the carbonyl group; therefore, the π electron
overlap might decrease in these groups. This small decrease in
activity might partly be counteracted by the incremental
contribution of the substituents in ortho position, suggesting that
an increase in the hydrophobicity either at that place or some
other might increase the biological property (see Figure 5).
We also observed an interesting fact in the second set of
compounds containing 3a-d; while the property increases, as

5178 J. Agric. Food Chem., Vol. 55, No. 13, 2007
Reino et al.
In summary, the activity displayed by the benzohydrazide
derivatives studied here demonstrates that in addition to N′,N′-
dibenzyl and N-aminoisoindoline substitution, the presence of
a substituent in the ortho position of benzoic acid is critical to
the antifungal activity of these molecules. On the other hand,
while N′-benzyl substitution seems to play an important role in
the inhibition mechanism by enhancing the antifungal activity
of these compounds, the aliphatic chain on the nitrogen of
benzohydrazide does not seem to have a significant effect on
their fungistatic activity, except when the aliphatic chain is
sufficiently large for it to be considered appropriately hydro-
phobic. Thus, the enhanced biological properties in that case
(comparing the structures that only differ in the size of the
aliphatic chain linked to nitrogen of the benzohydrazide) might
be caused by the necessary balance provided by hydrophobicity
to cross the lipophobic or lipidic membranes of plants. Work is
under way to study the mode of action of these compounds on
the metabolism of the fungus.
Figure 6. Contribution of fragments of compounds belonging to the second
set. Compounds 3a−d.
LITERATURE CITED
(1) Elad, Y., Williamson B., Tudzynsky, P., Delen, N., Eds.
Botrytis: Biology, Pathology and Control; Kluwer Academic
Publishers: Dordrecht, The Nertherlands, 2004.
(2) Kuhn, P. J. Mode action of carboximides. Symp. Br. Mycol. Soc.
1989, 9, 155-183.
(3) Pommer, E. H.; Zwick, W. Efficacy of systemic fungicides
against Basidiomycetes. Indian Phytopathol. 1974, 27, 53-58.
(4) Kataria, H. R.; Verma, P. R.; Racow, G. Fungicidal control of
damping-off and seedling root in Brassica species caused by
Rhizoctonia solani in the growth chamber. Ann. Appl. Biol. 1993,
123, 247-256.
(5) Waisser, K.; Houngbedji, N.; Odlerova, Z.; Thiel, W.; Mayer,
R. Tuberculostatics. Part 49. Thiohydrazides, potential tubercu-
lostatics. Pharmazie 1990, 45 (2), 141-142.
(6) Rando, D. G.; Sato, D. N.; Siqueira, L.; Malvezzi, A.; Leite, C.
Q. F.; do Amaral, A. T.; Ferreira, E. I.; Tavares, L. C. Potential
tuberculostatic agents. Topliss application on benzoic acid [(5-
nitro-thiophen-2-yl)-methylene]-hydrazide series. Bioorg. Med.
Chem. 2002, 10, 557-560.
(7) Eisa, H. M.; Tantawy, A. S.; El-Kerdawy, M. M. Synthesis of
certain 2-aminoadamantane derivatives as potential antimicrobial
agents. Pharmazie 1991, 46, 182-184.
(8) Parmar, S. S.; Gupta, A. K.; Gupta, T. K.; Stenberg, V. I.
Synthesis of substituted benzylidinohydrazines and their monoam-
ine oxidase inhibitory and anticonvulsant properties. J. Pharm.
Sci. 1975, 64, 154-157.
Figure 7. Contribution of fragments of compounds belonging to the third
set. Compounds 4a e.
−
(9) Thu-Cuc, N. T.; Buu-Hoy, N. P.; Xuong, N. D. Potential
antifungal benzohydrazides. Med. Pharm. Chem. 1961, 3, 361.
(10) Akagi, T.; Mitani, S.; Komyoji, T.; Nagatani, K. Quantitative
structure-activity relationships of fluazinam and related fungicidal
N-phenylpyridinamines. Preventive activity against Botrytis
cinerea. Nippon Noyaku Gakkaishi 1995, 20 (3), 279-290.
(11) Akagi, T.; Mitani, S.; Ito, K.; Shigehara, I.; Komyoji, T.; Matsuo,
N. Structure-activity relationships of pyridylcarbamates active
against both benzimidazole-sensitive and -resistant isolates of
Botrytis cinerea. Pestic. Sci. 1995, 44 (1), 39-48.
(12) Humeres, E.; Cantos, G. A.; Debacher, N. A.; Nunes, R. J.
Hydrophobicity constant of the dithiocarbamic fragment and
QSAR of dithiocarbamate fungicides. Org. React. 1997, 31 (1),
45-49.
(13) Kataoka, T. QSAR of 1-N-substituted azoles active against
Botrytis cinerea. In Rational Approaches to Structure, ActiVity
and Ecotoxicology of Agrochemicals; Draber, W., Fujita, T., Eds.;
CRC: Boca Raton, FL, 1992; pp 465-482.
(14) Gonza´lez-D´ıaz, H.; Prado-Prado, F. J.; Santana, L.; Uriarte, E.
Unify QSAR approach to antimicrobials. Part 1: Predicting
antifungal activity against different species. Bioorg. Med. Chem.
2006, 14, 5973-5990.
determined active place in the fungus, thus facilitating the latter’s
biotransformation and hence their antifungal effect due to a
degree of rigidity in the molecules.
Analysis of the last set of compounds supports the previous
explanation concerning aromatic ring contribution to biological
properties. It shows the important role of hydrophobicity in the
activity of these molecules due to the high contribution of the
piperidine ring. This influence requires the presence of a
hydrophobic substructure to enable the compound to cross the
lipophobic or lipidic membranes of plants. They could therefore
be considered more bioavailable in the exertion of their
antifungal function.
A behavior similar to that observed in the previous sets is
also observed when a substituent is introduced in the ortho
position in the aromatic ring. The contribution of this substituent
increases according to its size, and although in this case only a
moderate increase in activity is achieved, it shows the impor-
tance of a second aromatic ring linked to the N of the hydrazine
(Figure 7).

Benzohydrazides against Botrytis cinerea
J. Agric. Food Chem., Vol. 55, No. 13, 2007 5179
(15) Ara´n, V. J.; Asensio, J. L.; Ruiz, J. R.; Stud, M. The heterocy-
clization of N′,N′-disubstituted 2-halobenzohydrazides to 1,1-
disubstituted indazol-3-ylio oxides. J. Chem. Res. 1993, 1322-
1345.
(16) Noto, R.; Lamartina, L.; Arnone, C.; Spinelli, D. An analysis of
carbon-13 NMR substituent chemical shifts in 3-substituted
thiophene-2-carboxylic and 2-substituted benzoic acids by linear
free energy relationships. J. Chem. Soc. Perkin Trans. 1988, 2,
887-892.
(17) Ho¨nig, H. Improved 13C NMR shift prediction program for
polysubstituted benzenes and sterically defined cyclohexane
derivatives. Magn. Reson. Chem. 1996, 34, 395-406.
(18) Katritzky, A. R.; Rao, M. S. C. Chemistry of benzotriazole.
Preparation of 1,1-disubstituted hydrazines and their 2-acyl
derivatives. J. Chem. Soc. Perkin Trans. 1989, 1, 2297-2303.
(19) Katritzky, A. R.; Fan, W. Q. The chemistry of benzotriazole. A
novel and versatile synthesis of 1-alkyl-, 1-aryl-, 1-(alkylamino)-,
or 1-amido-substituted and of 1,2,6-trisubstituted piperidines from
glutaraldehyde and primary amines or monosubstituted hydra-
zines. J. Org. Chem. 1990, 55, 3205-3209.
(20) Yeung, J. M.; Knaus, E. E. Synthesis of 3,6-dihydro-1(2H)-
pyridinyl derivatives with hyperglycemic activity. Eur. J. Med.
Chem.sChim. Ther. 1986, 21, 181-185.
(21) Estrada, E. Spectral moments of the edge-adjacency matrix of
molecular graphs. 2. Molecules containing heteroatoms and
QSAR applications. J. Chem. Inf. Comput. Sci. 1997, 37, 320-
328.
(22) Estrada, E. Spectral moments of the edge adjacency matrix in
molecular graphs. 1. Definition and applications to the prediction
of physical properties of alkanes. J. Chem. Inf. Comput. Sci.
1996, 36, 844-849.
(23) Estrada, E.; Uriarte, E.; Gutierrez, Y.; Gonza´lez, H. Quantitative
structure-toxicity relationships using TOPS-MODE. 3. Structural
factors influencing the permeability of commercial solvents
through living human skin. SAR QSAR EnViron. Res. 2003, 14
(2), 145-163.
relationships. SAR QSAR EnViron. Res. 2000, 11 (3-4), 189-
209.
(29) Klein, D. J.; Randic´, M.; Babic´, D.; Lucˇic´, B.; Nikolic´, S.;
Trinajstic´, N. Hierarchical orthogonalization of descriptors. Int.
J. Quantum Chem. 1991, 63 (1), 215-222.
(30) Randic´, M. Correlation of enthalphy of octanes with orthogonal
connectivity indices. J. Mol. Struct. (Theochem) 1991, 233, 45-
59.
(31) Randic´, M. Orthogonal molecular descriptors. New J. Chem.
1991, 15 (7), 517-525.
(32) Randic´, M. Resolution of ambiguities in structure-property studies
by use of orthogonal descriptors. J. Chem. Inf. Comput. Sci. 1991,
31, 311-320.
(33) Lucˇic´, B.; Nikolic´, S.; Trinajstic´, N.; Juric´, D. The structure-
property models can be improved using the orthogonalized
descriptors. J. Chem. Inf. Comput. Sci. 1995, 35, 532-538.
(34) Hawkins, D. M. The problem of overfitting. J. Chem. Inf.
Comput. Sci. 2004, 44 (1), 1-12.
(35) Estrada, E.; Patlewicz, G.; Gutierrez, Y. From knowledge
generation to knowledge archive. A general strategy using TOPS-
MODE with DEREK to formulate new alerts for skin sensitiza-
tion. J. Chem. Inf. Comput. Sci. 2004, 44 (2), 688-98.
(36) Gonza´lez, M. P.; Teran, C.; Teijeira, M. A topological function
based on spectral moments for predicting affinity toward A3
adenosine receptors. Bioorg. Med. Chem. Lett. 2006, 16, 1291-
1296.
(37) Anthoni, U.; Larsen, C.; Nielsen, P. H. Derivatives of hydrazine.
V. Cis-trans isomerism, dipolar structure, and magnetic non-
equivalence of the isopropyl methyl 1H NMR signals of acyl
and thioacyl N,N-diisopropylhydrazines. Acta Chem. Scand.
1969, 23, 3513-3524.
(38) Walter, W.; Reubke, K. J. Structure of N,N-dialkylhydrazides.
Chem. Ber. 1970, 103, 2197-2207.
(39) Dewar, M. J. S.; Jennings, W. B. Conformational interchange
in acyclic hydrazines. J. Am. Chem. Soc. 1973, 95, 1562-1569.
(40) Golbraikh, A.; Tropsha, A. Beware of q2! J. Mol. Graphics
Modell. 2002, 20 (4), 269-276.
(41) Gonzalez-D´ıaz, H.; Vilar, S.; Santana, L.; Uriarte, E. On the
applicability of QSAR for recognition of miRNA bioorganic
structures at early stages of organism and cell development:
Embryo and stem cells. Bioorg. Med. Chem. 2007, 15, 2544-
2550.
(24) Gutierrez, Y.; Estrada, E. MODESLAB 1.0; MOlecular DEScrip-
tors LABoratory for Windows, 2002, http://www.modeslab.com.
(25) Statsoft, I. STATISTICA (Data Analysis Software System), version
6.0; Statistica: 2002.
(26) Vedani, A.; Dobler, M. Multi-dimensional QSAR in drug
research. Predicting binding affinities, toxicity and pharmaco-
kinetic parameters. Prog. Drug. Res. 2000, 55, 105-35.
(27) Tropsha, A.; Zheng, W. Identification of the descriptor phar-
macophores using variable selection QSAR: Applications to
database mining. Curr. Pharm. Des. 2001, 7 (7), 599-612.
(28) Hasegawa, K.; Funatsu, K. Partial least squares modeling and
genetic algorithm optimization in quantitative structure-activity
Received for review February 13, 2007. Revised manuscript received
April 12, 2007. Accepted April 12, 2007. Financial support from the
Spanish Science and Technology Ministery through project AGL2006-
13401-C02-01 is gratefully acknowledged.
JF0704211