Oxazolomycins: Natural product lead structures for novel antibacterials by click fragment conjugation

Article abstract of DOI:10.1016/j.bmcl.2010.02.066

Conjugation of amide and lactam subunits by a 'Click' type approach provides access to structural mimics of the oxazolomycin series of natural products, some of which exhibit antibacterial activity.

Full text of DOI:10.1016/j.bmcl.2010.02.066

Bioorganic & Medicinal Chemistry Letters 20 (2010) 2090–2094
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Oxazolomycins: Natural product lead structures for novel antibacterials
by click fragment conjugation
*
Claire L. Bagwell, Mark G. Moloney , Muhammad Yaqoob
Department of Chemistry, Chemistry Research Laboratory, University of Oxford, Mansfield Road, Oxford OX1 3TA, United Kingdom
a r t i c l e i n f o
a b s t r a c t
Article history:
Conjugation of amide and lactam subunits by a ‘Click’ type approach provides access to structural mimics
of the oxazolomycin series of natural products, some of which exhibit antibacterial activity.
Ó 2010 Elsevier Ltd. All rights reserved.
Received 13 January 2010
Revised 17 February 2010
Accepted 17 February 2010
Available online 20 February 2010
Keywords:
Antibacterials
Natural products
Click conjugation
The rapid emergence of virulent (e.g. Influenza A (H1N1))) and
resistant (e.g. penicillin-resistant Streptococcus pneumoniae, vanco-
mycin resistant enterococci, methicillin resistant S. aureus (MRSA),
multi-resistant salmonellae, and multi-resistant Mycobacterium
tuberculosis) bacterial and virus strains to widely used and effective
antimicrobials has led to a significant increase in persistent infec-
tions in the human population.1 The speed with which multi-drug
resistance emerged in a diverse range of organisms has recently led
to the recognition that the ‘Golden Age of Antibiotics’ is over, and
that substantial efforts in what has hitherto been considered to be
a ‘closed-book’ will be now required;2,3 a number of strategies for
the most effective way to achieve this have been proposed.4,5 But it
is not just high profile infectious diseases that are problematic;
while anti-infective agents constitute only 16% of the total budget
spent on drug development, they comprise the highest proportion
of the national drug therapy budget in many countries.6 Further-
more, substantial niches exist in the area of the treatment of ‘ne-
glected diseases’: of 1393 new chemical entities marketed
between 1975 and 1999, only 16 were for tropical diseases and
tuberculosis.7 Thus, there is ample opportunity for the develop-
ment both of wholly novel drug discovery methodology and of
anti-infective agents themselves in the race to ensure health provi-
sion in the 21st century. However, this needs to be done with care,
since concerns over the indiscriminate use of biocides, particularly
by their wide dispersal in the environment, have been raised.8
Unfortunately, the development of new antibiotic agents suitable
for therapeutic application has been significantly less successful
than anticipated,9 and the number of leads has remained approxi-
mately constant over the past two decades despite a tenfold in-
crease in R&D spending by the pharmaceutical industry.10,11 This
pressure on drug discovery implies that there is an urgent need
to create a new paradigm for antibiotic drug design, delivery and
therapy, and this endeavour is aided by improved recent under-
standing of the requirements of the physicochemical property
space12 and cytoplasmic entry and retention behaviour of antibac-
terials13 although it has also been suggested that the number of
possible antibacterial targets might be more limited than has hith-
erto been assumed.14 Examples of recently reported novel and
effective approaches for antibacterial/antimicrobial discovery in-
clude diversity orientated synthesis15 and whole animal (nema-
tode) bioassay.16,17
Re-examination of the function and availability of natural prod-
ucts18,19 has proved to be a key impetus in recent innovation,20 and
natural product inspired synthesis is gaining renewed accep-
tance.5,21–26 Although this approach is widely used within antican-
cer drug development,27 its application within anti-infective drug
discovery more generally has been strongly advocated, and a sense
of urgency to ensure that the benefits of such an approach become
available in suitable timeframes is emerging.10,28 That natural
products might in fact possess the required physicochemical prop-
erties for drug discovery and development has been demonstrated,
contrary to widely held assumptions.29 Libraries based upon such
leads have several key benefits which do not apply to combinato-
rially-derived systems: they will have benefited from the optimisa-
tion of bioactivity for a given receptor as a result of natural
selection; they will be expected to provide an enhanced rate of po-
sitive hits for a given library size; they will likely provide novel
* Corresponding author. Tel.: +44 1865 275656.
E-mail address: mark.moloney@chem.ox.ac.uk (M.G. Moloney).
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.02.066

C. L. Bagwell et al. / Bioorg. Med. Chem. Lett. 20 (2010) 2090–2094
2091
adduct 3c, but in very low yield (14%), and whose stereochemistry
was readily established by NOE analysis (Scheme 1). This compound
could be efficiently deprotected to give pyroglutaminol 4d in good
yield. Of these, 4d was inactive, and although 3c showed significant
bioactivity against E. coli (zone size 17 mm, 0.13% activity of Ceph-
alosporin C), these results along with a consideration of cheminfor-
matic data suggested that these molecules were of inappropriate
Log P and PSA values in comparison to known antibiotics to be
effective antibacterials.12 Since an obvious difference between these
small molecules and the natural product oxazolomycin was spacer
length between amide and lactam moieties, we sought to establish
direct access to analogues where this distance might be readily con-
trolled, and in which the poor yield of the direct coupling stage
leading to 3c might be avoided. Although we have published meth-
odology for the synthesis of the middle fragment of oxazolomycin,51
we elected to use a ‘Click’ strategy48 to join the relevant fragments,
since this would enable direct hybridisation with our earlier re-
ported antibacterials.43 This concept has been used recently to de-
liver antimicrobial properties in other series of compounds.52,53
Extension of this strategy for the addition of hexyn-1-ol permit-
ted the synthesis of alcohol 5a, which was readily converted to
bromide 5b and thence to azide 5c in excellent overall yield; of
these compounds, only bromide 5b was active, consistent with
structural chemotypes not currently in use in existing therapeutic
regimes; they would not be immediately susceptible to resistance-
conferring genes in the bacterial and DNA pools; and they are likely
to provide novel new target proteins and receptors.30,31 The case
for the re-initiation of natural product guided development proto-
cols therefore appears to be overwhelming.32
The curromycins,33,34 cinnabarimides35 and oxazolomycins36
(Scheme 1) are natural products with particularly interesting and
potent antibacterial activity; the bioactivity of the latter has been
proposed to arise from its protonophoric properties37 and its bio-
synthesis has been the recent focus of attention.38,39 In order to
understand the pharmacophoric source of its unusual antimicro-
bial activity, we initiated a programme to examine some of its
key skeletal subunits.40–42 We demonstrated that the 3-hydroxy-
2,2-dimethylpropanamide unit A induces a U-shaped conforma-
tion and that some analogues exhibit antibacterial activity against
S. aureus and Escherichia coli,43 but that neither simple tetramate
nor pyrrolidinone analogues B exhibited significant antibacterial
activity.44–47 Of interest to us was the possibility that reconstitu-
tion of a mimic of oxazolomycin by conjugation of its smaller com-
ponent units with a suitable linking unit might restore bioactivity,
and we report here the exploration of this concept using a Click-
type approach.48
Tetramate
2 is readily available by a Dieckmann cyclisation
the intrinsic lack of bioactivity of the lactam unit generally ob-
served with compounds 3 and 4 above. ‘Click’ reaction, which re-
quired the presence of stoichiometric copper(I) iodide, with
acetylides 6a–c, readily available using our published protocol,43
gave triazoles 7a–c, whose regiochemistry and stereochemistry
was readily assigned by NOE analysis (Scheme 1) and which could
be deprotected to give alcohols 8a–c; of interest was the unex-
pected low solubility of all triazole conjugates, even in DMSO. Bio-
assay against S. aureus and E. coli indicated complete absence of
activity against the former, and only weak activity of 7a–c, and
8a,c against the latter, with the deprotected compounds 8 more ac-
tive than the fully protected ones 7.
In order to probe the effect of both a more flexible side chain
and a less polar linking group, reduction of acetylene 5a (H-Cube)
to alcohol 9 followed by Swern oxidation and condensation with
hydroxylamine gave oxime 10b; cycloaddition with acetylides
6a–c gave isoxazoles 11a–c in modest yield, which could be readily
exploiting our previously published approach,40,41 and kinetically
controlled nucleophile delivery has been shown to deliver the
endo-isomer,49 an outcome in keeping with the reported diastere-
oselectivity of the hydride reduction of tetramates.50 In the case
of acetylide anions, we found that a similar process operated effi-
ciently, giving acetylenic alcohols 3a,b (R = Ph, (CH2)7CH3), whose
structure for 3a was confirmed by NOE analysis. This material could
be readily deprotected by application of Corey–Reichard conditions
giving 4a,b which in the case of the former was followed by alkaline
hydrolysis to give 4c; the stereostructure of 4a was again confirmed
by NOE analysis. Of significance is that all of 3b, 4a,b were inactive
by bioassay against S. aureus and E. coli (Table 1), although alcohol
3a showed weak activity (zone size 13 mm), and this is consistent
with earlier observations of low antibacterial activity of related tet-
ramates.45,49 Addition of acetylene 6a, a compound we had earlier
shown to exhibit activity against E. coli.,43 in a related manner gave

2092
C. L. Bagwell et al. / Bioorg. Med. Chem. Lett. 20 (2010) 2090–2094
Scheme 1.
deprotected to give alcohols 12a–c. These compounds exhibited
both higher Log P and lower PSA values than the corresponding tri-
azoles 7a–c (Table 1), consistent with their lower polarity, and
some were found to exhibit a higher level of bioactivity than the
triazoles (Table 1); compounds 11c and 12b proved to be the most
active found from this analysis, with 0.1% and 0.2%, respectively, of
the activity of the cephalosporin C standard. Of interest is the
different pattern of bioactivity of these libraries compared to the

C. L. Bagwell et al. / Bioorg. Med. Chem. Lett. 20 (2010) 2090–2094
2093
Table 1
Cheminformatic and bioassay data against S. aureus and E. coli of compound libraries
Compound
Log Pa
PSA
%PSA
MSA
Bioactivityb
S. aureus
E. coli
Zone diameter (mm)
Zone diameter (mm)
Relative potencyc (%)
3a
3b
3c
4a
4b
4c
4d
5a
5b
7a
7b
7c
8a
8b
8c
9
11a
11b
11c
12a
12b
12c
3.59
5.57
3.19
0.94
2.92
0.80
0.54
2.35
3.89
4.22
5.6
5.84
1.58
2.95
3.2
2.61
4.59
5.96
6.21
1.94
3.32
3.57
76.1
12.9
10.4
15.7
21.1
16.1
26.2
22.0
15.7
12.2
15.8
14.2
14.0
20.6
18.3
18.0
14.6
15.1
13.6
13.4
19.7
17.5
17.3
589.9
13
Inactive
Inactive
17
—
—
0.1
—
—
—
—
—
0.05
—
—
—
—
—
—
—
—
—
76.1
125.4
95.9
729.1
796.7
454.8
594.2
419.2
661.2
614.9
624.0
990.4
1097.3
1114.4
854.5
962.6
977.2
657.4
1003.3
1112.5
1125.9
867.8
977.7
990.6
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
Inactive
14
95.9
109.7
145.19
96.3
76.07
156.11
156.11
156.11
175.9
175.9
175.9
96.3
151.43
151.43
151.43
171.22
171.22
171.22
14d
14d,e
13d,e
15d
Inactive
14d
Inactive
Inactive
Inactive
15d,e
—
—
0.2
0.1
13d,e
17
14
a
b
c
Log P, PSA and MSA calculated using Marvin.54
Hole plate bioassay at 4 mg/ml (7:3 DMSO/H2O) with zone diameter in mm ( 1 mm).60
Expressed as zone diameter per M, of the analyte relative to cephalosporin C standard.
Halo only.
Very feint.
d
e
parent oxazolomycins, which display activity against S. aureus but
not E. coli.36 The significantly lower cell wall permeability of Gram-
negative bacteria, as a result of the presence of the lipopolysaccha-
ride outer membrane, is well known, making the development of
novel Gram negative antibacterials particularly challenging.12
Chemical informatics analysis is instructive (Table 1); Log P val-
ues confirm the lipophilic character of protected compounds 3a–c,
7a–c and 11a–c but the significantly more hydrophilic character of
the deprotected equivalents 4a–c, 8a–c and 12a–c, and that the
most acidic group of these molecules (either the tertiary hydroxyl
or the lactam NH) possessed pKa values of 11.6–13.1, so that they
are unlikely to be ionised under physiological conditions.54 All
compounds have van der Waals molecular surface area in the
range 600–1100 Å2,55 as would be expected from their common
structural skeleton. The polar surface area parameter (PSA), which
correlates the presence of polar atoms with membrane permeabil-
ity and therefore gives an indication of drug transport properties,56
has been reported to have an optimal value of 70 < PSA < 120 Å2 for
a non-CNS orally absorbable drug,57 and of interest is that, of the
most active compounds, 3c was only just outside this range, while
compounds 12b,c possessed PSA values significantly greater than
120 Å.2 As a proportion of MSA, %PSA values were in the range
10–26%, with the most bioactive nearer to 17%. Noteworthy was
that the three most active compounds 3c and 12b,c possessed sim-
ilar Log P (3.19–3.57) and %PSA (15.7–17.5) values, while the sim-
pler lactams 3a, 3b, 4a and 4c (which lack the amide subunit),
whose Log P and %PSA values differed markedly from these values,
were found to be completely inactive. Comparison indicates that
all compounds with a Log P value of approximately 3.2–4.2 exhibit
bioactivity, but outside this range bioactivity diminishes; this is
likely to reflect a balance between membrane permeability and
aqueous solubility.
properties, and without any bias of preselection of the most opti-
mal target of a pathway, although the downside is that further
optimisation will require identification of the target.58,59 However,
we currently have no information relating to the mode of action of
these compounds, although the oxazolomycins are reported to
possess protonophoric activity.37
We have shown that hybrid analogues of the oxazolomycins,
comprising lactam and amide subunits, exhibit anti-infective bio-
activity not shown by the lactam component alone, although the
activity is weak, at least in comparison to the standard, cephalo-
sporin C. Moreover, the selectivity against different organisms of
the analogues appears to differ from that of the natural products.36
However, the ready access to structurally unusual templates based
upon these natural products should permit convenient library gen-
eration for the purpose of optimisation against different bacterial
targets.
Acknowledgments
This work was supported by EPSRC grant GR/T20380/01 (M.Y.),
and an EPSRC DTA (C.L.B.).
References and notes
1. Davies, J. J. Ind. Microbiol. Biotechnol. 2006, 33, 496.
2. Kotra, L. P.; Golemi, D.; Vakulenko, S.; Mobashery, S. Chem. Ind. 2000, 341.
3. Niccolai, D.; Tarsi, L.; Thomas, R. J. Chem. Commun. 1997, 2333.
4. Wright, G. D.; Sutherland, A. D. Trends Mol. Med. 2007, 13, 260.
5. Baltz, R. H. J. Ind. Microbiol. Biotechnol. 2006, 33, 507.
6. Fasehun, F. Bull. World Health Organ. 1999, 77, 211.
7. Trouiller, P.; Olliaro, P. P.; Torreele, E. E.; Orbinski, J.; Laing, R.; Ford, N. Lancet
2002, 359, 2188.
8. McBain, A. J.; Gilbert, P.; Allison, D. G. Revi. Environ. Sci. Bio./Technol. 2003, 2,
141.
9. Payne, D. J.; Gwynn, M. N.; Holmes, D. J.; Pompliano, D. L. Nat. Rev. Drug Disc.
2007, 6, 29.
The value of forward chemical genetic (phenotypic) screens in
cell-based systems used in this work is that hits are automatically
selected for their combination of activity and cell-permeability
10. Butler, M. S. J. Nat. Prod. 2004, 67, 2141.
11. Newman, D. J.; Cragg, G. M.; Snader, K. M. J. Nat. Prod. 2003, 66, 1022.

2094
C. L. Bagwell et al. / Bioorg. Med. Chem. Lett. 20 (2010) 2090–2094
12. O’Shea, R.; Moser, H. E. J. Med. Chem. 2008, 51, 2871.
13. Silver, L. L. Expert Opin. Drug Discovery 2008, 3, 487.
14. Bumann, D. Curr. Opin. Microbiol. 2008, 11, 387.
15. Galloway, W. R. J. D.; Bender, A.; Welch, M.; Spring, D. R. Chem. Commun. 2009,
2446.
16. Moy, T. I.; Conery, A. L.; Larkins-Ford, J.; Wu, G.; Mazitschek, R.; Casadei, G.;
Lewis, K.; Carpenter, A. E.; Ausubel, F. M. ACS Chem. Biol. 2009, 4, 527.
17. Moy, T. I.; Ball, A. R.; Anklesaria, Z.; Casadei, G.; Lewis, K.; Ausubel, F. M. Proc.
Natl Acad. Sci. U.S.A. 2006, 103, 10414.
39. Zhao, C. H.; Ju, J. H.; Christenson, S. D.; Smith, W. C.; Song, D. F.; Zhou, X. F.;
Shen, B.; Deng, Z. X. J. Bacteriol. 2006, 188, 4142.
40. Andrews, M. D.; Brewster, A. G.; Moloney, M. G. Synlett 1996, 612.
41. Andrews, M. D.; Brewster, A. G.; Crapnell, K. M.; Ibbett, A. J.; Jones, T.; Moloney,
M. G.; Prout, K.; Watkin, D. J. Chem. Soc., Perkin Trans. 1 1998, 223.
42. Andrews, M. D.; Brewster, A. G.; Moloney, M. G. J. Chem. Soc., Perkin Trans. 1
2002, 80.
43. Bagwell, C. L.; Moloney, M. G.; Thompson, A. L. Biorg. Med. Chem. Lett. 2008, 18,
4081.
18. Butler, M. S. Nat. Prod. Rep. 2008, 25, 475.
44. Jeong, Y.-C.; Moloney, M. G. Synlett 2009, 2487.
19. von Nussbaum, F.; Brands, M.; Hinzen, B.; Weigand, S.; Habich, D. Angew.
Chem., Int. Ed. 2006, 45, 5072.
20. Gullo, V. P.; McAlpine, J.; Lam, K. S.; Baker, D.; Petersen, F. J. Ind. Microbiol.
Biotechnol. 2006, 33, 523.
45. Anwar, M.; Moloney, M. G. Tetrahedron Lett. 2007, 48, 7259.
46. Chandan, N.; Moloney, M. G. Org. Biomol. Chem. 2008, 6, 3664.
47. Hill, T.; Kocis, P.; Moloney, M. G. Tetrahedron Lett. 2006, 47, 1461.
48. Evans, R. A. Aust. J. Chem. 2007, 60, 384.
21. Rouhi, A. M. Chem. Eng. News 2003, 81, 77.
22. Breinbauer, R. B.; Vetter, I. R.; Waldmann, H. Angew. Chem., Int. Ed. 2002, 41,
2878.
49. Moloney, M. G.; Yaqoob, M. Tetrahedron Lett. 2008, 49, 6202.
50. Galeotti, N.; Poncet, J.; Chiche, L.; Jouin, P. J. Org. Chem. 1993, 58, 5370.
51. Wang, Z. Y.; Moloney, M. G. Tetrahedron Lett. 2002, 43, 9629.
52. Rogers, S. A.; Krayer, M.; Lindsey, J. S.; Melander, C. Org. Biomol. Chem. 2009, 7,
603.
53. Giffin, M. J.; Heaslet, H.; Brik, A.; Lin, Y.-C.; Cauvi, G.; Wong, C.-H.; McRee, D. E.;
Elder, J. H.; Stout, C. D.; Torbett, B. E. J. Med. Chem. 2008, 51, 6263.
54. Marvin was used for drawing, displaying and characterising chemical
structures, substructures and reactions, Marvin 5.2.1, 2009, ChemAxon
(http://www.chemaxon.com). Calculator Plugins were used for structure
property prediction and calculation, Marvin 5.2.5.1, 2009, ChemAxon (http://
www.chemaxon.com).
23. Ganesan, A. Curr. Opin. Biotechnol. 2004, 15, 584.
24. Koch, M. A.; Waldmann, H. Drug Discovery Today 2005, 10, 471.
25. Shang, S.; Tan, D. S. Curr. Opin. Chem. Biol. 2005, 9, 248.
26. Nören-Müller, A.; Prinz, H.; Rosenbaum, C.; Saxena, K.; Schwalbe, H. J.;
Vestweber, D.; Cagna, G.; Schunk, S.; Schwarz, O.; Schiewe, H.; Waldmann, H.
Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 10606.
27. Banerjee, S.; Wang, Z.; Mohammad, M.; Sarkar, F. H.; Mohammad, R. M. J. Nat.
Prod. 2008, 71, 492.
28. Newman, D. J.; Cragg, G. M. J. Nat. Prod. 2007, 70, 461.
29. Quinn, R. J.; Carroll, A. R.; Pham, N. B.; Baron, P.; Palframan, M. E.; Suraweera,
L.; Pierens, G. K.; Muresan, S. J. Nat. Prod. 2008, 71, 464.
30. Balamurugan, R.; Dekkerab, F. J.; Waldmann, H. Mol. BioSyst. 2005, 1, 36.
31. Koch, M. A.; Wittenberg, L. O.; Basu, S.; Jeyaraj, D. A.; Gourzoulidou, E.;
Reinecke, K.; Odermatt, A.; Waldmann, H. Proc. Natl. Acad. Sci. U.S.A. 2004, 101,
16721.
32. Newman, D. J. J. Med. Chem. 2008, 51, 2589.
33. Nakamura, M.; Honma, H.; Kamada, M.; Ohno, T.; Kunimoto, S.; Ikeda, Y.;
Kondo, S.; Takeuchi, T. J. Antibiot. 1994, 47, 616.
55. Ferrara, P. Proteins: Struct. Function Bioinformatics 2002, 46, 24.
56. Ertl, P.; Rohde, B.; Selzer, P. J. Med. Chem. 2000, 43, 3714.
57. Ertl, P. In Molecular Drug Properties; Mannhold, R., Ed.; Wiley-VCH: Weinheim,
2007; p 111.
58. Puri, A. A.; Bogyo, M. ACS Chem. Biol. 2009, 4, 603.
59. Frearson, J. A.; Collie, I. T. Drug Discovery Today 2009, 14, 1150.
60. Bioassay of penicillin products:61–63 Microbiological assays were performed by
the hole- plate method with the test organism Staphylococcus aureus N.C.T.C.
6571 or E. coli X580. Solutions (100 ll) of the compounds to be tested (4 mg/
34. Ogura, M.; Nakayama, H.; Furihata, K.; Shimazu, A.; Seto, H.; Otake, N. J.
Antibiot. 1985, 38, 669.
35. Stadler, M.; Bitzer, J.; Mayer-Bartschmid, A.; Muller, H.; Benet-Buchholz, J.;
Gantner, F.; Tichy, H.-V.; Reinemer, P.; Bacon, K. B. J. Nat. Prod. 2007, 70, 246.
36. Moloney, M. G.; Trippier, P. C.; Yaqoob, M.; Wang, Z. Curr. Drug Discovery
Technol. 2004, 1, 181.
37. Grigorjev, P. A.; Schlegel, R.; Grafe, U. Pharmazie 1992, 47, 707.
38. Dorrestein, P. C.; Van Lanen, S. G.; Li, W. L.; Zhao, C. H.; Deng, Z. X.; Shen, B.;
Kelleher, N. L. J. Am. Chem. Soc. 2006, 128, 10386.
ml) were loaded into wells in bioassay plates, and incubated overnight at 37 °C.
The diameters of the resultant inhibition zones were measured, and relative
potency estimated by reference to standards prepared with Cephalosporin C.64
61. Smith, B.; Warren, S. C.; Newton, G. G. F.; Abraham, E. P. Biochem. J. 1967, 103,
877.
62. Baldwin, J. E.; Coates, J. B.; Halpern, J.; Moloney, M. G.; Pratt, A. J. Biochem. J.
1989, 261, 197.
63. Baldwin, J. E.; Pratt, A. J.; Moloney, M. G. Tetrahedron 1987, 43, 2565.
64. Boyd, D. B.; Ott, J. L. Antimicrob. Agents Chemother. 1986, 29, 774.