Tetracycline analogues with a selective inhibitory effect on HIF-1α

Article abstract of DOI:10.1039/c4md00111g

As part of a programme to discover novel transcription factor inhibitors, a 40-membered library of tetracycline analogues was screened to identify potential HIF-1 inhibitors. Two novel analogues (5b and 5c) with significant HIF-1 modulation properties were identified. These molecules possess good cellular penetration properties, and provide significant down-regulation of VEGF in U251 cells.

Full text of DOI:10.1039/c4md00111g

MedChemComm
View Article Online
View Journal | View Issue
CONCISE ARTICLE
Tetracycline analogues with a selective inhibitory
effect on HIF-1a†
Cite this: Med. Chem. Commun., 2014,
5, 923
Y. Bendiabdellah,a K. M. Rahman,b B. Uranchimeg,‡c K. S. Nahar,b D. Antonow,a
c
d
e
b
*
*
R. H. Shoemaker,‡ G. Melillo,§ G. Zinzalla and D. E. Thurston
Received 10th March 2014
Accepted 7th May 2014
As part of a programme to discover novel transcription factor inhibitors, a 40-membered library of
tetracycline analogues was screened to identify potential HIF-1 inhibitors. Two novel analogues (5b and
5c) with significant HIF-1 modulation properties were identified. These molecules possess good cellular
penetration properties, and provide significant down-regulation of VEGF in U251 cells.
DOI: 10.1039/c4md00111g
www.rsc.org/medchemcomm
Transcription factors (TFs) are master regulators of central signalling and resistance to therapy.1,3–7 HIF-1 is a hetero-
biological processes relating to cell development and growth. dimeric protein complex consisting of the regulatory HIF-1a
TFs control the expression of a large number of genes critical for subunit, the level of which is mainly dependent on cellular
cell survival and proliferation, and that ultimately determine oxygen levels, and the ubiquitously expressed HIF-1b subunit.7,8
cellular phenotype. Over-expression or hyper-activity of TFs may In addition to tumour hypoxia, HIF-1a can be over-expressed
be required for survival, growth and metastatic behaviour of all due to genetic mutations such as gain of function in oncogenes
human cancers. Thus, oncogenic TFs are an increasingly (e.g., ERBB2), or loss of function in tumour suppressor genes
important target for novel anticancer therapies, and direct (e.g., VHL and PTEN).9
inhibition of their function is an attractive strategy. Hypoxia
There are a number of steps within this pathway that could be
within tumour cells is a common phenomenon, arising due to targeted for therapeutic purposes such as HIF-1 synthesis,
the avascularity of primary tumours and/or abnormality in formation of the HIF-1a:HIF-1b heterodimer, DNA-binding of
tumour microvasculature.1 It is associated with increased risk of the heterodimeric complex and transactivation.6,10 So far,
tumour invasion, metastasis, resistance to therapy and patient attempts have been made to reduce HIF-1a expression using
mortality.1,2 A key mechanism by which hypoxia in primary specic antisense oligonucleotides,11 and to block protein–
tumours or their metastases promotes disease progression is protein interactions between HIF-1a:HIF-1b5,12–14 or HIF-1a and
through induction of the transcription factor Hypoxia-Inducible the co-activator protein p300/CBP.15–18 For example, the small
Factor-1 (HIF-1) which controls the expression of genes encod- molecules chetomin16 and YC-118 have been reported to reduce
ing proteins that play important roles in tumour biology such as HIF-1a expression, and acriavine has been shown to down-
evasion of apoptosis, stem cell maintenance, autocrine growth, regulate HIF-1-dependent genes in human tumour xenogras by
tumour angiogenesis, metabolic reprogramming, invasiveness, inhibiting
HIF-1a:HIF-1b
heterodimerization
through
metastasis, enhancement of receptor tyrosine kinase-mediated interaction with the PAS-B domain.12 In addition, Cardoso and
co-workers have recently identied a covalent-binding small-
molecule inhibitor that alkylates a cysteine residue ofthe HIF-1a-
PasB domain, although the use of covalently binding PPI
inhibitors as potential drug candidates is controversial.13 Finally,
aSchool of Pharmacy, University College London, 29-39 Brunswick Square, London,
WC1N 1AX, UK
bInstitute of Pharmaceutical Science, King's College London, Britannia House, 7 Trinity
hypoxia-inducible transcription has been inhibited by a poly-
peptide15 and cyclopeptides14 acting asHIF-1a:HIF-1b inhibitors.
However, peptidic compounds are problematic lead candidates
for drug development due to their poor metabolic stability.
The tetracycline family of antibiotics have good drug-like
properties with a low toxicity prole. Interestingly, the semi-
synthetic tetracycline antibiotic rolitetracycline (1) has been
reported to possess HIF-1 inhibitory activity in a cell free envi-
ronment.5 However, it does not inhibit the HIF-1 signalling
pathway in cells, most likely due to its poor cellular penetration
properties.5 Building on this observation, and based on the
excellent drug-like characteristics of the tetracycline family, we
Street, London, SE1 1DB, UK. E-mail: david.thurston@kcl.ac.uk; Tel: +44 (0)207 848
7507
cScreening Technologies Branch, Developmental Therapeutics Program, Division of
Cancer Treatment and Diagnosis, National Cancer Institute, Frederick, Maryland, USA
dSAIC-Frederick, Frederick, Maryland, USA
e
¨
Centre for Advanced Cancer Therapies, MTC Dept., Karolinska Institutet, Nobelsvag
16, SE-17177 Stockholm, Sweden. E-mail: giovanna.zinzalla@ki.se; Tel: +46 8524
81262
† Electronic supplementary information (ESI) available: HTS screening protocol
and results, synthetic chemistry experimental details and analytical data, other
general experimental details. See DOI: 10.1039/c4md00111g
‡ Current address: Division of Cancer Prevention, National Cancer Institute, 9609
Medical Center Dr Rm 5-E452, Bethesda, Maryland, 20892, USA.
§ Current address: AstraZeneca Oncology, Gaithersburg, Maryland, 20878, USA.
This journal is © The Royal Society of Chemistry 2014
Med. Chem. Commun., 2014, 5, 923–926 | 923

View Article Online
MedChemComm
Concise Article
decided to synthesize and screen a library of novel tetracycline (6, NSC-2609699), an agent known to inhibit HIF-1a, was used
analogues in an attempt to identify potential lead molecules as a compound control. This assay identied seven “hit”
with HIF-1 inhibitory activity in cells. For this purpose, we molecules (Table 1), along with distinct structure activity rela-
synthesized a focused library of 40 molecules based on the tionships (SARs). The two most potent molecules, 5b and 5c,
tetracycline scaffold with diverse structural modications in the selectively inhibited HIF-1-dependent luciferase expression
C and D rings. Two molecules were identied, 5b and 5c (Fig. 1) with EC50 values of 2.3 mM and 1.4 mM, respectively (Table 2).
that can inhibit HIF-1a transcriptional activity in cell based With regard to SAR, good activity and selectivity were observed
assays. Crucially, they appear to selectively target HIF-1a for compounds with CF3 and H, or Br and C2H4COOH, at the R4
without affecting levels of constitutively expressed HIF-1b. They and R5 positions, respectively. For both of these R4/R5-
also selectively inhibit HIF-1-dependent down-stream genes substituted families of molecules (see ESI†), the most active had
such as VEGF.
R1 ¼ R2 ¼ H and R3 ¼ Me substituents. Interestingly, loss of
The design of the focused library involved retention of the activity was observed when R2 and R3 ¼ Me in the R4/R5 Br/
core molecular framework and spatial conformation of tetracy- C2H4COOH series (see compound YB78 in Table S1 of ESI†).
cline while modifying perimeter functional groups. Library Activity was lost for compounds with CF3 and C2H4COOH, or Br
members were synthesized by employing consecutive Diels– and H, at the R4 and R5 positions, respectively. Interestingly,
Alder cycloadditions and indium-mediated chemistry (Scheme replacement of the R4 Br atom in compound 5c with a F atom
1), prompted by our previously reported synthetic studies.19 The (i.e., compound 5g) led to a similar potency but signicantly less
synthetic route developed for these molecules required fewer selectivity.
steps and allowed more scope for variation than other routes to
Next, the effect of 5b and 5c on HIF-1a mRNA levels in U251-
tetracycline and related analogues published to date. The cross- wt cells under both normoxic and hypoxic conditions was
conjugated trienes (3) were prepared in two steps19 starting from evaluated aer both 8 and 16 hours exposure for two concen-
1,2-addition of 5-bromopenta-1,3-diene to the corresponding tration levels (Fig. 2). The greatest inhibition under hypoxic
para-substituted benzaldehydes (see ESI†), and then reacted conditions was observed at the 10 mM level, with 5b and 5c
with naphthoquinone dienophiles of type 2 in a Lewis acid-cat- providing decreases of 29% and 43%, respectively, aer 8 hours
alysed Diels–Alder reaction at low temperature (Scheme 1). The incubation (45% and 78%, respectively, aer 16 hours).
diene-containing products 2a were then subjected to a thermal
Diels–Alder reaction with various maleimides (4) to produce a expression in normoxia, probably due to a non-specic effect.
library of 40 racemic tetracyclic analogues of type 5 (see ESI†). There was no evidence that this was due to generalised cellular
The results show that both 5b and 5c also effect HIF-1a
The library members were screened initially in a luciferase- toxicity, or an effect on basal transcription. For example, even at
based plasmid reporter assay using U251-HRE cells containing the highest concentrations, 5b and 5c did not completely block
a HIF Response Element (HRE), and U251-pGL3 control
cells lacking
a
HRE20 (see Table S1, ESI†). Topotecan
Table 1 Structures of the most potent “hit” molecules identified
through the U251-HRE cell-based luciferase assay
Number
R1
R2
R3
R4
R5
5a
5b
5c
5d
5e
5f
OH
H
H
OMe
OMe
H
Me
H
H
Me
Me
Me
H
H
CF3
CF3
Br
CF3
Br
H
H
Me
Me
H
H
Me
Me
C2H4COOH
H
C2H4COOH
H
CF3
F
5g
H
C2H4COOH
Fig. 1 Structures of rolitetracycline (1) and analogues 5b and 5c.
Table 2 Results of the U251-HRE cell-based screen. Topotecan (6)
was included as a compound control. The U251-pGL3 cell line lacked a
HIF response element (HRE) and acted as a cellular control
EC50 HRE
Number
(mM)
EC50 pGL3 (mM)
pGL3/HRE
5a
5b
5c
5d
5e
5f
6.7
2.3
1.4
6.7
15.5
7.8
>20
>20
>20
>20
>20
>20
8.7
>3
>8.6
>14
>3
>1.4
>2.7
2.15
>1700
5g
6
4.1
0.012
>20
Scheme 1 Synthesis of tetracycline analogues of type 5.
924 | Med. Chem. Commun., 2014, 5, 923–926
This journal is © The Royal Society of Chemistry 2014

View Article Online
Concise Article
MedChemComm
Cells were always used when conuence was ꢀ60%, and con-
uency was never reached during the experiments. These cells
have a PTEN mutation and, based on the density of the culture
conditions, expression levels in normoxia have been previously
reported.21–24
Complete inhibition was observed for 5c at 5 mM, whereas 5b
was less potent, providing maximal inhibition at 10 mM. A
similar level of inhibition was observed under normoxic
conditions aer 16 hours, but only at the higher concentration
of 10 mM. Although we have not investigated the mechanism of
the latter effect, it may be due, in part, to non-specic effects at
the highest concentrations, or to increased degradation of HIF-
1a due to interference with protein–protein interactions.
Importantly, neither of the two molecules had any effect on HIF-
1b levels (Fig. 3), conrming their selectivity for processes
relating to HIF-1a.
Fig. 2 Inhibition of induction of HIF-1a mRNA expression in U251wt
cells by 1 mM and 10 mM concentrations of 5b and 5c under conditions
of normoxia and hypoxia after 8 h and 16 h. TPT ¼ topotecan (6).
Experiments were repeated in triplicate, with the SD values shown
being the average of the three experiments. MED ¼ Medium control.
It has been previously demonstrated that HIF-1 activity is
stronglylinkedtoVEGFmRNAexpression,25 andsoaHIF-1small-
molecule inhibitor might be expected to lower the level of VEGF
mRNA expression. Under hypoxic conditions, 5b and 5c were
shown to inhibit VEGF mRNA expression aer 8 and 16 hours at
two concentration levels. Aer 16 hours at 1 mM, 5b and 5c
reduced expression by 35% and 38% respectively, whereas inhi-
bition levels reached 59% and 69%, respectively, for 10 mM
(Fig. 4). Withregard toinhibitionbythecompounds innormoxia,
again these cells have a PTEN mutation, and basal expression of
HIF-1 in normoxia has been reported.21–24 Under these condi-
tions, basal expression of VEGF may be, in part, under HIF-1
control. In addition, expression and activity of HIF-1 under nor-
moxic conditions has been described even though HIF-1 protein
may not be detectable, due to rapid degradation by oxygen.
In summary, two novel tetracycline analogs, 5b and 5c, with
favourable cellular penetration characteristics and a selective
effect on HIF-1a have been identied. They have selective HIF-1a
andVEGFinhibitoryactivity inthelow micromolarrangewithout
affecting cellular levels of constitutively expressed HIF-1b.
Further studies are required to elucidate the mode of action
of 5b and 5c, and are currently underway in our laboratories.
VEGF expression (see Fig. 4) or effect basal transcription, both
of which might be expected in the case of general toxicity.
Further experiments included Western blotting analysis to
elucidate the effect of 5b and 5c on the accumulation of HIF-1a
protein in cells grown under hypoxia compared to those
growing in normoxic conditions. HIF-1b was used as a loading
control in these experiments as it is constitutively present and
abundantly expressed in cells, and is not modulated by oxygen.
Crucially, both molecules inhibited hypoxically-induced HIF-1a
protein accumulation in U251 wt cells in a dose-dependent
fashion (Fig. 3). Constitutive HIF-1a levels were observed in
total and nuclear extracts of normoxic U251 cells, and HIF-1a
binding (by EMSA) was detected in normoxic nuclear extracts.
Fig. 3 (a) The effect of 5b and 5c at 1 mM and 10 mM by Western Fig. 4 Inhibition of induction of VEGF mRNA expression in U251wt
blotting on HIF-1a and HIF-1b expression after 8 and 16 hours in U251 cells by 1 mM and 10 mM concentrations of 5b and 5c under conditions
wt cells under both normoxic and hypoxic conditions using topotecan of normoxia and hypoxia after 8 h and 16 h. Topotecan (TPT) was used
(TPT) as a control; (b) similar experiments at 1 mM, 5 mM and 10 mM for as a control compound, and the SD values shown were the average of
HIF-1a only.
three experiments.
This journal is © The Royal Society of Chemistry 2014
Med. Chem. Commun., 2014, 5, 923–926 | 925

View Article Online
MedChemComm
Concise Article
These experiments will guide the synthesis of focused libraries 14 E. Miranda, I. K. Nordgren, A. L. Male, C. E. Lawrence,
of analogues of 5b and 5c to further understand and validate the
SAR for the compounds reported in this study.
F. Hoakwie, F. Cuda, W. Court, K. R. Fox, P. A. Townsend,
G. K. Packham, S. A. Eccles and A. Tavassoli, J. Am. Chem.
Soc., 2013, 135, 10418–10425.
15 A. L. Kung, S. Wang, J. M. Klco, W. G. Kaelin and
D. M. Livingston, Nat. Med., 2000, 6, 1335–1340.
Acknowledgements
Cancer Research UK is acknowledged for providing nancial 16 A. L. Kung, S. D. Zabludoff, D. S. France, S. J. Freedman,
support for these studies (C180/A9327 to D.E.T.), and the Alger-
ian Government is thanked for providing a studentship (to Y.B.).
E. A. Tanner, A. Vieira, S. Cornell-Kennon, J. Lee, B. Wang,
J. Wang, K. Memmert, H.-U. Naegeli, F. Petersen, M. J. Eck,
K. W. Bair, A. W. Wood and D. M. Livingston, Cancer Cell,
2004, 6, 33–43.
Notes and references
17 A. Staab, J. Loeffler, H. M. Said, D. Diehlmann, A. Katzer,
M. Beyer, M. Fleischer, F. Schwab, K. Baier, H. Einsele,
M. Flentje and D. Vordermark, BMC Cancer, 2007, 7, 213–
219.
1 W. R. Wilson and M. P. Hay, Nat. Rev. Cancer, 2011, 11, 393–
410.
2 G. Melillo, Cancer Metastasis Rev., 2007, 26, 341–352.
3 E. Poon, A. L. Harris and M. Ashcro, Expert Rev. Mol. Med., 18 S. H. Li, D. H. Shin, Y.-S. Chun, M. K. Lee, M.-S. Kim and
2009, 11, e26.
J.-W. Park, Mol. Cancer Ther., 2008, 7, 3729–3738.
19 Y. Bendiabdellah, I. Villanueva-Margalef, A. Misale,
K. S. Nahar, M. R. Haque, D. E. Thurston and G. Zinzalla,
Synthesis, 2011, 14, 2321–2333.
4 G. L. Semenza, Drug Discovery Today, 2007, 12, 853–859.
5 E. J. Park, D. Kong, R. Fisher, J. Cardellina, R. H. Shoemaker
and G. Melillo, Cell Cycle, 2006, 5, 1847–1853.
6 J. A. Diaz-Gonzalez, J. Russell, A. Rouzaut, I. Gil-Bazo and 20 G. Melillo, Methods Enzymol., 2007, 435, 385–402.
L. Montuenga, Cancer Biol. Ther., 2005, 4, 1055–1062.
7 G. L. Semenza, Nat. Rev. Cancer, 2003, 3, 721–732.
8 A. L. Harris, Nat. Rev. Cancer, 2002, 2, 38–47.
9 Q. Ke and M. Costa, Mol. Pharmacol., 2006, 70, 1469–1480.
10 G. Melillo, Mol. Cancer Res., 2006, 4, 601–605.
11 E.-J. Yeo, Y.-S. Chun and J.-W. Park, Biochem. Pharmacol.,
2004, 68, 1061–1069.
21 A. Rapisarda and G. Melillo, Oncogene, 2007, 26, 6875–6884.
22 A. Rapisarda, B. Uranchimeg, O. Sordet, Y. Pommier,
R. H. Shoemaker and G. Melillo, Cancer Res., 2004, 64,
1475–1482.
23 W. Zundel, C. Schindler, D. Haas-Kogan, A. Koong, F. Kaper,
E. Chen, A. R. Gottschalk, H. E. Ryan, R. S. Johnson,
A. B. Jefferson, D. Stokoe and A. J. Giaccia, Genes Dev.,
2000, 14, 391–396.
12 K. Lee, H. Zhang, D. Z. Qian, S. Rey, J. O. Liu and
G. L. Semenza, Proc. Natl. Acad. Sci. U. S. A., 2009, 106, 24 A. Rapisarda, B. Uranchimeg, D. A. Scudiero, M. Selby,
17910–17915.
E. A. Sausville, R. H. Shoemaker and G. Melillo, Cancer
Res., 2002, 62, 4316–4324.
13 R. Cardoso, R. Love, C. L. Nilsson, S. Bergqvist, D. Nowlin,
J. Yan, K. K. C. Liu, J. Zhu, P. Chen, Y.-L. Deng, H. J. Dyson, 25 L. S. Ziemer, C. J. Koch, A. Maity, D. P. Magarelli, A. M. Horan
M. J. Greig and A. Brooun, Protein Sci., 2012, 21, 1885–1896.
and S. M. Evans, Neoplasia, 2001, 3, 500–508.
926 | Med. Chem. Commun., 2014, 5, 923–926
This journal is © The Royal Society of Chemistry 2014