Hybrid aminoglycoside antibiotics via tsuji palladium-catalyzed allylic deoxygenation

Article abstract of DOI:10.1021/ol2027703

Biosynthetically inspired manipulation of the antibiotic paromomycin led, in six high-yielding steps, to a ring A harboring an R,β-unsaturated 6′- aldehyde and an allylic 3′-methylcarbonate group. Tsuji deoxygenation in the presence of 5 mol % Pd₂(dba)₃ and Bu₃P granted access to a novel series of 3′,4′-dideoxy- 4′,5′-dehydro ring A hybrids. The neomycin-sisomicin hybrid exhibited superior in vitro antibacterial activity to the parent compound neomycin.

Full text of DOI:10.1021/ol2027703

ORGANIC
LETTERS
2011
Vol. 13, No. 24
6476–6479
Hybrid Aminoglycoside Antibiotics
via Tsuji Palladium-Catalyzed Allylic
Deoxygenation
Stephen Hanessian,*,† Juan Pablo Maianti,† Rowena D. Matias,‡ Lee Ann Feeney,‡ and
Eliana S. Armstrong‡
ꢀ
Department of Chemistry, Universite de Montreal, C. P. 6128, Succ. Centre-Ville,
Montreal, QC, Canada, H3C 3J7, and Achaogen Inc., 7000 Shoreline Court, Suite 371,
ꢀ
ꢀ
South San Francisco, California 94080, United States
stephen.hanessian@umontreal.ca
Received October 14, 2011
ABSTRACT
Biosynthetically inspired manipulation of the antibiotic paromomycin led, in six high-yielding steps, to a ring A harboring an R,β-unsaturated 60-
aldehyde and an allylic 30-methylcarbonate group. Tsuji deoxygenation in the presence of 5 mol % Pd2(dba)3 and Bu3P granted access to a novel
series of 30,40-dideoxy-40,50-dehydro ring A hybrids. The neomycinꢀsisomicin hybrid exhibited superior in vitro antibacterial activity to the parent
compound neomycin.
Aminoglycoside antibiotics have occupied an important
position inthe arsenalofantibacterialtherapeuticsforover
half a century.1 They are powerful broad-spectrum anti-
biotics which target an rRNA helix at the mRNA-tRNA
decoding center of the bacterial 30S ribosomal subunit,2
effecting their bactericidal action by inducing translation
inaccuracy and inhibition.2 As classical representatives,
the gentamicin complex, amikacin, neomycin, and tobra-
mycin have been used in clinical practice as intravenous,
topical, or nebulizer drugs.1 However, their widespread
and first-line use has been compromised by the emergence
of drug-resistant strains of bacteria, which express amino-
glycoside modifying enzymes, classified as acetyltrans-
ferases (AACs), phosphotransferases (APHs), and adenyl-
transferases (ANTs).1,3 Extensive efforts have been made
over the years to overcome the growing resistance problem
through chemical removal of susceptible functionalities
in the 4,6-substituted 2-deoxystreptamine aminoglyco-
sides, starting with the readily available kanamycins.1,4
These strategies have been successful, exemplified by the
marketed drugs amikacin and arbekacin.5 In contrast, the
†
ꢀ
ꢀ
Universite de Montreal.
‡ Achaogen Inc.
(3) (a) Magnet, S.; Blanchard, J. S. Chem. Rev. 2005, 105, 477.
(b) Shaw, K. J.; Rather, P. N.; Hare, R. S.; Miller, G. H Microbiol.
Rev. 1993, 57, 138. (c) Vakulenko, S. B.; Mobashery, S. Clin. Microbiol.
Rev. 2003, 16, 430. (d) Smith, C. A.; Baker, E. N. Curr. Drug Targets
Infect. Disord. 2002, 2, 143. (e) Davies, J. Science 1994, 264, 375.
(4) (a) Mingeot-Leclercq, M. P.; Glupczynski, Y.; Tulkens, P. M.
Antimicrob. Agents Chemother. 1999, 43, 727. (b) Li, J.; Chang, C.-W. T.
Anti-Infective Agents Med Chem 2006, 5, 255.
(1) (a) Arya, D. P. Aminoglycoside Antibiotics: From Chemical
Biology to Drug Discovery; Wiley: Hoboken, NJ, 2007. (b) Walsh, C
Antibiotics: Actions, Origins, Resistance; ASM Press: Washington, DC,
2003.
(2) (a) Ogle, J. M.; Ramakrishnan, V. Annu. Rev. Biochem. 2005, 74,
129. (b) Franc-ois, B.; Russell, R. J.; Murray, J. B.; Aboul-ela, F.;
Masquida, B.; Vicens, Q.; Westhof, E. Nucleic Acids Res. 2005, 33, 5677.
r
10.1021/ol2027703
Published on Web 11/15/2011
2011 American Chemical Society

4,5-disubstituted deoxystreptamine class remains relatively
less explored.5
detail is known of the 30 and 40 deoxygenation steps between
gentamicin precursors and the final fermentation products.8
However, recent reports of the surprising biosynthetic in-
volvement of an APH(30) kinase in these gene clusters11 lead
us to hypothesize that the 30-hydroxyl was being activated as
a leaving group to be removed profiting from the stabiliza-
tion from the neighboring electron-rich olefin, which could,
in turn, arise from a 60-aldehyde-equivalent pyridoxal-
catalyzed 40,50-dehydration.7 Therefore, we set out to find
ways in which abiotic versions of such activated inter-
mediates could be harnessed and controlled for reduction.7
Herein we report on our attaining these objectives
relying on the venerable Tsuji Pd-catalyzed deoxygenation
reaction,12,13 asappliedtothe aminoglycoside seriesfor the
first time.12ꢀ14
Paromomycin, a member of the 4,5-disubstituted 2-de-
oxystreptamine aminoglycosides, has been in clinical use
against amoebic dysentery and leishmaniasis in under-
developed regions of the world (Figure 1).1,6 Its 60-amino
congener, neomycin, is widely used as a topical agent
worldwide (e.g., Neosporin).1 However, the structural
features of paromomycin and neomycin make them highly
susceptible to inactivating enzymes such as APH(30) and
ANT(40) that target ring A hydroxyl groups (Figure 1).1ꢀ4
To explore the feasibility of introducing the combined
features of 40,50-unsaturation and 30-deoxygenation we
devised a practical protocol on a ring A model, starting
with known sugar 3(Scheme1).7 We installed a C3-methyl-
carbonate, followed by liberation of the 4,6-diol to 4.
Conversion to the 4-mesylate 5, followed by oxidation
and β-elimination, led to the aldehyde 6. During the course
of a substrate scope study for Tsuji deoxygenation (Table
S1),7 we discovered that, under unoptimized conditions,
treatment of aldehyde 6 with Pd2(dba)3 (20 mol %) in the
presence of Bu3P (40 mol %), Et3N, and formic acid in
degassed THF at 60 °C12 resulted in the reliable conversion
to the 3-deoxy-4,5-unsaturated aldehyde (7) in 92% yield.
Figure 1. Aminoglycosides of the paromomycin and sisomicin
families, and the designed target hybrid analogs 1 and 2.7ꢀ9
Scheme 1. Pd-Catalyzed Allylic Deoxygenation: Model Studies
Unlike the structurally diverse 4,6-disubstituted 2-deoxy-
streptamine subclass,1,7 fewer modified analogs of paro-
momycin and neomycin have been isolated1,7 or chem-
ically synthesized,10 and instead most of their congeners
feature a primordial 2-amino-2-deoxy-R-D-glucopyrano-
syl moiety as ring A.8
Sisomicin is a naturally derived congener of gentamicin
that features a 60-amino-30,40-dideoxy-40,50-dehydro ring A
(Figure 1).1,9 Intrigued by the biosynthetic transformation
of aminoglycoside precursors to sisomicin and gentamicin,8
we considered the synthesis of paromomycin and neomy-
cin analogs in which the enzyme-susceptible ring A was
exchanged for the same subunit found in sisomicin. Little
We hypothesize that Tsuji deoxygenation of 6 occurs by
hydride delivery at C5 toward a D-configuration, through
(5) For review of medicinal chemistry efforts 1975ꢀ85, see: Price,
K. E. Am. J. Med. 1986, 80, 182.
(6) Davidson, R. N.; den Boer, M.; Ritmeijer, K. Trans. R. Soc. Trop.
Med. Hyg. 2009, 103, 653.
(7) See Supporting Information for further details, complete listing
of aminoglycosides, and a proposed biosynthetic pathway.
(8) For review of aminoglycoside biosynthesis, see ref 1a and
(a) Flatt, P. M.; Mahmud, T. Nat Prod Rep. 2007, 24, 358. (b) Llewellyn,
N. M.; Spencer, J. B. Nat. Prod. Rep. 2006, 23, 864.
(11) (a) Dairi, T.; Ohta, T.; Hashimoto, E.; Hasegawa, M. Mol. Gen.
Genet. 1992, 236, 39. (b) Unwin, J.; Standage, S.; Alexander, D.; Hosted,
T., Jr.; Horan, A. C.; Wellington, E. M. J. Antibiot. (Tokyo) 2004, 57,
436. (c) Hong, W. R.; Ge, M.; Zeng, Z. H.; Zhu, L.; Luo, M. Y.; Shao, L.;
Chen, D. J. Biotechnol. Lett. 2009, 31, 449.
(12) For review, see: Tsuji, J.; Mandai, T. Synthesis 1996, 1, 1.
(13) For applications to natural product synthesis, see: (a) Wipf, P.;
Methot, J. L. Org. Lett. 2000, 2, 4213. (b) Wipf, P.; Rector, S. R.;
Takahashi, H. J. Am. Chem. Soc. 2002, 124, 14848. (c) Wipf, P.; Spencer,
S. R. J. Am. Chem. Soc. 2005, 127, 225.
(9) Testa, R. T.; Tilley, B. C. Jpn. J. Antibiot. 1979, 32, S47.
(10) (a) Franc-ois, B.; Szychowski, J.; Adhikari, S. S.; Pachamuthu,
K.; Swayze, E. E.; Griffey, R. H.; Migawa, M. T.; Westhof, E.;
Hanessian, S. Angew. Chem., Int. Ed. Engl. 2004, 43, 6735. (b) Hanessian,
ꢁ
S.; Pachamuthu, K.; Szychowski, J.; Giguere, A.; Swayze, E. E.; Migawa,
M. T.; Franc-ois, B.; Kondo, J.; Westhof, E. Bioorg. Med. Chem. Lett.
2010, 23, 7097 and references within.
(14) For an application of a similar reaction to carbohydrates, see:
Greenspoon, N.; Keinan, E. J. Org. Chem. 1988, 53, 3723.
Org. Lett., Vol. 13, No. 24, 2011
6477

Scheme 2. Application of the Tsuji Palladium-Catalyzed Allylic Deoxygenation to Paromomycin
an endo-face π-allyl complex (6a), followed by rapid olefin
isomerization in situ.7 In support of this mechanism,
Table 1. Optimization of Catalyst and Ligand Loadinga
identical Tsuji deoxygenation of analog substrates of the
C6-alcohol oxidation state provided only 3,4-unsaturated
products with a C5-D-configuration, exemplified by 8,
determined by X-ray crystallography.7 However, unlike
aldehyde 6, the supplemental substrate series we explored
underwent reduction sluggishly in low to moderate yields,
albeit at prohibitively high catalyst loadings.7 The proper-
ties of allylic R,β-unsaturated aldehydes suchasin6 appear
streamlined for Tsuji deoxygenation, yet are not repre-
sented in the literature.12
Pd2(dba)3
(mol %)
Bu3P
time
(h)
yield
entry
(mol %)
13 (%)b
1
2
4
5
6
20
10
5
40
20
10
20
10
3
9
82
73
46
97
80
24
3
5
2.5
5
a Conditions: 0.1 M in THF, 60 °C, 10 equiv of HCO2H, 11 equiv of
Et3N. b Isolated yields.
We were now poised to apply our Tsuji deoxygenation
protocol of allylic carbonates to highly functionalized
aminoglycoside derivatives, seeking dideoxygenated analogs
which could evade the APH(30) and ANT(40) modifying
enzymes. Experience in our group and others has shown
that paromomycin is a functionally malleable aminoglyco-
side for medicinal chemistry.1,10,15 Paromomycin sulfate
was protected with N-Cbz groups, followed by formation
of the 40,60-O-benzylidene derivative 9 as a functional
handle for ring A (Scheme 2).15 The remaining alcohol
functionalities in 9 were capped with methylcarbonates
using a transesterificaiton protocol, simultaneously pro-
viding the requisite 30-leaving group. Removal of the
benzylidene acetal afforded ring-A diol 10, which was
converted to 40-mesylate 11 and submitted to Parikhꢀ
Doering oxidation/β-elimination to give intermediate 12,
in excellent overall yield. The heavily functionalized ami-
noglycoside aldehyde 12 showed excellent concordance
with the reactivity of the ring-A model (Scheme 1). Follow-
ing optimization of the deoxygenation protocol, using 5
mol % Pd2(dba)3 and 20 mol % Bu3P provided a nearly
quantitative yield of the desired 30-deoxy-40,50-unsaturated
aldehyde 13 (Table 1). Changing the ligand stoichiometry
to a 1:4 ratio was indispensable for this improvement
(Table 1, entry 5).
Hence, 30,40-dideoxygenated intermediate 13 was pro-
duced in enabling gram scale and ∼60% overall yield from
paromomycin sulfate. We then directed our efforts to pro-
duce the first generation of novel 30,40-dideoxy-40,50-un-
saturated hybrid tetrasaccharide aminoglycosides with a
sisomicin ring-A motif (Scheme 3). The 60-position was
functionalized by reduction and introduction of a 60-azide,
followed by facile cleavage of the remaining methylcarbo-
nate groups to 14 (Scheme 3). The end-game global depro-
tection required Birch reduction to afford aminoglycoside
analogs 1 and 2, for it is well established that hydrogenation
would lead to reduction of the olefin predominantly toward
inactive C50-L-diastereomers.4
Our first generation of novel sisomicin-hybrid amino-
glycosides were tested against a panel of isolates represent-
ing susceptible wild-types and strains known to contain
inactivating enzymes (Table 2). Preliminary data indicate
(15) (a) Hanessian, S.; Takamoto, T.; Masse, R.; Patil, G. Can. J.
€
Chim. Acta 2005, 88, 2967.
(16) (a) Paulsen, H.; Jansen, R. Carbohydr. Res. 1981, 92, 305. (b)
Girodeau, J. M.; Pineau, R.; Masson, M.; Le Goffic, F. J. Antibiot.
(Tokyo) 1984, 37, 150.
Chem. 1978, 56, 1482. (b) Pathak, R.; Bottger, E. C.; Vasella, A. Helv.
6478
Org. Lett., Vol. 13, No. 24, 2011

Table 2. Antibacterial Assessment: Sisomicin Ring A Hybrids, Parent Antibiotics, and Controls; Minimum Inhibitory Concentrations
(μg/mL)a
entry
strain name
strain code/enzyme
Amk
Gent
Par
NeoB
Rib
Btr
1
2
16
1
2
3
4
5
S. aureus
ATCC 29213
ATCC 25922
ATCC 10031
ATCC 27853, APH(30)
ANT(40)
2
2
1
2
4
1
2
16
8
8
2
8
1
16
64
16
16
16
E. coli
0.5
0.25
0.5
0.5
>64
>64
>16
16
2
K. pnuemoniae
P. aeruginosa
S. aureus
0.5
1
2
0.5
32
2
1
0.5
>64
>64
>64
>64
>32
>32
e1
64
>64
1
a Abbreviations: Amk, amikacin; Gent, gentamicin complex; Par, paromomycin; NeoB, neomycin B; Rib, ribostamycin; Btr, butirosin.7
give 15 (Scheme 3).15a Using this protocol, we accessed
pseudotrisaccharide hybrid 16 related to the ribostamycin,
previously obtained from degradation of sisomicin fol-
lowed by ribosylation, but whose antimicrobial activity
has not been reported.16 The NMR spectrum was identical
to that of the literature compound,16 providing further
support to the structural assignment of our novel amino-
glycoside analogs. In general, we observed that ring D
truncated hybrids, such as 16 were 16- to 32-fold less active
compared to the tetrasaccharide congener 2, and ribosta-
mycin was greater than 2- to 16-fold less active than
neomycin B. We did not detect growth inhibition using
the 60-OH analog of 16.7
Scheme 3. Neomycin and Paromomycin Ring A Hybrid Ana-
logs
To conclude, we have demonstrated that a biosynthesis-
inspired paradigm relying on transition metal control of the
venerable Tsuji Pd-catalyzed deoxygenation in the aminogly-
coside series can provide a novel analog series with promising
antibacterial activities. Herein, we have determined that the
combination of 60-amino substitution and 30,40-dideoxygena-
tion are best tolerated in the neomycin series, while providing
evasion of APH(30) and ANT(40) enzymes shown in Table 2.
Our novel deoxygenated 4,5-disubstituted 2-deoxystrepta-
mine analogs profit from neomycin-like features such as the
L-idose ring D, which can explore an extended binding pocket
and provide additional beneficial RNA contacts.2,10 Mean-
while lacking the third accessory ring C of gentamicins and
kanamycins may hinder recognition by widespread subtypes
of inactivating enzymes.3,4 Further details on second genera-
tion analogs of sisomicinꢀneomycin hybrids 2 and 16 will be
reported in due course.
Acknowledgment. We thank Fonds de Recherche en
in vitro growth inhibition for neomycin analog2, with MIC
values ranging from 0.5 to 2 μg/mL in E. coli, K. pneumoniae,
P. aeruginosa, and S. aureus. The activity of analog 2 was not
affected by two resistance enzymes prevalent in P. aeruginosa
and S. aureus (from g32- and >64-fold improvement over
neomycin and amikacin, respectively). In contrast, analog 1
displayed from 4- to greater than 32-fold less potency in four
out of the five strains compared to the parent antibiotic
paromomycin.10b,15
ꢀ
ꢀ
Sante du Quebec (FRSQ) for a fellowship to J.P.M. and
Natural Sciences and Engineering Research Council of
Canada (NSERC) for continued support. We thank
Achaogen for financial support, fruitful discussions, and
^
the supply of 9. We thank Benoıˆt Deschenes-Simard,
ꢀ
Universite de Montreal, for the X-ray analysis of 8.
ꢀ
Supporting Information Available. Experimental pro-
cedures, supplemental substrate scope table, ORTEP of
8, and full spectroscopic data. This material is available
free of charge via the Internet at http://pubs.acs.org.
To extend our knowledge of the role played by ring D in
these hybrid molecules, we profited from the isolated diols
in 14, to perform oxidative cleavage and β-elimination to
Org. Lett., Vol. 13, No. 24, 2011
6479