Anticancer agent synthesis designed by artificial intelligence: Pd(OAc)2-catalyzed one-pot preparation of biphenyls and its application to a concise synthesis of various diazofluorenes

Article abstract of DOI:10.1016/j.tetlet.2020.152267

We successfully synthesized the antitumor agent 1-methoxydiazofluorene via an artificial intelligence (AI)-proposed design. The pivotal biphenyl scaffold of 1-methoxydiazofluorene was constructed by applying Pd(OAc)₂-catalyzed cycloaromatizatio

Full text of DOI:10.1016/j.tetlet.2020.152267

Tetrahedron Letters 61 (2020) 152267
Contents lists available at ScienceDirect
Tetrahedron Letters
journal homepage: www.elsevier.com/locate/tetlet
Anticancer agent synthesis designed by artificial intelligence:
Pd(OAc)2-catalyzed one-pot preparation of biphenyls and its
application to a concise synthesis of various diazofluorenes
Tetsuhiko Takabatake a, Hiroki Tomita b, Syo Okada b, Natsumi Hayashi c, Takashi Masuko c,
Masahiro Toyota b,
⇑
a Advanced Materials Development Laboratory, Sumitomo Chemical Co., Ltd., 1-98, Kasugadenaka, 3-chome, Konohana-ku, Osaka 554-8558, Japan
b Department of Chemistry, Graduate School of Science, Osaka Prefecture University, Sakai, Osaka 599-8531, Japan
c Faculty of Pharmacy, Kindai University, 3-4-1 Kowakae, Higashiosaka City, Osaka 577-8502, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
We successfully synthesized the antitumor agent 1-methoxydiazofluorene via an artificial intelligence
(AI)-proposed design. The pivotal biphenyl scaffold of 1-methoxydiazofluorene was constructed by
applying Pd(OAc)2-catalyzed cycloaromatization, a method developed by our group. AI-proposed func-
tional group conversion of the biphenyl intermediate provided 9-diazo-1-methoxy-9H-fluorene, which
inhibits the proliferation of HeLa cells, was achieved in reasonable chemical yield. In addition, various
diazofluorenes were synthesized using the above protocol and their antitumor effects were evaluated.
As a result, several novel diazofluorenes, which have a stronger cytotoxic activity than cisplatin against
various human epithelial cancer cells, were found.
Received 23 June 2020
Revised 14 July 2020
Accepted 16 July 2020
Available online 14 August 2020
Keywords:
Artificial intelligence
Antitumor agent
Biphenyl
Ó 2020 Elsevier Ltd. All rights reserved.
Diazofluorene
Introduction
added to the evaluation of each synthetic route. Furthermore, the
verification of experimental results significantly depends on the
Artificial intelligence (AI) is playing an increasingly significant
role in many fields [1]. AI is being utilized in self-driving cars [2],
humanoid robots [3], and to produce medical diagnoses [4], to
name a few examples. It will eventually develop to a stage where
machines will work as teammates with humans. How exactly will
AI evolve in the future? How quickly will AI transform the way
people live? While the answer to these questions cannot be known
at present, what is clear is that the field of synthetic organic chem-
istry is lagging behind in transitioning to the use of AI. One of the
major advantages of AI is its ability to make swift and accurate cal-
culations in determining the next best move, and this trait is lever-
aged in the use of AI in games and competitions. Similarly, if the AI
loses, it can be upgraded via feedback gained by the experience.
These concepts can be used to apply AI in the field of synthetic
organic chemistry, although additional layers of complication must
be considered. For example, verification experiments are essential
in order to evaluate synthetic routes proposed by AI, and there are
a number of factors that determine the superiority or inferiority of
AI-proposed synthetic routes. The total yield, number of steps,
reaction safety, and toxicity of intermediates and byproducts are
skill of the experimenters. All of these make the evaluation of AI
programs more complex. It therefore takes a long time to complete
the verification experiments of the proposed synthetic routes, and
this has contributed to the delay in the development and utiliza-
tion of AI in organic synthetic chemistry.
In our previous paper related to AI-proposed drug synthesis, we
demonstrated that our AI program, SYNSUP [5], is a useful tool in
synthetic organic chemistry [6]. The reaction conditions of the
key reaction (1 ? 2) were studied after tuning a part of the reac-
tion substrate structure proposed by the SYNSUP program
(Scheme 1). Although the reaction mechanism of the key step
was different from the one originally proposed, we found a novel
method for the stereoselective synthesis of trans b-lactams. The
usefulness of the protocol was verified by the demonstration of
the total synthesis of SCH 47949 (3), a cholesterol absorption inhi-
bitor [7]. It is worth noting that the functional group conversions
(2 ? 3) in this synthesis were all identical with the transforma-
tional processes proposed by AI.
As part of an effort to demonstrate the versatility of AI-pro-
posed target molecule synthesis, we herein report a Pd(OAc)2-cat-
alyzed one-pot preparation of biphenyls and its application to 9-
diazo-1-methoxy-9H-fluorene (4) synthesis. 1-Methoxydiazo-fluo-
⇑
Corresponding author.
https://doi.org/10.1016/j.tetlet.2020.152267
0040-4039/Ó 2020 Elsevier Ltd. All rights reserved.

2
T. Takabatake et al. / Tetrahedron Letters 61 (2020) 152267
rene 4, a simple diazofluorene analog of kinamycins, inhibits the
proliferation of HeLa cells [8].
Scheme 2 shows the SYNSUP-proposed retrosynthetic analysis
for 9-diazo-1-methoxy-9H-fluorene (4) (See Supporting Informa-
tion-1). Namely, the target molecule 4 could be constructed
through a series of functional group transformations of the biphe-
nyl 5a, which could be assembled to form the 1,3-dicarbonyl com-
pound 6a by the Pd(OAc)2-catalyzed cycloaromatization (7 ? 8)
method recently developed by our group (Scheme 3) [9]. For exam-
ple, methyl 2-hydroxy-6-methylbenzoate (8) was obtained catalyt-
ically from the acyclic unsaturated 3-keto ester 7 in a single step.
Results and discussion
Scheme 1. AI-designed total synthesis of SCH 47949 (3).
First, we examined whether our cycloaromatization process
could be adapted to use 1,3-dicarbonyl compound 6a to assemble
the target biphenyl compound 5a [10]. The requisite substrate 6a
was easily synthesized in a reasonable chemical yield from cin-
namyl chloride and dianion of methyl acetoacetate according to
the standard procedure [11]. A number of different reaction
parameters, such as concentration, temperature, and reaction time
were tested to optimize this cycloaromatization.
Scheme 2. Retrosynthetic analysis of 4 proposed by SYNSUP.
Since 8 was obtained in a 66% yield by means of Pd(OAc)2-cat-
alyzed cycloaromatization (Scheme 3), we first applied the above
reaction conditions to the present biphenyl synthesis. The catalyst
and reoxidant were fixed, and different reaction temperatures,
reaction times, and solvents were examined. Even at room temper-
ature, the desired biphenyl 5a was produced in a 20% yield; how-
ever, 65% of starting material 6a was recovered (Table 1, entry 1).
The catalytic reaction was performed at 45 °C for 12 h to afford 5a
(46%) together with 6a (17%), and when this reaction time was
extended to 18 h, the reaction gave 5a in a 48% yield (Table 1,
entries 2 and 3). Although 6a was consumed completely, the con-
version yield was not high. The 1H NMR spectra of the crude pro-
duct shows only 6a, likely because some organic molecules were
taken into the palladium complex. When this reaction was carried
Scheme 3. Pd(OAc)2-catalyzed cycloaromatization.
Table 1
Evaluation of the reaction conditions for cycloaromatization.
Entry
Temperature (°C)
Substrate concentration (mol/L)
Time (h)
Solvent
Yield (%)a
1
2
3
4
5
6
7
8
rt
3.6
3.6
3.6
3.6
3.6
3.6
3.6
3.6
3.6
3.6
5.0
2.5
2.5
2.5
24
12
18
16
18
18
18
18
18
18
18
24
24
24
DMSO
DMSO
DMSO
DMSO
toluene
DMF
1,2-DME
1,2-DCE
DMSO
DMSO
DMSO
DMSO
DMSO
DMSO
20 [65]
46 [17]
48
45
45
70
45
45
45
45
45
45
45
50
50
50
29
28 [23]
34 [16]
38 [30]
36 [31]
47
17
45
60
11
9b
10c
11
12
13d
14e
15
a
b
c
Isolated yields in parentheses represent the recovered starting material 6a.
10 mol % Pd(TFA)2 was used instead of 10 mol % Pd(OAc)2.
10 mol % PdCl2 was used instead of 10 mol % of Pd(OAc)2.
N2 was used instead of O2.
d
e
In the absence of 20 mol % Cu(OAc)2.

T. Takabatake et al. / Tetrahedron Letters 61 (2020) 152267
3
out at 70 °C, no other product was found; however, the isolation
yield was drastically reduced to 29% (Table 1, entry 4). Next, the
solvent effect was evaluated in this catalytic system. When this
catalytic reaction was carried out using toluene, DMF, 1,2-
dimethoxyethane (1,2-DME), or 1,2-dichloroethane (1,2-DCE) at
45 °C, a substantial amount of 6a was recovered together with
5a (Table 1, entries 5–8). When we examined the effect of the cat-
alyst, Pd(TFA)2 proved to be a useful catalyst; however, PdCl2 was
ineffective (Table 1, entries 9 and 10). Altering the reaction concen-
tration resulted in a higher concentration (approximately 5.0 mol/
L) of 6a giving rise to 5a in a 45% yield; however, when the reaction
concentration was reduced using a 2.5 mol/L, the desired product
5a was obtained with the highest yield (Table 1, entries 11 and
12). Finally, we found that when this reaction was carried out using
nitrogen instead of oxygen, only 11% of 5a was obtained, and only
15% of 5a was obtained in the absence of Cu(OAc)2. These results
suggest that the Pd(0) generated in the system is oxidized
smoothly to Pd(OAc)2 only when an oxygen and Cu(OAc)2 coexist
(Table 1, entries 13 and 14).
had little influence on the chemical yields of products 5c–e. In the
case of methyl-substituted substrates 5f–g, the product yields
were slightly decreased. If a halogen atom, such as chloride or flu-
oride, was placed as a substituent on the benzene ring of substrate
6, the chemical yield decreased slightly. In the case of substrate 6k,
which is substituted by fluoride, the yield of 5k decreased to 41%.
The desired products were obtained in reasonable yields when
using substrates 6m–n with a bulky naphthyl group as a sub-
stituent. Interestingly, each cyclization of substrate 6o with a phe-
nyl group at the 4 position and substrate 6p with a phenyl group at
both the 4 and 7 positions gives biphenyl 5o and terphenyl 5p in
moderate yields.
Scheme 5 shows a possible reaction mechanism of the one-pot
preparation of biphenyls. After activation of an isolated olefin of 6
by Pd(OAc)2, an enolate carbon in I attacks the terminal carbon,
forming the six-membered alkyl palladium intermediate II. In
our system, b-elimination of HPdOAc from II surpasses b-elimina-
tion of ArPdOAc to afford the intermediate III [12]. It is noticeable
that no trace of methyl salicylate was detected when the crude
product was analyzed by 1H NMR. The 1H NMR spectrum of the
reaction intermediate shows the possibility that intermediate III
may exist (See the Supporting Information-2). The addition–elimi-
nation of Pd(II) in this process is reversible. Since this reaction
intermediate (III) is so stable that it can be isolated, the b-elimina-
tion of ArPdOAc apparently does not proceed. The 1,4-diene III is
further oxidized by Pd(OAc)2 through the intermediate IV to pro-
vide V. Enolization of ketone V leads to the thermodynamically
stable aromatic compound 5. HPdOAc releases AcOH and becomes
Pd(0), which could be oxidized by Cu(OAc)2 and an oxygen to gen-
erate Pd(OAc)2.
To expand the versatility of this one-pot synthesis of biphenyls,
the influence of different substituent groups on the benzene ring
and side chain of 6 on the chemical yield was examined (Scheme 4).
tert-Butyl ester 5b was obtained in a 71% yield when the reaction
was performed at 60 °C. Changing the substituent group on the
benzene ring of reaction substrates 6c–e to a methoxy substituent
Scheme 4. One-pot synthesis of substituted biphenyls.
Scheme 5. Proposed reaction mechanism.

4
T. Takabatake et al. / Tetrahedron Letters 61 (2020) 152267
Scheme 6. Total synthesis of 9-diazo-1-methoxy-9H-fluorene (4).
Fig. 1. Structures of novel synthetic diazofluorenes 16–26.
After triflate formation of 5a, a Suzuki-Miyaura coupling reac-
tion [16] with phenylboronic acid afforded terphenyl 13 in an
excellent yield. Intramolecular Friedel-Crafts reaction of 13 in the
presence of 98% H2SO4 gave 1-phenyl-9H-fluoren-9-one (14),
which was transformed into the desired 1-phenydiazo-fluorene
15 (Scheme 7).
Diazofluorenes 16–24 were prepared from the corresponding
biphenyls 5 using the same synthetic route as for 4. On the other
hand, 1-aryldiazofluorenes 25 and 26 were synthesized by apply-
ing the same approach as for 15 (Fig. 1). Each of the synthetic
diazofluorenes were easily purified by flash column chromatogra-
phy. It should be noted that most synthetic diazofluorenes
including 4, and 15–26 are stable enough not to decompose even
if left in a nitrogen atmosphere at room temperature for about a
week.
Scheme 7. Total synthesis of 9-diazo-1-phenyl-9H-fluorene (15).
Since the biphenyl 5a was obtained in a 60% yield, began the
total synthesis of 1-methoxydiazofluorene 4 according to the pro-
posed functional group transformations by SYNSUP. Methylation of
the alcohol 5a with methyl iodide in the presence of K2CO3 gave
rise to 9 (95%), which was cyclized using 98% H2SO4 to furnish flu-
orenone 10 in an 84% yield [13]. Refluxing of 10 with hydrazine
monohydrate in the presence of acetic acid provided the hydrazone
11 in a 94% yield [14], which led to 9-diazo-1-methoxy-9H-fluo-
rene (4) in an 87% yield by MnO2 oxidation [15]. The spectroscopic
properties of synthetic material 4 were identical with those
reported for 4 [8] (Scheme 6).
Because a series of conversion reactions from the biphenyl com-
pound 5 seemed to be effective for a concise synthesis of various
diazofluorenes, we decided to carry out antitumor activity tests
of various novel diazofluorenes using human cancer cells. Com-
pound 15, having a phenyl group at the 1-position of compound
4, was synthesized as follows (Scheme 7) (See Supporting
Information-2).
Cytotoxicity test
Since natural kinamycins are not commercially available, we
selected cisplatin (CDDP) [17] as a standard to evaluate the
strength of antitumor activity of the novel diazofluorenes. Cytotox-
icity tests of synthetic diazofluorenes 4, and 15–26 using HeLa,
PK1, A549, BT474, and SW1116 human cancer cells were per-
formed. Compared with CDDP, compounds 24 and 25 showed
stronger cytotoxic activities against HeLa cervical cancer cells.
Compound 18 was very effective against PK1 pancreatic cancer
cells. Additionally, compounds 22–24 showed strong cytotoxic
activities against A549 lung, BT474 breast, and SW1116 colon
cancer cells [18]. Details with statistical analysis are described in
Supporting Information-3.

T. Takabatake et al. / Tetrahedron Letters 61 (2020) 152267
5

6
T. Takabatake et al. / Tetrahedron Letters 61 (2020) 152267

T. Takabatake et al. / Tetrahedron Letters 61 (2020) 152267
7
CAMEO: T.D. Salatin, W.T. Jorgensen, J. Org. Chem. 45 (1980) 2043–2051; (j)
SYNSUP: Ref. [5a]; (k) synchem: H.L. Gelernter, A.F. Sanders, D.L. Larsen, K.K.
Agarwal, R.H. Boivie, G.A. Spritzer, J.E. Searleman, Science 197 (1977) 1041–
049; (l) ARChem Route Designer: J. Law, Z. Zsoldos, A. Simon, D. Reid, D. Liu, S.
Y. Khew, A.P. Johnson, S. Major, R.A. Wade, H.Y. Ando, J. Chem. Inf. Model. 49
(2009) 593–602; (m) AIPHOS: K. Funatsu, S. Sasaki, Tetrahedron Comput.
Methodol. 1 (1988) 27–37; (n) KOSP: K. Satoh, K. Funatsu, J. Chem. Inf. Comput.
Sci. 39 (1999) 316–325; (o) SYNGEN: J.B. Hendrickson, Angew. Chem. Int. Ed.
29 (1990) 1286–1295; (p) CICLOPS: J. Blair, J. Gasteiger, C. Gillespie, P.D.
Gillespie, I. Ugi, Tetrahedron 30 (1974) 1845–1859; (q) EROS: J. Gasteiger, C.
Jochum, Topics Curr. Chem. 74 (1978) 93–126; (r) WODCA: J. Gasteiger, W.-D.
Ihlenfeldt, P. Roese, Red. Trav. Chim. Pays-Bas. 111 (1992) 270–290; (s)
CHIRON: S. Hanessian, J. France, G. Gagnon, D. Laramée, B.J. Larouche, Chem.
Inf. Comput. Sci. 30 (1990) 413–425; (t) Chematica: Ref. [6e].
[7] T. Takabatake, T. Yoneda, J. Otsuka, N. Kagawa, M. Toyota, Tetrahedron Lett. 60
(2019) 150942–150945.
Declaration of Competing Interest
The authors declare that they have no known competing finan-
cial interests or personal relationships that could have appeared
to influence the work reported in this paper.
Acknowledgement
We would like to thank Sumitomo Chemical for allowing us to
use SYNSUP.
Appendix A. Supplementary data
[8] (a) Selected total synthesis of kinamycins, see: T. Kumamoto, Y. Kitani, H.
Tsuchiya, Y. Yakaguchi, H. Seki, T. Ishikawa Tetrahedron 63 (2007) 5189–5199;
(b) K.C. Nicolaou, H. Li, A.L. Nold, D. Pappo, A. Lenzen, J. Am. Chem. Soc. 129
(2007) 10356–10357;
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.tetlet.2020.152267.
(c) KX. Lei, A.P. Porco Jr., J. Am. Chem. Soc. 128 (2006) 14790–14791, 1-
Methoxydiazo-fluorene synthesis, see: W. Zeng, T.E. Ballard, A.G. Tkachenko, V.A.
Burns, D.L Feldheim, C. Melander, Bioorg. Med. Chem. Lett. 16 (2006) 5148–
5151.
References
[9] T. Miyagi, S. Okada, N. Tada, M. Sugihara, N. Kagawa, T. Takabatake, M. Toyota,
Tetradehron Lett. 60 (2019) 1653–1657.
[10] (a) Some selected reviews: see G.P. McGlacken, L.M. Bateman Chem. Soc. Rev.
38 (2009) 2447–2467;
[1] S. Russel, P. Norvig, Artificial Intelligence. A Modern Approach, Third ed.,
Pearson, London, 2016.
[2] M. Campbell, A.J. Hoane, F.-H. Hsu, Artif. Intell. 134 (2001) 57–83.
[3] L.D. Burns, Nature 497 (2013) 181–182.
(b) D. Alberico, M.E. Scott, M. Lautens, Chem. Rev. 107 (2007) 174–238,
Selected biphenyl synthesis from 3-keto esters: see B.B. Snider, J.J. Patricia, J. Org.
Chem. 54 (1989) 38–46.
[4] Y. Chen, E. Argentinis, G. Weber, Clin. Ther. 38 (2016) 688–701.
[5] (a) Original program: M. Bersohn, Bull. Chem. Soc. Jpn. 45 (1972) 1897–1903;
(b) Selected papers related to the application: T. Takabatake, M. Bersohn, Pure
Appl. Chem. 62 (1990) 1977–1978; (c) A. Tanaka, T. Kawai, T. Takabatake, N.
Oka, H. Okamoto, M. Bersohn, Tetrahedron 63 (2007) 10226–10236.
[6] Some selected reviews: see (a) M. Bersohn, A. Esack, Chem. Rev. 76 (1976)
269–282; (b) I. Ugi, J. Bauer, K. Bley, A. Dengler, A. Deitz, E. Fontain, B. Gruber,
R. Herges, M. R. Knauer, N. Stein, Angew. Chem. Int. Ed. 32 (1993) 201–227; (c)
W.-D. hlenfeldt, J. Gasteiger, Angew. Chem. Int. Ed. 34 (1995) 2613–2633; (d)
M.H. Todd, Chem. Soc. Rev. 34 (2005) 247–266; (e) S. Szymkuc´, E.P. Gajewska,
T. Klucznik, K. Molga, P. Dittwald, M. Startek, M. Bajczyk, B.A. Grzybowski,
Angew. Chem. Int. Ed. 55 (2016) 5904–5937; Selected programs: see (f) OCSS:
E.J. Corey, T. Wipke, Science 166 (1969) 178–192; (g) LHASA: E.J. Corey, T.
Wipke, R.D. Cramer III, W.J. Howe, J. Am. Chem. Soc. 94 (1972) 421–430; (h)
SECS: T. Wipke, T.M. Dyott, J. Am. Chem. Soc. 96 (1974) 4825–4834; (i)
[11] S.N. Huckin, L. Weiler, J. Am. Chem. Soc. 96 (1974) 1082–1087.
[12] S.W. Youn, B.S. Kim, A.R. Jagdale, J. Am. Chem. Soc. 134 (2012) 11308–11311.
[13] G. Mroß, H. Reinke, C. Fischer, P. Langer, Tetrahedron 65 (2009) 3910–3917.
[14] M. Minabe, Y. Takabayashi, Y. Setta, H. Nakamura, T. Kimura, M. Tsubota, Bull.
Chem. Soc. Jpn. 69 (1996) 3633–3638.
[15] Y. Natori, M. Ito, M. Anada, H. Nambu, S. Hashimoto, Tetrahedron Lett. 56
(2015) 4324–4327.
[16] N. Miyaura, A. Suzuki, Chem. Rev. 95 (1995) 2457–2483.
[17] R.A. Alderden, M.D. Hall, T.W. Hambley, J. Chem. Educ. 83 (2006) 728–734.
[18] K.A. O’Hara, X. Wu, D. Patel, H. Liang, J.C. Yalowich, N. Chen, V. Goodfellow, O.
Adedayo, G.I. Dmitrienko, B.B. Hasinoff, Free Radical Biol. Med. 43 (2007)
1132–1144.