Second basic pKa: An overlooked parameter in predicting phospholipidosis-inducing potential of diamines

Article abstract of DOI:10.1016/j.bmcl.2019.126933

In this paper, we present the phospholipidosis-inducing potential (PLIP) of forty fragment-sized diamines derived from N-benzyl-4-(methylamino)piperidine and discuss the relationship between their PLIP and the physicochemical properties. Our results demonstrate that the previously reported methods are not suitable for predicting the PLIP of fragment-sized diamines; the second basic pKa can distinguish PLIP-positive diamines from PLIP-negative diamines more accurately than ClogP or most basic pKa. To the best of our knowledge, this is the first report describing the relationship between PLIP and second basic pKa.

Full text of DOI:10.1016/j.bmcl.2019.126933

Journal Pre-proofs
Second basic pKa: An overlooked parameter in predicting phospholipidosis-
inducing potential of diamines
Hiroki Sakai, Hidekazu Inoue, Kenji Murata, Tetsuya Toba, Naohiro Takemoto,
Takahiro Matsumoto, Takeo Kawabata
PII:
S0960-894X(19)30916-3
DOI:
https://doi.org/10.1016/j.bmcl.2019.126933
BMCL 126933
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
8 October 2019
12 December 2019
23 December 2019
Please cite this article as: Sakai, H., Inoue, H., Murata, K., Toba, T., Takemoto, N., Matsumoto, T., Kawabata, T.,
Second basic pKa: An overlooked parameter in predicting phospholipidosis-inducing potential of diamines,
Bioorganic & Medicinal Chemistry Letters (2019), doi: https://doi.org/10.1016/j.bmcl.2019.126933
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will
undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing
this version to give early visibility of the article. Please note that, during the production process, errors may be
discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
© 2019 Elsevier Ltd. All rights reserved.

Second basic pKa: An overlooked parameter in predicting phospholipidosis-inducing potential of
diamines.
Hiroki Sakai,a,* Hidekazu Inoue,b Kenji Murata,a Tetsuya Toba,a Naohiro Takemoto,a Takahiro
Matsumoto,c and Takeo Kawabatad
a-c
Asubio Pharma Co., Ltd., 6-4-3 Minatojima-Minamimachi, Chuo-ku, Kobe, Hyogo 650-0047,
Japan
d Institute for Chemical Research, Kyoto University Uji, Kyoto 611- 0011, Japan
-------------------------------------------------------------
Present address:
a Daiichi Sankyo Co. Ltd, 1-2-58, Shinagawa R&D center, Hiromachi, Shinagawa-ku, Tokyo 140-
8710, Japan
b Daiichi Sankyo Co. Ltd, 1-12-1, Shinomiya, Hiratsuka-shi, Kanagawa, 254-0014, Japan
c Daiichi Sankyo Co. Ltd, 3-5-1, Head Office, Nihonbashi Honcho, Chuo-ku, Tokyo 103-8426, Japan
*Corresponding author. E-mail address: sakai.hiroki.xu@daiichisankyo.co.jp (H. Sakai)

Abstract
In this paper, we present the phospholipidosis-inducing potential (PLIP) of forty fragment-sized
diamines derived from N-benzyl-4-(methylamino)piperidine and discuss the relationship between
their PLIP and the physicochemical properties. Our results demonstrate that the previously reported
methods are not suitable for predicting the PLIP of fragment-sized diamines; the second basic pKa
can distinguish PLIP-positive diamines from PLIP-negative diamines more accurately than ClogP or
most basic pKa. To the best of our knowledge, this is the first report describing the relationship
between PLIP and second basic pKa.
Keywords: phospholipidosis, diamine, second basic pKa
We have identified a novel heterocyclic pyrimidine derivative SUN13837, which exhibits biological
activities similar to that of basic fibroblast growth factor (bFGF). This is possible because of the
modulation of the signal transduction pathway of the fibroblast growth factor receptor by the
pyrimidine derivative.1 Although SUN13837 exhibited significant pharmacological activities that
could aid the treatment of neurodegenerative diseases, in vitro examination showed that one of its
major metabolites, N-benzyl-4-(methylamino)piperidine (BMP, 1), had high phospholipidosis-
inducing potential (PLIP) (Figure 1). Phospholipidosis is a pathological condition in which
phospholipids accumulate in tissues, and this is characterized by the formation of lysosomal lamellar
bodies. The accumulated phospholipids interfere with cellular functions and may sometimes prove
fatal.2 In our efforts to develop the follow-up compounds of SUN13837, we realized that it is necessary
to identify the factors influencing the PLIP of BMP-like diamines, so that alternative diamines with
reduced PLIP can be developed.
H2N
N
H2N
Carboxylesterase
N
HN
N
N
O
O H
+
N
O
N
O
N
O
SUN13837
BMP (1)
Figure 1. Carboxylesterase-catalyzed hydrolysis of SUN13837.
A number of central nervous system (CNS) drugs, such as SUN13837, that can produce diamines
by hydrolysis have been reported (Figure 2). Therefore, we also considered that studying the PLIP of
fragment-sized diamines should help assess the risk of phospholipidosis caused by these drugs and
guide the selection and development of a drug candidate.

O
O
O
O
N
H
O
N
N
N
N
O
O
Bavisant
Trimethobenzamide
H3 receptor antagonist
D2 receptor antagonist
N
O
O
O
N
N
H
Cl
S
N
H
Cl
Encenicline
Moclobemide
α7-nAChR agonist
monoamine oxidase A inhibitor
Figure 2. CNS drugs that can produce diamines by hydrolysis
Physicochemical approaches are a useful way of predicting the PLIP of newly designed compounds,
and various methods have been reported for this purpose.3 One of the simplest models was described
by Ploemen et al., in which they used only two descriptors, ClogP and the most basic pKa, to represent
the lipophilicity and basicity, respectively, of a compound.4 Tomizawa et al. modified Ploemen’s
model to include the net charge at the lysosomal pH, instead of the most basic pKa, and suggested that
diamines are highly likely to exhibit high PLIP.5 It should, however, be noted that each prediction
model was constructed using different datasets, and hence, the application of these models to different
datasets may not give the same accuracy as originally reported.
In this study, we evaluated the PLIP of forty fragment-sized diamines derived from 1 and found that
there were diamines with low PLIP. Furthermore, a new criterion based on the second basic pKa was
introduced, which could distinguish PLIP-positive diamines from PLIP-negative diamines more
accurately than ClogP or most basic pKa.
According to the study conducted by Kasahara et al,6 the PLIP can be evaluated using normalized
value (NV), which indicates the amount of phosphatidylcholine-conjugated dye taken up per living
cell. NV is correlated to the degree of pathological changes and compounds with NV greater than 1.5
are likely to induce cytoplasmic lamellar bodies. With the aim of obtaining diamines with lower PLIP
even at higher concentrations, we measured NV at the concentration of 300 μM, which is ten times
higher than that in Kasahara’s study. Thus, in this study, high-PLIP diamines are defined as those with
NV greater than 1.5 at 300 μM.
We began our study by examining the PLIP of diamines 1–7,7,8 in order to understand the behavior
of the PLIP of BMP-like compounds and also to investigate the relationships between their PLIP and
physicochemical properties such as ClogP, pKa1, and pKa2 (Table 1).9 Furthermore, we examined the
scope of application of the previously reported prediction methods.10 The NV of 1 was more than 1.5,

indicating that 1 has high PLIP. The NV of the tert-butyl derivative 2 was lower than that of 1 but still
higher than 1.5. The NV of the cyclopropyl derivative 3 was as high as that of 1, despite its very low
lipophilicity. The NVs of 4 and 5, which are regioisomers of 1, were as high as that of 1. However,
the NVs of 3-aminopiperidine 6 and fused-ring piperidine 7 were less than 1.5, despite their high
lipophilicity. The PLIP of 3–7 observed in the in vitro assay could not be accurately predicted by
previously reported models, suggesting that these methods are not suitable for predicting the PLIP of
BMP-like diamines. Based on the above results, we hypothesized that pKa2 would be a more accurate
predictor of the PLIP, especially for fragment-sized diamines, because the pKa2 values of 6 and 7 were
lower than those of compounds 1–5. Thus, to verify our hypothesis, we examined other BMP-like
diamines which have low pKa2 values.
Table 1. Normalized values and physicochemical properties of BMP-like diamines.
Cmpd Structure
.
Clog pKa1 pKa2 Normali Prediction Prediction Prediction
d
d
P c
zed
by
by
by
value
Ploemen’s modified
Tomizawa
model
Ploemen’s ’s model
model
BMP
1.3
1.2
0.3
0.7
10.2
10.3
10.2
9.8
6.3
7.6
7.5
6.4
3.3
2.3
3.2
3.0
positive
positive
negative
negative
negative
negative
negative
negative
positive
positive
positive
positive
HN
N
(1)
2 a
HN
N
3 a
HN
N
4 a,b
HN
N
5 a,b
0.7
9.8
6.4
2.7
negative
negative
positive
HN
N
6
2.8
2.0
10.2
9.3
5.0
2.8
1.4
1.1
positive
positive
positive
positive
positive
positive
HN
N
7 a,b
HN
N
a 2 HCl salt.
b Racemate.

c Calculated using ChemBioDraw Ultra 12.0.
d Calculated using ADMET predictor, version 7.2.0001.
The pKa2 value of 1 corresponds to the basicity of the piperidyl nitrogen, because the benzyl group
acts as an electron-withdrawing group (EWG). Generally, introduction of EWGs in the vicinity of an
amine center or relocation of EWGs closer to an amine center reduces its basicity.11 Thus, we designed
and synthesized some diamines with reduced basicity of the piperidyl nitrogen, according to the
options shown in Figure 3.
Option 2 : Introduction of EWGs to the piperidine ring
R
R
R = F, O or H
Option 1 : Replacement of the benzene ring
with stronger EWGs
Option 3 : Replacement of the methyl group
with EWGs
X
X
X
X
X = N or C
Ro
Ro = O-containing alkyl group
X
H N
H N
Rf = F-containing alkyl group
Ro = O-containing alkyl group
Rf
N
R
O
R = Alkyl group
less basic amine
Ro
H N
Option 4 : Relocation of diamines within three sigma bonds
(1,3-diaminopropane moiety is depicted with thick line)
H
N
H
N
H
N
H N
N
N
N
N
Figure 3. Design of diamines with lower pKa2 values
Table 2 shows the NVs and physicochemical properties of the diamines in which EWGs stronger than
the phenyl group are introduced (Option 1 in Figure 3). It is worth noting that the pKa1 values of the
newly designed compounds 8–24 are almost similar to that of compound 1, while the pKa2 values are
different from each other. For the heteroaromatic derivatives such as pyridine (8, 9), pyrazine (10),
pyridazine (11), and pyrimidine (12, 13), the pKa2 values are lower than that of 1, and the NVs are
lower than 1.5. Introduction of alkyne functionality as an EWG (14, 15) also reduced the pKa2 values,
and the NVs were lower than 1.5. For the fluoroalkyl derivatives, the pKa2 value changes depending
on the number of fluorine atoms and the distance between fluorine and the piperidyl nitrogen atom.
At the γ-position of the piperidyl nitrogen (16, 17), at least two fluorine atoms were required to make
the pKa2 value lower than that of 1. The NV was lower than 1.5 only for 17. At the β-position of the
piperidyl nitrogen (18-20), only one fluorine atom was enough to make the pKa2 value lower than that
of 1. The NVs of these diamines were not higher than 1.5. Since the capacity of oxygen to lower the
basicity is weaker than that of fluorine, the pKa2 values of the oxyalkyl derivatives (21-24) were higher
than that of 1. The NVs of 21 and 22 were higher than 1.5, despite their very low lipophilicity,

indicating that lowering the lipophilicity alone does not decrease the NV. Contrary to our expectation,
the NVs of 23 and 24 were less than 1.5.
Table 2. Normalized values and physicochemical properties of diamines shown in Option 1 (Figure 3).
HN
N
e
e
Compd. Structure
ClogP d pKa1
pKa2
6.3
Normalized value
3.3
1
1.3
0.1
0.9
10.2
8 a
9.9
5.1
5.4
1.0
1.2
F
N
9
10.0
F
F
F
N
N
10
-1.1
-1.4
-1.1
-0.4
0.0
9.8
5.4
5.3
5.2
5.1
6.1
4.9
6.8
5.9
6.2
4.7
1.1
1.1
0.9
0.8
0.8
1.1
1.7
1.0
1.4
1.0
N
11
9.7
N
N
12
9.8
N
N
Cl
13
9.6
N
N
14 b
10.1
10.1
10.2
10.1
10.2
10.1
H
15 b
0.3
F
F
F
16 b
0.6
F
γ
17 b,c
-0.3
0.6
γ
F
F
18
β
F
19 b
0.0
F
F

20 b
21
0.2
10.0
10.2
10.2
9.8
3.2
7.2
7.0
6.4
6.8
0.9
2.7
2.5
1.2
1.0
F
F
F
-0.2
-0.5
-1.0
-0.5
O H
22 b
23
O H
O
24
9.9
Me
O
a 3 HCl salt.
b 2 HCl salt.
c Racemate
d Calculated using ChemBioDraw Ultra 12.0.
e Calculated using ADMET predictor, version 7.2.0001.
Table 3 shows the NVs and physicochemical properties of the diamines in which EWGs are introduced
on the piperidine ring (Option 2 in Figure 3). Introduction of one fluorine atom reduced the pKa2 value
by 1.4, and the NV of 25 was below 1.5. Difluorination of piperidine ring further reduced the pKa2
value, and the NV of 26 was less than 1.5. In cyclic ether 27, the basicity of the piperidyl nitrogen is
reduced due to the two σ-bonds connected to the nitrogen atom, while the basicity of the methylamino
group at 4-position is mostly unaffected. Therefore, the pKa2 value of 27 could be rendered lower than
that of 1, with the pKa1 value remaining unchanged. The pKa2 value of 27 was similar to that of mono
fluorinated compound 25, and the NV was lower than 1.5.
Table 3. Normalized values and physicochemical properties of diamines shown in Option 2 (Figure 3).
e
e
Compd.
Structure
ClogP d pKa1
pKa2
6.3
Normalized value
3.3
1
1.3
10.2
HN
N
25 a,b,c
1.6
8.6
4.9
3.8
1.2
1.1
F
HN
N
26 b,c
2.5
7.4
F F
HN
N

27 b
0.3
9.7
4.6
0.9
HN
N
O
a A mixture of cis/trans isomers (cis/trans = 4/1)
b Racemate
c HCl salt
d Calculated using ChemBioDraw Ultra 12.0.
e Calculated using ADMET predictor, version 7.2.0001.
Table 4 shows the NVs and physicochemical properties of the diamines exemplified in Option 3 in
Figure 3, in which methyl group is replaced with an EWG to increase the inductive effect of the amino
group at the 4-position of the piperidine ring. In silico calculations suggest that the protonation mode
of these diamines are different from that of 1 (Figure 4).12 The nitrogen atoms of 1 are protonated in a
stepwise manner, because the difference in the basicities of the two nitrogen atoms are relatively large.
On the contrary, those in 28 or 29 will be protonated almost simultaneously, because their basicities
are similar to each other.13 In 30, protonation will occur in a stepwise manner, though the order will
be reverse of that in 1. The decrease in both pKa2 values and NVs, regardless of the protonation
behavior, suggests the universality of the pKa2-based hypothesis.
Table 4. Normalized values and physicochemical properties of diamines shown in Option 3 (Figure 3).
N
c
c
Compd.
Structure
ClogP b pKa1
pKa2
6.3
Normalized value
3.3
1
1.3
10.2
Me
HN
28
29
1.6
8.6
5.8
5.9
2.9
1.0
1.5
0.8
O
HN
2.2
1.5
8.6
8.1
O
HN
30 a
O
HN
a 2 HCl salt.
b Calculated using ChemBioDraw Ultra 12.0.
c Calculated using ADMET predictor, version 7.2.0001.

Net Charge = 1
H2N+
N
H2N+
HN
99%
N+
H
1
N
pKa1 = 10.2
pKa2 = 6.3
HN
N+
H
1%
O
O
O
H2N+
H2N+
N
HN
28
N+
H
O
62%
N
pKa1 = 8.6
pKa2 = 5.8
HN
N+
H
38%
O
H2N+
O
O
HN
H2N+
N
N+
H
30
0%
N
pKa1 = 8.1
pKa2 = 2.9
O
HN
N+
H
100%
Figure 4. Protonation behaviors of diamines 1, 28, and 30
Table 5 shows the NVs and physicochemical properties of the diamines in which the intramolecular
nitrogen atoms are closer to each other (Option 4 in Figure 3). Replacement of piperidine (1) with
pyrrolidine (31) reduced both the pKa2 value and NV, while no significant reduction in ClogP and
pKa1 were observed. This supports our pKa2-based hypothesis. Introduction of a fluorine atom or
hydroxyl group on the pyrrolidine ring (32, 33) further reduced the pKa2 values, and their NVs
decreased below 1.5. Methylation of the hydroxyl group in 33 increases its lipophilicity by 0.9, but no

changes were observed in the pKa2 value or NV. For the azetidine derivative 35, both pKa2 and the
NV were lower than that of 1. Introduction of a fluorine atom on the benzene ring (36) reduced both
pKa2 and the NV, although there were no significant changes in ClogP and pKa1 as compared with
35. These observations also support our pKa2-based hypothesis. Furthermore, the pKa2 values of ring-
expansion analogs such as homopiperazine derivative 37, [3,2,0]diazabicycloheptane derivatives 38
and 39, and spiro compound 40 were also lower than that of 1, and the NVs of these diamines were
less than 1.5.
Table 5. Normalized values and physicochemical properties of diamines shown in Option 4 (Figure 3).
d
d
Compd.
Structure
ClogP c pKa1
pKa2
6.3
Normalized value
3.3
1
1.3
10.2
HN
N
31 a,b
1.7
10.0
5.2
3.8
4.6
4.4
1.6
1.0
0.9
0.8
HN
N
32 b
2.2
1.4
2.3
8.7
9.4
9.4
HN
N
F
33 a,b
HN
N
HO
34 a,b
HN
N
O
35
2.1
2.3
1.9
9.5
9.3
9.9
4.5
3.9
5.3
1.5
1.0
1.3
HN
N
36 a
37
F
HN
N
H
N
N
38 b
1.8
9.6
5.1
1.3
H
HN
N
H

39
1.6
2.2
9.5
9.7
4.1
4.2
1.1
1.4
H
N
H
H
N
N
40 a
N
H
a 2 HCl salt.
b Racemate.
c Calculated using ChemBioDraw Ultra 12.0.
d Calculated using ADMET predictor, version 7.2.0001.
To further verify our hypothesis, we analyzed the relationship between the NV and the
physicochemical properties for the forty diamines examined in this study (Figure 5). No definite
correlation between NV and ClogP or pKa1 was observed. However, it was clear that the NVs were
greater than 1.5 when the pKa2 values exceed 6.2 with the exception of 23 and 24. Since both these
compounds have an oxetane moiety, we speculate that the basicity-lowering effect of oxetane moiety
may be underestimated in the in silico calculations performed in this study.

Figure 5. Relationships between NV and ClogP, pKa1 or pKa2
In order to verify the above speculation and confirm the accuracy of the in silico predicted pKa values,
we compared the calculated and experimentally obtained basic pKa values of the representative
aliphatic amines (Table 6).14 For the amines with and without fluorine or oxygen atom, the ΔΔpKa
values were 0.4 and 0.3, respectively, indicating that the in silico calculations can predict the basicity-
lowering effects of these atoms almost accurately. On the other hand, the ΔΔpKa value for oxetanyl
group was as high as 1.9, indicating that the basicity-lowering effect of an oxetanyl group was grossly
underestimated in the in silico calculations performed in this study. Assuming that the pKa2 values of
23 and 24 are reduced by 1.9, the corrected pKa2 values would be 4.5 and 4.9, respectively; thus, these
compounds will no longer be an exception to our hypothesis.
Table 6. Difference in pKa-lowering effect as obtained from the calculated and experimental values.

Compd. Structure
pKacalculated ΔpKacalculated pKaexperimental ΔpKaexperimental ΔΔpKa
a
b
c
42
10.4
10.7
-1.3
-0.9
-1.7
-1.2
0.4
NH2
43
44
45
46
9.1
9.8
8.9
9.6
9.0
10.2
9.0
9.9
F
N
0.3
1.9
N
O
N
Ph
Ph
-0.8
-2.7
N
47
8.8
7.2
O
a calculated using ADMET predictor, version 7.2.0001.
b experimentally measured pKa values reported in ref. 13
cΔΔpKa is defined as ΔpKacalculated – ΔpKaexperimental. It gives the difference in basicity-lowering effects
as obtained from the calculated and the experimental values.
Phospholipidosis is caused by lysosomal accumulation of a compound.2 For a compound to be
accumulated into lysosomes, it first need to enter the lysosomes, which is typically driven by the
lipophilicity of a compound. Once incorporated into the lysosome, the compound is protonated under
lysosomal pH and cannot leave the lysosome, because cationic species are unlikely to passage across
biological membrane. Therefore, the more basic a compound is, the more accumulated in lysosomes.
Indeed, almost all prediction models are based on lipophilicity and basicity of a compound. That is
why ClogP and pKa1 are considered to be the key physicochemical attributes of phopholipidosis
inducers. However, our results showed that there are little correlation between NV and ClogP or pKa1.
Although we have no convincing explanation, we speculate that the differences in the dataset would
be one of the reasons, That is, we focus on the fragment-sized diamines, while the previous studies
have investigated the marketed drugs and their derivatives. Some monocationic amines have been
reported to passively traverse the lysosome membranes if they are small and not very hydrophilic.15
In such a situation where monocations could escape from lysosomes, lysosomal accumulation of
diamines will depend on their tendency toward the formation of dicationic state, since dications are
much less permeant than monocations. This further predicts that pKa2 is the dominant predictor for
the PLIP of the fragment-sized diamines.

In conclusion, we have revealed that the previously reported methods to predict the PLIP are not
suitable for fragment-sized diamines. Then, we have proposed a new descriptor, pKa2, which can
distinguish PLIP-positive diamines from PLIP-negative diamines more accurately than ClogP or pKa1.
To the best of our knowledge, this is the first report focusing the relationship between pKa2 and the
PLIP. The pKa2-based prediction might be adopted only when examining small diamines; however,
their toxicological behavior is worth investigating in the drug discovery process. Synthetic methods
for conformationally restricted small diamines have been frequently reported recently,16 and some of
these diamines are commercially available.17 Under these circumstances, it is expected that
compounds that release a small diamine as a metabolite are frequently synthesized during drug
discovery programs,18 and our study can contribute in evaluating the toxicological potentials of these
compounds.
1 (a) Benjamin, L.; James, L.; Hubert C.; Dennis M. Clinical Neurology and Neuroscience, 2018, 2,
1. (b) Sakai, H., Inoue, H., Toba, T., Murata, K., Narii, N., Shimmyo, Y., Igawa, Y., Matsumoto, T.,
Takemoto, N. Bioorg. Med. Chem. Lett. 2019, 29, 2332.
2 Hanumegowda U. M.; Regueiro-Ren A. Top. Med. Chem. 2015, 9, 261.
3 See ref. 2 and references therein.
4 According to Ploemen’s model, PLIP is predicted to be positive if (most basic pKa)2 + (ClogP)2 ≥
90 and most basic pKa ≥ 8 and ClogP ≥ 1. It is predicted to be negative if even one of these
requirements is not satisfied. Ploemen, J. P.; Kelder, J.; Hafmans, T.; van de Sandt, H.; van Burgsteden,
J. A.; Salemink, P.J.; van Esch, E. Exp. Toxicol. Pathol. 2004, 55, 347.
5 Compounds with NC equal to 1.0 are more likely to cause phospholipidosis as ClogP increases. For
those with NC above 1.0, phospholipidosis is induced regardless of ClogP. Tomizawa, K.; Sugano,
K.; Yamada, H.; Horii, I. J. Toxicol. Sci. 2006, 31, 315.
6
Kasahara, T.; Tomita, K.; Murano, H.; Harada, T.; Tsubakimoto, K.; Ogihara, T.; Ohnishi, S.;
Kakinuma, C; Toxicol Sci. 2006, 90, 133.
7
The stoichiometric ratio of amine and HCl has not been experimentally determined. We however
consider that all basic nitrogen atoms are protonated under strong acidic conditions because previous
studies have confirmed by elemental analysis that diamines form dihydrochloride when treated with
hydrochloric acid(a) or saturated solution of hydrogen chloride in ethanol.(b),(c)
(a) Reiter, J.;Trinka, P.; Bartha, F.L.; Pongo,L.;Volk,B.; Simig, G. Org. Process Res. Dev. 2012, 16,
1279.
(b) Veselov, I.S.; Trushkov, I.V.; Zefirov, N.S.; Grishina, G.V.; Russ. J. Org. Chem. 2009, 45, 1050.
(c) Ganina, O.G.; Veselov, I.S.; Grishina, G.V.; Fedorov, A.Y.; Beletskaya, I.P. Russ. Chem. Bull.,Int.
Ed. 2006, 55, 1642.
8
Culture media used in the PLIP assay contains inorganic salts such as sodium bicarbonate or
disodium phosphate, which compose acid-base buffering systems to keep the pH constant. The media

also contains Phenol Red as a pH indicator. We have confirmed that the color of the media didn’t
change significantly between when free base was used and when HCl salt was used, indicating the
difference between free base and HCl salt does not affect physiological pH of evaluated conditions.
Detailed assay conditions are described in the supplementary material.
9 Diamines have two basic pKa. The first and second basic pKa of diamines are expressed as pKa1 and
pKa2, respectively (see the scheme below). Unless otherwise noted, pKa values were calculated using
ADMET predictor, Simulation.plus.inc. Version 7.2.0001.
2+
MH+
MH2
second basic pKa
(pKa2)
M
first basic pKa
(pKa1)
10 In addition to Ploemen’s model (see ref. 4) and Tomizawa’s model (see ref. 5), modified Ploemen’s
model(a) was used.
(a) Pelletier D.J.; Gehlhaar, D.; Tilloy-Ellul, A.; Johnson, T.O.; Greene, N. J. Chem. Inf. Model. 2007,
47, 1196.
11
Morgenthaler, M.; Schweizer, E.; Hoffmann-Röder, A.; Benini, F.; Martin, RE.; Jaeschke, G.;
Wagner, B.; Fischer, H.; Bendels, S.; Zimmerli, D.; Schneider, J.; Diederich, F.; Kansy, M.; Müller
K. ChemMedChem. 2007, 2, 1100.
12 The protonation behaviors are simulated by ADMET predictor, version 7.2.0001. All pKa1 and pKa2
values listed in Table 1 to 5 have been specified as shown in Figure 4.
13 Even though the basicities of two nitrogen atoms are similar to each other, pKa1 and pKa2 values
are not necessarily close to each other. This is because pKa1 and pKa2 values do not represent the
basicity of a particular nitrogen atom but the pH at which the whole positive charge of the compound
is 0.5 or 1.5, respectively.
14
Experimentally
measured
pKa
values,
cited
from
the
tabulation
at
http://www.zirchrom.com/organic.htm and references therein. [Accessed: 1-Sep-2019]
15 (a) Andrew, C. L.; Klemm, A.R.; Lloyd, J. B.; Biochim Biophys Acta. 1997, 1330, 71. (b) Ohkuma,
S.; Poole, B.; J Cell Biol. 1981, 90, 656.
16 Grygorenko, O. O.; Radchenko, D.S.; Volochnyuk D. M.; Tolmachev A. A.; Komarov, I. V. Chem
Rev. 2011, 111, 5506.
17
http://www.enamine.net/download/catalogs/Enamine_BicyclicAmines_2014.pdf [Accessed: 1-
Sep-2019]
18 Roughley, S. D.; Jordan, A. M. J Med Chem. 2011, 54, 3451.
Declaration of interests
☐ The authors declare that they have no known competing financial interests or
personal relationships that could have appeared to influence the work reported in this
paper.

☒The authors declare the following financial interests/personal relationships
which may be considered as potential competing interests:
This research was conducted with a research fund from Asubio pharma, which the authors belonged to.