Development of a technique to determine bicyclomycin-rho binding and stoichiometry by isothermal titration calorimetry and mass spectrometry

Article abstract of DOI:10.1021/ja046441q

Bicyclomycin (1) is the only natural product inhibitor of the transcription termination factor rho. Rho is a hexameric helicase that terminates nascent RNA transcripts utilizing ATP hydrolysis and is an essential protein for many bacteria. The paucity of

Full text of DOI:10.1021/ja046441q

Published on Web 02/03/2005
Development of a Technique to Determine Bicyclomycin-Rho
Binding and Stoichiometry by Isothermal Titration Calorimetry
and Mass Spectrometry
Andrew P. Brogan,† William R. Widger,*,‡ Dalila Bensadek,§ Isabel Riba-Garcia,§
Simon J. Gaskell,*,§ and Harold Kohn*,†
DiVision of Medicinal Chemistry and Natural Products, School of Pharmacy, UniVersity of
North Carolina, Chapel Hill, North Carolina 27599-7360, Department of Biology and
Biochemistry, UniVersity of Houston, Houston, Texas 77204-5001, and Michael Barber Centre
for Mass Spectrometry, UMIST, P.O. Box 88, Manchester M60 1QD, United Kingdom
Received June 16, 2004; Revised Manuscript Received December 15, 2004; E-mail: harold_kohn@unc.edu; widger@uh.edu;
simon.gaskell@umist.ac.uk
Abstract: Bicyclomycin (1) is the only natural product inhibitor of the transcription termination factor rho.
Rho is a hexameric helicase that terminates nascent RNA transcripts utilizing ATP hydrolysis and is an
essential protein for many bacteria. The paucity of information concerning the 1-rho interaction stems
from the weak binding affinity of 1. We report a novel technique using imine formation with rho to enhance
the affinity of a bicyclomycin analogue and determine the binding stoichiometry by isothermal titration
calorimetry (ITC) and mass spectrometry (MS). Our designed bicyclomycin ligand, 5a-(3-formyl-phenyl-
sulfanyl)-dihydrobicyclomycin (2) (apparent I50 ) 4 µM), inhibits rho an order of magnitude more efficiently
than 1 (I50 ) 60 µM). MS shows that 2 selectively forms an imine with K181 in rho. We found that despite
the heterogeneity of ATP binding (three tight and three weak) imposed on the rho hexamer, the nearby
bicyclomycin binding pocket is not affected, and both 1 and 2 bind with equal affinity to all six subunits.
Introduction
RNA binding sites are located in the N-terminus and span all
six subunits.11 It is the initial site for RNA binding and
recognizes rut (rho utilizing) sites on the nascent transcript,
characterized by high cytosine content. The primary RNA
binding sites will also bind single-stranded DNA sequences with
high cytosine content. The secondary RNA binding sites line
the central hole and consist of basic residues (Lys, Arg)
positioned at the N- and the C-terminal direction from the ATP
binding site on the individual rho monomers.12-14 Binding at
the secondary sites are RNA specific, unlike the primary sites.
Sequential binding of the nascent transcript at the secondary
sites of the individual monomers allows rho to translocate along
the mRNA.10 RNA binds to the secondary sites less tightly than
it does to the primary sites.15,16
Cell viability requires the orderly expression of genes. In
Escherichia coli, RNA transcripts are terminated either by a
spontaneous process or by the intervention of the transcription
termination factor rho.1 It has been estimated that rho actively
terminates transcription in approximately half of the open
reading frames in E. coli.2 Rho is a homohexameric protein
arranged in a toroid shape.3-9 Rho translocates along the mRNA
and mediates the release of transcripts by disrupting RNA
polymerase polynucleotide biosynthesis.10
Several ligand sites in rho are essential for protein function.
These include two distinct RNA binding sites, termed primary
and secondary, and an ATP binding site.10 The primary, tight
† University of North Carolina.
Secondary site RNA binding stimulates ATP hydrolysis that,
in turn, fuels rho translocation. The ATP binding site is located
in the C-terminal domain at the interface of adjacent subunits.
The nucleotide-binding pocket is distinguished by Walker-A
and Walker-B sequence motifs with the signature P-loop at
residues 175-183 in E. coli rho. The stoichiometry of ATP
binding to rho has been investigated. Two models, a three tight-
site model17-19 and a six-site model where three of the six sites
‡ University of Houston.
§ Michael Barber Centre for Mass Spectrometry, UMIST.
(1) Yager, T. D.; von Hippel, P. H. In The Molecular and Cell Biology of E.
coli. and S. typhimurium; Neidhardt, F., Ed.; The American Society of
Microbiology: Washington, DC, 1987; pp 1241-1275.
(2) Zsu, A. Q.; von Hippel, P. H. Biochemistry 1998, 37, 11202-11214.
(3) Oda, T.; Takanami, M. J. Mol. Biol. 1972, 71, 799-802.
(4) Shigesada, K.; Tsurushita, N.; Imai, M. Ann. Rep. Inst. Virus Res., Kyoto
UniV. 1978, 21, 72-73.
(5) Bear, D. G.; Hicks, P. S.; Escudero, K. W.; Andrews, C. L.; McSwiggen,
J. A.; von Hippel, P. H. J. Mol. Biol. 1988, 199, 623-635.
(6) Gogol, E. P.; Seifried, S. E.; von Hippel, P. H. J. Mol. Biol. 1991, 221,
1127-1138.
(11) Richardson, J. P. J. Biol. Chem. 1982, 257, 5760-5766.
(12) Wei, R. R.; Richardson, J. P. J. Biol. Chem. 2001, 276, 28380-28387.
(13) Wei, R. R.; Richardson, J. P. J. Mol. Biol. 2001, 314, 1007-1015.
(14) Xu, Y.; Kohn, H.; Widger, W. R. J. Biol. Chem. 2002, 277, 30023-30030.
(15) Geiselmann, J.; Yager, T. D.; von Hippel, P. H. Protein Sci. 1992, 1, 861-
873.
(7) Geiselmann, J.; Yager, T. D.; Gill, S. C.; Calmettes, P.; von Hippel, P. H.
Biochemistry 1992, 31, 111-121.
(8) Yu, X.; Horiguchi, T.; Shigesada, K.; Egelman, E. H. J. Mol. Biol. 2000,
299, 1279-1287.
(9) Skordalakes, E.; Berger, J. M. Cell 2003, 114, 135-146.
(10) Richardson, J. P. Biochim. Biophys. Acta 2002, 1577, 251-260.
(16) Wang, Y.; von Hippel, P. H. J. Biol. Chem. 1993, 268, 13947-13955.
9
10.1021/ja046441q CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 2741-2751
2741

A R T I C L E S
Brogan et al.
bind ATP tightly,20-25 have been proposed. Recently, we
reported the stoichiometry of adenine nucleotides with rho using
a sensitive fluorescence-quenching method that allowed detec-
tion of both the tight and the weak binding sites.25 Using the
rho mutant F355W, we showed that ATP binds to three sites
tightly (Kd1 ) 3 ( 0.3 µM) and three sites loosely (Kd2 ) 58
( 3 µM), while ADP, the product of ATP hydrolysis, binds
loosely (Kd ) 92 µM) to essentially all six sites. The
heterogeneity in ATP binding sites was observed in the absence
and in the presence of single-stranded RNA (poly(C)) and DNA
(poly (dC)).
the drug-binding site, increasing 2 binding affinity. We docu-
ment, using mass spectrometry (MS) and isothermal titration
calorimetry (ITC), that 2 binds to all six subunits, that the
binding affinity of 2 is the same at all subunits, and that binding
is not appreciably affected by the absence or presence of adenine
nucleotides. Furthermore, reversible covalent bond formation
to enhance the binding of low affinity ligands has not been
previously reported, and we examine the broader implications
of this novel methodology.
Bicyclomycin (1), a commercial antibiotic, is the only known
selective inhibitor of rho.26 Mechanistic studies have shown that
1 is a mixed inhibitor of secondary site RNA binding27 and is
a reversible, noncompetitive inhibitor of ATP hydrolysis.28
Bicyclomycin binding occurs at a unique site in the C-terminus
of the rho protomer; this site is situated near the ATP binding
site and overlaps with a portion of the secondary RNA binding
sites. Rho mutational studies29,30 and bicyclomycin affinity
labels24,31-33 suggest that 1 binds in a cleft between the R7 and
the R8 helices (nomenclature based on the Skordalakes and
Berger X-ray crystal structure of rho9), with the C(1) triol group
in 1 pointed toward the outside of the rho hexamer and the
terminal C(5)-C(5a) exomethylene group pointed toward the
terminal γ-phosphate of ATP and the P-loop on rho.34 Signifi-
cantly, bicyclomycin inactivation of one to two subunits in rho
is sufficient to halt rho ATP hydrolysis.24 In vitro kinetic assays
measuring ATP hydrolysis (not to be confused with transcription
termination assays35) provide a Ki of 21 µM for 1, indicating
that the antibiotic binds to rho with low affinity.28 This affinity
has contributed to limited information concerning the stoichi-
ometry of the 1-rho hexamer complex. It is not known whether
1 binds to all six rho subunits and if there is binding
heterogeneity similar to ATP.
Results
1. Design Considerations for ITC and Choice of Bicyclo-
mycin Aldehyde. The thermodynamic binding constants and
stoichiometry for ligand-protein complexes can be measured
by ITC.36-40 An integral parameter in ITC measurements is the
Wiseman c value41-43 (eq 1), where [M]t is the receptor protein
concentration, n is the number of binding sites per receptor,
and Kd is the dissociation constant. Accurate binding isotherms
and the corresponding binding stoichiometry can be determined
if curve fitting is in the experimental window of 10 e c e
500.41-43
n[M]t
c )
(1)
Kd
According to eq 1, low-affinity ligands require higher protein
concentrations to fit within the experimental window. The Kd
for the 1-rho complex is not known, but estimates derived from
Ki (21 µM) and I50 (60 µM) in the rho poly(C)-dependent
ATPase assay indicate it is >10 µM. Thus, to obtain c g 10
and measure 1 binding to rho by ITC, the rho monomer
concentration (n[M]t) needs to exceed 100 µM. This corresponds
to protein levels g5 mg/mL. Attempts to maintain an aqueous
rho solution at 5 mg/mL (100 µM, based on monomer) for the
duration of an ITC experiment (≈5 h, 26 °C) were unsuccessful
(data not shown). However, we were able to achieve stable,
aqueous rho solutions at 24 µM monomer (4 µM hexamer, 1.1
mg/mL) using relatively high KCl concentrations (400 mM) and
the addition of the cofactors, poly(dC) or poly(C).
In a recent study, Turnbull and Daranas reported that binding
isotherms using ITC can be determined for low-affinity systems
when c < 10, if four criteria are met: (1) a sufficient portion
of the binding isotherm is used for analysis, (2) the binding
stoichiometry is known, (3) the concentrations of ligand and
receptor are known with accuracy, and (4) adequate signal-to-
noise is observed during the titration.44 We measured the binding
isotherm for 1 with the 24 µM rho (based on monomer) solution
Herein, we report the synthesis and evaluation of 5a-(3-
formyl-phenylsulfanyl)-dihydrobicyclomycin (2), a bicyclomy-
cin analogue designed to efficiently inhibit rho. We show that
the attachment of an aryl aldehyde moiety to 1 enhances binding
to rho through imine formation with a lysine residue located at
(17) Stitt, B. L. J. Biol. Chem. 1988, 263, 11130-11137.
(18) Stitt, B. L.; Xu, Y. M. J. Biol. Chem. 1998, 273, 26477-26486.
(19) Stitt, B. L. Biochemistry 2001, 40, 2276-2281.
(20) Geiselmann, J.; von Hippel, P. H. Protein Sci. 1992, 1, 850-860.
(21) Kim, D. E.; Shigesada, K.; Patel, S. S. J. Biol. Chem. 1999, 274, 11623-
11628.
(22) Kim, D. E.; Patel, S. S. Biochemistry 2000, 39, 1551-1552.
(23) Kim, D. E.; Patel, S. S. J. Biol. Chem. 2001, 276, 13902-13910.
(24) Vincent, F.; Widger, W. R.; Openshaw, M.; Gaskell, S. J.; Kohn, H.
Biochemistry 2000, 39, 9067-9076.
(25) Xu, Y.; Johnson, J.; Kohn, H.; Widger, W. R. J. Biol.Chem. 2003, 278,
13719-13727.
(26) Zwiefka, A.; Kohn, H.; Widger, W. R. Biochemistry 1993, 32, 3564-3570.
(27) Magyar, A.; Zhang, Y. D.; Kohn, H.; Widger, W. R. J. Biol. Chem. 1996,
271, 25369-25374.
(28) Park, H. G.; Zhang, X. D.; Moon, H. S.; Zwiefka, A.; Cox, K.; Gaskell, S.
J.; Widger, W. R.; Kohn, H. Arch. Biochem. Biophys. 1995, 323, 447-
454.
(29) Magyar, A.; Zhang, X. D.; Abdi, F.; Kohn, H.; Widger, W. R. J. Biol.
Chem. 1999, 274, 7316-7324.
(30) Moyse, K. A.; Knight, J. S.; Richardson, J. P. J. Mol. Biol. 2000, 302,
565-579.
(31) Cho, H. J.; Park, H. G.; Zhang, X. D.; Riba, I.; Gaskell, S. J.; Widger, W.
R.; Kohn, H. J. Org. Chem. 1997, 62, 5432-5440.
(32) Riba, I.; Gaskell, S. J.; Cho, H. J.; Widger, W. R.; Kohn, H. J. Biol. Chem.
1998, 273, 34033-34041.
(36) Jelesarov, I.; Bosshard, H. R. J. Mol. Recognit. 1999, 12, 3-18.
(37) Ladbury, J. E.; Chowdhry, B. Z. Chem. Biol. 1996, 3, 791-801.
(38) Leavitt, S.; Freire, E. Curr. Opin. Struct. Biol. 2001, 11, 560-566.
(39) Ward, W. H.; Holdgate, G. A. Prog. Med. Chem. 2001, 38, 309-376.
(40) Holdgate, G. A. BioTechniques 2001, 31, 164-166, 168, 170.
(41) Wiseman, T.; Williston, S.; Brandts, J. F.; Lin, L. N. Anal. Biochem. 1989,
179, 131-137.
(33) Vincent, F.; Openshaw, M.; Trautwein, M.; Gaskell, S. J.; Kohn, H.; Widger,
W. R. Biochemistry 2000, 39, 9077-9083.
(34) Vincent, F.; Srinivasan, J.; Santillan, A.; Widger, W. R.; Kohn, H. J. Org.
Chem. 2001, 66, 2251-2264.
(35) Wu, A. M.; Christie, G. E.; Platt, T. Proc. Natl. Acad. Sci. U.S.A. 1981,
78, 2913-2917.
(42) Fisher, H. F.; Singh, N. Methods Enzymol. 1995, 259, 194-221.
(43) Indyk, L.; Fisher, H. F. Methods Enzymol. 1998, 295, 350-364.
9
2742 J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005

Bicyclomycin-Rho Binding and Stoichiometry
A R T I C L E S
site to give the bound imine (Figure 2, A h C) and that imine
formation was reversible.24,31 We further demonstrated that
NaBH4 treatment of the bicyclomycin bound imine gave the
irreversibly modified amine (Figure 2, C f D). ESI-MS was
used to analyze the reductive amination rho products obtained
from 3 and 4. We found that 3 (1 mM) bound ≈45% of the
available monomers,31 and MS/MS analysis of trypsin-treated
3-bound rho showed that the lysine residue K181 was selectively
modified.32 Correspondingly, 4 (400 µM) modified approxi-
mately five of the six subunits,24 and adduction occurred at
K336.33 In the present study, we examined whether we could
exploit the ease with which bicyclomycin aldehydes undergo
reversible imine formation to promote bicyclomycin binding
to rho. We anticipated that imine adduction would improve the
binding affinity of the bicyclomycin derivative.
We required a bicyclomycin aldehyde that efficiently bound
all six subunits with high affinity. Neither 3 or 4 was satis-
factory. Bicyclomycin aldehyde 3 (1 mM) modified only half
of the subunits after NaBH4 reduction.31 Furthermore, the
apparent I50 for 3 in the poly(C)-dependent ATPase assay was
70 µM, placing this compound on par with 1 (60 µM).31
Correspondingly, 4 (400 µM) modified five of the six sites after
NaBH4 treatment.24 However, when the inhibition of 4 was
measured in the poly(C)-dependent ATPase assay, we obtained
an apparent I50 of 35 µM and a Ki of 62 µM.24 These findings
indicated that the binding affinities of 3 and 4 did not
substantially differ from 1, making both compounds unsuitable
for ITC stoichiometric measurements.
Bicyclomycin aldehyde 3 is a substituted dihydrobicyclomy-
cin where an aniline substituent is tethered to the 5a site. We
previously documented that 5a-thio-substituted dihydrobicyclo-
mycins exhibited excellent rho inhibitory activities and that their
activities typically exceeded their nitrogen counterparts.45,46 For
example, the I50 for 5a-(phenylsulfanyl)-dihydrobicyclomycin
(5) was 75 µM while 5a-(anilino)-dihydrobicyclomycin (6) was
120 µM. These findings led us to choose 2, the sulfur analogue
of 3, for synthesis and evaluation.
Figure 1. Example binding isotherm for the titration of 1 into a rho solu-
tion (24 µM, based on monomer; 4 mM, based on hexamer) containing
poly(dC) (1.33 µM) and ATP (200 µM). (A) Raw calorimetric trace. Each
peak corresponds to the thermal power evolved from the addition of 5.0
µL of 1 (1 mM) into a 1.4 mL rho solution. (B) Hyperbolic plot of
normalized titration isotherm. The solid line represents the best fitting curve
calculated from assuming six equal binding sites for 1 within the rho
hexamer.
containing poly(dC) (1.3 µM) and ATP (200 µM) using ITC.
Complete saturation of all rho primary RNA binding sites occurs
with 1.3 µM poly(dC) since three to four rho hexamers (4 µM)
will bind to one poly(dC) molecule given an average poly(dC)
length of ≈450 nt. As expected, we obtained a hyperbolic
binding isotherm (rather than sigmoidal), indicative of c < 10,
that could only be analyzed assuming the stoichiometry of the
1-rho complex (Figure 1). Assuming a stoichiometry of six,
where all six sites were identical, yielded a Kd of 40.7 ( 5.0
µM and a corresponding c value of 0.6 using the Origin 5.0
fitting program (Table 1). It is noteworthy that fitting of the 1
binding isotherms to a two-type binding site model (two sets
of three equal binding sites per hexamer) and assuming a
maximum of six total binding sites led to two Kd values (≈40
µM) that were approximately equal to each other. Attempts to
fit the data to the other two binding site models with various
stoichiometries using the Origin 5.0 software (e.g., five tight,
one weak) led to poor fits that did not converge well. This
provided the preliminary information that 1 bound to each of
the subunits within the rho hexamer with equal binding affinity.
Nonetheless, we needed to enhance bicyclomycin-rho binding
(lower Kd) to provide c values that would allow independent
determinations of the Kd and stoichiometry of the bicyclomy-
cin-rho complex.
2. Synthesis and Characterization of 5a-(Phenylsulfanyl)-
dihydrobicyclomycin (2). Synthesis of 2 proceeded in four steps
(Scheme 1). Treatment of 1 with 2,2-dimethoxypropane and
p-toluenesulfonic acid gave 2′,3′-acetonide 7.47 Dissolution of
We previously reported that bicyclomycin aldehydes 3 and
4 reacted with lysine residues near the bicyclomycin binding
(45) Park, H.; Zhang, Z. M.; Zhang, X. D.; Widger, W. R.; Kohn, H. J. Org.
Chem. 1996, 61, 7764-7776.
(46) Brogan, A. P.; Widger, W. R.; Kohn, H. J. Org. Chem. 2003, 68, 5575-
5587.
(44) Turnbull, W. B.; Daranas, A. H. J. Am. Chem. Soc. 2003, 125, 14859-
14866.
(47) Kamiya, T.; Maeno, S.; Kataura, Y. Belgium Patent (847-475), 1976.
9
J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005 2743

A R T I C L E S
Brogan et al.
Table 1. Isothermal Titration Calorimetry. Titration of Hexameric Rho (4 µM Rho, Based on Hexamer) with 1 or 2 at 26 °C
ligand (titrant)
poly-nucleotide
adenosine nucloteide
Kd
(µ
M)
∆G
b (kcal/ mol)
∆H
b (kcal/mol)
T
∆S
b (kcal/mol)
na
cb
1
2
2
poly(dC)
poly(dC)
poly(dC)
ATP
ATP
none
40.7 ( 5.0
2.5 ( 1.8
6.3 ( 0.2
-6.01 ( 0.08
-7.70 ( 0.44
-7.12 ( 0.02
-0.79 ( 0.72
-2.74 ( 0.94
-1.34 ( 0.62
5.22 ( 0.80
4.95 ( 0.24
5.78 ( 0.26
6c
6.1 ( 2.2
6c
0.6
10
4
a n ) stoichiometry of binding, based on hexameric rho. b Wiseman c value.41-43 c Stoichiometry of six was fixed for this value prior to fitting.44 All
values reported are the average of two independent experiments ( two times the standard deviation.
Scheme 1. Preparation of
5a-(3-Formyl-phenylsulfanyl)-dihydrobicyclomycin (2)
Figure 2. Proposed mode of action of bicyclomycin reductive amination
probes 2-4.
7 with 3-mercaptobenzyl alcohol48 in 50% aqueous methanol
adjusted to pH 10.5 yielded the Michael addition adduct 8 as a
mixture of diastereomers (≈3:1, NMR analysis). Deprotection
of the acetonide group in 8, using TFA, gave 9. Careful
oxidation of the benzyl alcohol unit with MnO2 provided
displayed little or no antimicrobial activity against W3350 E.
aldehyde 2 without competitive oxidation of the sulfide linkage,
coli.34,45,46,52 Accordingly, we tested 2 in this assay and found
in 18% overall yield (four steps).
it to be the most efficient inhibitor of rho in vitro activity to
1
The spectroscopic data (IR, H NMR, 13C NMR, low- and
date. It had an apparent I50 of 4 µM, an order of magnitude
high-resolution mass spectrometry) for 2 were in agreement with
lower than the parent compound 1 (I50 ) 60 µM) and the
its proposed structural assignment. Diagnostic NMR signals for
previous bicyclomycin reductive amination probes 3 (apparent
2 were the appearance of the aldehyde proton at δ 9.94 in the
I50 ) 70 µM) and 4 (apparent I50 ) 35 µM). The improved
1H NMR49 and the detection of the aldehyde carbon at 193.8
inhibitory activity of 2 as compared with 3 was consistent with
ppm in the 13C NMR.50 Aldehyde 2 was isolated as a
expectation.45,46 The factors, however, that contributed to this
diastereomeric mixture. We were unable to separate the isomers
improved bioactivity remain unclear. Recently, we obtained an
by PTLC. For 6, X-ray crystallographic analysis showed that
X-ray crystallographic structure for 2 bound to rho.53 The aryl
the major isomer corresponded to the C(5)-S isomer.45 Bicy-
unit in 2 extended out of the bicyclomycin binding pocket and
clomycin aldehyde 2 was routinely stored as a solid (-80 °C)
nestled between the hydrophobic side chain of P180 of one
for months without apparent decomposition (TLC and NMR
monomer and the aliphatic portion of the side chain of K336
analysis).
from the adjacent monomer. The terminal aldehyde of this
3. Biochemical and Biological Studies. The rho poly(C)-
dependent ATPase assay51 serves as a reliable test of rho
(50) Stothers, J. B. Carbon-13 NMR Spectroscopy; Academic Press: New York,
1972.
inhibition, and any compound that was ineffective in this assay
(51) Sharp, J. A.; Galloway, J. L.; Platt, T. J. Biol. Chem. 1983, 258, 3482-
3486.
(48) Thurber, T. C.; Prince, A.; Halpern, O. J. Heterocycl. Chem. 1982, 19,
961-965.
(49) Jackman, L. M.; Sternhell, S. Applications of Nuclear Magnetic Resonance
Spectroscopy in Organic Chemistry; Permagon Press: Oxford, 1969.
(52) Santillan, A.; Zhang, X. D.; Widger, R.; Kohn, H. J. Org. Chem. 1998,
63, 1290-1298.
(53) Skordalakes, E.; Brogan, A. P.; Park, B. S.; Kohn, H.; Berger, J. M.
Structure, 2005, 13, 99-109.
9
2744 J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005

Bicyclomycin-Rho Binding and Stoichiometry
A R T I C L E S
Figure 4. Histogram depicting the reversibility of 2-rho imine formation
by measuring percent of rho ATPase activity at various time points after
passage through a Bio-Spin 6 column using a fixed concentration of 2 (10
µM). Control is without 2.
findings indicate that 2 is a more efficient reductive amination
probe with rho than its predecessors 331 and 4,24 with 100%
loss of ATPase activity at 2 concentrations of g40 µM and
maximum imine formation occurring at 2 h when rho concen-
trations are 1 µM (based on monomer).
Analysis of 3 and 4 in previous studies24,31 indicated that the
imines formed with rho are transient and that full recovery of
rho ATPase activity occurs upon dissolution in buffer solutions
at 1 h for 331 and at 24 h for 4.24 These findings indicate that
the 3-rho imine was less stable than the 4-rho imine. We
recognized that the reversibility of the 2-rho complex as an
important parameter to consider when designing our ITC
experiments. Accordingly, we determined the reversibility of
the 2-rho transformation by measuring the incubation time
required for recovery of rho poly(C)-dependent ATPase activity.
Solutions containing rho (1 µM, based on monomer), poly(C)
(40 nM), and a fixed concentration of 2 (10 µM) were incubated
for 2 h, and unbound 2 was removed by passage through a
Biospin-6 column. The rho ATPase activity was measured after
various incubation times following removal of unbound 2 (0-4
h) (Figure 4). We found that after spin column treatment, 2
forms a moderately stable rho-imine and that the 2-rho
interaction is a reversible process with full recovery of ATPase
activity after 2 h. Thus, the apparent I50 for 2 likely corresponds
to a combination of structural complementarity with the bicy-
clomycin binding pocket in rho and 2-rho imine formation.
We proceeded to determine the Ki for 2 using the kinetics of
the rho poly(C)-dependent ATPase assay. The Lineweaver-
Burk plot of 1/V versus [ATP] at various 2 concentrations
shown in Figure 5A was obtained using the Enzyme Kinetics
Module 1.1 for SigmaPlot 2001 and revealed noncompetitive
inhibition (R2 ) 0.984) with a Ki ) 6 µM. Similar to the I50
values observed for 1-4, the Ki for 2 is significantly lower than
that of 1 (Ki ) 21 µM), 3 (Ki ) 27 µM), and 4 (Ki ) 62 µM).
To ensure that 2-rho imine formation inhibited rho ATPase
activity by a reversible noncompetitive pathway with respect
to ATP, we measured the velocity of ATP hydrolysis (Vmax) as
a function of total rho concentration (125-1000 nM, based on
monomer) in the presence of 2 (5 µM). We compared these
results with 1 (60 µM) and reactions performed in the absence
of inhibitor. Plots of Vmax versus [rho] without inhibitor (control),
1 (60 µM), and 2 (5 µM) yielded straight lines that all intersected
Figure 3. (A) Histogram depicting percent of rho ATPase activity after
preincubation (2-4 h) with 2 followed by reductive amination with NaBH4
(60 mM, final concentration) and centrifugation through a Bio-spin 6 column
at various concentrations of 2. (B) Histogram depicting the required
preincubation time for maximum imine formation. Addition of NaBH4 (60
mM, final concentration) at various preincubation times at a fixed 2
concentration (10 µM). Control is without 2.
moiety appeared to be able to hydrogen bond to K181, but the
poor electron density for this section of the molecule suggested
that the thiobenzaldehyde unit is mobile. The C(5) stereochem-
istry in 2 could not be assigned because of the resolution (2.90
Å) of the X-ray crystallographic structure. Since we do not have
a comparable image for 3 with rho, we do not know if the same
binding interactions seen in the 2-rho structure exist for 3. Little
antimicrobial activity was observed for 2 when measured against
W3350 E. coli in a filter disk assay.54 The minimum inhib-
itory concentration (MIC) for 2 was only 16 mg/mL, while the
MIC for 1 in a control experiment was 0.21 mg/mL. The lack
of antimicrobial activity for 2 was in agreement with pre-
vious findings for this class of 5a-substituted dihydrobicyclo-
mycins.45
Next, we determined whether 2 served as a reductive
amination probe. Incubation of rho (1 µM, based on monomer)
and poly(C) (40 nM) with various concentrations of 2 (0-40
µM) followed by treatment with NaBH4 led to permanent loss
of rho ATPase activity (Figure 3A). Increasing loss of ATPase
activity was observed with increasing 2 concentrations and 100%
loss of activity at 40 µM. The addition of 1 (5 mM) prohibited
the irreversible inactivation of rho ATPase activity by 2,
indicating that 1 and 2 bind to the same site in the protein (data
not shown). We also examined the incubation time required for
maximum reductive amination by 2. When solutions containing
rho (1 µM, based on monomer), poly(C) (40 nM), and a fixed
concentration of 2 (10 µM) were preincubated for various times
(0-4 h) followed by NaBH4 reduction, we observed a time to
maximum imine formation of 2 h (Figure 3B) if we assume
that NaBH4 reduction of the 2-rho imine is rapid. These
(54) Ericsson, H. M.; Sherris, J. C. Acta Pathol. Microbiol. Scand. B: Microbiol.
Immunol. 1971, 217, Suppl.
9
J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005 2745

A R T I C L E S
Brogan et al.
the spectrum of native rho (MALDI data not shown). When
ESI-MS analysis was performed, the ions at m/z 737.80 and
492.23 were observed in the modified 2-rho spectra and
corresponded to the doubly and triply protonated species from
the singly charged ion at m/z 1474.99 observed by MALDI. If
the ion at m/z 1474.99 (MH+) is a modified fragment, then the
mass of the unmodified fragment must be 1473.99-424.13 )
1049.86 Da. This mass is unique to the tryptic fragment T23-
24 (residues 174-184; GLIVAPPKAGK), making it a good
candidate for modification by 2 since it contains an internal
lysine residue.
With this hypothesis in mind, the site of 2 modification was
further investigated by tandem MS experiments. The product
ion spectra were recorded for the doubly and the triply charged
precursor ions, m/z 737.80 and 492.23, respectively. The
fragment ions generated in the collision induced dissociation
(CID) experiments allowed the identification of the following
N-terminal partial sequence tag (G,L)IVAXn (Figure 7). Since
this sequence occurs only once in the complete rho sequence,
it allowed the unequivocal identification of the modified tryptic
peptide as T23-24. This modified fragment contains two lysine
residues (K181 and K184). However, since a modified lysine
would not be recognized by trypsin as a cleavage site, K181 is
identified as the site of modification. Furthermore, the y1 ion
corresponding to the unmodified K184 residue was observed
in both tandem MS analyses. The higher y-series fragment ions
(y4-y8) have a mass shift of 424 Da, consistent with K181
modification by reductive amination with 2. Other y-series ions
(y5-y8) with a mass increment of 314 Da were also observed,
suggesting additional fragmentation of the adducted moiety.
Together, these results allowed us to identify K181 as the site
of modification, in keeping with previous studies with 3.32
Figure 5. (A) Lineweaver-Burk plot of the poly(C)-dependent ATPase
activity of rho at varying 2 concentrations obtained from the Enzyme
Kinetics Module 1.1 in SigmaPlot 2001. (B) Plot of Vmax vs [rho]. All lines
intersect the origin indicating reversible inhibition.
5. Isothermal Titration Calorimetry (ITC). We have shown
the time to maximum imine formation of a solution containing
10 µM 2 and 1 µM rho (based on monomer) was 2 h (Figure
3B). The formation of the reversible 2-rho imine is likely a
second-order process that depends on the concentrations of both
2 and rho. We took this into consideration when designing our
ITC experiments and recognized that by using 24 µM rho (based
on monomer), the reaction parameters for imine formation would
be altered from those previously measured (Figures 3 and 4).
Accordingly, we varied the time between injections (400-1200
s) and found that a 1200 s interval between injections of 2 was
sufficient for the heat evolved to return to the baseline value
from this reversible binding/imine formation process (a priori).
An adequate return to the baseline value was not observed with
shorter time intervals (data not shown). By comparison, the time
between injections with 1 was 400 s. We obtained a sigmoidal
binding isotherm, indicative of c g10, by titrating 2 (400 µM,
5 µL injections) into a solution of rho (24 µM, based on
monomer) containing poly(dC) (1.3 µM) and ATP (200 µM)
(Figure 8). Using the Origin 5.0 software and the rho hexamer
concentration, the binding isotherm for 2 converged and fit best
to a single type of binding site model with a Kd of 2.5 ( 1.8
µM, a binding stoichiometry of 6.1 ( 2.2, and a corresponding
c value of 10 (Table 1). Similar to the ITC data obtained from
1, fitting of the 2 binding isotherms to a two-type binding site
model led to two Kd values ≈3 µM.
the origin, indicating reversible noncompetitive inhibition
(Figure 5B).55
Collectively, our findings from the inhibition profile, imine
formation, and reductive amination of 2 are consistent with
previous bicyclomycins where 1, 3, and 4 were found to be
reversible, noncompetitive inhibitors of rho ATP hydrolysis.
4. Mass Spectrometry. Previous examination of 3 and 4
reductive amination with rho by ESI-MS indicated that 3
modified ≈45% of the rho monomers,31 while 4 modified five
out of six rho monomers.24 Comparative analysis of the ESI-
MS maximum entropy processed spectra showed a peak that
corresponds to a mass of 47 002 Da for native rho and a mass
of 47 428 Da for the irreversible 2-rho adduct (Figure 6). This
426 Da mass shift for 2-modified rho correlates well with the
expected monoisotopic shift of m/z 425.1322. The ESI-MS
results indicate modification of five and two-tenths out of the
six rho subunits that comprise the hexamer using 100 µM 2.
The modification site within the rho protomer was identified
by analyzing the tryptic digests of both unmodified and
2-modified rho. Comparative analyses of the tryptic digests by
MALDI-MS revealed the presence of a singly charged ion at
m/z 1474.99 in the 2-rho adduct sample, which was absent in
We further examined the 2-rho interaction by changing the
conditions of the rho solution by omitting the nucleotide (Table
(55) Segel, I. H. Enzyme Kinetics; John Wiley & Sons: New York, 1975; pp
127-128.
9
2746 J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005

Bicyclomycin-Rho Binding and Stoichiometry
A R T I C L E S
Figure 6. Maximum entropy processed spectrum obtained from ESI-QTOF-MS of (A) unmodified rho protein (1.5 µM) and (B) modified rho (1.5 µM)
subjected to reductive amination using 2 (100 µM). The mass shift of 426 Da corresponds to 2-modified rho.
1). We observed modest increases in Kd (decreased affinity)
for 2 binding when rho solutions (24 µM, based on monomer)
containing poly(dC) (1.3 µM) and no nucleotide (Kd ) 6.3 (
0.2 µM) were compared with the initial ATP (200 µM) (Kd )
2.5 ( 1.8 µM) conditions. We assumed a six equal binding
site model for these binding isotherms since c < 10 from the
increase in Kd.
binding equally to all six subunits since in our ITC studies with
2 we concluded that there are a total of six bicyclomycin binding
sites within the rho hexamer and that all subunits bind 2 with
the same affinity. In a recent X-ray crystallographic study, we
provided structural data for 1, 2, and 4 bound to rho.53 This
study further validates the equality of the bicyclomycin binding
sites within the rho hexamer. Current crystallization conditions
only permit rho to be crystallized in the lock washer conforma-
tion. This conformation is an inactive ATPase with a 10-12 Å
gap between two of the rho monomers. Electron density for 1,
2, and 4 was observed in five of the six subunits. The inhibitor
was absent from the protomer that did not have a completely
formed binding pocket as a result of the gap in the hexameric
ring. Crystallization of the closed ring form of the rho hexamer
is ongoing; nonetheless ITC, MS, and X-ray crystallographic
studies all point toward bicyclomycin binding equally to all six
subunits within the rho hexamer. This is an intriguing discovery
considering the heterogeneity of binding observed for the nearby
ATP binding pocket. Previously, we reported a bicyclomycin
fluorescent probe (BFP) that allowed us to measure BFP and
ATP binding to the three tight ATP binding sites using fluor-
escence resonance energy transfer (FRET).46 FRET measure-
ments are distance dependent, and we were also able to sense
a conformational change in rho induced by ATP binding. This
conformational change is implicated in rho tracking along the
RNA transcript. Average movements ≈20 Å in the C-ter-
minus were observed upon ATP binding. Apparently, the nearby
R7 and R8 helices involved in 1 binding53 are minimally affected
by this nucleotide-induced movement or the heterogeneity of
ATP binding within the rho hexamer. We were unable to
examine the heterogeneity of BFP and ATP binding in our
FRET studies since we did could not measure binding at the
weak ATP binding subunits. This unanswered question was
examined in the present study, and we were able to assign the
binding of 1 and 2 equally to all six subunits.
A unique attribute of ITC is the ability to measure the
thermodynamic binding parameters ∆Gb, ∆Hb, and ∆Sb. In our
case, we can compare the thermodynamics of ligand 1, which
binds to the protein noncovalently, with a similar ligand 2, which
binds and also forms a reversible imine bond with the protein.
When examining the binding thermodynamics from the ITC
measurements of 1 and 2 under the same conditions (Table 1,
poly(dC) and ATP), we find a difference in ∆∆Gb ≈ 1.7
kcal/mol that corresponds to a similar difference in ∆∆Hb
(≈2.0 kcal/mol) and only a small change in T∆Sb (∆T∆Sb ≈
0.3 kcal/mol). Furthermore, the ∆Hb for 1 is only -0.79
kcal/mol, indicating that 1 binding is driven by the entropy term.
A larger ∆Hb was observed for 2 (∆Hb ) -2.74 kcal/mol),
suggesting that the difference in binding from 1 and 2 is driven
by the enthalpy term that likely corresponds to the thermody-
namics of imine bond formation of 2 with K181 in rho. Further
examination of the thermodynamic parameters of 2 binding in
the absence of adenine nucleotide revealed only modest changes
in ∆Gb. These results indicate that, in the presence of poly-
(dC), bicyclomycin binding is not appreciably affected by either
the absence or the presence of nucleotide.
Discussion
Bicyclomycin is the only natural product inhibitor of the
transcription termination factor rho. The weak binding of 1 has
led to limited information on the 1-rho interaction. We report
the design, synthesis, and evaluation of 5a-(3-formyl-phenyl-
sulfanyl)-dihydrobicyclomycin (2), the most efficient inhibitor
of the rho in vitro activity to date. Examination of the 2-rho
interaction by MS and ITC indicated that 2 binds equally to all
six rho hexamer subunits in the presence and absence of ATP.
Further extrapolation of the ITC data allowed us to assign 1
In addition to defining the binding and stoichiometry of the
1-rho interaction, we outline a novel technique that may have
broader implications. ITC is a powerful tool to determine Kd
9
J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005 2747

A R T I C L E S
Brogan et al.
Figure 7. Product ion spectrum derived from the doubly charged ion at m/z 737.89 for the tryptic fragment T23-24 (174-184, GLIVAPPKAGK) incorporating
modified K181. The y fragment ions labeled with * have a 424 Da mass increment in comparison with masses expected for the unmodified sequence; y ions
labeled with ++ have a 314 Da mass increment. Note that for clarity, the spectrum is presented as three overlapping segments, with separate normalization
of each segment.
and stoichiometry and is the only technique that provides direct
measurement of the thermodynamics of binding. A limitation
of ITC is the high protein concentrations required to stay within
the experimental window 10 e c e 500. The formation of a
reversible covalent bond enhanced the affinity of our ligand
and increased our c value from 0.6 to 10, permitting accurate
measurements of the binding stoichiometry for 2. There are no
previous reports of ITC measurements of reversible covalent
bond formation. Further analysis using the technique for low
affinity ligands outlined by Turnbull and Daranas44 enabled us
to assign 1 binding equally to all six subunits of the protein.
Imine bond formation is ideally suited for the use in this
reversible, covalent modification process to enhance ligand
binding necessary for ITC stoichiometric measurements, but
other transient bonds such as disulfide exchange, hemiacetals,
and hemithioacetals could also be used. These studies allowed
us to define the parameters for a broader use of this technique.
First, the covalent modification of the ligand to the receptor
should not perturb the binding process since distortion of this
interaction may affect the ligand binding affinities. This criterion
requires that the functional groups on the ligand and the receptor
site be properly aligned to permit covalent adduction without
alteration of the binding site. Thus, we predict that the site of
covalent modification should be at a locus not involved in
receptor binding and that the appended site should not affect
binding of the ligand. Adherence to this criterion increases the
likelihood that the ligand analogue and the ligand bind to the
receptor in the same fashion. Second, the dynamics for reversible
ligand covalent modification must be complete within the time
interval between ITC injections allowing the system to reach
equilibrium after each incremental addition of ligand. Third,
the covalent modification site should not affect the binding of
other ligands or cofactors necessary for receptor function, which
can influence the binding of the ligand under study. In our case,
2 meets these criteria. In particular, we tethered the 3-formyl-
phenylsulfanyl component to the 5a-position of bicyclomycin.
The aryl aldehyde projects away from the cleft between the R7
and R8 helices toward the site of ATP hydrolysis.53 Structure-
activity relationship studies have shown that a wide range of
substituents can be attached at this site without loss of inhibitory
activity.34,45,52 Aiding in our choice of 2 was the site of covalent
amino acid attachment. Compound 2 reacts with the ꢀ-amino
group of K181. The four methylene units that separate the termi-
nal amine from the peptide backbone R-carbon in K181 provide
9
2748 J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005

Bicyclomycin-Rho Binding and Stoichiometry
A R T I C L E S
in vacuo, and the residue was purified by preparative TLC (10%
MeOH-CHCl3) to afford 8 as a mixture of diastereomers (≈3:1): yield,
80 mg (53%); Rf 0.44 (10% MeOH-CHCl3); FT-IR (KBr) 3304 (br),
1
2934 (br), 1690, 1395, 1045 cm-1; H NMR (CD3OD) for the major
diastereomer, δ 1.36 (s, 3 H, C(2′)CH3), 1.45 (s, 6 H, C(CH3)2), 1.88-
2.30 (m, 3 H, C(4)HH′, C(4)HH′, C(5)H), 2.53 (dd, J ) 11.6, 13.9 Hz,
1 H, C(5a)HH′), 3.65-3.88 (m, 3 H, C(3)HH′, C(3′)HH′, C(5a)HH′),
3.94-4.07 (m, 1 H, C(3)HH′), 4.09 (s, 1 H, C(1′)H), 4.45 (d, J ) 8.7
Hz, 1 H, C(3′)HH′), 4.56 (s, 2 H, C(3′′)CH2OH), 7.13-7.19 (m, 1 H,
Ar), 7.24-7.31 (m, 2 H, Ar), 7.34 (s, 1 H, C(2′′)H); 1H NMR
(CD3OD) for the minor diastereomer, δ 1.44 (s, 6 H, C(CH3)2), 2.77
(dd, J ) 11.6, 13.9 Hz, 1 H, C(5a)HH′), 4.46 (d, J ) 8.7 Hz, 1 H,
C(3′)HH′), 4.57 (s, 2 H, C(2′′)CH2OH), the remaining peaks overlapped
with nearby signals and were not resolved; the structural assignments
were in agreement with the 1H-1H COSY experiment; 13C NMR
(CD3OD) for the major diastereomer, 24.9, 26.9, 28.3, 30.4, 33.4, 51.9,
63.5, 64.9, 73.2, 73.4, 83.7, 86.3, 88.7, 111.7, 125.6, 128.2, 128.5, 130.1,
137.4, 143.8, 168.0, 171.6 ppm; 13C NMR (CD3OD) for the minor
diastereomer, 25.0, 28.4, 31.7, 63.7, 83.7, 86.4, 89.0, 111.7, 125.7,
128.7, 130.1, 143.9 ppm, the remaining peaks overlapped with nearby
signals and were not resolved; MS (+CI) 483 [M + 1]+; Mr (+CI)
483.181 01 [M + 1]+ (calcd for C22H31N2O8S 483.180 11).
5a-(3-Hydroxymethyl-phenylsulfanyl)-dihydrobicyclomycin (9).
To a 50% aqueous methanolic solution (10 mL) containing 8 was added
TFA (eight drops), and then the solution was stirred at room temperature
(20 h) until no starting material remained (TLC analysis). The solvent
was removed in vacuo, and the residue was purified by preparative
TLC (20% MeOH-CHCl3) to afford 9 (65 mg, 0.13 mmol) as a mixture
of diastereomers (≈3:1): yield, 45 mg (78%); Rf 0.32 (20% MeOH-
Figure 8. Example binding isotherm for the titration of 2 into a rho solution
(24 µM, based on monomer; 4 µM, based on hexamer) containing poly-
(dC) (1.33 µM) and ATP (200 µM). (A) Raw calorimetric trace. Each peak
corresponds to the thermal power evolved from the addition of 5.0 µL of
2 (400 µM) into 1.4 mL of rho solution. (B) Sigmoidal plot of normalized
titration isotherm. The solid line represents the best fitting curve calculated
from one type of binding site model (all sites are equal within the rho
hexamer).
CHCl3); FT-IR (KBr) 3300 (br), 2908 (br), 1688, 1405, 1042 cm-1
;
1H NMR (CD3OD) for the major diastereomer, δ 1.33 (s, 3 H, C(2′)-
CH3), 1.93-2.30 (m, 3 H, C(4)HH′, C(4)HH′, C(5)H), 2.58 (dd, J )
11.6, 13.9 Hz, 1 H, C(5a)HH′), 3.52 (d, J ) 11.3 Hz, 1 H, C(3′)HH′),
3.61-4.06 (m, 4 H, C(3)HH′, C(3)HH′, C(5a)HH′, C(3′)HH′), 4.04 (s,
1 H, C(1′)H), 4.56 (s, 2 H, C(3′′)CH2OH), 7.12-7.17 (m, 1 H, Ar),
7.23-7.29 (m, 2 H, Ar), 7.37 (s, 1 H, C(2′′)H); 1H NMR (CD3OD) for
the minor diastereomer, δ 1.34 (s, 3 H, C(2′)CH3), 2.75 (dd, J ) 11.6,
13.9 Hz, 1 H, C(5a)HH′), 3.55 (d, J ) 11.3 Hz, 1 H, C(3′)HH′), 4.06
(s, 1 H, C(1′)H), 4.57 (s, 2 H, C(3′′)CH2OH), the remaining peaks
overlapped with nearby signals and were not resolved; the structural
conformational flexibility and help ensure that the bicyclomy-
cin-rho complex is not distorted upon imine formation.
Experimental Procedures
1
assignments were in agreement with the H-1H COSY experiment;
General Methods. Low- and high-resolution (CI) mass spectral
studies were run at the University of Texas at Austin by Dr. M. Moini.
Thin-layer chromatographies were run on precoated silica gel slides
(20 × 20 cm; Sigma Z12272-6). Bicyclomycin was purified by three
successive silica gel chromatographies using 20% MeOH-CHCl3 as the
eluant prior to use in biological experiments. Rho protein was isolated
from E. coli AR 120 containing the overexpressing plasmid p39-ASE,56
which has the wild-type E155 residue seen in the original p39-AS
plasmid.57 Rho purity was assessed by SDS-PAGE, and protein
concentration was measured according to the bicinchoninic acid (BCA)
method.58 [γ-32P]ATP was purchased from Perkin-Elmer (Boston, MA),
and nucleotides were obtained from Sigma. Polyethyleneimine (PEI)
thin-layer chromatography (TLC) plates used for ATPase assays were
purchased from J. T. Baker, Inc. (Phillipsburg, NJ).
5a-(3-Hydroxymethyl-phenylsulfanyl)-dihydrobicyclomycin 2′,3′-
Acetonide (8). To a methanolic solution (10 mL) of 747 (105 mg, 0.32
mmol) under Ar was added 3-mercaptobenzyl alcohol (300 mg, 2.14
mmol), and then the pH was adjusted to 10.5 with aqueous 1.0 M
NaOH. The mixture was stirred at room temperature (8 h) until no
starting material remained (TLC analysis). The pH of the mixture was
adjusted to 7.0 with aqueous 0.1-1.0 M HCl. The solvent was removed
13C NMR (CD3OD) for the major diastereomer, 24.2, 30.1, 33.0, 51.8,
62.2, 64.9, 68.5, 72.2, 78.2, 83.7, 89.4, 125.6, 128.2, 128.6, 130.1, 137.4,
143.7, 168.6, 172.1 ppm; 13C NMR (CD3OD) for the minor diastere-
omer, 30.4, 32.0, 63.2, 72.3, 78.2, 83.6, 89.6, 125.6, 128.1, 130.1, 137.5,
143.8, 167.6, 174.0 ppm, the remaining peaks overlapped with nearby
signals and were not resolved; MS (+CI) 443 [M + 1]+; Mr (+CI)
443.147 17 [M + 1]+ (calcd for C19H27N2O8S 443.148 81).
5a-(3-Formyl-phenylsulfanyl)-dihydrobicyclomycin (2). To an
acetone solution containing 9 (9 mg, 0.02 mmol) was added MnO2 (50
mg, 0.58 mmol) in five portions (10 mg each) over 7 h. The mixture
was stirred at room temperature until no starting material remained
(TLC analysis). The mixture was filtered and concentrated in vacuo.
The residue was redissolved in MeOH and isolated by preparative TLC
(20% MeOH-CHCl3) to afford 2 as a mixture of diastereomers (≈3:
1): yield, 4 mg (45%); Rf 0.42 (20% MeOH-CHCl3); FT-IR (KBr)
3266 (br), 2935 (br), 1691, 1403, 1043 cm-1 1H NMR (CD3OD)
;
for the major diastereomer, δ 1.33 (s, 3 H, C(2′)CH3), 1.91-2.32 (m,
3 H, C(4)HH′, C(4)HH′, C(5)H), 2.65 (dd, J ) 11.6, 13.9 Hz, 1 H,
C(5a)HH′), 3.53 (d, J ) 11.3 Hz, 1 H, C(3′)HH′), 3.63-4.07 (m, 4 H,
C(3)HH′, C(3)HH′, C(5a)HH′, C(3′)HH′), 4.03 (s, 1 H, C(1′)H), 7.45-
7.55 (m, 1 H, Ar), 7.65-7.73 (m, 2 H, Ar), 7.88-7.93 (m, 1 H,
C(2′′)H), 9.94 (s, 1 H, C(3′′)C(H)O); 1H NMR (CD3OD) for the minor
diastereomer, δ 2.82 (dd, J ) 11.6, 13.9 Hz, 1 H, C(5a)HH′), 3.55 (d,
J ) 11.3 Hz, 1 H, C(3′)HH′), 4.07 (s, 1 H, C(1′)H), 9.95 (s, 1 H,
C(3′′)C(H)O), the remaining peaks overlapped with nearby signals and
(56) Nehrke, K. W.; Seifried, S. E.; Platt, T. Nucleic Acids Res. 1992, 20, 6107.
(57) Mott, J. E.; Grant, R. A.; Ho, Y. S.; Platt, T. Proc. Natl. Acad. Sci. U.S.A.
1985, 82, 88-92.
(58) Smith, P. K.; Krohn, R. I.; Hermanson, G. T.; Mallia, A. K.; Gartner, F.
H.; Provenzano, M. D.; Fujimoto, E. K.; Goeke, N. M.; Olson, B. J.; Klenk,
D. C. Anal. Biochem. 1985, 150, 76-85.
9
J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005 2749

A R T I C L E S
Brogan et al.
were not resolved; the structural assignments were in agreement with
the 1H-1H COSY experiment; 13C NMR (CD3OD) for the major
diastereomer, 24.2, 30.1, 32.8, 51.8, 62.1, 68.5, 72.3, 78.2, 83.7, 89.4,
The Biospin-6 columns were centrifuged at 4000 rpm (4 °C) for 5 min.
Aliquots (40 µL) from the centrifuged filtrates were diluted into buffer
(40 mM Tris HCl (pH 7.9), 50 mM KCl, and 12 mM MgCl2) containing
poly(C) (300 nM) and were assayed for poly(C)-dependent ATPase
activity using General Procedure 1.
127.7, 130.2, 130.8, 135.0, 138.7, 139.6, 168.7, 172.0, 193.8 ppm; 13
C
NMR (CD3OD) for the minor diastereomer, 30.6, 31.9, 53.5, 63.2, 83.6,
89.7, 127.8, 130.0, 130.9, 139.7 ppm, the remaining peaks overlapped
with nearby signals and were not resolved; MS (+CI) 441 [M + 1]+;
Mr (+CI) 441.134 59 [M + 1]+ (calcd for C19H25N2O8S 441.133 16).
General Procedure 1. Inhibition of Rho Poly(C)-Dependent
ATPase Activity.51 The ribonucleotide-stimulated ATPase activity of
rho at 32 °C was assayed by the amount of 32P-labeled inorganic
phosphate hydrolyzed from ATP after separation on PEI-TLC plates
(prerun with water and dried) using 0.75 M potassium phosphate (pH
3.5) as the mobile phase. Reactions (100 µL) were initiated by ATP
(250 µM) and 0.5 µCi [γ-32P]ATP being added to solutions containing
inhibitor (0-400 µM), poly(C) (300 nM), and rho (300 nM, based on
monomer) in buffer (40 mM Tris HCl (pH 7.9), 50 mM KCl, and 12
mM MgCl2). Solutions were preincubated for 2 h at 25 °C and 90 s at
32 °C prior to the addition of ATP. Five aliquots (0.7 µL) were removed
at 15 s intervals during the assay and spotted onto PEI-TLC plates.
The TLC plates were exposed to PhosphorImager plates (Fuji and
Molecular Dynamics) (15 h), scanned on a Storm 860 PC Phosphor-
Imager, and analyzed using Molecular Dynamic’s ImageQuant 5.0. The
initial rates of the reactions were determined by plotting the amount
of ATP hydrolyzed against time. Each assay was performed in duplicate,
and the results were averaged.
Preincubation Time for Rho-Imine Formation with 5a-(3-
Formyl-phenylsulfanyl)-dihydrobicyclomycin (2). Using General
Procedure 2 and 10 µM 2, solutions were allowed to preincubate for
various times (5, 30, 60, 120, 240 min) before NaBH4 was added. After
reductive amination, 2 µL of 10 mg/mL bovine serum albumin (BSA)
(0.1 mg/mL BSA, final concentration) was added to the reaction
mixture, and 100 µL aliquots were added to Biospin-6 columns
preequilibrated with buffer (40 mM Tris HCl (pH 7.9), 50 mM KCl,
and 12 mM MgCl2) containing 0.1 mg/mL BSA. The Biospin-6 columns
were centrifuged at 4000 rpm (4 °C) for 5 min. Aliquots (40 µL) from
the centrifuged filtrates were diluted into buffer (40 mM Tris HCl (pH
7.9), 50 mM KCl, and 12 mM MgCl2, total volume: 100 µL) containing
poly(C) (300 nM) and were assayed for poly(C)-dependent ATPase
activity using General Procedure 1.
5a-(3-Formyl-phenylsulfanyl)-dihydrobicyclomycin (2)-Rho Imi-
ne Adduct: Stability versus Time. A solution (200 µL) containing
buffer (40 mM Tris HCl (pH 7.9), 50 mM KCl, and 12 mM MgCl2),
rho (1 µM, based on monomer), ATP (1 mM), poly(C) (40 nM), and
2 (10 µM) was incubated at 25 °C (2 h). After incubation, 2 µL of 10
mg/mL bovine serum albumin (BSA) (0.1 mg/mL BSA, final concen-
tration) was added to the reaction mixture, and 100 µL aliquots were
added to Biospin-6 columns preequilibrated with buffer (40 mM Tris
HCl (pH 7.9), 50 mM KCl, and 12 mM MgCl2) containing 0.1 mg/mL
BSA. The Biospin-6 columns were centrifuged at 4000 rpm (4 °C) for
5 min. Aliquots (40 µL) from the centrifuged filtrates were di-
luted into buffer (40 mM Tris HCl (pH 7.9), 50 mM KCl, and 12 mM
MgCl2, total volume: 100 µL) containing poly(C) (300 nM), allowed
to incubate at various times (0, 15, 40, 60, 90, 120, 240 min), and
then assayed for poly(C)-dependent ATPase activity using General
Procedure 1.
Kinetics of Rho Poly(C)-Dependent ATPase Activity. ATPase
assays were carried out as described in General Procedure 1. Reactions
(100 µL) were initiated by various concentrations of ATP (3.75, 7.5,
15, 30, 60, and 120 µM) and 0.5 µCi [γ-32P] ATP being added to
solutions containing poly(C) (100 nM) and rho (100 nM, based on
monomer) in buffer (40 mM Tris HCl (pH 7.9), 50 mM KCl, and 12
mM MgCl2). Five aliquots (0.7 µL) were taken at various times (5-20
s) depending upon the ATP concentrations and spotted onto PEI-TLC
plates. Each series was repeated in the presence of 2 (0-20 µM).
Solutions were preincubated for 2 h at 25 °C and 90 s at 32 °C prior
to the addition of ATP. The initial rates for each ATP concentration
were plotted using the Enzyme Kinetics Module 1.1 for SigmaPlot 2001.
Each assay was performed in duplicate, and the results were averaged.
Assay Testing Bicyclomycin (1) and 5a-(3-Formyl-phenylsul-
fanyl)-dihydrobicyclomycin (2) Reversibility. A modified version of
General Procedure 1 was employed by the ATP hydrolysis being
measured at various concentrations of rho (125-1000 nM, based on
monomer) with poly(C) (500 nM) and either 1 (60 µM), 2 (5 µM), or
no inhibitor in buffer (40 mM Tris HCl (pH 7.9), 50 mM KCl, and 12
mM MgCl2). Reactions (100 µL) were initiated by ATP (250 µM) and
0.5 µCi [γ-32P]ATP being added, and five aliquots (0.7 µL) were
removed at 5-15 s intervals during the assay and spotted onto
PEI-TLC plates. Plots of the initial rates versus [rho, based on monomer]
were made to determine if the inhibitors were reversible. Each assay
was performed in duplicate, and the results were averaged.
Mass Spectrometry: Detection of 5a-(3-Formyl-phenylsulfanyl)-
dihydrobicyclomycin (2)-Modified Rho. The reductive amination of
rho (1 µM, based on monomer, 0.5-1 mL) by 2 (0-100 µM) was
conducted using General Procedure 2. After reductive amination, the
samples were dialyzed (4 °C, 20 h) against buffer (10 mM Tris HCl
(pH 7.9), 100 mM NaCl, 5% glycerol, 0.1 mM EDTA, and 0.1 mM
DTT). Aliquots (40 µL) from the dialysis were diluted into buffer (40
mM Tris HCl (pH 7.9), 50 mM KCl, and 12 mM MgCl2) containing
poly(C) (300 nM) and assayed for poly(C)-dependent ATPase activity
using General Procedure 1. The remainder of the samples was sent to
UMIST in dialysis buffer at room temperature. The samples were
desalted using ZipTipC18 pipet tips (Millipore, Billerica, MA), using
sequential elution with acetonitrile/water (20:80 by volume with 0.1%
formic acid) in the first instance and acetonitrile/water (80:20 by volume
with 0.1% formic acid) in the second elution. The protein fractions
were pooled together to give a final concentration of 1.5 µM protein
in acetonitrile/water (50:50 by volume with 0.1% formic acid). A
portion of the eluate was used directly for MS analysis. Electrospray
MS analyses were performed using a hybrid quadrupole time-of-flight
(Q-TOF) mass spectrometer (Micromass, Ltd., Manchester, UK). The
capillary potential was held at 3.5 kV, and the cone was set at 45-46
V. Samples were introduced via a syringe driver (Harvard Apparatus,
South Natick, MA) at a constant flow rate of 0.5 µL/min. The m/z
range of 400-1600 was recorded. Spectra were accumulated for 7 min
under the control of Masslynx software (Micromass) and were processed
using maximum entropy.
General Procedure 2. Reductive Amination of Rho.24,31-33
A
solution (200 µL) containing buffer (40 mM Tris HCl (pH 7.9), 50
mM KCl, and 12 mM MgCl2), rho (1 µM, based on monomer), ATP
(1 mM), poly(C) (40 nM), and 2 (0-100 µM) was incubated at 25 °C
(2-4 h). An aqueous solution (20 µL) of NaBH4 (600 mM, freshly
made) was added, and the reaction was allowed to incubate at 25 °C
(20 min).
Reductive Amination of Rho using Various Concentrations of
5a-(3-Formyl-phenylsulfanyl)-dihydrobicyclomycin (2). General Pro-
cedure 2 for the reductive amination of rho was employed using 2
(0-40 µM). After reductive amination, 2 µL of 10 mg/mL bovine serum
albumin (BSA) (0.1 mg/mL BSA, final concentration) was added to
the reaction mixture, and 100 µL aliquots were added to Biospin-6
columns (Bio-Rad, Inc.) preequilibrated with buffer (40 mM Tris HCl
(pH 7.9), 50 mM KCl, and 12 mM MgCl2) containing 0.1 mg/mL BSA.
Mass Spectrometry: Full Tryptic Digestion of the 5a-(3-Formyl-
phenylsulfanyl)-dihydrobicyclomycin (2)-Modified Rho. The reduc-
tive amination of rho (3 µM, based on monomer, 0.5-1 mL) in the
presence of poly(C) (120 nM) by 2 (0, 50 µM) was conducted using
General Procedure 2. After reductive amination, the samples were
9
2750 J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005

Bicyclomycin-Rho Binding and Stoichiometry
A R T I C L E S
dialyzed (4 °C, 20 h) against buffer (10 mM Tris HCl (pH 7.9), 100
mM NaCl, 5% glycerol, 0.1 mM EDTA, and 0.1 mM DTT). Aliquots
(40 µL) from the dialysis were diluted into buffer (40 mM Tris HCl
(pH 7.9), 50 mM KCl, and 12 mM MgCl2) containing poly(C) (300
nM) and assayed for poly(C)-dependent ATPase activity using General
Procedure 1. The remainder of the samples was sent to UMIST in
dialysis buffer at room temperature. A portion of the dialysate was
treated with a solution of trypsin in 2 mM CaCl2 to give an enzyme/
substrate ratio of 1:50. The solution was incubated at 37 °C for 18 h,
after which the digestion was terminated by addition of formic acid to
a final concentration of 5 vol %. The peptide products of the digestion
were desalted using ZipTipC18 pipet tips (Millipore, Billerica, MA),
using sequential elution with acetonitrile/water (20:80 by volume with
0.1% formic acid) followed by acetonitrile/water (80:20 by volume
with 0.1% formic acid). The two eluates were pooled together, and a
portion was used for MALDI-MS analysis performed on a Voyager
DE-STR (Applied Biosystems, Framingham). A portion (10 µL) of the
eluate was analyzed by ESI-MS and tandem MS using a Micromass
Q-TOF instrument equipped with an electrospray source and nanospray
source, respectively. For MS analysis, the sample was introduced using
a syringe driver (Harvard Apparatus) set to deliver a flow rate of 0.5
µL/min. For tandem MS, a 3 µL portion of the desalted digest was
introduced into a gold-coated borosilicate glass capillary with a 1 µm
tapered tip (Micromass). To obtain the spectrum shown in Figure 7, a
doubly charged ion corresponding to an m/z value of 737.80 was
selected as the precursor ion and subjected to collisional activation,
using a laboratory collision energy of 48-58 eV and an indicated
(manifold) pressure of argon collision gas of 5 × 10-5 mbar. Fragment
ions are labeled according to the nomenclature of Biemann.59 C-terminal
(y-type) fragment ions, which retain the modified lysine residue, are
accompanied by satellite ions arising from partial loss of the moiety
derived from 2. The product ion spectrum of the triply charged precursor
ion ([M + 3H]3+) analogue (m/z ) 492.23) was also recorded and was
entirely consistent with the proposed structure.
of the test compound (mg/mL), yielded linear slopes to provide the
minimal inhibitory concentrations (MIC) for bicyclomycin and bicy-
clomycin derivatives. Each assay was performed in duplicate, and the
results were averaged.
Isothermal Titration Calorimetry. Purified rho was dialyzed two
times against 500 mL of buffer (40 mM Tris HCl (pH 7.9), 400 mM
KCl, 0.1 mM EDTA, and 0.1 mM DTT) at 4 °C. After dialysis, the
protein concentration was determined by the BCA method. The rho
solution from dialysis was diluted (24 µM, based on monomer) into
the same buffer solution that contained poly(dC) (1.3 µM) or poly(C)
(4 µM) and either ATP (200 µM) or without nucleotide. The
concentration of the ligand in the syringe was 400 µM for 2 or 1
mM for 1 in the identical buffer from the dialysis to minimize the heats
of dilution. ITC experiments were performed using the high precision
VP-ITC (MicroCal, Inc.) at 26 °C. Prior to titration, all the solutions
were degassed. The titration schedule consisted of 50-60 consecutive
injections of 5 µL with a 400 or 1200 s interval between injections.
The values for the last five injections were averaged and subtracted
from each point of the experimental trace isotherm to correct for the
heats of dilution and nonspecific binding of the bicyclomycin to rho.
Curve fitting was undertaken in Origin 5.0 using the standard
noninteracting one or two site model supplied by MicroCal using the
hexamer concentration of rho (4 µM). In experiments where c < 10,
the stoichiometry parameter was fixed to six, or for the two site model,
two integers with a sum of six. The values reported in Table 1 are the
average of two experiments ( two times the standard deviation.
Acknowledgment. We thank Dr. Y. Itoh and the Fujisawa
Pharmaceutical Co., Ltd., Japan, for the gift of 1, Dr. T. Platt
(University of Rochester) for the overproducing strain of rho,
Dr. A. Tripathy of UNC’s Macromolecular Interactions Facility
for assistance with the ITC experiments, and Drs. Verna Frasca
and John F. Brandts (MicroCal, LLC) for review of our ITC
experiments. The American Chemical Society Division of
Medicinal Chemistry and Bristol-Myers Squibb are gratefully
acknowledged for support of a predoctoral fellowship to A.P.B.
This investigation was supported by NIH Grant GM37934 and
the Robert A. Welch Foundation Grant E1381 (W.R.W.).
Inhibitory Properties of Bicyclomycin (1) and 5a-(3-Formyl-
phenylsulfanyl)-dihydrobicyclomycin (2) in the Antimicrobial As-
say.54 Aliquots (200 µL) from suspensions of overnight LB broth
cultures (E. coli W3350) were diluted into LB broth (2 mL). The
suspension was poured onto 15 mL volume LB agar plates. The solution
was gently rocked to distribute the cells evenly over the surface of the
plates, and any excess cell solution was removed by pipet. The plates
were incubated at 37 °C (30 min), and an antibiotic-assay disk (0.25
in. diameter) containing 20 µL of the test compound (1, 2, 4, 8, 16, 32
mg/mL in 50-100% DMSO) was placed on the agar surface. The plates
were incubated at 37 °C (18 h). Data plots of the zone of inhibited
bacterial growth (cm3) versus log(1000C), where C is the concentration
Supporting Information Available: Binding isotherm for the
titration of 2 into a rho solution containing poly(dC). This
material is available free of charge via the Internet at
http://pubs.acs.org.
JA046441Q
(59) Biemann, K. Biomed. EnViron. Mass Spectrom. 1988, 16, 99-111.
9
J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005 2751