Successful application of serum shift prediction models to the design of dual targeting inhibitors of bacterial gyrase B and topoisomerase IV with improved in vivo efficacy

Article abstract of DOI:10.1016/j.bmcl.2014.03.022

A series of dual targeting inhibitors of bacterial gyrase B and topoisomerase IV were identified and optimized to mid-to-low nanomolar potency against a variety of bacteria. However, in spite of seemingly adequate exposure achieved upon IV administration,

Full text of DOI:10.1016/j.bmcl.2014.03.022

Bioorganic & Medicinal Chemistry Letters 24 (2014) 2177–2181
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Successful application of serum shift prediction models to the design
of dual targeting inhibitors of bacterial gyrase B and topoisomerase
IV with improved in vivo efficacy
a
a
a
a
Emanuele Perola a, , Dean Stamos , Anne-Laure Grillot , Steven Ronkin , Tiansheng Wang ,
Arnaud LeTiran a, , Qing Tang a, David D. Deininger a, Yusheng Liao a, Shi-Kai Tian a,à, Joseph E. Drumm a,§
David P. Nicolau b, Pamela R. Tessier b, Nagraj Mani a, Trudy H. Grossman a,–, Paul S. Charifson a
⇑
,
a Vertex Pharmaceuticals Incorporated, 50 Northern Avenue, Boston, MA 02210, United States
b Center for Anti-Infective Research and Development, Hartford Hospital, 80 Seymour Street, Hartford, CT 06102, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of dual targeting inhibitors of bacterial gyrase B and topoisomerase IV were identified and opti-
mized to mid-to-low nanomolar potency against a variety of bacteria. However, in spite of seemingly
adequate exposure achieved upon IV administration, the in vivo efficacy of the early lead compounds
was limited by high levels of binding to serum proteins. To overcome this limitation, targeted serum shift
prediction models were generated for each subclass of interest and were applied to the design of prospec-
tive analogs. As a result, numerous compounds with comparable antibacterial potency and reduced pro-
tein binding were generated. These efforts culminated in the synthesis of compound 10, a potent
inhibitor with low serum shift that demonstrated greatly improved in vivo efficacy in two distinct rat
infection models.
Received 4 March 2014
Accepted 7 March 2014
Available online 19 March 2014
Keywords:
Gyrase B
Topoisomerase IV
Antibacterial
Protein binding
Serum shift
Ó 2014 Elsevier Ltd. All rights reserved.
Predictive models
The efficacy of a drug is driven by two main factors: the drug’s
intrinsic potency and its available concentration at the site of ac-
tion. One important factor, which affects both the percentage of
drug molecules available for interaction with the target and the
drug’s pharmacokinetic behavior, is binding to plasma or tissue
proteins. High levels of protein binding can severely limit the con-
centration of unbound drug, thus affecting its ability to interact
with the intended target. At the same time, binding to plasma or
tissue proteins reduces the availability of free drug to the target or-
gans of clearance, thus slowing the elimination process and pro-
longing the exposure at the site of action.1 The balance of these
two effects ultimately determines the efficacy of the drug, and
their relative importance may depend on the drug’s mechanism
and on the extent of protein binding. The impact of protein binding
on antibiotic efficacy has been the object of numerous studies.
Trends of reduced efficacy in connection with increased protein
binding have been reported for a number of beta-lactams,2–4 and
the importance of considering the unbound concentration rather
than the total concentration as a predictor of efficacy has been
shown for other antibiotics and antibiotic classes.5–7 While large
amounts of proteins are also present in organs and tissues, the im-
pact of protein binding is usually estimated based on the measured
binding to serum proteins.8
In the gyrase B/topoisomerase IV antibacterial program conducted
at Vertex, the benzimidazole urea lead series was optimized to low
nanomolar levels of enzyme potency and mid-to-low nanomolar
levels of cell potency against a number of bacterial strains.9 How-
ever, measures of minimal inhibitory concentration (MIC) against
Staphylococcus aureus in the presence of 50% human serum showed
that the effective concentration of the most potent molecules in
this class was substantially reduced by protein binding, thus
resulting in a significant increase in MIC values. The structures of
representative early leads in this class and the corresponding S.
aureus MIC values in the presence and absence of 50% human ser-
um are reported in Table 1. The ratio between the MIC values mea-
sured in the presence and in the absence of 50% human serum will
be referred to as ‘serum shift’ in the remainder of this Letter.
⇑
Corresponding author. Tel.: +1 617 341 6646.
E-mail address: emanuele_perola@vrtx.com (E. Perola).
Present address: Institut de Recherches Servier, 125 Chemin de Ronde, 78290
Croissy sur Seine, France.
à
Present address: Department of Chemistry, University of Science and Technology
of China, 96 Jinzhai Road, Hefei, Anhui 230026, China.
§
Retired.
–
Present address: Tetraphase Pharmaceuticals, 480 Arsenal Street, Watertown, MA
02472, United States.
http://dx.doi.org/10.1016/j.bmcl.2014.03.022
0960-894X/Ó 2014 Elsevier Ltd. All rights reserved.

2178
E. Perola et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2177–2181
Table 1
Structures of representative compounds in the benzimidazole urea class and corresponding antimicrobial activities against Staphylococcus aureus, strain ATCC 29213
Compound #
1
2
3
4
N
N
N
N
F
N
N
N
N
N
N
N
NH
O
NH
O
Structure
N
NH
O
N
HN
HN
HN
HN
N
NH
O
HN
HN
HN
HN
MIC (medium)a
(l
g/ml)
0.125
4
0.047
2
0.032
2
0.031
0.625
MIC (serum)b
(lg/ml)
a
Minimal inhibitory concentration measured in standard assay medium.
b
Minimal inhibitory concentration measured in standard medium + 50% human serum. The experimental protocols for the MIC measurements were previously
reported.9,10
In general, protein binding levels above 95% and serum shifts
for the compound classes of interest is likely to be required to cap-
ture the complexity of the process and build reliable protein bind-
ing prediction models.
Our first approach to address the problem was the introduction
of basic functionalities in our lead molecules. A large number of
analogs bearing basic amines were synthesized, and a significant
reduction of serum shift was observed in most cases (e.g., see com-
pounds 5 and 6 in Table 2).
However, the introduction of a basic functionality led to less
favorable rat IV PK for these compounds, as the more potent ana-
logs designed with this approach exhibited increased clearance.
Eventually, it became apparent that a more subtle approach was
required to design neutral analogs with reduced protein binding.
Aside from the general considerations about the impact of high
ClogP values on protein binding, no accurate predictive model
was available for compounds of intermediate lipophilicity. We
therefore initiated attempts to generate a local model to predict
the serum shift of the aminobenzimidazole class using the avail-
able data from our gyrase program to derive a training set. The
choice to model serum shift instead of protein binding was
ranging from 16 to 128-fold were common among these com-
pounds. While the reduced free concentration may be counterbal-
anced by the protective effect of protein binding, which slows the
elimination process and leads to more extended exposure, the free
concentration of a drug at the site of infection must be at or above
the MIC at least some of the time during treatment to achieve effi-
cacy. Given the degree of protein binding and the levels of expo-
sure observed for our early leads upon IV or oral administration,
the doses required to achieve free concentrations above the MIC
even for a limited time were exceedingly high, resulting in limited
activity in vivo. As an example, compound 2 only achieved 20% sur-
vival in rat when dosed intravenously (IV) at 25 mg/kg in a sys-
temic S. aureus sepsis model,11 in spite of its potent in vitro
activity and a favorable pharmacokinetic profile. As a point of com-
parison, a 30% survival rate was achieved in the same model by the
marketed antibiotic linezolid, which has inferior rat IV PK, much
higher S. aureus MIC (1.56 lg/ml) but no serum shift. These results
prompted us to explore targeted design approaches to reduce pro-
tein binding for this class of compounds.
Binding to serum proteins is governed both by non-specific
hydrophobic interactions and by specific interactions with plasma
proteins. Due to the importance of both components, the genera-
tion of reliable models to predict the fraction of a given drug that
is sequestered by serum proteins has been challenging, and at-
tempts to generate predictive models that could be applicable to
diverse compound classes have produced mixed results.12–14
Hydrophobicity is generally believed to play a major role, and an
analysis of the available data for marketed drugs shows that drugs
with ClogP >4.5 tend to be over 90% protein bound, while for drugs
with ClogP >5 protein binding tends to exceed 95% (data not
shown). Unfortunately, while high ClogP values are generally asso-
ciated with high levels of protein binding, lower ClogP values do
not always lead to reduced protein binding. Due to the contribu-
tion of both specific and non-specific interactions, more polar mol-
ecules can still be sequestered in large proportions, only leaving a
small fraction available for interaction with the target of choice.
Another important factor is the presence of ionizable groups. When
molecules of similar logP are considered, acidic compounds tend to
be more highly protein bound than neutral compounds, which in
turn tend to be more highly protein bound than basic compounds
and zwitterions.13 While a ClogP reduction strategy may be
Table 2
Structures and S. aureus MIC values for two representative benzimidazole ureas
containing basic amines
Compound #
5
6
O
N
N
N
N
N
F
N
Structure
N
N
NH
N
N
NH
HN
HN
O
HN
HN
O
MIC (medium) (
MIC (serum) (
l
g/ml)
g/ml)
0.125
0.125
0.5
0.125
successfully applied to lower protein binding in specific cases,15
detailed analysis of properties, structural features and existing data
a
l

E. Perola et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2177–2181
2179
dictated by two factors: (1) serum shift provides a direct measure
of the impact of protein binding on antibacterial activity; (2) at the
high levels of protein binding observed for our series, small varia-
tions in protein binding lead to relatively large variations in serum
shift, thus making measurements of the latter more discriminant.
As an example, if compound A is 99.5% protein bound and com-
pound B is 98% protein bound, the free fraction of compound B is
4-fold higher than the free fraction of compound A, and the corre-
sponding serum shift will be 4-fold lower. A 4-fold difference in
serum shift can be reliably measured, while the difference between
99.5% and 98% in protein binding is likely within experimental er-
ror. Serum shift data was available for 338 compounds sharing the
substructure in Figure 1 below.
Since the objective was to build a model for neutral compounds,
all the charged compounds (mostly basic amines) were removed,
and the initial attempts at model building were made on the
remaining set of 202 compounds. Additional experiments had
shown that binding to human serum albumin was the dominant
cause of the observed serum shift. In our initial approaches we ana-
lyzed the correlation between serum shift and global properties
such as molecular weight, ClogP and polar surface area. We then
attempted to combine ClogP, which gave a weak but tangible sig-
nal, with 2D or 3D similarity to warfarin and other known high
affinity ligands of albumin. We also built a binary model using sup-
port vector machine with different sets of 2D descriptors and we
docked our compounds into the benzodiazepine and warfarin
binding sites of albumin. Although some of these approaches pro-
vided hints on the factors affecting serum shift for this class of
compounds, none of them resulted in a convincing model that
could bear sufficient predictive power. At that time the two sub-
classes illustrated in Figure 2 were being further investigated.
A pairwise analysis across the two subclasses showed that com-
pounds carrying a pyrimidine ring at the 7 position of the benz-
imidazole core (class B) consistently showed a lower serum shift
relative to the corresponding 7-fluoropyridines (class A). This real-
ization prompted us to perform a separate analysis for each sub-
class. Serum shift data was available for 62 compounds in class A
and 47 in class B. Given the findings that emerged from the broader
set, we focused our analysis on the correlation between serum shift
and global properties such as ClogP and polar surface area (PSA).
The plots in Figure 3 illustrate the correlation between serum shift
and each of these two properties for class A compounds.
While there was no clear correlation between serum shift and
PSA, there was a clear dependency in this subclass between serum
shift and ClogP. A ClogP value of 3.1 provided a good separation
between high and low shifters: 79% of compounds with ClogP
<3.1 had serum shifts of 8 or less, while 97% of the compounds with
ClogP >3.1 had serum shifts of 16 or more. This trend suggested
that prospective compounds with ClogP >3.1 would almost cer-
tainly be high shifters, while compounds with ClogP <3.1 would
likely be low shifters. Figure 4 illustrates the correlation between
serum shift and the same two properties for class B compounds.
In this case both ClogP and PSA played a significant role, and
the plots clearly indicated that the vast majority of the compounds
with serum shift of 4 or less had ClogP <2.1 and PSA >120. A more
detailed analysis of the individual data showed that using the two
parameters in combination had the potential to minimize both the
false positives and the false negatives: 92% of the compounds with
ClogP <2.1 and PSA >120 had serum shifts of 4 or less, while 91% of
the compounds with ClogP >2.1 and/or PSA <120 had serum shifts
of 8 or more. This trend suggested that prospective compounds
meeting both criteria would almost certainly be low shifters, while
compounds violating at least one of the two would almost cer-
tainly be high shifters.
The findings of this analysis were applied to the design of pro-
spective analogs by prioritizing compounds that met the following
criteria:
C7-fluoropyridines (class A): ClogP <3.1 (expected serum shift
68)
C7-pyrimidines (class B): ClogP <2.1 and PSA >120 (expected
serum shift 64)
This approach, combined with the SAR-driven and structure-
based methods previously described for this series,9 led to the syn-
thesis of compounds with low serum shift and a greater potential
to be efficacious in an animal infection model. Overall, 61 addi-
tional compounds between the two subclasses were synthesized
according to the general procedure in Scheme 1. Additional details
on the synthesis of these two subseries can be found in our previ-
ous publications.9,18
R2
Low nanomolar potencies against gyrase B and topoisomerase
IV were generally maintained, and MIC values between 0.016 and
0.25 lg/ml against S. aureus were observed for 55 of the 61 com-
R1
N
NH
pounds. The serum shift values were consistent with the predic-
tions based on the above criteria in 46/61 cases (75.4%). In all the
remaining 15 cases the shifts were only 2-fold outside of the pre-
dicted range. This method proved especially effective in filtering
out compounds with high serum shift, as only two of the com-
pounds predicted to have high shift turned out to have low shift.
Overall, the compounds with low shift were predicted with 62%
accuracy while the compounds with high shift were predicted with
92% accuracy. Table 3 illustrates the structures of four optimized
compounds with their antimicrobial activities both in the absence
and in the presence of human serum.
HN
O
X = O, NH
X
Figure 1. General substructure for compounds in the benzimidazole series with
available serum shift data.
R
R
F
N
7
7
As both models included an upper limit for ClogP of low serum
shifters, their application led to the design of compounds with re-
duced lipophilicity. It has been observed that reducing the lipophil-
icity of compounds within a series often leads to a reduction in
intrinsic clearance, which in turn can result in increased levels of
exposure in vivo.1,19 Such an effect could have been a contributing
factor to the desired improvements in efficacy for our series. How-
ever, a clear correlation between ClogP and intrinsic clearance for
the synthesized compounds could not be established, and no corre-
lation was observed between the same parameter and in vivo
exposure upon IV or oral dosing (data not shown).
N
N
N
N
NH
O
NH
O
HN
HN
HN
HN
Class A
Class B
Figure 2. Substructures of the benzimidazole urea subclasses investigated in this
study.

2180
E. Perola et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2177–2181
Figure 3. Serum shift versus ClogP and polar surface area (PSA) for compounds in class A. The ClogP calculations were performed using the ClogP program from Daylight.16
The PSA calculations were performed using an internal implementation of the method described by Ertl and colleagues.17
Figure 4. Serum shift versus ClogP and polar surface area (PSA) for compounds in class B.
Pd(PPh3)4
1. Br2, NaOAc, HOAc
2. HNO3, TFA, H2SO4
3. 6N HCl, reflux
F
O
B
F
F
Na2CO3, aq DME
O
100 oC
Cl
N
+
Br
N
NH
N
NH
NH2
O
O
NO2
O
S
O
1.
N
H
N
N
H
N
H
Bis(pinacolato)-
diboron
F
ArBr
Pd(PPh3)4
Na2CO3, DMF
aq. buffer (pH3.5)
F
F
O
B
Pd(PPh3)4
Ar
N
KOAc, dioxane
2. MeSO3H
Ar
O
N
N
NH
NH2
N
NH2
O
NO2
NO2
HN
HN
Scheme 1. General procedure for the synthesis of the compounds in class A. The same scheme applies to class B with the 2-(3-fluoro)pyridyl group at the 7-position of the
benzimidazole core replaced by a 2-pyrimidyl group. Ar = aryl group.
Compound 10 had adequate exposure in rat both upon IV and
upon oral administration and was evaluated in vivo in two rat
infection models. The experimental protocols for these models
exhibited modest efficacy in a systemic S. aureus infection model
when dosed IV at 25 mg/kg, similarly to linezolid. Compound 10
was also dosed orally in a Streptococcus pneumoniae rat lung infec-
tion model where it achieved a reduction in bacterial density of 5.8
and 3.7 logs in CFU when dosed at 30 and 3 mg/kg, respectively.
Linezolid achieved a reduction of 4.5 and 0.8 logs at the same
doses, while compound 2 achieved a reduction of 3.0 logs when
dosed IV at 38.5 mg/kg in the same model.
To best understand the impact of serum shift on the in vivo effi-
cacy of these compounds, these results should be viewed in light of
their antibacterial activities, serum shift values and rat IV PK
parameters (Table 4).
have been described in our previous publication.9
A single
30 mg/kg IV dose of compound 10 administered in a S. aureus rat
thigh infection model caused a reduction in bacterial density of
2.1 logs in CFU (colony forming units) 8 h after dosing, while a
3 mg/kg dose caused a reduction of 1.7 logs. Linezolid caused a
reduction of 1.4 logs in CFU when dosed at 30 mg/kg and no mea-
surable reduction when dosed at 3 mg/kg in the same model. Com-
pound 4, which has an equal S. aureus MIC in the absence of serum
and comparable IV PK but a 2.5-fold higher serum shift relative to
compound 10, caused a reduction of 1.5 logs in CFU after 6 hours
when dosed at 50 mg/kg. As mentioned above, compound 2, which
has an equal S. aureus MIC in the absence of serum and slightly bet-
ter IV PK relative to 10 but an 8-fold higher serum shift (64 vs 8),
Compounds 2, 4 and 10 have similar antibacterial potencies
against both S. aureus and S. pneumoniae; compound 2 has slightly
better IV PK while 4 and 10 have comparable PK parameters.
Linezolid is considerably less potent and has comparable IV PK

E. Perola et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2177–2181
2181
Table 3
Structures, calculated properties and S. aureus MIC values in the absence and in the presence of human serum for a representative set of optimized compounds in the
benzimidazole urea class
Compound #
7
8
9
10
-O
H2N
H2N
N
O
HO
N
N+
N
N
F
N
F
N
Structure
N
NH
N
N
N
N
N
NH
O
HN
HN
NH
O
N
NH
O
N
O
HN
HN
HN
HN
HN
HN
ClogP
1.73
134
0.125
0.25
1.56
139
0.064
0.25
2.64
110
0.5
1.96
129
0.032
0.25
PSA (Å2)
MIC (medium) (
MIC (serum) (
l
g/ml)
g/ml)
l
0.5
Table 4
References and notes
Antimicrobial activities, serum shifts and rat IV pharmacokinetic parameters for
compounds 2, 4, 10 and linezolid
1. Smith, D. A.; Di, L.; Kerns, E. H. Nat. Rev. Drug Disc. 2010, 9, 929.
2. Merrikin, D. J.; Briant, J.; Rolinson, G. N. J. Antimicrob. Chemother. 1983, 11, 233.
3. Tawara, S.; Matsumoto, S.; Kamimura, T.; Goto, S. Antimicrob. Agents Chemother.
1992, 36, 17.
4. Wise, R. Clin. Pharmacokinet. 1986, 11, 470.
5. Lee, B. L.; Sachdeva, M.; Chambers, H. F. Antimicrob. Agents Chemother. 1991, 35,
2505.
6. Crandon, J. L.; Banevicius, M. A.; Nicolau, D. P. Antimicrob. Agents Chemother.
2009, 53, 1165.
7. Scaglione, F.; Mouton, J. W.; Mattina, R.; Fraschini, F. Antimicrob. Agents
Chemother. 2003, 47, 2749.
S. a MICa
S. p MICb
SSc
CLd
t1/2
Vssf
AUCg
e
Compound
2
4
0.047
0.031
0.031
1.56
0.004
0.004
0.004
0.4
43
20
8
12.5
14
23
2.1
0.7
1.2
0.7
1.1
1.6
2.1
1
1.6
1.1
0.8
1.1
10
Linezolid
1
14.5
a
b
c
d
e
f
Minimal inhibitory concentration versus S. aureus (
l
g/ml).
Minimal inhibitory concentration versus Streptococcus pneumoniae (
l
g/ml).
Serum shift.
8. Zeitlinger, M. A.; Derendorf, H.; Mouton, J. W.; Cars, O.; Craig, W. A.; Andes, D.;
Theuretzbacher, U. Antimicrob. Agents Chemother. 2011, 55, 3067.
9. Charifson, P. S.; Grillot, A. L.; Grossman, T. H.; Parsons, J. D.; Badia, M.; Bellon,
S.; Deininger, D. D.; Drumm, J. E.; Gross, C. H.; LeTiran, A.; Liao, Y.; Mani, N.;
Nicolau, D. P.; Perola, E.; Ronkin, S.; Shannon, D.; Swenson, L. L.; Tang, Q.;
Tessier, P. R.; Tian, S. K.; Trudeau, M.; Wang, T.; Wei, Y.; Zhang, H.; Stamos, D. J.
Med. Chem. 2008, 51, 5243.
Clearance (ml/min/kg).
Half-life (hours).
Volume of distribution (L/kg).
g
Area under the curve (l
g⁄h/ml). The AUC values are dose-normalized to 1 mg/
kg. The actual doses in the PK studies ranged from 2.8 to 9.5 mg/kg.
10. Mani, N.; Gross, C. H.; Parsons, J. D.; Hanzelka, B.; Muh, U.; Mullin, S.; Liao, Y.;
Grillot, A. L.; Stamos, D.; Charifson, P. S.; Grossman, T. H. Antimicrob. Agents
Chemother. 2006, 50, 1228.
11. The Staphylococcal neutropenic sepsis model utilized specific-pathogen-free,
male Sprague-Dawley rats. Neutropenia was induced by cyclophosphamide
intraperitoneal (IP) injections of 150 and 50 mg/kg on day 4 and 1 day prior to
bacterial inoculation. Infection was established via an IP injection of 1.5 ml of
109 CFU/ml Staphylococcus aureus ATCC 29213. Antibacterial test agents were
administered intravenously 1 h after bacterial inoculation and survival was
assessed over 120 h.
relative to both 4 and 10. If protein binding and the consequent
serum shift were not a factor, compound 2 should have compara-
ble or slightly better efficacy relative to compounds 4 and 10 and
all three compounds should have better efficacy relative to linezo-
lid. Instead, the large serum shift increases the effective MIC of
compound 2 and makes it comparable to that of linezolid, which
does not experience any shift. The comparable efficacy of com-
pound 2 and linezolid reflects their serum shifted MIC values.
The lower serum shift experienced by compound 10 relative to
compounds 2 and 4 resulted in a lower effective MIC, which led
to its superior in vivo efficacy.
Overall, this study showed the effectiveness of a divide-and-
conquer approach that enabled us to derive accurate predictive
models for the serum shift of specific subclasses when clear trends
could not be identified for the entire class. It also showed that, in
some cases, simple local models can have greater predictive value
than more sophisticated global models, consistent with a retro-
spective analysis reported in a previous study.13 The application
of these models contributed to the design of compounds with re-
duced protein binding and improved in vivo efficacy.
12. Yamazaki, K.; Kanaoka, M. J. Pharm. Sci. 2004, 93, 1480.
13. Gleeson, M. P. J. Med. Chem. 2007, 50, 101.
14. Ma, C. Y.; Yang, S. Y.; Zhang, H.; Xiang, M. L.; Huang, Q.; Wei, Y. Q. J. Pharm.
Biomed. Anal. 2008, 47, 677.
15. Ellsworth, B. A.; Sher, P. M.; Wu, X.; Wu, G.; Sulsky, R. B.; Gu, Z.; Murugesan, N.;
Zhu, Y.; Yu, G.; Sitkoff, D. F.; Carlson, K. E.; Kang, L.; Yang, Y.; Lee, N.; Baska, R.
A.; Keim, W. J.; Cullen, M. J.; Azzara, A. V.; Zuvich, E.; Thomas, M. A.; Rohrbach,
K. W.; Devenny, J. J.; Godonis, H. E.; Harvey, S. J.; Murphy, B. J.; Everlof, G. G.;
Stetsko, P. I.; Gudmundsson, O.; Johnghar, S.; Ranasinghe, A.; Behnia, K.;
Pelleymounter, M. A.; Ewing, W. R. J. Med. Chem. 2013, 56, 9586.
16. CLOGP v. 4.1. Daylight Chemical Information Systems.
17. Ertl, P.; Rohde, B.; Selzer, P. J. Med. Chem. 2000, 43, 3714.
18. Charifson, P. S.; Deininger, D. D.; Grillot, A. L.; Liao, Y.; Ronkin, S. M.; Stamos, D.;
Perola, E.; Wang, T.; LeTiran, A.; Drumm, J. US Patent 7,495,014 B2; 2009.
19. Lewis, D. F.; Jacobs, M. N.; Dickins, M. Drug Discovery Today 2004, 9, 530.