Dioxonaphthoimidazoliums are Potent and Selective Rogue Stem Cell Clearing Agents with SOX2-Suppressing Properties

Article abstract of DOI:10.1002/cmdc.201600262

Pluripotent stem cells are uniquely positioned for regenerative medicine, but their clinical potential can only be realized if their tumorigenic tendencies are decoupled from their pluripotent properties. Deploying small molecules to remove remnant undifferentiated pluripotent cells, which would otherwise transform into teratomas and teratomacarcinomas, offers several advantages over non-pharmacological methods. Dioxonapthoimidazolium YM155, a survivin suppressant, induced selective and potent cell death of undifferentiated stem cells. Herein, the structural requirements for stemotoxicity were investigated and found to be closely aligned with those essential for cytotoxicity in malignant cells. There was a critical reliance on the quinone and imidazolium moieties but a lesser dependence on ring substituents, which served mainly to fine-tune activity. Several potent analogues were identified which, like YM155, suppressed survivin and decreased SOX2 in stem cells. The decrease in SOX2 would cause an imbalance in pluripotent factors that could potentially prompt cells to differentiate and hence decrease the risk of aberrant teratoma formation. As phosphorylation of the NF-κB p50 subunit was also suppressed, the crosstalk between phospho-p50, SOX2, and survivin could implicate a causal role for NF-κB signaling in mediating the stem cell clearing properties of dioxonaphthoimidazoliums.

Full text of DOI:10.1002/cmdc.201600262

DOI: 10.1002/cmdc.201600262
Full Papers
Dioxonaphthoimidazoliums are Potent and Selective
Rogue Stem Cell Clearing Agents with SOX2-Suppressing
Properties
Si-Han Sherman Ho,[a] Azhar Ali,[b] Yi-Cheng Ng,[a] Kuen-Kuen Millie Lam,[c] Shu Wang,[c, d]
Woon-Khiong Chan,[c] Tan-Min Chin,[b] and Mei-Lin Go*[a]
Pluripotent stem cells are uniquely positioned for regenerative
medicine, but their clinical potential can only be realized if
their tumorigenic tendencies are decoupled from their pluripo-
tent properties. Deploying small molecules to remove remnant
undifferentiated pluripotent cells, which would otherwise
transform into teratomas and teratomacarcinomas, offers sev-
eral advantages over non-pharmacological methods. Dioxo-
napthoimidazolium YM155, a survivin suppressant, induced se-
lective and potent cell death of undifferentiated stem cells.
Herein, the structural requirements for stemotoxicity were in-
vestigated and found to be closely aligned with those essential
for cytotoxicity in malignant cells. There was a critical reliance
on the quinone and imidazolium moieties but a lesser depend-
ence on ring substituents, which served mainly to fine-tune ac-
tivity. Several potent analogues were identified which, like
YM155, suppressed survivin and decreased SOX2 in stem cells.
The decrease in SOX2 would cause an imbalance in pluripotent
factors that could potentially prompt cells to differentiate and
hence decrease the risk of aberrant teratoma formation. As
phosphorylation of the NF-kB p50 subunit was also sup-
pressed, the crosstalk between phospho-p50, SOX2, and survi-
vin could implicate a causal role for NF-kB signaling in media-
ting the stem cell clearing properties of dioxonaphthoimidazo-
liums.
Introduction
The clinical potential of stem cell therapy in regenerative medi-
cine is based on two unique characteristics of stem cells: their
capacity to proliferate indefinitely while retaining cellular iden-
tity (self-renewal) and the ability to differentiate into tissue-
specific functional cells when challenged by appropriate differ-
entiation signals (pluripotency).[1] Unfortunately, these traits
predispose stem cells toward tumorigenicity, a major hurdle
that must be overcome if human pluripotent stem cells
(hPSCs) are to be safely deployed for clinical applications.[2,3]
The tumorigenic potential of hPSCs is closely linked to the
presence of residual pluripotent cells that failed to differenti-
ate.[4] No more than a few thousand cells in differentiated cul-
tures were reportedly sufficient to induce benign teratomas in
immunodeficient mice.[5] More menacing is the risk of remnant
hPSCs transforming into malignant teratocarcinomas, due to
genetic aberrations induced by culture adaptations.[1] To miti-
gate the problem of aberrant teratogenicity, many conceptual-
ly different strategies have been pursued. These include the in-
troduction of suicide genes,[6] genetic selection to remove un-
desired cell types,[7] extending the time allowed for cells to dif-
ferentiate,[8] immunodepletion,[9] and the use of cytotoxic anti-
bodies.[10] The deployment of small molecules to selectively
remove undifferentiated pluripotent cells has received atten-
tion as a viable alternative.[11–13] This pharmacological approach
has several advantages: it is robust, rapid, and cost-effective,
maintains genetic stability of cells, minimizes cell attrition, and
has good translational potential, as it can be applied to com-
plex differentiation protocols. Figure 1A shows small molecules
that have been reported to inhibit the formation of stem-cell-
derived teratomas.
[a] Dr. S.-H. S. Ho, Y.-C. Ng, Prof. M.-L. Go
Department of Pharmacy, National University of Singapore,
18 Science Drive 4, Singapore 117543 (Singapore)
E-mail: phagoml@nus.edu.sg
The acylphenylhydrazine PluriSin #1 prevented teratoma for-
mation by a mechanism that involved inhibition of stearoyl-
CoA desaturase, a key enzyme in the biosynthesis of monosa-
turated fatty acids.[11] Disruption of this pathway led to ER
stress, unfolded protein response, translational attenuation in
PSCs, and ultimately, apoptotic cell death. The N-benzylnona-
namide JC011, a structural analogue of capsaicin, also induced
ER stress by activating the PERK/AT4/DDIT3 pathway.[12] The fla-
vonoid quercetin and dioxonaphthoimidazolium analogue
YM155 inhibited survivin, an anti-apoptotic protein whose
gene (BIRC5) is strongly linked to teratoma formation.[13,14] The
rationale for targeting survivin rests on a balance of pro- and
anti-apoptotic genes in undifferentiated stem cells.[13] These
[b] Dr. A. Ali, Dr. T.-M. Chin
Cancer Science Institute of Singapore,
National University of Singapore,
14 Medical Drive, Singapore 117599 (Singapore)
[c] Dr. K.-K. M. Lam, Prof. S. Wang, Prof. W.-K. Chan
Department of Biological Sciences, National University of Singapore,
14 Science Drive 4, Singapore 117543 (Singapore)
[d] Prof. S. Wang
Institute of Bioengineering and Nanotechnology,
31 Biopolis Way, Singapore 138669 (Singapore)
Supporting information for this article can be found under http://
dx.doi.org/10.1002/cmdc.201600262.
ChemMedChem 2016, 11, 1 – 13
1
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
scaffold. Substitution on the scaffold served to fine-tune po-
tencies, and our investigations revealed a striking preference
for small, compact alkyl substituents on the imidazolium ring.
Consequently, we identified two promising analogues—IV-
1 and IV-7—which were similar to or exceeded YM155 in
terms of growth inhibitory potencies using a panel of NSCLC
cells (Figure 1B). In this context, we asked if these structural re-
quirements were also required for potent stemotoxic activity.
From the discovery perspective, a consensus would identify
the dioxonaphthoimidazolium ring system as a hit scaffold for
stemotoxic activity. Furthermore, with growing evidence that
YM155 acts on other bona fide molecular targets besides survi-
vin,[18] overlapping structural requirements allude to the inter-
ception of one or more pathways that are common to these
different cell types. Scrutiny of these pathways should provide
valuable insight into aspects of stem cell biology that are relat-
ed to cell fate and survival.
Results and Discussion
Figure 1. A) Small molecules that have been reported to prevent teratoma
formation in stem cell cultures.[11–13] B) Structures of IV-1 and IV-7, which are
potent cytotoxic analogues of YM155.[17]
Growth inhibitory SAR of dioxonaphthoimidazolium
analogues in embryonic carcinoma cells
The synthesis and characterization of the library of YM155-re-
lated analogues investigated in this report have been de-
scribed previously.[17] Briefly, the compounds were divided into
four series (Series I–IV) based on the structural variations made
to the scaffold (Figure 2). Series I compounds were modified at
the dioxonaphthoimidazolium core. Series II–IV retained the tri-
cyclic scaffold, and were systematically varied at the substitu-
ents attached to N3 (Series II), N1 (Series III), and both N1 and
N3 (Series IV).
cells express a large number of pro-apoptotic (pro-death)
genes and relatively few anti-apoptotic (pro-survival) genes rel-
ative to their differentiated counterparts. The survival of stem
cells is thus critically dependent on a small number of resident
anti-apoptotic genes, and inhibition of these genes or their
products would induce apoptosis of undifferentiated stem
cells, hence preventing their transition into teratomas. Collater-
al evidence from YM155 and quercetin confirmed the feasibili-
ty of exploiting the differential expression of pro- and anti-
apoptotic proteins as a means of suppressing the tumorigenic
traits of hPSCs.[13] YM155 (nanomolar IC50 value) was signifi-
cantly more potent than quercetin in this regard by at least
1000-fold.
The compounds were evaluated for growth inhibitory activi-
ties in two embryonal carcinoma (EC) cell lines (NCCIT and
NTERA-2). EC cells are the malignant counterparts of embryon-
ic stem (ES) cells and are widely employed as models to inves-
tigate stem cells, because their growth requirements are less
demanding than those of ES cells.[19,20] Tables 1–4 list the
growth inhibitory IC50 values of the Series I–IV compounds in
these cell lines and in non-malignant lung fibroblast IMR-90
cells. IC50 values from the two EC cell lines were strongly corre-
lated (Spearman correlation coefficient 1=0.973, Table S1), im-
plicating a common SAR for both cell lines. We also compared
the IC50 values from the EC cell lines with those previously re-
ported in malignant NSCLC cells[17] and found, yet again, a sig-
The outstanding stemotoxic potency of YM155 closely paral-
lels its potent growth inhibitory activity in human malignant
cells.[15,16] We previously reported a comprehensive structure–
activity relationship (SAR) analysis of the in vitro cytotoxic ac-
tivity of YM155 and its analogues in non-small-cell lung cancer
(NSCLC) cells.[17] Briefly, activity was critically dependent on the
quinone moiety and, to a lesser extent, on the positively
charged imidazolium ring in the tricyclic naphthoimidazolium
Figure 2. Structural motifs of Series I–IV compounds. Substituents at N1 (R1) and N3 (R3) were varied in Series II–IV. In Series I, changes were made to the tricy-
clic scaffold. In Series II, the N3 side chain of YM155 was modified with no change to the 2’-methoxyethyl side chain at N1. In Series III, the N1 side chain was
modified while retaining the N3 2’-pyrazinylmethyl side chain. Series IV combines optimal N1 and N3 groups identified in Series II and III.
&
ChemMedChem 2016, 11, 1 – 13
www.chemmedchem.org
2
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Table 1. IC50 values of Series I compounds.
Compound
IC50 [nm][a]
NTERA-2
SI[b]
NCCIT
IMR-90[17]
YM155
2.09ꢀ0.07
1.49ꢀ0.08
247ꢀ37
142
I-1
441ꢀ32
318ꢀ20
1360ꢀ150
3.68
I-2
I-3
I-4
24000ꢀ1000
34300ꢀ5700
>100 mm
28900ꢀ3400
8290ꢀ1190
>100 mm
55400ꢀ4600
73000ꢀ4500
>100 mm
2.11
5.47
–
I-5
I-6
I-7
>100 mm
>100 mm
>100 mm
>100 mm
>100 mm
>100 mm
>100 mm
>100 mm
>100 mm
–
–
–
I-8
I-9
>100 mm
>100 mm
>100 mm
–
34.8ꢀ1.9
19.6ꢀ2.0
1900ꢀ110
75.8
[a] Evaluated by MTT assay, 72 h incubation, 378C, 5% CO2; values are the meanꢀSD of n=3 separate determinations. [b] Selectivity index: (IC50 IMR-90)/
(mean IC50 NCCIT and NTERA-2).
ChemMedChem 2016, 11, 1 – 13
www.chemmedchem.org
3
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
nificant correlation between these activities (1=0.855–0.920,
Table S1). Thus, we anticipated structural requirements for EC
and malignant cells to be broadly similar. However, a compari-
son of the median IC50 values showed that the compounds
were at least five- to tenfold more potent in EC cells and had
a wider range of potencies (10000-fold in terms of measurable
IC50 values) as compared to NSCLC cells (1000-fold). In the fol-
lowing paragraphs, SAR is discussed with reference to NCCIT
cells and focuses on comparing structural trends deduced
from NCCIT and NCSLC cells.
ing analogues (II-15 to II-21) had IC50 values that fell within
a narrow threefold range (26–60 nm), in contrast to the wider
tenfold variation (29–280 nm) for the same compounds in
NSCLC cells.[17] The tighter activity range meant that it was
harder to identify an optimal alkyl substituent. The N3-methyl
analogue II-15 is illustrative. Compound II-15 was the most
potent Series II analogue in NSCLC cells, but in NCCIT cells, it
was no better than the ethyl (II-16), n-propyl (II-17) or cyclo-
propyl (II-19) homologues.
Several SAR similarities were also detected in Series III. These
were losses in activity when the 2’-methoxyethyl side chain at
N1 was modified by demethylation (III-1), homologation (III-2),
or functionalization with substituted amino groups (III-3 and
III-4). There was also a limited tolerance for ring-bearing sub-
stituents at N1, as seen from the tetrahydrofuranylmethyl (III-
5), benzyl (III-6), and pyrazinylmethyl (III-7) compounds, and
a preference for N1-alkyl substituents that were unbranched
(III-10 versus III-14; III-11 versus III-15) and sterically compact
(cyclopropyl III-12 versus cyclopentyl III-13). In a departure
from Series II, homologation from a methyl (III-8) to an n-butyl
(III-11) moiety in Series III identified the n-propyl to be optimal.
Cyclization of n-propyl (III-10) to a cyclopropyl (III-12) did not
yield a clear advantage. No outstanding differences were ob-
served between the SAR trends for EC and NSCLC cells for this
series.
Series I compounds were designed to probe the importance
of the intact tricyclic dioxonaphthoimidazolium core. Previous-
ly, we showed that any attempt to reduce the tricyclic scaffold
to a bicyclic or monocyclic entity invariably led to significant
losses in activity, and that of the various motifs in the scaffold,
the quinone was the most critical, followed by the positively
charged imidazolium, and lastly, the distal benzene ring.[17]
These requirements were similarly observed in EC cells. Thus, I-
1, which lacked the distal phenyl ring but retained the quinone
and positively charged imidazolium entities, had nanomolar ac-
tivity (IC50 =440 nm). Analogues with an intact quinone but
lacking the positive charge (I-2 and I-3) were strikingly less
potent (micromolar IC50 values), while the reverse modification
(retaining the positive charge without the quinone) gave com-
pounds (I-4, I-5, and I-6) that were conspicuously inactive.
Indole analogues I-7 and I-8, which lacked both features, were
predictably without activity.
Series IV incorporated highly ranked substituents identified
from Series II (pyridin-3- or 4-ylmethyl, pyridimin-5-ylmethyl at
N3) and Series III (ethyl or cyclopropyl at N1), based on growth
potencies determined in NSCLC cells.[17] These substituents
were also optimal in EC cells and were thus investigated to de-
termine if their concurrent presence on the same molecule
would result in superior activity. Disappointingly, this was not
the case, and only one analogue (IV-1, IC50 =2.6 nm) narrowly
approached the activity of YM155 (IC50 =2.1 nm). The N1,N3-di-
methyl analogue IV-7 should be mentioned, because it was
the most potent analogue in NSCLC cells (IC50 =8–18 nm
versus 14–36 nm for YM155)[17] but had mediocre activity in EC
cells (IC50 =6.1–24 nm versus 1.5–2.1 nm for YM155).
Modifying the substituents on the scaffold, as observed here
with the C2-ethyl homologue of YM155 (I-9), resulted in only
a moderate (tenfold) loss in potency. This was further con-
firmed in Series II and III, where more extensive side chain var-
iations were made at the imidazolium nitrogen atoms (Tables 2
and 3). IC50 values ranged from 2 to 2400 nm for Series II
(median 24 nm) and from 5 to 1400 nm (median 37 nm) for
Series III. Tellingly, even the weakest analogues in both series
retained single digit micromolar growth inhibitory activities,
underscoring a largely differentiating role for substituents on
the scaffold.
Analysis of the Series II compounds revealed several structur-
al requirements that were common for NSCLC cells. We noted
a preference for azinylmethyl side chains at N3. Analogues
with non-azinylmethyl residues (benzyl II-1, cyclohexylmethyl
II-2, thienylmethyl II-3, and imidazolylmethyl II-4) were less
active. Non-ring-bearing side chains (II-15 to II-22) were also
not favored. We further noted a regioisomeric bias against
azines with ortho-azomethine nitrogen atoms (II-5 and II-8
versus II-6, II-7, and II-9–II-11). Interestingly, the adverse effect
of an ortho-nitrogen could be overruled by the concurrent
presence of a para-nitrogen (II-9) or a meta-nitrogen (YM155).
The importance of the para-nitrogen was further highlighted
in II-12, where a methyl substituent at the para position of the
pyrazinylmethyl ring elicited a steep loss in activity. Additional-
ly, we noted the importance of maintaining a methylene linker
between N3 and pyrazine, as seen from the diminished activi-
ties of homologues II-13 and II-14.
In summary, we found a striking overlap in the structural re-
quirements for potent growth inhibitory activities in dioxo-
naphthoimidazolium analogues in both EC and NSCLC cell
lines. In both instances, reducing the tricyclic scaffold to a bicy-
clic or monocyclic entity resulted in significant losses in activi-
ty. Thus, the intact dioxonaphthoimidazolium scaffold was obli-
gatory for potent activity. Of the various motifs in the scaffold,
the quinone was the most critical, followed by the charged
imidazolium ring, and lastly, the distal benzene ring. Function-
alization at N1 and N3 served mainly to fine-tune activity, and
potency differences in the resulting compounds were narrow
and still within the nanomolar range. Azinylmethyl side chains
were preferred at N3, whereas unbranched non-cyclized alkyl
groups were favored at N1. Several compounds (II-7, II-9, III-
12, and IV-1) were comparable to YM155 in terms of activity
but none exceeded it by twofold or more.
Differences in structural requirements were also recognized,
but these were mainly quantitative. Notably, the N3-alkyl-bear-
&
ChemMedChem 2016, 11, 1 – 13
www.chemmedchem.org
4
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Table 2. IC50 values of Series II compounds.
Compd
R3
IC50 [nm][a]
NTERA-2
SI[b]
NCCIT
IMR-90[17]
II-1
II-2
43.7ꢀ3.2
31.9ꢀ3.9
30.8ꢀ3.4
1160ꢀ70
31.4
75.6ꢀ11.8
53.6ꢀ1.5
5270ꢀ960
1450ꢀ50
120
II-3
44.2ꢀ7.3
29.9
27.2
II-4
II-5
II-6
II-7
II-8
II-9
II-10
76.6ꢀ8.6
19.7ꢀ3.3
5.46ꢀ0.32
1.99ꢀ0.26
25.7ꢀ4.7
2.23ꢀ0.11
6.22ꢀ0.58
71.6ꢀ1.5
13.8ꢀ1.0
3.79ꢀ0.55
1.55ꢀ0.15
9.36ꢀ1.83
1.98ꢀ0.40
4.62ꢀ0.22
2010ꢀ110
2520ꢀ110
611ꢀ44
176ꢀ9
155
137
101
232
3180ꢀ40
205ꢀ14
276ꢀ27
97.7
52.1
42.7
II-11
II-12
5.51ꢀ0.99
20.6ꢀ1.98
3.63ꢀ0.43
10.6ꢀ0.8
187ꢀ12
2000ꢀ130
143
II-13
II-14
104ꢀ8
52.0ꢀ2.4
2810ꢀ520
40.5
2411ꢀ175
2803ꢀ506
8820ꢀ1210
3.40
II-15
II-16
II-17
II-18
26.2ꢀ0.9
22.2ꢀ2.2
24.2ꢀ3.5
45.1ꢀ4.0
12.3ꢀ1.6
12.9ꢀ1.5
16.4ꢀ3
345ꢀ38
1230ꢀ130
1720ꢀ83
2640ꢀ540
20.6
75.4
88.0
17.3ꢀ1.0
106
II-19
II-20
II-21
24.0ꢀ0.6
60.3ꢀ2.8
57.1ꢀ7.0
12.0ꢀ2.3
35.3ꢀ5.9
23.2ꢀ3.7
1160ꢀ140
3890ꢀ290
6790ꢀ290
72.5
87.3
206
ChemMedChem 2016, 11, 1 – 13
www.chemmedchem.org
5
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Table 2. (Continued)
Compd
R3
IC50 [nm][a]
NTERA-2
SI[b]
NCCIT
IMR-90[17]
II-22
17.8ꢀ1.94
13.2ꢀ0.9
743ꢀ89
49.0
[a] Evaluated by MTT assay, 72 h incubation, 378C, 5% CO2; values are the meanꢀSD of n=3 separate determinations. The range of measurable values
was 1.99 nm–2.4 mm for NCCIT (median 25 nm) and 1.55 nm–2.8 mm for NTERA-2 (median 13.5 nm). [b] Selectivity index: (IC50 IMR-90)/(mean IC50 NCCIT and
NTERA-2).
Table 3. IC50 values of Series III compounds.
Compd
R1
IC50 [nm][a]
NTERA-2
SI[b]
NCCIT
IMR-90[17]
2480ꢀ320
1430ꢀ260
III-1
III-2
25.6ꢀ2.8
21.9ꢀ1.04
21.1ꢀ3.3
13.8ꢀ0.7
107
84.5
14.6
III-3
III-4
501ꢀ90
512ꢀ30
919ꢀ74
7420ꢀ760
3730ꢀ110
1364ꢀ71
3.40
III-5
III-6
188ꢀ9
149ꢀ5
56.4ꢀ9.6
105ꢀ14
9890ꢀ170
5610ꢀ380
114
45.5
III-7
59.1ꢀ5.7
19.0ꢀ1.8
3570ꢀ710
124
III-8
III-9
13.8ꢀ1.8
8.86ꢀ0.85
5.71ꢀ0.82
10.2ꢀ1.1
7.62ꢀ1.37
3.63ꢀ0.56
3.45ꢀ0.67
4.69ꢀ0.26
582ꢀ89
480ꢀ22
59
93.2
III-10
III-11
565ꢀ27
131
252
75.3
1620ꢀ140
III-12
III-13
4.88ꢀ0.57
158ꢀ3
1.99ꢀ0.30
109ꢀ23
213ꢀ20
6580ꢀ180
50.6
86.7
III-14
III-15
20.9ꢀ2.1
48.4ꢀ1.4
13.5ꢀ2.8
31.4ꢀ2.9
1420ꢀ150
7970ꢀ180
209
[a] Evaluated by MTT assay, 72 h incubation, 378C, 5% CO2; values are the meanꢀSD of n=3 separate determinations. Range of measurable values was
4.88 nm–1.36 mm for NCCIT (median 25.6 nm) and 1.99 nm–919 nm for NTERA-2 (median 19.0 nm). [b] Selectivity index: (IC50 IMR-90)/(mean IC50 NCCIT and
NTERA-2).
&
ChemMedChem 2016, 11, 1 – 13
www.chemmedchem.org
6
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Table 4. IC50 values of Series IV compounds.
Compound
IC50 [nm][a]
NTERA-2
SI[b]
NCCIT
IMR-90[17]
IV-1
IV-2
IV-3
2.46ꢀ0.20
2.56ꢀ0.25
4.44ꢀ0.35
34.6ꢀ0.8
287ꢀ59
115
7.61ꢀ1.03
40.5ꢀ2.4
20.1ꢀ2.1
310ꢀ51
649ꢀ79
440ꢀ7
55.3
17.4
25.4
IV-4
IV-5
IV-6
15.2ꢀ0.9
9.15ꢀ0.92
5.72ꢀ0.48
10.4ꢀ1.4
484ꢀ64
475ꢀ91
49.7
69.9
8.38ꢀ0.54
IV-7
24.2ꢀ2.0
6.08ꢀ0.60
1.49ꢀ0.08
159ꢀ16
247ꢀ37
16.4
YM155
2.09ꢀ0.07
142
[a] Evaluated by MTT assay, 72 h incubation, 378C, 5% CO2; values are the meanꢀSD of n=3 separate determinations. [b] Selectivity index: (IC50 IMR-90)/
(mean IC50 NCCIT and NTERA-2).
Selective cytotoxicity for embryonic carcinoma cells
from 2 (I-2) to 250 (III-11). Tellingly, they were significantly
Tables 1–4 also list the selectivity ratios of Series I–IV com-
pounds, based on a comparison of IC50 values derived from
human lung fibroblast IMR-90 cells and the mean IC50 values
from NCCIT and NTERA cells. YM155 and its analogues were se-
lectively more cytotoxic toward EC cells, with ratios ranging
more selective against EC cells than NSCLC cells, in which no
more than a 20-fold variation (2 to 38) was observed.[17] We
were further encouraged to find that potent analogues like II-
7, II-9, III-12, and IV-1 were at least 70-fold more cytotoxic
toward EC cells. More intriguing were the outstanding selectiv-
ChemMedChem 2016, 11, 1 – 13
www.chemmedchem.org
7
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
ities (ꢁ200) of II-8, II-21, III-11, and III-15 (IC50 values of 5–
57 nm), which were admittedly less potent than YM155 but
promising stemotoxic candidates in view of their exceptionally
selective cytotoxicities.
Growth inhibitory activity of YM155 and selected analogues
toward other stem cell lines
We next proceeded to determine if the potent growth inhibi-
tory activities of the YM155 analogues would be replicated in
other stem cell types. Hence, YM155 and seven analogues (II-7,
II-13, III-6, III-13, IV-1, IV-2, and IV-7) were evaluated in
a human ES cell line (H9) and an induced pluripotent stem
(iPS) cell line (HCT-83.11). HCT-83.11 was derived from its non-
stem-cell progenitor HCT-8, a colorectal carcinoma cell line, by
transfection of the Yamanaka factors (SOX2, OCT4, KLF4, and c-
MYC).[21,22] Results are given in Table 5.
Figure 3. Morphology of H9–MEF co-cultures following 24 h treatment with
5 and 50 nm A) YM155 and B) IV-1. Control co-cultures were treated with
0.5% DMSO in media. The characteristic clear demarcation of H9 colonies
were lost at 50 nm YM155 and IV-1. Images presented are representative of
triplicate experiments performed under each treatment condition. Scale bar
(500 mm) for YM155 control applies to other figures.
Table 5. IC50 values of YM155 and analogues on embryonic stem cell line
H9, induced pluripotent stem cell line HCT-83.11, and non-stem cell pro-
genitor HCT-8 cells.
the boundaries demarcating the H9 colonies became progres-
sively blurred, indicating that the H9 cells had lost their viabili-
ties. This was particularly marked in cells treated with 50 nm
YM155 or IV-1. In contrast, the MEF cells retained their mor-
phology and remained viable under these conditions. Thus,
the cytotoxicities of IV-1 and YM155 were selectively directed
against H9 cells, as was observed earlier in EC cells.
Compd
IC50 [nm][a]
H9[b]
HCT-83.11[c]
HCT-8[c]
NTERA-2/NCCIT[d]
YM155
II-7
II-13
III-6
III-13
IV-1
IV-2
10.0ꢀ1.6
11.6ꢀ1.6
174ꢀ7
6.43ꢀ0.97
7.26ꢀ1.32
21.8ꢀ2.6
11.8ꢀ0.5
1.5/2.1
1.6/2.0
52/104
110/150
110/160
2.5/2.6
4.4/7.6
6.1/24
215ꢀ41
1210ꢀ152
2130ꢀ367
1250ꢀ269
15.4ꢀ2.3
28.9ꢀ1.7
32.9ꢀ1.5
739ꢀ148
156ꢀ31
604ꢀ80
367ꢀ71
4.21ꢀ0.49
4.31ꢀ0.54
24.9ꢀ0.2
5.47ꢀ0.80
7.66ꢀ1.00
13.9ꢀ1.5
IV-7
YM155, IV-1, and IV-7 decrease SOX2 expression in an EC
cell line
[a] Values are the meanꢀSD of n=3 separate determinations. [b] Deter-
mined by MTT assay after 24 h incubation. [c] Determined by MTT assay
after 72 h incubation. [d] IC50 values of EC cells were included for compar-
ison.
To inhibit the formation of stem cell-derived teratomas, an
agent should potently and selectively eliminate residual stem
cells that are intermixed with terminally differentiated cell pop-
ulations designated for clinical use. If the remnant cells could
also be induced to differentiate, this would further repress
their tumorigenic potential.[1] Pluripotency is regulated by tran-
scription factors such as OCT4, SOX2, MYC, KLF4, NANOG, and
LIN28, combinations of which have been successfully used to
induce the reversion of somatic cells to a stem-cell-like
state.[21–23] Many of these transcription factors are oncogenes
or are highly expressed in various types of cancer. Many onco-
genic signaling pathways are intercepted by YM155.[24,25] Thus,
we asked if YM155 and its analogues could augment their cy-
totoxicity toward stem cells by suppressing pluripotent mark-
ers, promoting cell differentiation, and thus efficiently obviat-
ing teratoma formation. Hence, we examined the effects of
YM155 and its potent analogue, IV-1, on OCT4, SOX2, and
NANOG, which are core components of the regulatory network
responsible for maintaining the pluripotent state of stem
cells.[26,27]
The analogues were selected to cover a range of potencies,
as determined by IC50 values in EC cells: II-7 and IV-1 (potent,
IC50 =2–3 nm, comparable to YM155); IV-2 and IV-7 (moderate-
ly potent, IC50 =5-25 nm); and II-13, III-6, and III-13 (weakly
potent, IC50 =50–160 nm).[17] The results showed that the test
compounds, including YM155, were generally less potent
toward ES (H9) and iPS (HCT-83.11) cells. Changes in the rank
order of potencies were also detected and were most evident
among the more potent compounds. Thus, only IV-1 and not
II-7 was highly ranked against H9 and HCT-83.11. No change
was found in the rank order of the weakly potent analogues
which remained poorly active in both H9 and HCT-83.11 cells.
The analogues displayed differential growth activities toward
HCT-8 and HCT-83.11, with greater potencies (1.5- to fivefold
higher) in the iPS cell line.
Figure 3 shows the effects of IV-1 and YM155 on the mor-
phology of a co-culture of H9 and mouse embryonic fibro-
blasts (MEF), which were feeder cells for H9. In untreated cells,
the spindle-shaped MEF cells were readily distinguished from
the colony-forming H9 cells. When treated with YM155 or IV-1,
First, we probed the mRNA expression of these transcription
factors using qRT-PCR in NCCIT, H9, and HCT-83.11 cells treated
with YM155, IV-1, and IV-7 at their IC50 concentrations. The
analysis was carried out at 12 h and 24 h for NCCIT and H9
&
ChemMedChem 2016, 11, 1 – 13
www.chemmedchem.org
8
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Taken together, we deduced that SOX2 was most suscepti-
ble to perturbation. It was reduced by all three test com-
pounds in treated NCCIT cells, by YM155 and IV-7 in H9 cells,
and by YM155 only in HCT-83.11 cells. In contrast, OCT4 levels
were unchanged in NCCIT and H9 cells but increased in HCT8
3.11. The susceptibility of NANOG to perturbation was inter-
mediate, as it was observed in two cell lines, namely H9
(YM155 caused a reduction) and HCT-83.11 (all test com-
pounds resulted in increases).
As the transcription of SOX2 mRNA was diminished in most
of the treated cell lines, we proceeded to determine if SOX2
protein levels would be similarly affected. To this end, we treat-
ed NCCIT and HCT-83.11 cells with YM155 and IV-1 at their IC50
and 2ꢁIC50 concentrations for 48 h and probed for SOX2 by
immunoblotting. As shown in Figure 5, SOX2 levels were
Figure 4. Expression of SOX2, OCT4, and NANOG mRNAs following 12 or
24 h of treatment with YM155, IV-1, or IV-7, quantified by qRT-PCR. IC50
values at 24 h were found to be: YM155: 10 nm (NCCIT and H9) and 15 nm
(HCT-83.11); IV-1: 30 nm (NCCIT), 4 nm (H9), and 20 nm (HCT-83.11); and IV-
7: 150 nm (NCCIT), 25 nm (H9), and 40 nm (HCT-83.11). mRNA levels of the
housekeeping gene GAPDH were used for normalization. Bars represent the
fold change of cDNA normalized against GAPDH in the same sample. Error
bars represent the SD of three separate experiments. Significant statistical
difference from vehicle control: *p<0.05, **p<0.01, ***p<0.001 (Tukey
post-hoc test of respective populations of treated groups vs. control).
cells but limited to 12 h for the iPS cell line. As shorter treat-
ment times were deployed, the IC50 values of these com-
pounds were re-determined at 24 h to ensure that pharmaco-
logically relevant concentrations were used.
Figure 5. Levels of cleaved caspase 3, survivin, SOX2, and NF-kB subunits
p50, p65, and p105 and their phosphorylated forms (p.p50, p.p65) in NCCIT
and HCT-83.11 cells after treatment with YM155 and IV-1 for 48 h. GAPDH
was used as a loading control. YM155 and IV-1 were tested at their IC50 and
2ꢁIC50 concentrations. IC50 values of YM155 and IV-1 in NCCIT and HCT-8
3.11 cells are mentioned in Table 5.
Figure 4 shows the fold change in mRNA expression of
SOX2, OCT4, and NANOG in treated cells after incubation. Vari-
able trends were observed depending on cell type and test
compound. On the basis of cell type, we noted that SOX2
mRNA levels in treated NCCIT cells were significantly reduced
at both time points. In contrast, OCT4 and NANOG levels were
unchanged. In H9 cells, only YM155 significantly reduced the
transcription of SOX2 and NANOG within 12 h. The transcrip-
tion of SOX2 in H9 was also reduced by IV-7, but only after
24 h. None of the transcription factors in H9 were perturbed
by IV-1. In HCT-83.11, we found that only YM155-treated cells
showed a decrease in SOX2 mRNA. Thus, YM155 consistently
reduced SOX2 levels in all cell lines (NCCIT, H9, and HCT83.11).
In HCT-83.11 cells, we noted increases in OCT4 and NANOG,
a trend that was not found with the other stem cell lines. As
a significant proportion of OCT4 and NANOG expression in
HCT-83.11 was due to transfected genes, we speculated that
these genes may not be regulated in the same way as the
native gene, which could have given rise to the observed in-
creases.
indeed reduced by YM155 and IV-1 in both cell lines. We also
monitored the levels of the apoptotic marker proteins survivin
and cleaved caspase 3. The losses in survivin and appearance
of cleaved caspase 3 recapitulate the induction of apoptosis by
YM155 and IV-1.
The pluripotent state of human embryonic stem cells is
maintained by a network of transcription factors, of which
SOX2, OCT4, and NANOG have been widely identified as core
components.[28,29] The prevailing model advocates cooperativi-
ty between the pluripotency factors as central to self-renewal
and pluripotency,[27] while a contrarian view proposes a modu-
lar pluripotency network in hESCs in which each transcription
factor controls specific cell fates, and there is limited overlap in
the regulation of their gene expression.[30] Clearly, there are
gaps in our understanding of the molecular mechanisms and
interactions underlying the roles of these factors in pluripoten-
ChemMedChem 2016, 11, 1 – 13
www.chemmedchem.org
9
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
cy. Our results showed that SOX2 was selectively targeted by
YM155 and its potent analogues in different stem cell types,
with significant decreases elicited at both transcriptional and
protein levels. In our view, the ensuing imbalance in pluripo-
tency factors (unchanged OCT4 and NANOG, decrease in
SOX2) would invariably affect the ability of the cells to remain
in their pluripotent states.
action would influence cell fate and, if cells are thus prompted
to differentiate, the risk of remnant undifferentiated cells trans-
forming into the malignant phenotype would be lowered. Di-
oxonaphthoimidazoliums would then have a dual approach in
overcoming the tumorigenicity hurdle: inducing apoptosis and
promoting terminal differentiation of residual stem cells.
The overlapping structural requirements for stemotoxic and
cytotoxic activities implicate the involvement of similar mecha-
nisms in the induction of cell death by the dioxonaphthoimi-
dazolium analogues. In this regard, the interception of NF-kB
by YM155 and IV-1 in both stem cells and malignant cells was
interesting. NF-kB signaling is augmented in stem cells and is
important for maintaining their self-renewal and pluripotency.
Disruption of this pathway was reported to significantly de-
crease SOX2, NANOG, and OCT4 in ES and iPS cells.[32,33] Survi-
vin overexpression is the biological sequela of NF-kB activation
in several malignancies.[34] Taken together, the ability of YM155
and IV-1 to concurrently suppress survivin, phospho-p50, and
SOX2 in stem cells (NCCIT, HCT-83.11) implicate a causal role
for the NF-kB pathway. Inhibition of p50 phosphorylation
would disrupt gene transcription, and if the affected gene
products are involved in cell differentiation and survival (like
SOX2 and survivin), teratoma formation would be effectively
curtailed.
YM155 and IV-1 suppress activation of NF-kB subunit p50 in
NCCIT and HCT-83.11 cells
We have previously shown that YM155 and IV-1 intercepted
the phosphorylation of the NF-kB subunit p50, which regulates
the binding of NF-kB dimers to DNA and transcription of NF-
kB-controlled genes in NSCLC cell lines.[31] p50 forms a hetero-
dimer with p65, another NF-kB subunit, which then binds to
DNA sequences on NF-kB response elements to trigger tran-
scription of NF-kB-controlled genes. The phosphorylation of
p50 controls the binding process, whereas phosphorylation of
p65 affects transcription. Interception of either process inhibits
the signaling pathway. We proposed that the pleiotropy that is
increasingly cited for YM155 may be attributed in part to the
intervention of this pathway. NF-kB signaling has wide-ranging
effects on the expression of apoptotic proteins like survivin,
Mcl-1, and Bcl-xl, which are downregulated by YM155 and its
analogues.[31]
The role of NF-kB signaling in maintaining pluripotency of
iPSCs was highlighted by Takase et al.,[32] who showed that the
pathway was upregulated in undifferentiated iPSCs. When sup-
pressed by p65 siRNA, expression of OCT4 and NANOG was re-
duced. Furthermore, disrupting the NF-kB pathway in hESCs
led to the downregulation of SOX2, OCT4, and NANOG.[33]
Within this context, we were curious if our earlier observations
on the interception of p50 phosphorylation by YM155 and IV-
1 in malignant cells extended to hESCs as well. Thus, we
probed the levels of p65, p50, p105 (a precursor of p50), phos-
pho-p65, and phospho-p50 in NCCIT and HCT83.11 cells treat-
ed with YM155 and IV-1 under conditions described in the pre-
ceding section (Figure 5). Our results showed no change in
p65 and its activated state (phospho-p65) in the treated cells.
Levels of p50 and p105, the inactive precursor of p50, were
similarly unaltered. On the other hand, phospho-p50 was sig-
nificantly reduced by both compounds, which coincided with
our earlier findings in malignant cells.
Experimental Section
Synthesis and characterization of Series I–IV compounds: Syn-
thesis and characterization of compounds in Series I–IV were de-
scribed previously.[17] Compounds were characterized by 1H NMR
and 13C NMR spectroscopy, nominal or high-resolution masses
were determined by LC–MS, and purity was assessed by reverse-
phase HPLC on two solvent systems and found to be ꢁ95%. A
brief overview of the synthetic schemes for Series II, III, and IV and
spectroscopic data for potent analogues IV-1 and IV-7 are present-
ed in the Supporting Information.
Cell culture conditions for NCCIT, NTERA-2, HCT-8, and HCT-
83.11: Human pluripotent EC cell lines (NCCIT and NTERA-2) and
human colorectal carcinoma cells line HCT-8 were obtained from
American Type Culture Collection (ATCC, Manassas, VA, USA). HCT-
83.11, an induced pluripotent stem cell line, was generated from
HCT-8 using a baculovirus vector to transfect Yamanaka factors
into the parental line. RPMI-1640 and DMEM were supplemented
with 10% heat-inactivated fetal bovine serum (FBS) and 0.01% w/v
penicillin G–streptomycin before culturing. NCCIT cells were main-
tained in RPMI-1640, while NTERA-2 and HCT-8 cells were main-
tained in DMEM. The DMEM/F-12 (1:1) media (Hyclone, GE Health-
care, Buckinghamshire, UK) was supplemented with 1% FBS,
0.005% penicillin G–streptomycin, 5 mL GlutaMax (Gibco, Life Tech-
nologies Corporation, Carlsbad, CA, USA), and epidermal growth
factor (Peprotech, Rocky Hill, NJ, USA) for the growth of HCT-83.11
cells. Cells were passaged upon reaching 80% confluence.
Conclusions
Our investigations provided insight into the potential of
YM155 and its analogues as agents for the eradication of rem-
nant stem cells from differentiated cell populations. The effi-
cient and selective induction of cell death of human pluripo-
tent stem cells, first observed with YM155, was shown here to
be closely associated with other dioxonaphthoimidazolium an-
alogues. A definitive SAR that closely parallels that established
for normal malignant cells was evident. YM155 and its potent
analogue, IV-1, induced an imbalance in pluripotent factors by
downregulating SOX2 at gene and protein levels. Such an
Cell culture conditions for H9: The H9 and MEF cell lines originat-
ed from WiCell Research Institute (Madison, WI, USA). H9 cells were
maintained in DMEM/F-12 (1:1) (Gibco), prepared according to the
formulation described in the Supporting Information (Table S2).
MEF cells were maintained in high glucose DMEM (Gibco), supple-
mented with 10% FBS, 2 mm l-glutamine, and 0.1 mm nonessen-
tial amino acids solution. Culture dishes (35 mm) were coated with
&
ChemMedChem 2016, 11, 1 – 13
www.chemmedchem.org
10
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
0.1% gelatin (1 mL) and incubated at 378C for 1 h. Gamma-irradiat-
ed MEF cells were seeded in each gelatin-coated dish (3ꢁ105 cells
per dish) and incubated for 24 h for adhesion to the dish. Post-ad-
hesion, the MEF layer was washed once with phosphate-buffered
saline (PBS) and ~100–120 H9 colonies, suspended in H9 media,
were seeded. Media was changed daily, and passaging was per-
formed every 5–7 days, depending on the size of the colonies.
During passaging, each large colony was carefully separated into
200–300 mm pieces using a small scalpel. Roughly 100–120 pieces
were then collected and seeded on another culture dish previously
seeded with MEF cells.
2.11, with concentrations ranging from ~400 to ~1200 ngmLꢂ1. Re-
sults from an Agilent Bioanalyzer 2000 analysis (Agilent Technolo-
gies, Santa Clara, CA, USA) showed that all samples had RNA integ-
rity numbers (RINs) of 9.4–10. RNA (1.7 mg) was withdrawn from
each sample for cDNA synthesis using Promega M-MLV reverse
transcriptase following the recommended protocol (Promega Life
Sciences, Madison, WI, USA). Five primers, corresponding to SOX2,
NANOG, OCT4, and GAPDH, were obtained from AIT Biotech (Sin-
gapore) based on sequences sourced from PrimerBank (Harvard
Medical School, Boston, MA, USA). PCR was carried out in a volume
of 20 mL using SYBR Green dye on an Applied Biosystem 7500 Fast
Real Time PCR system (Thermo Fisher Scientific, Waltham, MA,
USA). Data analysis was performed using a comparative CT (thresh-
old cycle) method and normalized to GAPDH expression levels.
Growth inhibitory MTT assay: MEFs were seeded in each well of
a 24-well plate (5ꢁ104 cells per well) and left overnight for attach-
ment. H9 cells were then seeded at a density of 8ꢁ104 cells per
well. The co-culture was incubated for 48 h, with media changed
every 24 h, before treatment with test compounds. Other cell lines
were seeded in 96-well plates at the following densities and incu-
bated for 24 h before treatment: 3ꢁ103 (HCT-8 and HCT-83.11), 4ꢁ
103 (NTERA-2, IMR-90) and 6ꢁ103 (NCCIT) cells per well. Thereafter,
media in each well was aspirated and replaced with 497.5 mL (H9)
or 199 mL (other cell lines) of media and 2.5 mL (H9) or 1 mL (other
cell lines) of test compound (prepared in DMSO stock solution at
a 200-fold higher concentration). After 24 h (H9) or 72 h (other cell
lines), media was aspirated before adding an aliquot of
0.5 mgmLꢂ1 MTT (Alfa Aesar, Lancashire, UK) in media and incubat-
ing for 2–3 h. The resulting formazan crystals were dissolved in
DMSO and quantified at 570 nm on a microplate reader (Tecan In-
finite M2000 Pro, Mꢂnnedorf, Switzerland). Viable cells were deter-
mined from the following equation for HCT-8, HCT-83.11, NCCIT,
and NTERA-2 [Eq. (1)]:
Western blotting: Cells were seeded at the following densities
(per dish) in 100 mm petri dishes and incubated for 24 h for at-
tachment: NCCIT=6ꢁ105 cells, HCT-8 3.11=1.2ꢁ106 cells. Media
was aspirated and replaced by 10 mL fresh media containing test
compound. After 48 h incubation, cells were harvested and lysed
in Cellytic M buffer (Sigma–Aldrich, St. Louis, MO, USA). Protein
content was assessed by Bradford assay (Bio-Rad, Hercules, CA,
USA) and subjected to SDS-PAGE after standardization of protein
content. Proteins were blocked in 5% nonfat milk and probed with
antibodies against cleaved caspase 3, survivin, native and phos-
phorylated forms of p50/p105, p65 (Cell Signaling Technology,
Danvers, MA, USA), and SOX2 (Santa Cruz Biotech, Santa Cruz, CA,
USA). Anti-GADPH antibody (Santa Cruz Biotech) was used as
a loading control. WesternBright ECL or Quantum (Advansta,
Menlo Park, CA, USA) were added following incubation with horse-
radish peroxidase-conjugated anti-mouse, anti-rabbit, and anti-
goat secondary antibodies (Santa Cruz Biotech) to visualize bands.
Acompound ꢂ Ablank
ð1Þ
Percentage viability ¼
ꢃ 100 %
Acontrol ꢂ Ablank
Acknowledgements
for which Acompound is the average absorbance of compound-treated
cells, Acontrol is the average absorbance of untreated (control) cells,
and Ablank is the average of absorbance of DMSO.
For H9, a set of background controls were prepared by seeding
only MEF cells in the well. Each H9–MEF well was compared with
a MEF control well (both subjected to the same treatment condi-
tions) to isolate the response of H9 to treatment using Equa-
tion (2):
This work was supported by a Ministry of Education Faculty Re-
search Grant (R148000188112) to M.L.G. S.H. gratefully acknowl-
edges scholarship support from the Ministry of Education, Singa-
pore.
Keywords: antiproliferation
·
dioxonaphthoimidazoliums
·
factor SOX2 · stem cell clearing agents · structure–activity
relationships
At ꢂ Ablank;t
Percentage viability ¼
ꢃ 100 %
ð2Þ
Acontrol ꢂ Ablank; control
for which At is the average absorbance of compound-treated H9–
MEF wells, Ablank,t is the average absorbance of compound-treated
MEF-only wells, Acontrol is the average absorbance of untreated (con-
trol) H9–MEF wells, and Ablank,control is the average absorbance of un-
treated (control) MEF-only wells.
The IC50 value (the concentration of test compound required to in-
hibit cell growth by 50%) was determined by plotting the percent-
age viability against the logarithmic concentration of test com-
pound using GraphPad Prism 6 (GraphPad Software, Inc., USA).
[1] U. Ben-David, N. Benvenisty, Nat. Rev. Cancer 2011, 11, 268–277.
[2] P. S. Knoepfler, Stem Cells 2009, 27, 1050–1056.
[3] J. M. Brickman, T. G. Burdon, Nat. Genet. 2002, 32, 557–558.
[4] K. Miura, Y. Okada, T. Aoi, A. Okada, K. Takahashi, K. Okita, M. Nakagawa,
M. Koyanagi, K. Tanabe, M. Ohnuki, D. Ogawa, E. Ikeda, H. Okano, S. Ya-
manaka, Nat. Biotechnol. 2009, 27, 743–745.
[5] A. S. Lee, C. Tang, F. Cao, X. Xie, K. van der Bogt, A. Hwang, A. J. Connol-
ly, R. C. Robbins, J. C. Wu, Cell Cycle 2009, 8, 2608–2612.
[6] O. Naujok, J. Kaldrack, T. Taivankhuu, A. Jorns, S. Lenzen, Stem Cell Rev.
2010, 6, 450–461.
qRT-PCR: NCCIT, H9, and HCT-83.11 cells were seeded at 8.0ꢁ105
cells per well in six-well plates for 24 h followed by exposure to
test compound for 12 h or 24 h. RNA extraction was performed fol-
lowing the manufacturer’s protocol (RNeasy Mini Kit, Qiagen,
Venlo, Netherlands). RNA content was quantified using a ND-1000
spectrophotometer (NanoDrop Technologies, Wilmington, DE,
USA). All samples showed OD260/OD280 ratios between 1.73 and
[7] S. Chung, B. S. Shin, E. Hedlund, J. Pruszak, A. Ferree, U. J. Kang, O. Isac-
son, K. S. Kim, J. Neurochem. 2006, 97, 1467–1480.
[8] D. Scadden, A. Srivastava, Cell Stem Cell 2012, 10, 149–150.
[9] C. Tang, A. S. Lee, J. P. Volkmer, D. Sahoo, D. Nag, A. R. Mosley, M. A.
Inlay, R. Ardehali, S. L. Chavez, R. R. Pera, B. Behr, J. C. Wu, I. L. Weissan,
M. Drukker, Nat. Biotechnol. 2011, 29, 829–834.
[10] A. B. Choo, H. L. Tan, S. N. Ang, W. J. Fong, A. Chin, J. Lo, L. Zheng, H.
Hentze, R. J. Philp, S. K. Oh, M. Yap, Stem Cells 2008, 26, 1454–1463.
ChemMedChem 2016, 11, 1 – 13
www.chemmedchem.org
11
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
[11] U. Ben-David, Q. F. Gan, T. Golan-Lev, P. Arora, O. Yanuka, Y. S. Oren, A.
Leikin Frenkel, M. Graf, R. Garippa, M. Boehringer, G. Gromo, N. Benve-
nisty, Cell Stem Cell 2013, 12, 167–179.
[23] J. Yu, M. A. Vodyanik, K. Smuga-Otto, J. Antoslewicz-Bourget, J. L. Frane,
S. Tian, J. Nie, G. A. Jonsdottir, V. Routti, R. Stewart, I. I. Slukvin, J. A.
Thompson, Sci. 2007, 318, 1917–1920.
[12] M. Richards, C. W. Phoon, G. T. W. Goh, E. K. Seng, X. M. Guo, C. M. F. Tan,
W. K. Chan, J. M. K. Lee, PLOS ONE 2014, 9, e85039.
[24] M. Asahi, S. Ito, Y. Ishida, Clin. Lymphoma Myeloma Leuk. 2015, 15, e237.
[25] V. Wagner, D. Hose, A. Seckinger, L. Weiz, T. Meibner, T. Reme, I. Breitk-
reutz, K. Podar, A. D. Ho, H. Goldschmidt, A. Kramer, B. Klein, M. S. Raab,
Oncotarget 2014, 5, 10237–10250.
[26] K. M. Loh, B. Lim, Cell Stem Cell 2011, 8, 363–369.
[27] R. Jaenisch, R. Young, Cell 2008, 132, 567–582.
[28] K. Takahashi, S. Yamanaka, Nat. Rev. Mol. Cell. Biol. 2016, 17, 183–193.
[29] S. Zhang, W. Cui, World J. Stem Cells 2014, 6, 305–311.
[30] Z. Wang, E. Oron, B. Nelson, S. Razis, N. Ivanova, Cell Stem Cell 2012, 10,
440–454.
[13] M. O. Lee, S. H. Moon, H. C. Jeong, J. Y. Yi, T. H. Lee, S. H. Shim, Y. H.
Rhee, S. H. Lee, S. J. Oh, M. Y. Lee, M. J. Han, Y. S. Cho, H. M. Chung, K. S.
Kim, H. J. Cha, P. Natl. Acad. Sci. USA 2013, 110, E3281–3290.
[14] B. Blum, O. Bar-Nur, T. Govan-Lev, N. Benvenisty, Nature Biotech. 2009,
27, 281–287.
[15] T. Nakahara, M. Takeuchi, I. Kinoyama, T. Minematsu, K. Shirasuna, A.
Matsuhisa, A. Kita, F. Tominaga, K. Yamanaka, M. Kudoh, M. Sasamata,
Cancer Res. 2007, 67, 8014–8021.
[16] T. Nakahara, A. Kita, K. Yamanaka, M. Mori, N. Amino, M. Takeuchi, F. To-
minaga, I. Kinoyama, A. Matsuhisa, M. Kudoh, M. Sasamata, Cancer Sci.
2011, 102, 614–621.
[17] S. H. S. Ho, M. Y. Sim, W. L. S. Yee, T. M. Yang, S. P. J. Yuen, M. L. Go, Eur. J.
Med.Chem. 2015, 104, 42–56.
[18] A. Rauch, D. Hennig, C. Schꢂfer, M. Wirth, C. Marx, T. Heinzel, G. Schneid-
er, O. H. Krꢂmer, Biochim. Biophys. Acta Rev. Cancer 2014, 1845, 202–
220.
[31] S. H. Ho, A. Ali, T. M. Chin, M. L. Go, Oncotarget 2016, 7, 11625–11636.
[32] O. Takase, M. Yoshikawa, M. Idei, J. Hirahashi, T. Fujita, T. Takato, T. Isaga-
wa, G. Nagae, H. Suemori, H. Aburatani, K. Hishikawa, PLOS ONE 2013,
8, e56399.
[33] L. Armstrong, O. Hughes, S. Yung, L. Hyslop, R. Stewart, I. Wappler, H.
Peters, T. Walter, P. Stojkovic, J. Evans, M. Stojkovic, M. Lako, Hum. Mol.
Genet. 2006, 15, 1894–1913.
[34] D. C. Altieri, Cancer Lett. 2013, 332, 225–228.
[19] S. A. Przyborski, V. B. Christie, M. W. Hayman, R. Stewart, G. M. Horrocks,
Stem Cells Dev. 2004, 13, 400–408.
[20] R. Pal, G. Ravindran, Cell Prolif. 2006, 39, 585–598.
[21] K. Takahashi, K. Tanabe, M. Ohnuki, M. Narita, T. Ichisaka, K. Tomoda, S.
Yamanaka, Cell 2007, 131, 861–872.
Received: May 22, 2016
Revised: June 27, 2016
[22] K. Takahashi, S. Yamanaka, Cell 2006, 126, 663–676.
Published online on && &&, 0000
&
ChemMedChem 2016, 11, 1 – 13
www.chemmedchem.org
12
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

FULL PAPERS
Removing the rogues: The safety of
stem cell therapy depends on the re-
moval of undifferentiated rogue cells,
which would otherwise transform into
a malignant phenotype. The potent and
selective stemotoxic properties of diox-
onaphthoimidazoliums highlight their
potential as stem cell clearing agents
that could be deployed for this pur-
pose.
S.-H. S. Ho, A. Ali, Y.-C. Ng, K.-K. M. Lam,
S. Wang, W.-K. Chan, T.-M. Chin,
M.-L. Go*
&& – &&
Dioxonaphthoimidazoliums are Potent
and Selective Rogue Stem Cell
Clearing Agents with SOX2-
Suppressing Properties
ChemMedChem 2016, 11, 1 – 13
www.chemmedchem.org
13
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ