Synthesis, biological evaluation and SAR of 3-benzoates of ingenol for treatment of actinic keratosis and non-melanoma skin cancer

Article abstract of DOI:10.1016/j.bmcl.2013.11.073

Ingenol 3-benzoates were investigated with respect to chemical stability, pro-inflammatory effects, cell death induction and PKCδ activation. A correlation between structure, chemical stability and biological activity was found and compared to ingenol mebutate (ingenol 3-angelate) used for field treatment of actinic keratosis. We also provided further support for involvement of PKCδ for induction of oxidative burst and cytokine release. Molecular modeling and dynamics calculations corroborated the essential interactions between key compounds and C1 domain of PKCδ.

Full text of DOI:10.1016/j.bmcl.2013.11.073

Bioorganic & Medicinal Chemistry Letters 24 (2014) 54–60
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Synthesis, biological evaluation and SAR of 3-benzoates of ingenol
for treatment of actinic keratosis and non-melanoma skin cancer
Gunnar Grue-Sørensen a, Xifu Liang a, Kristoffer Månsson a, Per Vedsø a, Morten Dahl Sørensen a, Anke Soor a,
Martin Stahlhut b, Malene Bertelsen b, Karen Margrethe Engell c, Thomas Högberg a,
⇑
a Chemical Research, LEO Pharma A/S, 55 Industripaken, DK-2750 Ballerup, Denmark
b Biological Research, LEO Pharma A/S, 55 Industriparken, DK-2750 Ballerup, Denmark
c Product Reprofiling, LEO Pharma A/S, 55 Industriparken, DK-2750 Ballerup, Denmark
a r t i c l e i n f o
a b s t r a c t
Article history:
Ingenol 3-benzoates were investigated with respect to chemical stability, pro-inflammatory effects, cell
death induction and PKCd activation. A correlation between structure, chemical stability and biological
activity was found and compared to ingenol mebutate (ingenol 3-angelate) used for field treatment of
actinic keratosis. We also provided further support for involvement of PKCd for induction of oxidative
burst and cytokine release. Molecular modeling and dynamics calculations corroborated the essential
interactions between key compounds and C1 domain of PKCd.
Received 23 October 2013
Revised 24 November 2013
Accepted 29 November 2013
Available online 4 December 2013
Keywords:
Ó 2013 Elsevier Ltd. All rights reserved.
Ingenol 3-benzoates
PKCd activation
Oxidative burst
Cytokine release
Cell death induction
Molecular modeling
Ingenol mebutate (PEP005, ingenol 3-angelate, 1), the active
ingredient in PicatoÒ, a new drug for field treatment of actinic (so-
lar) keratosis (AK),1 is isolated from Euphorbia peplus plants
(Fig. 1).1,2 Ingenol mebutate and other ingenol 3-acylates are po-
tent activators of protein kinase C (PKC) responsible for a number
of essential cellular signal transduction responses.3,4 The clinical
efficacy of ingenol mebutate against actinic keratosis is believed
to be linked to a dual mechanism of action, that is (i) induction
of keratinocyte cell death and (ii) induction of a lesion-directed im-
mune response, at least partially mediated by PKC.5 This profile ac-
counts for high clearance rates after only 2–3 days of topical field
treatment of AK lesions with a predictable onset and short duration
of local skin responses.1 In addition, patients treated with ingenol
mebutate gel also experience sustained clearance of AK lesions
one year later.6
For decades natural products have been a great source of inspira-
tion for the pharmaceutical industry and minor changes of complex
structures isolated from nature have led to a number of potentdrugs.7
In our search for novel analogues of ingenol mebutate with improved
properties, such as chemical stability, potency and efficacy towards
non-melanoma skin neoplasia such as actinic keratosis, basal cell car-
cinoma (BCC), squamous cell carcinoma (SCC) and Bowen’s disease,8
____________R groups__________
Me
H
H
Me
Ph
Me
H
O
H
H
3
5
ingenol mebutate (1)
H
O
Me
Me
O
Me
HO
HO
R
OH
Me
Me
Me
2
4
6
Figure 1. 3-Acyl ingenol derivatives 2–6 with favorable properties comparable to
ingenol mebutate (1).9
wehavepreviouslyinvestigatedtheconsequencesofminorstructural
changesoftheingenolscaffoldof1aswellasthereplacement ofthe3-
angelate group with other aliphatic acylates.9 The rigid ingenol scaf-
foldconveyseveralstructuralfeatures,includingthe5-and20-hydro-
xyl groups, essential for biological activity. Only a small number of
compounds 2–6 with comparable or slightly improved overall profile
over 1 were identified (Fig. 1). As a continuation of this work we now
report on the preparation and the chemical and biological character-
ization of novel ingenol 3-benzoates.10 We had also earlier provided
support for the involvement of PKC activation in the pro-inflamma-
tory effects such as release of cytokines and reactive oxygen species
(ROS).9 In this work we have extended this aspect further by showing
⇑
Corresponding author. Tel.: +45 44 94 58 88.
E-mail address: thomas.hogberg@leo-pharma.com (T. Högberg).
0960-894X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2013.11.073

G. Grue-Sørensen et al. / Bioorg. Med. Chem. Lett. 24 (2014) 54–60
55
H
correlations between these parameters as well as modeling and dy-
namic studies of key molecules interacting with PKCd.
O
H
O
O
O
O
O
O
The preparation of new 3-benzoyl derivatives of ingenol follows
previously described methods shown in Scheme 1.9,11,12 Ingenol
5,20-acetonide (7) was isolated after crystallization, and standard
procedures, such as Steglich coupling conditions or using acyl chlo-
rides, were applied in the preparation of I.9 As will become clear in
the following discussion we were introducing increasingly steri-
c
H
b
O
R'
N
R'
NH
NH
O
HO
O
R'
O
O
(IV)
(V)
(VI)
c
a
H
O
cally hindered acids which also called for use of microwave (lw)
H
d
O
O
O
assisted reactions in pressure vials at short reaction times. The final
acid catalyzed deprotection step to produce the benzoate esters II
was based on our modified procedure.9,12 For the preparation of
anthranilic acid esters VII, IX and X we first converted the anthra-
nilic acids III into the isatoic anhydrides IV, which were N-methyl-
ated to give V. The isatoic anhydrides IV and V were reacted with
H
H
R'
NH2
H
O
H
H2N
O
O
HO
NH
O
O
HO
HO
(III)
R'
OH
R'
ingenol 5,20-acetonide
7 to give VIII and VI, respectively
(VIII)
(VII)
e,d
f, d
(Scheme 2). Subsequent N-acylation with acyl chlorides or N-alkyl-
ation via reductive aminations followed by deprotection provided
the modified anthralinate esters VII, IX and X. Likewise, the prep-
aration of salicylic acid ester 32 required O-protection with an allyl
group and conversion to 2-O-allyl-6-methyl-salicylic chloride
which was reacted with ingenol 5,20-acetonide 7 as shown in
Scheme 3. Palladium catalyzed removal of the allyl group followed
by the normal acidic acetonide deprotection provided the salicy-
late 32.
H
H
O
O
H
H
H
O
H
O
O
R''
NH
NH
R'''
O
HO
HO
O
HO
OH
HO
OH
R'
R'
The biological activity profile was evaluated as previously
described.9,13 The induction of necrotic cell death was determined
in HeLa cells. The activation of the immune response was mea-
sured as induction of oxidative burst (i.e., generation of ROS) in
polymorphonuclear leukocytes and the release of the cytokines
(X)
(IX)
Scheme 2. General synthetic route from ingenol 5,20-acetonide to ingenol 3-(2-
amino)benzoates VII, IX and X. Reagents and conditions: (a) triphosgen (1.1 equiv),
THF/DCM, rt; (b) (MeO)2SO2, K2CO3, DMF, rt; (c) ingenol 5,20-acetonide 7, DMAP,
MeCN, 160 °C (lw); (d) aq HCl/THF; or aq HCl/MeOH, rt; (e) aldehyde, NaBH(OAc)3,
IL8 and TNF
a in human primary keratinocytes. Furthermore, the
AcOH, DCM, rt; (f) acyl chloride, K2CO3, MeCN, rt.
activation of PKCd was determined for several compounds to
explore the correlation to ROS and cytokine release potency.14
Chemical stability is essential to ensure sufficient shelf-life of
drugs. Ingenol mebutate (1) is most stable at pH 2.0–4.5 and, as
previously described, the 3-angelate undergoes acyl migration in
aqueous solution to form an equilibrium mixture between the 3-,
5- and 20-mono-angelates in ratios of about 1:1:18 with no imme-
diate hydrolysis.9 The 5-hydroxyl group is critical for this
pH-dependent degradation mechanism,15 but this group is also a
prerequisite for biological activity.9 Bulkiness (steric hindrance)
of the 3O-acyl groups has been shown to be an important factor
for stability.9 In order to quickly rank the stability of new ingenol
3-benzoates we determined the recovery at pH 7.4 after 16 h at
room temperature. Under these conditions 60% ingenol mebutate
was recovered. Acyl groups where the corresponding acids are
weaker than angelic acid (pKa 4.3)16 are also thought to be less
prone to acyl migration, but our previous study did not indicate
this factor to be of major significance.9 For example the less acidic
isomeric tiglic acid (pKa 5.0) ester was less stable (45% recovered)
than the angelate ester and the less acidic cyclohexane carboxylic
acid (pKa 4.9) ester was found to be equally stable as the benzoic
acid (pKa 4.2) ester (30–35%, recovered).9,16 Hence, we focused
mainly on steric hindrance rather than thermodynamic acidity fac-
tors to influence the acyl migration and gain stability.
H
H
H
Ingenol 3-benzoate (10)11 was less stable (30% recovery at our
standard conditions) than ingenol 3-angelate (1) (Table 1). It can
be noted that pKa of the corresponding acids, benzoic and angelic
acid, are almost identical (4.2 and 4.3, respectively).16 The benzo-
ate had slightly higher oxidative burst potency, whereas the cyto-
kine release potency was lower. This lower potency may be caused
by the disappearance of the active 3-ester via acyl migration dur-
ing the assay period as seen with other unstable 3-acylates.9 The
3-benzoate 10 has previously been subject to molecular modeling
studies with the crystal structure of the zinc finger domain C1 of
PKCd determined for a phorbol ester complex.17 This domain is in-
volved in the PKC activation induced by diacylglycerol (DAG) to re-
lease the blocking pseudosubstrate from the catalytic site, an
action shared with phorbol esters.18 Several essential hydrogen
bond interactions such as with the 5- and 20-hydroxyl groups,
which we have shown to play critical roles for the activity of 1,9
could be accounted for. Furthermore, the ester carbonyl oxygen
was indicated to be engaged as a hydrogen bonding acceptor to
the amide NH in Gly23 in the back-bone of the C1 domain of
PKCd.17 To support the binding hypothesis we have prepared the
O
O
O
H
H
H
H
a
H
b
H
Ar
O
HO
HO
O
HO
HO
HO
HO
O
O
OH
O
O
7
(I)
ingenol
c
d
H
H
H
O
O
O
H
H
H
d
H
H
H
Ar
O
Ph
HO
O
O
HO
HO
Ph
HO
HO
O
O
OH
OH
O
9
8
(II)
Scheme 1. General synthetic route from ingenol to ingenol 3-benzoates II and
benzyl ether 9. Reagents and conditions: (a) acetone, PTSA (cat), rt, 0.5 h, 73%; (b)
ArCO2H, DCC, DMAP, MeCN, rt ? 140 °C (
rt ? 140 °C ( w); (c) benzyl bromide (5 equiv), Cs2CO3 (6 equiv), MeCN, 150 °C
w), 5 min. 62%; (d) aq HCl/THF; or aq HCl/MeOH, rt.
lw); or ArCOCl, DMAP, DIPEA, MeCN,
l
(l

56
G. Grue-Sørensen et al. / Bioorg. Med. Chem. Lett. 24 (2014) 54–60
H
O
O
H
OH
O
O
O
H
a, b,c
d
O
O
O
Cl
O
HO
O
H
O
H
O
H
O
H
e
H
f
O
H
HO
O
H O
O
HO
O
HO
HO
O
OH
32
Scheme 3. Synthesis of salicylate 32. Reagents and conditions: (a) allyl bromide (3 equiv), K2CO3, acetone, 90 °C (pressure tube), 3 days; (b) NaOH (2.3 equiv), MeOH/H2O
(3:1), 100 °C (pressure tube), 5 h, 80 °C, 14 h; (c) SOCl2, reflux, 1 h; (d) acyl chloride to ingenol 5,20-acetonide 7 (2:1), DMAP, DIPEA (2 equiv), MeCN, 100 °C (pressure tube),
1 h; (e) Pd(Ph3P)4 (0.034 equiv), Et2NH, dioxane, rt, 1 h; (f) aq HCl/MeOH, rt.
benzyl ether 9 by reacting the 5,20-acetonide 7 with benzyl bro-
mide and cesium carbonate as base catalyst to provide
A series of mono-substituted benzoates (Table 1) showed a
clear trend that more bulkiness in the ortho-position favored sta-
bility. 3-Ingenol esters of ortho-mono-substituted benzoic acids
were more stable than the unsubstituted benzoate 10 with recov-
eries in our standard test up to 80%. The lipophilic ingenol 3-o-
hexadecanylamino-benzoate 23 had a relatively high recovery of
90%, but this measurement could not be carried out under our
usual conditions due to the very poor solubility of the substance
in aqueous buffer. We assume that highly lipophilic ingenol 3-acy-
lates will be more stable towards acyl migration because the lipo-
philic acyl group will shield the ester group from the polar catalytic
action of water and its dissociation products (H+ and OHÀ) (cf. Ref.
15). As stated above, the influence of acidity of the corresponding
acid is subordinate to steric hindrance as the ortho-methoxy ben-
zoic acid (pKa 4.1) ester 14 is considerably more stable than ben-
zoic acid (pKa 4.2) ester 10 even if the ortho bromo benzoic acid
(pKa 2.9) ester 11 has intermediate stability (Table 1). Hence, ki-
netic data are likely to reflect the migration ability better.
We explored a diverse set of ortho-substituted benzoates 11–23
containing ‘neutral’ (Me, i-Pr, Ph), electron-withdrawing (Br) and
electron-donating groups of highly variable size (MeO, i-PrO, and
H2N, MeHN versus PhO, PhNH, BnHN, C12H25HN and C16H33HN).
The anilinic compounds are also capable of forming internal hydro-
gen bonds with the ester carbonyl group which will influence the
orientation of the aniline substituent and favoring a coplanar con-
formation and affecting the ability of the carbonyl (benzoate
group) to engage in the postulated interaction in PKCd. Notably,
the potency of oxidative burst activation did not change signifi-
cantly within the series of the ortho-mono-substituted derivatives
compared to the potency of ingenol 3-benzoate 10 (Table 1). Only
ingenol 3-o-methoxy-benzoate (14) appeared to be somewhat less
potent in the cytokine release assay, contrasting to the slightly
more bulky o-isopropoxy derivative 15 with a 10-fold higher po-
tency. The methoxy derivative also displayed the lowest PKCd acti-
vation potency (25.7 nM) of the investigated ortho-mono-
substituted benzoates (Table 1). Otherwise, most of the o-mono-
substituted benzoates tested were superior, i.e., 11, 13, 15–21, to
ingenol mebutate. In terms of chemical stability, several com-
pounds had somewhat improved stability over the angelate 1, with
the phenyl 16 and phenoxy 17 being the most stable with ꢀ80%
recovered (Table 1). The most lipophilic aniline 23 had good stabil-
ity, but was an outlier with poor potency in the oxidative burst and
cytokine release assays. Poor solubility of this compound made
determination of necrotic potential impossible.
8
(Scheme 1). The usual deprotection protocol produced the benzyl
ether 9 in an overall yield of 40% from 7. As expected, it showed
complete chemical stability during the screening conditions and
it displayed very low activity in the oxidative burst and IL8 release
assays (Table 1). In accordance with these observations no activa-
tion of PKCd was observed even at a concentration of 10 lM of 9 in
contrast to the benzoate ester 10 having an EC50 of 6.1 nM (Table 1),
which supports the reported binding predictions and the impor-
tance of the ester carbonyl for activity.
To further investigate the binding of the 3-benzoate 10 to PKCd
we performed 2 ns molecular dynamics (MD) calculations of both
the 3-benzoate and the 3-benzyl ether.19,21 The MD starting con-
formation of 3-benzoate was obtained by docking the ligand into
the crystal structure of the C1 domain of PKCd (1PTR)20,21 and it
is in excellent agreement with the binding model proposed by
Pak et al.17 Thus, hydrogen bonds were observed between the hy-
droxyl groups of the ligand and Thr12, Leu21, and Gly23 in PKCd
and of particular interest between the ester carbonyl in 10 and
the Gly23 NH. Furthermore, our docking model suggests a hydro-
phobic interaction between the side chain of Leu24 and the aro-
matic ring of the benzoate group (Fig. 2).
Our docking model did not indicate an interaction between the
benzoate group and the side chain of Trp22 (Fig. 2a). However, an
analysis of the MD trajectories strongly suggests that a rotation of
the Trp22 side chain allows the aromatic ring of the benzoate
group to form a stable hydrophobic sandwich with the side chains
of Trp22 and Leu24 (Fig. 2b). It is also noteworthy, that the car-
bonyl and the phenyl ring of the benzoate is in a coplanar orienta-
tion in this hydrophobic sandwich enabling favorable
with the indole ring of Trp22 (Fig. 2c).
p–p stacking
Similarly, we performed a MD calculation for the benzyl ether 9,
using the benzoate 10 starting conformation but with the carbonyl
reduced to a methylene.19 Also in the case of the benzyl ether 9,
Trp22 rotates to form a hydrophobic sandwich, but the p–p stack-
ing is not as stable as observed for the benzoate as the spread in
torsion angles of the phenyl ring is in the range of ꢀ70° compared
to a more narrow span of ꢀ40° for the ester 10 (Fig. 3). The major
difference between the compounds is the lack of a hydrogen bond
to Gly23 NH supporting the importance of the ester carbonyl for
activity. Thus, we concentrated on the benzoate esters even if sta-
bility needed to be substantially improved. It can be noted that the
benzyl ether 9 showed a reasonable necrotic effect, which indicates
different mode of actions for immunostimulatory and necrotic
effects.
We then added extra steric hindrance (cf. Ref. 22) to the benzo-
ates with the introduction of two flanking ortho-substituents.

G. Grue-Sørensen et al. / Bioorg. Med. Chem. Lett. 24 (2014) 54–60
57
Table 1
Chemical stability and biological activity of ingenol 3-benzoates compared to ingenol mebutate (1) and benzyl ether 9
H
O
H
H
O
X
O
HO
HO
OH
No.
X
Stabilitya
Oxidative burstb
EC50 (nM) Emax (%)
TNF
a
releasec
IL8 releasec
Necrosisd
PKCde
% Recovered
EC50 (nM)
Emax (%)
EC50 (nM)
Emax (%)
LC50
(lM)
EC50 (nM)
Angelate 1
Benzyl 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
NA
NA
H
2-Br
2-Me
2-iPr
2-MeO
2-iPrO
2-Ph
2-PhO
2-PhNH
2-NH2
2-MeNH
2-BnNH
2-C12H25NH
2-C16H33NH
2,6-Me2
60
>95
30
44
73
70
70
65
77
80
58
58
68
8.7
4540
6.0
4.3
5.7
11
6.8
8.7
16
9.4
11
15
6.7
8.6
25
1340
13
113
—
110
99
11.2
Nd
88
8.2
13
98
—
60
96
64
10.3
4520
182
2.3
15
95
58
54
93
58
230
378
Nd
141
115
110
220
192
77
4.1
>10,000
6.1
3.8
Nd
Nd
25.7
Nd
3.3
13.1
4.2
107
120
107
113
105
98
129
116
113
127
78
5.7
38
96
4.4
42
86
100
101
91
100
105
87
4.0
1.6
1.4
5.0
5.5
1.0
1.5
14
2.7
1.5
1.1
3.7
3.4
0.79
1.2
16
86
87
83
120
112
134
128
150
120
61
113
42
52
103
101
99
115
111
125
66
107
45
52
118
111
104
104
105
127
105
145
122
117
92
307
199
95
168
400
Nd
97
400
59
Nd
2.5
Nd
Nd
65
Ndf
90
94
200
83
61
62
Nd
>95
>95
>95
>95
>95
>95
93
123
116
123
130
117
109
161
133
125
136
124
123
103
112
113
116
10.8
156
19.8
Nd
8.0
Nd
Nd
Nd
13.4
Nd
Nd
Nd
Nd
Nd
988
3.7
2,6-MeO2
2,6-Cl2
84
21
57
44
278g
54
113g
51
2-Me, 6-Cl
2-Me, 6-NH2
2-MeO, 6-NH2.
2-Cl, 6-NH2
2-F, 6-NH2
2-Me, 6-OH
2-Me, 6-MeNH
2-Cl, 6-MeNH
2-F, 6-MeNH
2-Me, 6-BnNH
2-MeO, 6-BnNH
2-Me, 6-AcNH
1-Naphthoyl
6.6
8.6
25
7.1
12
9.4
46
21
9.3
42
23
538
11
126
146
Nd
12
108
28
10
98
108
83
8.2
41
20
8.6
71
12
108
104
156
143
97
174
126
120
92
185
201
199
105
145
152
187
400
Nd
66
>95
>95
>95
74
>95
>95
82
16
21
18
4.1
7.6
19
632
3.3
3.0
3.8
6.6
534
2.3
42
119
97
87
a
The chemical stability over 16 h was evaluated in an aqueous buffer with less than 30% organic solvent at pH 7.4. Reported as % recovered material.25
PMN respiratory burst after 40 min incubation of test compound was quantified by measuring fluorescence expressed in relative light units.26 EC50 denotes the test
b
compound concentration producing 50% of the maximum effect given by 1. Emax indicates the maximal response in relation to 1.
c
Cytokine secretions were measured in human adult log-phase primary epidermal keratinocytes after incubation of test compound for 6 h at 37 °C.27 The EC50 and Emax
calculations according oxidative burst protocol.
d
HeLa cells were treated with test compounds for 30 min at 37 °C and then measured remaining the metabolic activity. LC50 denotes concentration giving 50% loss of
metabolic activity.28
Activation potency of PKCd was measured by the use of an in vitro phosphorylation assay using a PKC-peptide substrate (40 min, room temperature).29
Nd: not determined.
Concentrations yielding maximum responses are given.
e
f
g
ortho-Methyl, o-methoxy, o-chloro and o-amino substituents
provided enough steric bulkiness, provided both ortho-positions
were substituted as in compounds 24–30, to give stable ingenol
3-benzoates with ‘complete’ stability, that is >95% recovery under
our assay conditions (Table 1). The importance of substituting
both ortho-positions in order to gain stability is clearly shown
when comparing the two isomers ingenol 3-(2-methylamino-
benzoate)23 (20) and 3-(2-amino-6-methyl-benzoate) (28). Intro-
duction of o-hydroxy as another group than o-amino capable of
internal hydrogen bonding provides a compound that is compara-
ble on oxidative burst but 10-fold less active on cytokine release
(32 vs 28).
As expected, introduction of an o-fluor did not contribute signif-
icantly to the stability, and cytokine release potency dropped
slightly compared to the hydrogen analogues (31 vs 19 and 35 vs
20). Notably, an even larger drop in cytokine release potency is
seen for the corresponding o-chloro derivatives (30 and 34). The
2-naphtoate ester (39) did not provide enough steric hindrance
to give a stable ester (42% recovery), albeit cytokine release po-
tency was superior to that of ingenol mebutate and the unsubsti-
tuted benzoate 10.
The o,o-disubstituted benzoates were, in general, more stable
than the o-mono-substituted benzoates, but this increase in stabil-
ity was apparently associated with a decrease in oxidative burst
and cytokine release potency. The most promising disubstituted
benzoates with respect to cytokine release potency contained an
o-amino group, either unsubstituted (28, 31) or N-mono-substi-
tuted (33–37), of which the 2-fluoro-6-methylamino-benzoate 35
and 2-methyl-6-benzylamino-benzoate 37 were slightly superior
to ingenol mebutate in the cytokine release. As observed for the
mono-substituted benzoates an o-methoxy substituent had a neg-
ative effect on potency compared to o-methyl (29 vs 28 and 37 vs
36), albeit the difference is smaller for the larger benzylamine
compounds than for the primary anilines.

58
G. Grue-Sørensen et al. / Bioorg. Med. Chem. Lett. 24 (2014) 54–60
Figure 3. Analysis of the co-planarity of the ligand phenyl ring and the indole ring
of Trp22. (a) Ring plane angles from the MD calculation of ingenol-3-benzoate 10
bound to PKCd. (b) Ring plane angles from the MD calculation of ingenol-3-benzyl
ether 9 bound to the C1 domain of PKCd.19,21
preferred coplanar orientation of the benzoate ring and the car-
bonyl for optimal effects on ROS and cytokine release. This led us
to speculate that the favorable combination of stability and biolog-
ical effects seen with some 6-substituted 2-anilinic derivatives,
such as 28, is driven by a closer coplanar arrangement via the
intramolecular hydrogen-bond tendency. Quantum mechanical
(QM) optimization24 of different 2,6-disubstituted benzoate sys-
tems confirmed that the 2-anilinic substituent does indeed favor
a co-planar orientation whereas for example 2,6-dimethyl substi-
tutions result in a twisted orientation as shown in Figure 4.
To further explore this relationship, the PKCd activity was mea-
sured for a number of compounds (Table 1) and the correlations
between oxidative burst and cytokine release on the one side
and PKCd activation on the other side were investigated (Fig. 5). Ta-
ken the considerations into account that the data set is somewhat
skewed containing mostly potent compounds the following con-
Figure 2. (a) Ingenol 3-benzoate 10 (cyan) docked into the crystal structure of the
zinc finger C1 domain of PKCd (1PTR). Residues Trp22 and Leu24 are shown in
orange. (b) Snapshot taken every 100 ps from a 2 ns MD calculation of 10 in PKCd.
(c) Binding interactions between the benzoate 10 and PKCd with the essential
residues.19–21 (The figure was made in Maestro, Schrödinger, LLC, New York, NY,
2013.)
Ingenol 3-(2-methyl-6-methylamino)-benzoate (33) is moder-
ately active in the cytokine release assays, but insertion of a car-
bonyl group in the methylamino moiety to give an acetamido
group (38) had a strong negative effect on potency. The acetamido
compound 38 also showed a very modest PKCd activation with an
clusions can be reached. As expected
a strong correlation
(r2 = 0.999) was found between PKCd activation and induction of
oxidative burst (Fig. 5a).14 The potency for release of the two cyto-
kines IL8 and TNF
only a moderate correlation was reached between PKCd and IL8/
TNF release as the significance was driven by one or two com-
a were also very closely correlated (Fig. 5b), but
EC50 of ꢀ1
lM (Table 1) supporting the link between PKC activa-
tion and ROS and cytokine release.
a
pounds (cf. IL8 data in Fig. 5c). Accordingly, a less impressive cor-
relation was found between IL8 and oxidative burst as the high
correlation coefficient (r2 = 0.96) is once again biased by two com-
pounds as shown in Figure 5d. These differences may be related to
different cellular PKC profile and the fact that different ingenol
derivatives could display differences in their PKC selectivity. Be-
sides, the stability of the compounds could influence the cytokine
release potency as the assay is run for 6 h which might give rise to
a lower cytokine release potency than in the in vitro kinase assay
or in respiratory burst which are measured after 40 min, as high-
lighted with benzoate 10 indicated with an arrow in Figure 5c
By increasing the stability of ingenol 3-acylates by increasing
the steric hindrance of the acid, there is a priori a risk that the ben-
zoate carbonyl group and aromatic ring will be forced out of the
co-planar orientation which seems optimal for interaction with
the target protein as illustrated with PKCd (Fig. 2). In fact, as shown
above, there is a tendency for the bulky and stable 2,6-disubsti-
tuted benzoates towards lower potency, which indicates that the
fixation of the carbonyl group away from a favorable position for
acyl migration at the same time is sub-optimal for PKC activation.
In accordance with the MD calculations, our SAR data indicate a

G. Grue-Sørensen et al. / Bioorg. Med. Chem. Lett. 24 (2014) 54–60
59
and d, respectively. Even if we have mostly determined PKCd acti-
vation on relatively potent compounds, the compounds 25 and 38
support that poor oxidative burst and cytokine release potency re-
lates to poor PKCd activation.
For most compounds in Table 1, induction of cell death takes
place at concentrations of about 60–200 lM that are more than
four orders of magnitude higher than the concentrations of about
1–20 nM needed for PKCd activation (Table 1). All benzoate com-
pounds in Table 1 (with the exception of the methoxy-substituted
benzoates 25 and 37 and the lipophilic compounds 22 and 18) ap-
pear to have a higher necrotic potential than ingenol mebutate,
and this is probably an indication of a mechanism for the necrosis
that is not primarily linked to PKCd activation. The mechanism for
cell death induction is currently under investigation.5d
Figure 4. (a) Quantum mechanical (QM) optimization of a twisted conformation of
2-Me, 6-NH2 benzoate (28) results in co-planar orientation (cyan). (b) QM
optimization of a planar conformation of 2,6-Me2 benzoate (24) results in a twisted
orientation (cyan). The starting conformations are shown in orange.21,24 (The figure
was made in Maestro, Schrödinger, LLC, New York, NY, 2013).
a
We conclude that it is possible to prepare chemically stable
ingenol 3-acylates, albeit in general, with lower potency in ROS
and cytokine release. However, the potent monosubstituted o-phe-
nyl and o-phenoxy derivatives 16 and 17, respectively, also have
significantly improved chemical stability and 10-fold higher po-
tency on cytokine release compared to ingenol mebutate (1). Fur-
thermore, we found a narrow span of 2,6-substituted anilinic
derivatives, for example 28, 30, and 34, with an attractive profile
of very good stability and reasonable potency in ROS and cytokine
release as well as with a good necrotic potential. Our investigation
also provided strong support for a causative link between PKC acti-
vation and ROS and cytokine release in accordance with our previ-
ous SAR studies.9 These findings were corroborated by modeling
studies indicating a crucial role of the carbonyl oxygen as well as
a preferable relatively coplanar orientation of the carbonyl and
the phenyl ring of the benzoates when interacting with PKC.
Pharmacological in vivo experiments are ongoing to test the
utility of these molecules in the treatment of actinic keratosis
and possibly non-melanoma carcinomas (e.g., BCC, SCC and Bo-
wen’s disease)8 as they display pronounced effects on inducing ne-
crotic cell death combined with the desirable immunostimulatory
properties and improved chemical stability.
10000
a
1000
100
10
1
1
10
100
1000
10000
EC50 PKCdelta (M)
10000
1000
100
10
b
c
d
1
1
10
100
1000
10000
Acknowledgments
EC50 IL-8 release (M)
10000
1000
100
10
We acknowledge Thomas Vifian, Karin Hvidtfeldt Hansen, Svitl-
ana Tkach, Christa Marvig, Carsten Ryttersgaard, Nina Ravn Borch,
Ninette Winther Hansen, Nannette Svendsen for skillful assistance.
References and notes
1. (a) Lebwohl, M.; Swanson, N.; Anderson, L. L.; Melgaard, A.; Xu, Z.; Berman, B.
N. Engl. J. Med. 2012, 366, 1010; (b) Lebwohl, M.; Sohn, A. Expert Rev. Dermatol.
2012, 7, 121; (c) Berman, B. Clin. Cosmet. Invest. Dermatol. 2012, 5, 111; (d) Ali,
F. R.; Wlodek, C.; Lear, J. T. Dermatol. Ther. 2012, 2, 8; (e) Martin, G.; Swanson, N.
J. Am. Acad. Dermatol. 2013, 68, S39; (f) Keating, G. M. Drugs 2012, 72, 2397.
2. (a) Vasas, A.; Rédei, D.; Csupor, D.; Molnár, J.; Hohmann, J. Eur. J. Org. Chem.
2012, 5115; (b) Hohmann, J.; Evanics, F.; Berta, L.; Bartók, T. Planta Med. 2000,
66, 291.
3. (a) Winkler, J. D.; Hong, B.-C.; Bahador, A.; Kazanietz, M. G.; Blumberg, P. M.
Bioorg. Med. Chem. Lett. 1993, 3, 577; (b) Winkler, J. D.; Kim, S.; Harrison, S.;
Lewin, N. E.; Blumberg, P. M. J. Am. Chem. Soc. 1999, 121, 296.
4. (a) Bosco, R.; Melloni, E.; Celeghini, C.; Rimondi, E.; Vaccarezza, M.; Zauli, G.
Mini-Rev. Med. Chem. 2011, 11, 185; (b) Reyland, M. Front. Biosci. 2009, 14,
2386; (c) Steinberg, S. Physiol. Rev. 2008, 88, 1341.
5. (a) Hampson, P.; Kavanagh, D.; Smith, E.; Wang, K.; Lord, J. M.; Ed Rainger, G.
Cancer Immunol. Immunother. 2008, 57, 1241; (b) Ersvaer, E.; Kittang, A. O.;
Hampson, P.; Sand, K.; Gjertsen, B. T.; Lord, J. M.; Bruserud, Ø. Toxins 2010, 2,
174; (c) Rosen, R. H.; Gupta, A. K.; Tyring, S. K. J. Am. Acad. Dermatol. 2012, 3,
486; (d) Stahlhut, M.; Bertelsen, M.; Hoyer-Hansen, M.; Svendsen, N.; Eriksson,
A. H.; Lord, J. M.; Scheel-Toellner, D.; Young, S. P.; Zibert, J. R. J. Drugs Dermatol.
2012, 11, 1181.
1
1
1
10
100
1000
10000
EC50 PKCdelta (M)
0,1
10000
1000
100
10
1
1
10
100
1000
10000
EC50 IL-8 release (M)
6. Lebwohl, M.; Shumack, S.; Gold, L. S.; Melgaard, A.; Larsson, T.; Tyring, S. K.
JAMA Dermatol. 2013, 149, 666.
7. Newman, D. J.; Cragg, G. M. J. Nat. Prod. 2007, 70, 461.
Figure 5. (a) Plot of EC50’s for PKCd activation versus oxidative burst; (b) Plot of
EC50’s for release of IL8 versus release of TNF ; (c) Plot of EC50’s for PKCd activation
versus IL8 release; (d) Plot of EC50’s for IL8 release versus oxidative burst. Arrows
8. (a) Chen, A. C.; Halliday, G. M.; Damian, D. L. Pathology 2013, 45, 331; (b) Lee, E.
H.; Klassen, A. F.; Nehal, K. S.; Cano, S. J.; Waters, J.; Pusic, A. L. J. Am. Acad.
Dermatol. 2013, 69, e59.
a
indicate the unstable ingenol 3-benzoate (10).

60
G. Grue-Sørensen et al. / Bioorg. Med. Chem. Lett. 24 (2014) 54–60
9. Liang, X.; Grue-Sørensen, G.; Månsson, K.; Vedsø, P.; Soor, A.; Stahlhut, M.;
22. Nicholls, A. W.; Akira, K.; Lindon, J. C.; Farrant, R. D.; Wilson, I. D.; Harding, J.;
Killick, D. A.; Nicholson, J. K. Chem. Res. Toxicol. 1996, 9, 1414.
23. Mainieri, F.; Paganini, A.; Ech-Chahad, A.; Appendino, G. Nat. Prod. Commun.
2007, 2, 375.
Bertelsen, M.; Engell, K. M.; Högberg, T. Bioorg. Med. Chem. Lett. 2013, 23, 5624.
10. Grue-Sørensen, G.; Liang, X.; Högberg, T.; Månsson, K.; Vedsø, P. WO 2012/
083954 A1, 2012; PCT Int. Appl. 2012.
11. Sorg, B.; Hecker, E. Z. Naturforsch 1982, 37b, 748.
24. The quantum mechanical optimisations were performed with Jaguar. DFT was
used with default settings (B3LYP, 6–31⁄⁄). Solvent was set to Water (PBF
model). Jaguar was used as implemented in the 2012–2 software Suite
(Schrödinger, LLC, New York, NY, 2012).
12. Liang, X.; Grue-Sørensen, G.; Petersen, A. K.; Högberg, T. Synlett 2012, 2647.
13. Grue-Sørensen, G.; Liang, X.; Högberg, T. WO 2012/085189 A1, 2012; PCT Int.
Appl. 2012.
14. (a) Tauber, A. Blood 1987, 69, 711; (b) Lambeth, J. D. J. Bioenerg. Biomembr.
1988, 20, 709.
15. Bundgaard, H.; Hansen, J. Int. J. Pharm. 1981, 3, 197.
25. DMSO stock solution containing a known amount of compound was diluted
with pre-heated (37 °C) aqueous buffer pH 7.4 to an organic content 630% v/v.
After thorough shaking, the solution was placed in the HPLC auto sampler
(37°C) and injected within 5 min and then repeatedly injected over a period of
16 hours. Based on the decrease of area of the compound signal (UV detection
at normally 270 nm) the recovery of the compound over time was assessed.
26. Polymorphonuclear leukocytes (PMN) were isolated and purified from fresh
buffy coats and incubated for 40 min at ambient temperature with titrated test
16. Brown, H. C.; McDaniel, D. H.; Hafliger, O. Determ. Org. Struct. Phys. Methods
1955, 567–662. http://research.chem.psu.edu/brpgroup/pKa_compilation.pdf.
17. Pak, Y.; Enyedy, I. J.; Varady, J.; Kung, J. W.; Lorenzo, P. S.; Blumberg, P. M.;
Wang, S. J. Med. Chem. 2001, 44, 1690.
18. (a) Blumberg, P. M.; Kedei, N.; Lewin, N. E.; Yang, D.; Czifra, G.; Pu, Y.; Peach, M.
L.; Marquez, V. E. Curr. Drug Targets 2008, 9, 641; (b) Gennäs, G. B.; Talman, V.;
Yli-Kauhaluoma, J.; Tuominin, R. K.; Ekokoski, E. Curr. Top. Med. Chem. 2011, 11,
1370.
compound (highest test concentration 10 lM) pre-mixed with hydroethidine.
The PMN respiratory burst was quantified by measuring fluorescence
expressed in relative light units at 579 nm (excitation: 485 nm) using an
Envision plate reader. EC50 was calculated as the concentration of test
compound producing 50% of the maximum effect given by ingenol 3-angelate
(1). Emax indicates the maximal response in relation to 1.
19. The molecular dynamics (MD) calculations of PKCd bound to compounds 10
and
9 were performed with MacroModel (OPLS-2005, water, TNCG
minimization, 25 ps equilibrium followed by 2 ns simulation, 1.5 fs time
step). The ligand and all atoms in residues within 5 Å of the ligand were
27. Human adult log-phase primary primary epidermal keratinocytes were sub-
cultured in 96-well plates at 10,000 cells/well and incubated with titrated test
allowed to move freely, while residues in
a second shell of 5 Å were
constrained (force constant 200) and the remaining atoms frozen. The
docked pose (Fig. 2a) was used as the starting conformation for 10. The
starting conformation of the benzylether 9 was obtained by reducing the ester
compound for 6 h at 37 °C in humidified air/CO2 (95%/5%). Secreted TNFa levels
were quantified using Meso Scale Discovery 4-spot cytokine plates. Secreted
IL8 levels were quantified using Meso Scale Discovery 4-spot cytokine plates,
or based on homogeneous time-resolved fluorescence (Human IL-8 HTRFÒ kit,
CisBio). EC50 and Emax were calculated as described in Ref. 26.
carbonyl to
simulation.
a methylene group. 500 structures were sampled for each
20. The docking of ingenol-3-benzoate 10 to the 1PTR crystal structure of the PKCd
C1 domain was performed in two steps. First, low-energy conformations (less
than 21 kJ/mol) were obtained from a MacroModel conformational search
(OPLS-2005, Water, TNCG minimization, Mixed torsional/Low-mode sampling,
1000 conformations). Secondly, the resulting conformations were docked
rigidly with Glide XP to PKCd with no scaling of the protein atoms. Default
Glide settings were used for the ligand scaling and maximum 5 poses per
conformation were kept. The highest scoring docked pose is shown in Fig. 2a,
the MacroModel conformational energy of the ligand is 6.7 kJ/mol.
28. HeLa cells were treated with increasing concentrations of ingenol esters up to
400 lM for 30 min at 37°C. Subsequently, metabolic activity was quantified by
the use of the resazurin-based dye formulation PrestoBlue (Invitrogen). LC50
was calculated as the concentration of test compound producing 50% loss of
metabolic activity.
29. PKCd was tested in the KinaseProfiler™ assay (Millipore) using 0.05 mg/mL
phosphatidylserine in the absence of diacylglycerol to allow for observations of
any stimulatory effects. The compounds were serially diluted, at
semilogarithmic intervals, for twelve points starting from 10 lM. Assays
21. Glide and MacroModel Were Used as Implemented in the 2012-2 Software
Suite; Schrödinger, LLC: New York, NY, 2012.
were started by the addition of ATP and run for 40 min at room temperature.
Data points for the EC50 determinations were performed in duplicate.