ppGpp analogues inhibit synthetase activity of Rel proteins from Gram-negative and Gram-positive bacteria

Article abstract of DOI:10.1016/j.bmc.2010.04.064

A prominent feature of the stringent response is the accumulation of two unusual phosphorylated derivatives of GTP and GDP (pppGpp: 5′-triphosphate-3′-diphosphate, and ppGpp: 5′-3′-bis-diphosphate), collectively called (p)ppGpp, within a few seconds after the onset of amino-acid starvation. The synthesis of these 'alarmone' compounds is catalyzed by RelA homologues. Other features of the stringent response include inhibition of stable RNA synthesis and modulation of transcription, replication, and translation. (p)ppGpp accumulation is important for virulence induction, differentiation and antibiotic resistance. We have synthesized a group of (p)ppGpp analogues and tested them as competitive inhibitors of Rel proteins in vitro. 2′-Deoxyguanosine-3′-5′-di(methylene bisphosphonate) [compound (10)] was found as an inhibitor that reduces ppGpp formation in both Gram-negative and Gram-positive bacteria. In silico docking together with competitive inhibition analysis suggests that compound (10) inhibits activity of Rel proteins by competing with GTP/GDP for its binding site. As Rel proteins are completely absent in mammalians, this appears to be a very attractive approach for the development of novel antibacterial agents.

Full text of DOI:10.1016/j.bmc.2010.04.064

Bioorganic & Medicinal Chemistry 18 (2010) 4485–4497
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
ppGpp analogues inhibit synthetase activity of Rel proteins from
Gram-negative and Gram-positive bacteria
b
b
Ezequiel Wexselblatt a, Jehoshua Katzhendler a, , Raspudin Saleem-Batcha , Guido Hansen ,
*
Rolf Hilgenfeld b, Gad Glaser c, Roee R. Vidavski c
a The Institute for Drug Research, School of Pharmacy, The Hebrew University of Jerusalem, Room 414, Jerusalem 91120, Israel
b Institute of Biochemistry, Center for Structural and Cell Biology in Medicine, University of Lübeck, Lübeck, Germany
c Institute for Medical Research Israel-Canada, Department of Developmental Biology and Cancer Research, Hebrew University of Jerusalem, Jerusalem, Israel
a r t i c l e i n f o
a b s t r a c t
Article history:
A prominent feature of the stringent response is the accumulation of two unusual phosphorylated deriv-
atives of GTP and GDP (pppGpp: 50-triphosphate-30-diphosphate, and ppGpp: 50-30-bis-diphosphate), col-
lectively called (p)ppGpp, within a few seconds after the onset of amino-acid starvation. The synthesis of
these ‘alarmone’ compounds is catalyzed by RelA homologues. Other features of the stringent response
include inhibition of stable RNA synthesis and modulation of transcription, replication, and translation.
(p)ppGpp accumulation is important for virulence induction, differentiation and antibiotic resistance.
We have synthesized a group of (p)ppGpp analogues and tested them as competitive inhibitors of Rel
proteins in vitro. 20-Deoxyguanosine-30-50-di(methylene bisphosphonate) [compound (10)] was found
as an inhibitor that reduces ppGpp formation in both Gram-negative and Gram-positive bacteria. In silico
docking together with competitive inhibition analysis suggests that compound (10) inhibits activity of
Rel proteins by competing with GTP/GDP for its binding site.
Received 14 January 2010
Revised 20 April 2010
Accepted 21 April 2010
Available online 28 April 2010
Keywords:
ppGpp
RelA
Stringent response
Antibacterial
As Rel proteins are completely absent in mammalians, this appears to be a very attractive approach for
the development of novel antibacterial agents.
Ó 2010 Elsevier Ltd. All rights reserved.
1. Introduction
essential processes, namely, transcription, replication and transla-
tion.3,6 In addition, (p)ppGpp accumulation was also found to be
Theneedfornewantibioticdrugswithnoveltargetshasbeenrec-
ognized by health organizations, industry and academia alike.1 At
present, most of the drugs administered to treat bacterial infections
are aimed to one of the following four processes: protein synthesis,
nucleic-acid synthesis, cell-wall synthesis, or folate synthesis.2
Duetotheextensiveknowledgeaccumulatedinthefield, substantial
efforts have been implemented to interfere with these essential syn-
thetic pathways. Careful examination of other processes in bacteria
revealed that targeting the stringent response has potential to lead
to the development of a novel class of antibacterial agents.
important for virulence induction,7–12 differentiation,13 and devel-
opment of antibiotic resistance.14
In E. coli, the mutation causing the relaxed phenotype, which
fails to accumulate (p)ppGpp during amino-acid starvation, was
mapped to the relA gene which encodes a 84-kDa protein, RelA.15
RelA is a ribosome-associated (p)ppGpp synthetase that is activated
in response to amino-acid starvation.3 During amino-acid starva-
tion, the binding of uncharged tRNAs to the ribosomal ‘A’ site stalls
protein synthesis, stimulating a reaction in which RelA transfers a
pyrophosphoryl group comprising the b- and c-phosphates of an
The pleiotropic response, called the ‘stringent response’, was
initially described for Escherichia coli cells more than 40 years
ago.3 The first observed feature of the stringent response was the
intracellular accumulation of guanosine tetraphosphate (ppGpp)
and guanosine pentaphosphate (pppGpp) (collectively known as
(p)ppGpp or ‘alarmones’) within a few seconds after the advent
of amino-acid starvation.4,5 Other features of the stringent re-
sponse include inhibition of stable RNA synthesis and changes in
ATP donor to the 30-hydroxy group of GTP or GDP to form pppGpp
and ppGpp, respectively.3 In cell extracts, RelA is found to be asso-
ciated only with a small fraction of the ribosomes (about 1%).16 De-
spite this low abundance of RelA, up to mM levels of (p)ppGpp are
rapidly produced. This is feasible since (p)ppGpp synthesis evokes
dissociation of RelA from the ribosome, allowing the enzyme to
shuttle to another stalled ribosome and repeat the reaction.17
In addition to amino-acid starvation, (p)ppGpp synthesis in
Gram-negative bacteria can be induced by other stress conditions
such as deprivation of phosphorus, iron, carbon source, and fatty
acids in a SpoT-dependent mode.3,18–20 Equally important, SpoT
* Corresponding author. Tel.: +972 2 6758693; fax: +972 2 6757076.
E-mail address: katzhe@cc.huji.ac.il (J. Katzhendler).
0968-0896/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2010.04.064

4486
E. Wexselblatt et al. / Bioorg. Med. Chem. 18 (2010) 4485–4497
Figure 1. ppGpp and its pcpGpcp analogues 5a: R1 = OH, R3 = ibu; 5b: R2 = OH, R3 = ibu; 5c: R1 = OTBDMS, R3 = ibu; 5d: R2 = OTBDMS, R3 = ibu; 6a: R1 = OH, R3 = H 6b: R2 = OH,
R3 = H; 10: R2,R3 = H.
also acts as a ribosome-independent (p)ppGpp hydrolase that de-
grades (p)ppGpp to GDP/GTP and pyrophosphate.21 Residual
the pyrophosphate donor, ATP.27 Support for a critical role of
Glu323 in the synthetase mechanism derives from a follow-up
mutational study: a Relseq385 mutant bearing a Glu323Gln substi-
tution exhibited severely defective synthetase activity.27
In this paper, we report the synthesis of a variety of (p)ppGpp
analogues and the study of their inhibitory properties on RelA
and its homologue, Relseq385. The compounds include two main
features: (1) the pyrophosphate moieties at positions 50 and 30
are replaced by non-hydrolyzable methylene-bisphosphonate
groups and (2) the sugar moiety consists of either ribose or 20-
(p)ppGpp synthesis found in
DrelA is abolished in a DrelADspoT
(‘double null’) mutant.22 Cells with this double deletion show a
complex phenotype that includes loss of ability to grow on ami-
no-acid-free minimal medium and morphological alterations.22 It
was also verified that Gram-positive bacteria, which lack SpoT
homologues, have a RelA homologue with both (p)ppGpp-synthe-
tase and (p)ppGpp-hydrolase activities.23–25 Deletion of the relA-
spoT integrated gene in the Gram-positive Bacillus subtilis was
thought to generate a phenotype resembling that of ‘double null’
in E. coli.25 However, recently two additional factors (YjbM and
YwaC) were found to have (p)ppGpp-synthetase activity in B. sub-
tilis.26 Molecular genetic analysis revealed that ywaC is induced by
alkaline shock and it has been suggested that both YjbM and YwaC
might have more specialized functions in (p)ppGpp metabolism.
Recently, the crystal structure of the catalytic N-terminal frag-
ment (NTD, residues 1–385, termed Relseq385) of the bifunctional
RelA homologue from the Gram-positive bacterium Streptococcus
equisimilis was determined at 2.1 Å resolution.27 In contrast to
Gram-negative bacteria, where RelA and SpoT are distinct proteins,
this fragment displays both the hydrolase and synthetase active
sites residing more than 30 Å apart from each other. Futile recy-
cling of the substrate is avoided by a complex regulatory mecha-
nism that involves subtle conformational changes in the protein.
The crystallographic analysis revealed two conformations for Re-
lseq385 that, respectively, typify the opposing hydrolase-OFF/syn-
thetase-ON and hydrolase-ON/synthetase-OFF states.27
deoxyribose. Analogues carrying
a methylene-bisphosphonate
group at the 20 position are also reported. Figure 1 depicts the alar-
mone ppGpp and the schematic structure of the inhibitors bearing
double methylene-bisphosphonate substitutions.
2. Results and discussion
2.1. Chemical synthesis
The synthesis of (5a), (5b), (6a) and (6b) is outlined in Scheme 1.
We have selected the isobutyryl group for the protection of the
exocyclic amine. Treatment of guanosine with trimethylsilyl chlo-
ride in pyridine as a transient protection of the ribose hydroxyl
groups followed by the addition of isobutyric anhydride afforded
N2 ibu nucleoside (1) in 79% yield. Compound (1) was re-crystal-
lized from hot water.
50-O-Dimethoxytritylation of (1) gave (2) in 73% yield. Com-
pound (2) precipitates from a mixture of diethyl ether and petro-
leum ether (3:1).
Subsequent silylation of the latter with TBDMS (tert-butyl di-
methyl silyl)28 resulted in a mixture of positional isomers (3a)
and (3b) that were separated by preparative HPLC on a C18 column
The crystal structure also suggested a mechanism for the attack
of the 30-OH group of GDP (or GTP) onto the b-phosphorus atom of
ATP. Most likely, Glu323 of Relseq385 activates the hydroxyl group
by proton abstraction; it will also coordinate the essential Mg2+ ion
which is not present in the crystals and is expected to be bound to

E. Wexselblatt et al. / Bioorg. Med. Chem. 18 (2010) 4485–4497
4487
Scheme 1. Reagents and conditions: (i) isobutyric anhydride, pyridine, room temperature, 5 h (79%); (ii) DMT–Cl pyridine, room temperature, 4 h (73%); (iii) TBDMS–Cl,
DCM, room temperature, overnight (34–63%); (iv) 5% DCA/DCM, room temperature, 15 min (95%); (v) (a) methylene bis-(phosphonic dichloride), trimethyl phosphate, 0 °C to
room temperature, overnight; (b) 0.5 M ammonium bicarbonate, 0 °C, 15 min (44–73%); (vi) ammonium hydroxide, 60 °C, overnight (97%).
employing isocratic elution (70% acetonitrile/water). The products
were identified by 2D NMR.
After detritylation of isomers (3a) and (3b) with 5% dichloroace-
tic acid (DCA) in methylene chloride (DCM) followed by flash-chro-
matography purification, the pure products (4a) and (4b) were
then bisphosphonylated (step v) employing methylene-bis-(phos-
phonic dichloride) in trimethyl phosphate. Upon completion of
the phosphonylation, the solution was treated with a concentrated
solution of ammonium bicarbonate to afford the di-bisphosphony-
lated products (5a) and (5b). Compound (5b) precipitated and was
isolated by filtration while (5a) remained completely soluble in the
reaction mixture. The TBDMS protecting group is sensitive to

4488
E. Wexselblatt et al. / Bioorg. Med. Chem. 18 (2010) 4485–4497
Scheme 2. Reagents and conditions: (i) (a) mMethylene bis-(phosphonic dichloride), trimethyl phosphate, 0 °C to room temperature, 2 h; (b) 0.5 M ammonium bicarbonate,
0 °C, 15 min (67–70%); (ii) acetone, room temperature, 70 min (83%).
Scheme 3. Reagents and conditions: (i) (a) methylene bis-(phosphonic dichloride), trimethyl phosphate, 0 °C to room temperature, overnight; (b) 0.5 M ammonium
bicarbonate, 0 °C, 15 min (73%).
highly acidic conditions and was almost completely removed dur-
ing the phosphonylation step; a reaction that involves the genera-
tion of acidic conditions.
In addition to the 30-50-bis-methylene-bisphosphonate adducts
(6a) and (6b) and to the 50-bisphosphonate 7, the 30-methylene-
bisphosphonate derivative was also prepared. The synthetic route
for the preparation of deoxyguanosine-30-yl-bisphosphonate 14 is
shown in Scheme 4. In this synthesis, an appropriate protecting
group is required for the 50-OH group. Under the reaction condi-
tions employing trimethyl phosphate as the solvent of choice,
HCl evolves. Thus, acid-labile DMT and TBDMS are not stable pro-
tecting groups under these reaction conditions. An attempt to add
base such as triethylamine or pyridine to the reaction mixture did
not lead to the desired product, nor did the use of pyridine as sol-
vent. This is probably due to the role played by trimethylphosphate
as described by Yoshikawa et al.30 Acetyl as protecting group was
selected and found to be adequate. 50-Acetyl-20-deoxyguanosine
was obtained by acylation of 20-deoxyguanosine with acetic anhy-
dride in dry pyridine. The di-acetylated product was treated with
hydrazine hydrate to yield a partial deacetylated mixture.31 The
mixture of 50-acetyl- and 30-acetyl-20-deoxyguanosine was sepa-
rated on RP-C18 preparative HPLC and the isolated compounds
were identified by 2D 1H NMR. Figure 2 displays a typical HPLC
chromatogram highlighting the identification of 50-acetyl-20-deox-
yguanosine. Phosphonylation was carried out as described (vide
supra) and the final step of acetyl deprotection was achieved by
The products were purified on a reverse-phase preparative
HPLC system. Peaks were separated and identified by mass spec-
troscopy. Small amounts of compounds (5c) and (5d) and traces
of the mono-phosphonylated compounds were also detected.
Removal of the isobutyryl protecting group on the exocyclic
amine at position 2 of the nucleobase of (5a) and (5b) was achieved
by treating these compounds with concentrated ammonium
hydroxide at 60 °C in sealed tubes overnight affording compounds
(6a) and (6b), respectively.
The overall yields (six steps) for (6a) and (6b) were 8.5% and
13.6%, respectively.
Compounds (7) and (9) were synthesized by direct phosphonyla-
tion with methylene-bis-(phosphonic dichloride) (Scheme 2) start-
ing with guanosine and isopropylidene guanosine, respectively.29
The synthesis of 10 followed the same procedure as for (5a) and
(5b) and is depicted in Scheme 3. Thus, the desired product precip-
itated after the addition of ammonium bicarbonate to form the
phosphonic acid. The precipitate was collected and purified in
the same manner as described for compounds (5a) and (5b) afford-
ing compound (10) in 73% yield.

E. Wexselblatt et al. / Bioorg. Med. Chem. 18 (2010) 4485–4497
4489
Scheme 4. Synthesis of 30-bisphosphonyl-20-deoxyguanosine. Reagents and conditions: (i) acetic anhydride, pyridine, room temperature, 45 min (90%); (ii) H2NNH2ꢀH2O,
pyridine, overnight (47%); (iii) (a) methylene bis-(phosphonic dichloride), trimethyl phosphate, 0 °C to room temperature, overnight; (b) 0.5 M ammonium bicarbonate, 0 °C,
15 min (67%); (iv) ammonium hydroxide, 60 °C, 15 h (91%).
Figure 2. (A) RP-C18 HPLC separation of compound (12) using a linear gradient from 0% to 90% acetonitrile in water over 50 min with a flow rate of 5 mL/min. (B) 2D NMR of
50-acetyl-20-deoxyguanosine. The triplet signal at 6.18 ppm was identified as 10H and used as reference for the assignment of the remaining hydrogen atoms of the sugar ring.
treating compound (13) with ammonium hydroxide yielding the
desired product.
tional N-terminal domain of the S. equisimilis RelA homologue
harboring (p)ppGpp-synthetase and hydrolase activities in sepa-
rate catalytic sites; this recombinant protein construct lacks the
C-terminal ribosome-binding domain.
2.2. ppGpp analogues possess the ability to inhibit the synthetic
activity of both E. coli RelA and Relseq385 in vitro
All of the synthesized compounds possess the ability of
inhibiting (p)ppGpp accumulation. Furthermore, a dose-response
correlation was observed. In general, analogues bearing the bis-
phosphonate substitution at positions 50 and 30 (compounds (5d),
(5b), (6b), and (10)) show more potent inhibition than others. It
can be seen (Fig. 3A) that for these four inhibitors, substitution at
the 20 position plays a role in the inhibitory characteristic of all
The biological activity of the synthesized ppGpp analogues was
evaluated in vitro (Fig. 3) with E. coli RelA and Relseq385. ‘RelA’ de-
notes the full-length RelA protein from E. coli, comprising the N-
terminal domain with (p)ppGpp-synthetase activity and the C-ter-
minal ribosome-binding domain. ‘Relseq385’ denotes the bifunc-

4490
E. Wexselblatt et al. / Bioorg. Med. Chem. 18 (2010) 4485–4497
Figure 4. Inhibition of Rel-C638F by compound (10). Lanes 1–2: activity of Rel-
C638F without compound (10); lanes 3–8: activity of Rel-C638F with increasing
concentrations of compound (10); lanes 3–4: 1 mM; lanes 5–6: 5 mM and lanes 7–
8: 10 mM.
The mutant E. coli strain relC, carrying a mutation in the struc-
tural gene for the ribosomal protein L11 (rplK), presents a relaxed
phenotype under amino-acid starvation conditions.33 RelA dissoci-
ation from ribosomes of relC was also examined (Fig. 5). It was evi-
dent that the addition of compound (10) did not affect the
dissociation of RelA from L11 mutated ribosomes.
These data reinforce the notion that (p)ppGpp plays an impor-
tant role in the dissociation of RelA from the ribosome.
Figure 3. Rel protein inhibition by ppGpp analogues (shown in Schemes 1–3).
Gray: 10 mM, white: 5 mM, black: 1 mM. (A) RelA. (B) Relseq385.
In order to assess the inhibitory effect of compound (10) in
greater detail, its capability to inhibit the synthetic activity of Re-
lseq385 and RelA was tested using a wider range of concentra-
tions (0.1–10 mM). It was found that compound (10) inhibits
(p)ppGpp accumulation by Relseq385 even at concentrations as
low as 0.2 mM (Fig. 6A). The compound exerted more than 50%
inhibition of Relseq385 at 0.5 mM, while more than 1 mM was
needed in order to reach a similar level of RelA inhibition
(Fig. 6B). Lowering the concentrations of compound (10) to sub-
millimolar concentrations did not affect the in vitro accumulation
of (p)ppGpp by RelA.
compounds. Among compounds (5d), (5b), and (6b), the less bulky
the substituent is, the more potent the inhibitory activity. When
the 20-hydroxyl is totally removed and replaced by a hydrogen
atom (compound (10)), the highest levels of inhibition are ob-
served. Also, substitution at the N2 nitrogen of the guanine base re-
duces the inhibitory ability. On the other hand, compounds bearing
only one bisphosphonate moiety at position 50 (compounds (7) and
(9)) or at position 30 (compound (14)) show relatively moderate
levels of inhibition. The same tendency is seen for the inhibition
of Relseq385 by the selected analogues (Fig. 3B).
2.3. Compound (10) competes with GTP for its binding site
Even though compound (6b) shows stronger activity at higher
concentrations (5 and 10 mM), compound (10) was selected for fur-
ther investigation as it was more potent at lower concentrations.
Since the ribosome-independent Relseq385 was strongly inhib-
ited by compound (10), we wanted to examine whether a ribo-
some-independent mutant of RelA, namely Rel-C638F,32 could
also be inhibited. As shown in Figure 4, compound (10) indeed
inhibited the synthetic activity of Rel-C638F in vitro.
Attempting to explain the rapid accumulation of high levels of
ppGpp during amino-acid starvation, Wendrich et al.17 proposed
a model suggesting that RelA is able to ‘hop’ between ribosomes
as a consequence of (p)ppGpp synthesis.
Based on this model, we hypothesized that inhibition of
(p)ppGpp synthesis would slow down RelA dissociation from the
ribosome. In order to provide evidence of such an hypothesis, the
amount of RelA proteins bound to the ribosome was quantified fol-
lowing the addition of increasing concentrations of compound (10)
(Fig. 5). As clearly seen, lowering the levels of (p)ppGpp produced
by RelA resulted in a remarkable decrease in RelA0s ability to disso-
ciate from the ribosome. The amount of bound protein (RelA) was
increased up to threefold. Another protein tested was Rel-G251E, a
RelA variant unable to synthesize (p)ppGpp.32 In this case, the
addition of increasing concentrations of compound (10) did not af-
fect the dissociation of Rel-G251E from the ribosome.
Based on the fact that compound (10) is able to inhibit both
ribosome-dependent and independent Rel enzymes, it was as-
sumed that the inhibitory mechanism through which the
accumulation of (p)ppGpp is affected involves a direct inhibi-
tor–enzyme interaction. As compound (10) inhibited both mono-
functional RelA and bifunctional Relseq385, the interaction was
suspected to take place at the catalytic synthetase site of the
two enzymes. To support this idea, the influence of the natural
substrates of Rel proteins (GTP and ATP) on the inhibitory effect
of compound (10) was tested. In vitro accumulation of (p)ppGpp
via both ribosome-dependent (RelA) and ribosome-independent
(Relseq385) enzymes was measured in the presence of a constant
concentration of compound (10) with increasing concentrations
of either GTP or ATP. Figure 7A shows the influence of GTP on
the inhibiton of RelA by compound (10). A distinct reversed linear
correlation between the inhibitory activity and the GTP concen-
tration is observed (R2 = 0.909). A similar tendency can be seen
for Relseq385, although linearity is less obvious (R2 = 0.708)
(Fig. 7B).
No such correlation was found while examining the influence of
compound (10)’s activity with increasing levels of ATP (data not
shown). This is in agreement with the crystal structure of Re-
lseq385, where GTP is found in the catalytic site while ATP is not.27

E. Wexselblatt et al. / Bioorg. Med. Chem. 18 (2010) 4485–4497
4491
Figure 5. Compound (10) effect on the dissociation of RelA from the ribosome. Increasing amounts of compound (10) were added to in vitro reactions containing ribosomes
and RelA (A) or Rel-G251E (B). (C) represents reactions where ribosomes produced from the mutant E. coli strain relC were used together with RelA.
Figure 7. Rel protein inhibition by 5 mM of compound (10) in the presence of
increasing concentrations of GTP. (A) RelA; (B) Relseq385.
2.4. Prediction of the binding mode of compound (10)
Using the crystal structure of Relseq385,27 in silico docking of
compound (10) into the catalytic synthetase site of the enzyme
was initially carried out. Detailed analysis of key interactions of
compound (10) and amino-acid residues of the (p)ppGpp-synthe-
tase site resulted in the identification of a putative binding mode
Figure 6. Inhibition by compound (10). (A) Relseq385; (B) RelA.
of compound (10) (Fig. 8). Interactions of GDP and Relseq385 iden-
tified by X-ray crystallography27 were used as constraints for the
These results strengthen the hypothesis that the inhibitor inter-
acts with the enzyme at the catalytic synthetase site. Moreover, a
direct competition between the inhibitor and GTP for the same
binding site is likely. However, we have to keep in mind that the
bifunctional Relseq385 has two active sites that can accommodate
(p)ppGpp and potentially its analogues: (1) the synthetase site
where (p)ppGpp is the product of the reaction between GDP
(GTP) and ATP and (2) the hydrolase site, where (p)ppGpp is the
substrate, yielding GDP (GTP) and pyrophosphate. To examine
the possible binding modes, we employed automated docking of
compound (10) to both of these sites, each in its ON and OFF
conformations.27
docking of compound (10). Based on experimental results, it is very
likely that the guanine moiety of compound (10) binds to Re-
lseq385 in a similar way as observed for GDP. Therefore, the hydro-
gen bond between the backbone carbonyl of Ala335 of Relseq385
and N1 of compound (10) as well as the one between the side-
chain amide of Asn306 of Relseq385 and O6 of compound (10)
were used as constraints for docking. Compound (10) was docked
into the synthetase active site of Relseq385 in both the synthetase-
ON and synthetase-OFF conformations as observed in the crystal
structure.27
Interestingly, docking solutions obtained for the synthetase-ON
site show that the 50-bisphosphonate of compound (10) likely

4492
E. Wexselblatt et al. / Bioorg. Med. Chem. 18 (2010) 4485–4497
Figure 8. Predicted binding mode of compound (10) in the synthetase active site of Relseq385 (synthetase-ON conformation) by automatic docking. (A) 2D-representation of
the interactions between compound (10) and Relseq385. Hydrogen bonds shown as dotted lines, van-der-Waals interactions as half-circles. (B) Surface representation of the
synthetase active site (brown) with compound (10) shown as sticks with atoms colored as follows: carbon—gray, oxygen—red, nitrogen—blue, phosphorus—orange.
nate moiety. This is reflected in low docking scores, showing a
reduction by more than 50% compared to solutions for the synthe-
tase site in the ON conformation. Solutions obtained show unfavor-
able van-der-Waals clashes between compound (10) and the more
restricted synthetase site in the OFF conformation.
The crystal structure of Relseq385 suggests that intramolecular
crosstalk between synthetase and hydrolase domains regulate the
activity of the two opposing catalytic sites. To avoid futile cycling
of synthesis and hydrolysis of (p)ppGpp, nucleotide binding in
the synthetase-ON state leads to conformational changes that
switch off the hydrolase activity and vice versa. The inhibitory ef-
fect of non-reactive ATP analogues (ApNHpp or ApCH2pp) on both
synthetase and hydrolase activity of Relseq385 provided further
evidence for this reciprocal regulation mechanism.27 It is therefore
possible that the observed inhibitory effect of compound (10) on
Figure 9. Overview of predicted binding sites for compound (10) in the synthetase-
(p)ppGpp synthesis is the result of a binding event in the hydrolase
ON active site of Relseq385, based on the top-ten solutions obtained from
site of Relseq385, leading to allosteric signaling and subsequent
automated docking. Surface representation of the synthetase active site (brown),
inactivation of the synthetase site. Indeed, docking studies re-
vealed that compound (10) is able to bind to the hydrolase site
of Relseq385 (as does the substrate of that site, (p)ppGpp). In all
solutions identified, the 30- or 50-bisphosphonate groups of com-
pound (10) participate in the coordination of the Mn2+ ion located
in the hydrolase site, whereas multiple orientations were obtained
for the guanine and ribose moieties. Although the docking results
should be interpreted with caution, slightly higher scores for the
best docking solution for the hydrolase-ON conformation com-
pared to the best solution for the hydrolase-OFF conformation of
the enzyme might indicate an increased affinity of compound
(10) to this state of Relseq385.
Our results from the docking studies provide a mechanistically
plausible model of the inhibitor orientation in the ON conforma-
tion of the synthesis active site, whereas binding of the inhibitor
to the synthesis site in the OFF conformation seems unlikely. Sur-
prisingly, the 50-bisphosphonate of compound (10) adopts a unique
conformation strikingly different from the 50-diphosphate of GDP
in the crystal structure of Relseq385. It binds to a deep, well-de-
fined pocket of strong positive charge (Site 2) present only in the
synthesis-ON conformation of Relseq385. This site is well suited
to accommodate phosphonate or phosphate moieties. In fact, in
previous docking studies using ATP as a ligand, we predicted that
during (p)ppGpp synthesis, the triphosphate of ATP might be lo-
cated in Site 2 (data not shown). Therefore, the mode of action of
compound (10) might provide means to block the binding of GTP
and to interfere with the binding of ATP, both substrates needed
positions of binding site 1 (blue), site 2 (green), and alternative sites 3–6 (red)
shown as circles. GDP (carbons—blue, heteroatoms colored as in Fig. 8B) from the
crystal structure of Relseq385 is shown for orientation.
binds to a unique, positively charged subpocket (Site 2) next to Site
1 (Fig. 8). All docking solutions show an orientation of the 50-bis-
phosphonate that is different from the orientations found for the
50-diphosphate of GDP in the synthetase-ON and -OFF states of Re-
lseq385 in the crystal structure. The interaction is stabilized by
hydrogen bonding and polar contacts with Arg241, Lys243,
Ser247, His244 and Trp185. An inspection of the top-ten docking
solutions showed that the 30-bisphosphonate moiety of compound
(10) most probably adopts the conformation shown in Figure 8.
This conformation was found in three of the top-ten docking solu-
tions. Still, scoring results are very similar among the top-ten solu-
tions (GOLD fitness score 93 for the best solution and 84 for the
solution ranked number 10). Taking all top-ten solutions into ac-
count, there are four distinct potential interaction sites for the 30-
bisphosphonate group of compound (10) (Fig. 9; Sites 3–6). This re-
sult might reflect possible alternative binding modes of this moi-
ety. One binding mode shows a similar orientation of the 30-
bisphosphonate of compound (10) and the 50-diphosphate of GDP
(Site 3; Fig. 9). In contrast, compound (10) does not seem to pos-
sess high binding affinity for the synthetase active site in the OFF
conformation. In this conformation, Relseq385 lacks a well-defined
Site 2 and is therefore unable to accommodate the 50-bisphospho-

E. Wexselblatt et al. / Bioorg. Med. Chem. 18 (2010) 4485–4497
4493
for (p)ppGpp synthesis. Although our docking studies together
4.3. Mass spectrometry (HRMS)
with results from competition experiments of compound (10)
and GTP strongly suggest that compound (10) preferably binds to
the synthetase active site of Relseq385, we cannot rule out that
additional interactions at the hydrolase site influence the synthe-
tase activity by allosteric signaling. To assess if conformational
crosstalk between hydrolase and synthetase plays a role in inhibi-
tion of Relseq385 by compound (10), more experimental evidence
is needed.
Mass spectrometry was carried out with Orbi-trap (Thermo
Finnigen) using nanospray attachment. MRFA was used as internal
standard. Data were processed using BIOWORKS 3.3 package.
4.4. UV–vis spectrometry
UV–vis spectrophotometric data were collected with an Ultro-
spec 2100 Pro Spectrophotometer. Data were processed using the
Swift II Wavescan™ software (Biochrom Ltd).
3. Conclusions
4.5. HPLC
ppGpp analogues were synthesized and their inhibitory effects
on Rel proteins from Gram-negative and Gram-positive bacteria
were tested in vitro. It was found that both RelA and Relseq385
were inhibited mostly by analogues bearing bisphosphonate
groups at positions 50 and 30. Inhibition of Relseq385 was more
effective than that of RelA, either due to differences in the structure
of the synthetase site, or additional binding of the compound to the
Relseq385 hydrolase site, thereby inhibiting the synthetase site
through an allosteric effect.
Analytical HPLC was performed using a Shimadzu LC-2010 HT
HPLC system equipped with a UV detector using a Merck RP-C18
25 ꢁ 4.6-mm column. Preparative and semi-preparative chroma-
tography fractions were collected on a SpectraSystem SCM 1000
HPLC system equipped with a Gilson 117 UV detector, using Phe-
nomenex RP-C18 25 ꢁ 10-mm and 25 ꢁ 22-mm columns.
4.6. Inhibitor docking into the synthetase and hydrolase sites of
Relseq385
An inverse correlation was found between the degree of inhibi-
tion of the synthetic activity of RelA and its dissociation from the
ribosome.
The crystal structure of Relseq38527 (pdb code: 1VJ7), which
shows the protein in the synthetase-ON/hydrolase-OFF conforma-
tion (molecule A) and in the synthetase-OFF/hydrolase-ON confor-
mation (molecule B), was used as a target for docking of compound
(10). Atom types and protonation states were assigned using SYBYL
(Tripos, St. Louis, MO, USA). For docking into the synthetase active
site, hydrogen-bonding interactions between the guanine moiety
of GDP and the protein in the crystal structure were introduced
as constraints, assuming they would be preserved in the complex
It was shown that compound (10) competes with GTP but not
with ATP. In silico docking studies with Relseq385 suggest that this
competition likely occurs at the binding pocket of GDP/GTP situ-
ated at the active site of the synthetase domain. The identification
of distinct binding sites for all moieties of compound (10) will al-
low rational modification of the molecule to further improve prop-
erties important for efficient and selective inhibition of (p)ppGpp
synthesis by Rel proteins.
By inhibiting the synthetic activity of Rel proteins, we may in
fact prevent bacteria from sensing conditions where amino-acids
are absent in their environment. This, in turn, may ultimately lead
to bacterial self-starvation and death. In order to confirm these
assumptions, bacterial uptake of negatively charged inhibitors is
currently investigated at our laboratories.
with compound (10). Docking calculations were performed using
34
the program GOLD
.
Docking solutions were ranked according to
the internal scoring function implemented in GOLD
.
4.7. Cell-growth media
Very recently, we succeeded in determining the crystal struc-
ture of Relseq385 in complex with compound (10). Although this
structure requires further refinement and improvement of resolu-
tion, it clearly shows that compound (10) binds exclusively to the
synthetase catalytic site in its ON conformation. The binding mode
seen in the crystals (at this stage of refinement) is very similar to
the one shown in Figure 8. There is no large-scale conformational
change of the protein visible, compared to our previous crystal
structure of Relseq385.27
The growth media used was Luria–Bertani (LB) (1.6% tryptone,
1% yeast extract, 0.5% NaCl). Antibiotics used for selections were
added at the following concentrations: 100
and 50 g of kanamycin/mL.
lg of ampicillin/mL
l
4.8. Protein purification
Starter cultures of relA E. coli strain from our laboratory col-
D
lection32 over-expressing either pQE30-RelA (pQE30-XA (QIAGEN)
carrying E. coli RelA) or pENH385 (pET-21(+) (Novagen) carrying
Relseq38535) were grown overnight at 37 °C with shaking. The next
day, cultures were diluted 1:50 in 400 mL LB containing the proper
antibiotic and cells were grown at 37 °C until they reached an
OD600 = ꢂ0.6. Then, isopropyl-b-thio-galactopyranoside (IPTG)
(1 mg/mL) was added and the cells were grown at the same condi-
tions for an additional 2–3 h. Subsequently, cells were harvested
for 10 min at 4000 rpm and the pellet was frozen at ꢃ70 °C over-
night. The pellet was resuspended in 20 mL of lysis buffer contain-
ing 0.5 M NaCl, 20 mM NaH2PO4 (pH 8) and 10 mM imidazole.
Lysozyme was added to a final concentration of 3 mg/mL together
with a tablet of protease-inhibitor cocktail (Complete™, EDTA-
free; Roche) and the cells were then subjected to 3.5 min of sonica-
tion in order to break their outer membrane. The resuspended pel-
let was centrifuged at 10,000 rpm at 4 °C for 10 min to remove the
broken membrane. A sample of the soluble fraction was taken and
the supernatant was gently mixed with Ni-NTA beads (QIAGEN) at
4 °C for 1 h. The beads were loaded onto a column and washed
4. Experimental
4.1. Materials
All chemicals and reagents were purchased from Sigma–Aldrich
Ltd and used without further purification. Ultra-dry solvents stored
with molecular sieves under an argon atmosphere were purchased
from Acros Ltd. HPLC solvents were acquired from Bio-Lab Ltd.
Radioactive a-
32P-GTP was purchased from Isotope Ltd.
4.2. NMR
NMR data were collected with a Varian VXR-300 MHz spec-
trometer equipped with a 5-mm switchable probe. Data were pro-
cessed using the VNMR software. NMR abbreviations are as
follows: s (singlet), d (doublet), t (triplet), q (quartet), m (multi-
plet), br (broad).

4494
E. Wexselblatt et al. / Bioorg. Med. Chem. 18 (2010) 4485–4497
with 50 mL of washing buffer containing 0.5 M NaCl, 20 mM NaH2-
PO4 (pH 8) and 20 mM imidazole and a sample was taken. Elution
of the bound protein was achieved by the addition of 30 mL of elu-
tion buffer, containing 0.5 M NaCl, 20 mM NaH2PO4 (pH 8) and
250 mM imidazole. The elution fractions were run on a 12% acryl-
amide/bisacrylamide gel and the fractions containing the highest
amount of protein were loaded onto a cellulose tubular membrane,
mwco 12,000–14,000 (Cellu Sep), and stirred in dialysis buffer con-
taining 100 mM Tris–acetate (pH 8.5), 10 mM EDTA, 1 mM DTT and
25% glycerol (two times overnight at 4 °C). After dialysis, the con-
centration of the purified proteins was determined using Bradford
reagent (OD595).
centrifuged for 4 h at 32,000 rpm at 4 °C and each reaction
was divided into two fractions: soluble or ribosomal. The ribo-
somal samples were separated by 12% SDS–polyacrylamide gel
electrophoresis, transferred to a PVDF membrane (Millipore, Bed-
ford, MA), and processed for immunoreaction using anti-His anti-
bodies (1:10,000, GE Healthcare). Immunoreactive proteins were
detected using a chemiluminescence kit (Biological Industries,
Bet Haemek, Israel) according to manufacturer’s protocol and
quantified using the TINA 2.0 software (Raytest, Strauben-Hardt,
Germany).
4.12. Chemical synthesis
4.9. Ribosome preparation
4.12.1. N2-Isobutyrylguanosine (1)
Guanosine hydrate (10 g, 35.3 mmol) was dried by co-evaporat-
ing its suspension in dry pyridine (3 ꢁ 100 mL). The residue was
suspended in dry pyridine (250 mL) under nitrogen atmosphere,
and chlorotrimethylsilane (28.8 g, 265 mmol) was added. The reac-
tion mixture was stirred at ambient temperature for 2 h, cooled to
0 °C, and isobutyryl chloride (11.3 g, 106 mmol) was added drop-
wise over 20 min. The mixture was allowed to warm to room tem-
perature and stirred for 3 h. The reaction mixture was cooled to
0 °C and quenched by addition of H2O (30 mL). After stirring for
5 min at 0 °C and 5 min at room temperature, concentrated aque-
ous NH4OH (65 mL) was added. The solution was stirred for an
additional 15 min at room temperature. The mixture was diluted
with H2O (500 mL) and washed with DCM (200 mL). The aqueous
layer was concentrated by evaporation and the residue was recrys-
tallized from hot H2O to afford N2-isobutyryl guanosine (9.90 g,
79%) as a white solid.
Ribosomes were prepared as described by Block and Haseltine36
with the following modifications: cells were grown in LB medium
with shaking at 37 °C. At an OD600 of 1.5, 2 L of the cell culture was
pelleted at 4 °C (6000 rpm for 20 min) and frozen overnight at
ꢃ70 °C. Next, the pellet was resuspended in 60 mL of cold ribo-
some buffer containing 100 mM Tris–acetate (pH 8.0), 10 mM
Mg(OAc)2, and 1 mM dithiothreitol. Lysozyme was added to a final
concentration of 3 mg/mL together with a tablet of protease-inhib-
itor cocktail (Complete™, EDTA-free; Roche), and cells were lysed
by vigorous vortexing. Cells were then sonicated on ice for
3.5 min and pelleted at 4 °C (15,000 rpm for 40 min). A sample of
the supernatant representing the soluble proteins before fraction-
ation was saved for further analysis, while the rest of the superna-
tant was centrifuged in a Beckman Ti-65 rotor (30,000 rpm) at 4 °C
for 4 h. The pellet was dissolved and kept stirred at 4 °C in 10 mL of
ribosome buffer overnight. On the following day, the supernatant
was pelleted at 10,000 rpm at 4 °C for 15 min in order to remove
excess membrane residues and stored on ice. A ribosome buffer
containing 40% sucrose was prepared and 5 mL were loaded into
30 mL polycarbonate Ultra bottles (Sorvall Instruments). Thereaf-
ter, the supernatant was loaded on top of the sucrose buffer and
HRMS: Calcd: 354.1408. Found: 354.1404.
1H NMR (300 MHz, D2O-d): 8.10 (s, H8), 5.90 (d, J = 4.7 Hz, H10),
4.68 (t, J = 5.0 Hz, H20), 4.45 (t, J = 5 Hz, H30), 3.95 (m, H40), 3.86 (m,
J = 12 Hz, H50 H500), 2.77 (m, 1H (ibu)), 1.25 (d, 6H, J = 6.8 Hz, (ibu)).
4.12.2. N2-Isobutyryl-50-O-dimethoxytrityl guanosine (2)
Compound (1) was dried three times by co-evaporation with
dry pyridine. To a stirred suspension of dry N2-isobutyryl guano-
sine (5 g; 14.15 mmol) in pyridine (100 mL), a solution of dime-
thoxytrityl chloride (5 g; 14.8 mmol) in pyridine (30 mL) was
added dropwise over a period of 60 min. The reaction mixture
was stirred for 4 h at room temperature, cooled to 0 °C by immer-
sion in an ice/water bath, quenched with 5% NaHCO3 (100 mL), and
extracted with ethyl acetate (3 ꢁ 100 mL). The organic layers were
combined, dried over magnesium sulfate, concentrated in vacuum
and the residue was co-evaporated with toluene. The gum oil res-
idue was dissolved in a minimum amount of DCM and added drop-
wise to a mixture of diethyl ether and petroleum ether (2000 mL
3:1) while stirring. Pure N2-isobutyryl-50-O-dimethoxytrityl gua-
nosine precipitated from the solution, collected by filtration and
dried to obtain a white powder. Yield: 75%.
the bottles were centrifuged using
a Beckman Ti-65 rotor
(32,000 rpm) at 4 °C for 4 h.
The pellet was then suspended in 2 mL of ribosome buffer and
the ribosomes were frozen and stored at ꢃ70 °C. Protein concen-
trations in the ribosomal fraction were determined based on RNA
measurements in an ND-1000 Spectrophotometer (Nano-Drop).
4.10. In vitro (p)ppGpp accumulation assay
5ꢁ reaction mixture containing 2.5 mM GTP, 20 mM ATP,
200 mM Tris–HCl (pH 7.4), 5 mM DTT, 50 mM MgCl2, 50 mM KCl,
135 mM (NH4)2SO4 and
prepared. 0.5–1
g of either RelA or Relseq385 together with 1ꢁ
reaction mixture and 30 g of ribosomes (where necessary) were
mixed in a total volume of 20 L. Each reaction also contained a
ppGpp analogue at concentration ranging between and
a- lL per reaction) was freshly
32P-GTP (0.1
l
l
l
a
0
HRMS: Calcd: 656.2714. Found: 656.2713.
10 mM. The reactions were incubated at room temperature for a
period of either 5 min (Relseq385) or 60 min (RelA). The reactions
were stopped by the addition of 5 lL formic acid. Five microliters
of each reaction were loaded onto Cellulose PEI (Merck) and run
for 2 h in 1.5 M KH2PO4. The data was analyzed using image reader
1000 V1.8 and the (p)ppGpp content was determined using the
TINA 2.0 software (Raytest, Strauben-Hardt, Germany).
1H NMR (300 MHz, DMSO-d): 8.10 (s, H8), 7.0–7.6 (m, 13H, aro-
matic DMT), 5.82 (d, J = 7, H10), 5.48 (d, J = 5.6, OH 20), 5.18 (d,
J = 5.6, OH 30), 4.54–4.49 (q, J = 14.6 Hz, H20), 4.22–4.17 (q,
J = 15.5, H30), 4.08–4.06 (q, J = 12 Hz, H40), 3.82 (s, 6H, DMT–
OCH3), 3.2 (m, H50 H500), 2.75 (m, 1H (ibu)), 1.13 (d, J = 7 Hz, 6H
(ibu)).
4.12.3. N2-Isobutyryl-50-O-dimethoxytrityl-20(30)-O-tert-
butyldimethylsilyl guanosine (3a,b)
4.11. Measurements of dissociation of Rel proteins from
ribosomes
These compounds were prepared according to Sung et al.28
The crude isomers (20-OTBDMS and 30-OTBDMS) were sepa-
rated using preparative RP-C18 HPLC, applying 70% acetonitrile
in water as the mobile phase at a flow rate of 4 mL/min.
In vitro reactions were carried on as mentioned above, but
without the presence of labeled 32P-GTP with increasing concen-
trations of compound (10) added. The reactions were then

E. Wexselblatt et al. / Bioorg. Med. Chem. 18 (2010) 4485–4497
4495
4.12.3.1. Compound (3a): N2-Isobutyryl-50-O-dimethoxytrityl-30-
O-tert-butyldimethylsilyl guanosine. White powder. Yield: 34%.
HRMS: Calcd: 770.3579. Found: 770.3577.
and (6a–b)) in trimethyl phosphate (10 mL) cooled to 0 °C was
added to a suspension of 1 equiv of an adequately protected nucle-
oside in trimethyl phosphate (10 mL) at 0 °C. The reaction mixture
was stirred at 0 °C. After 20 h (2 h for compounds (7, 9)), 0.5 M
aqueous ammonium bicarbonate (pH 7.0) was added. Chromato-
graphic purification was carried out first on HPLC DEAE–Sephadex,
using a 0–1 M gradient of ammonium bicarbonate followed by RP-
C18 HPLC, resulting in white powders after lyophilization.
1H NMR (300 MHz, DMSO-d): 8.10 (s, H8), 7.0–7.6 (m, 13H, aro-
matic DMT), 5.85 (d, J = 6.1 Hz, H10), 5.11 (d, J = 6.3 Hz, OH 20),
4.63–4.6 (q, J = 17.1 Hz, H20), 4.23–4.21 (t, J = 8 Hz, H30), 3.95–
3.93 (q, J = 12 Hz, H40), 3.78 (s, 6H, DMT–OCH3), 3.2–3.4 (m, H50
H500), 2.78 (m, 1H (ibu)), 1.10 (d, J = 7 Hz, 6H (ibu)), 0.83 (s, 9H,
tBu–Si), 0.02 (s, 6H, (CH3)2–Si).
4.13.1. Guanosine 50-methylene bisphosphonate (7)
Starting material: guanosine, 70% yield.
4.12.3.2. Compound (3b): N2-Isobutyryl-50-O-dimethoxytrityl-20-
O-tert-butyldimethylsilyl guanosine. White powder. Yield: 63%.
HRMS: Calcd: 770.3579. Found: 770.3573.
HRMS: Calcd: 442.0523. Found: 442.0524.
1H NMR (300 MHz, D2O-d): 8.10 (s, H8), 5.90 (d, J = 5.8 Hz, H10),
4.68 (t, J = 8.2 Hz, H20), 4.45 (t, J = 8 Hz, H30), 4.2 (m, H40), 3.98 (m,
H50 H500), 2.2 (t, P–CH2–P).
1H NMR (300 MHz, DMSO-d): 8.12 (s, H8), 7.1–7.7 (m, 13H, aro-
matic DMT), 5.88 (d, J = 5.8 Hz, H10), 5.48 (d, J = 5.4 Hz, OH 30), 4.6
(t, J = 10.7 Hz, H20), 4.17–4.14 (q, J = 14.2 Hz, H30), 4.08–4.06 (q,
J = 11.7 Hz, H40), 3.82 (s, 6H, DMT–OCH3), 3.1–3.3 (m, H50 H500),
2.75 (m, 1H (ibu)), 1.13 (d, J = 7 Hz, 6H (ibu)), 0.87 (s, 9H, tBu–Si),
0.03 (s, 6H, (CH3)2–Si).
31P NMR (163 MHz): d 19.15, 16.37.
4.13.2. 20,30-Isopropylideneguanosine-50-methylene
bisphosphonate) (9)
Starting material: (8), 67% yield.
4.12.4. N2-Isobutyryl-20(30)O-tert-butyldimethylsilyl guanosine
(4a,b)
HRMS: Calcd: 482.0836. Found: 482.0837.
1H NMR (300 MHz, D2O): 7.93 (s, 1H), 5.90 (d, J = 5.8 Hz, H10),
5.17 (t, J = 8.2 Hz, H20), 4.94 (t, J = 8 Hz, H30), 4.09 (m, H40), 3.80
(m, H50 H500), 2.3 (t, J = 19.8 Hz, P–CH2–P), 1.49 (s, 3H, CH3), 1.29
(s, 3H, CH3).
1 g of (3a or 3b) was dissolved in a solution of 5% dichloroacetic
acid in dry CH2Cl2 (20 mL). After stirring for 15 min, the solvent
was evaporated and the resulting oily residue was subjected to sil-
ica-gel column chromatography. The elution was conducted with
CH2Cl2/MeOH (93:7 v/v). The appropriate fractions were pooled
and concentrated to give pure (4a–b) as a white powder.
31P NMR (163 MHz): d 18.2, 16.1.
4.13.3. N2-Isobutyryl-guanosine-20-50-di(methylene
bisphosphonate) (5a)
4.12.4.1. Compound (4a): N2-Isobutyryl-30-O-tert-butyldimeth-
ylsilyl guanosine. White powder. Yield: 90%.
Starting material: (4a), 53% yield.
HRMS: Calcd: 670.0476. Found: 670.0477.
1H NMR (300 MHz, D2O): 8.12 (s, H8), 6.08 (d, J = 5.8 Hz, H10),
5.31 (m, H20), 4.57 (t, J = 8.9 Hz, H30), 3.95 (m, H40), 3.60–3.72 (m,
H50 H500), 2.75 (m, 1H (ibu)), 2.1–2.3 (t, J = 19.7 Hz, P–CH2–P),
1.13 (d, 6H (ibu)).
HRMS: Calcd: 468.2272. Found: 468.2273.
1H NMR (300 MHz, DMSO-d): 8.12 (s, H8), 5.88 (d, J = 5.8 Hz,
H10), 5.48 (d, J = 6.3 Hz, OH 20), 5.31 (q, J = 14.3 Hz, H20), 5.15 (t,
J = 11.5 Hz, OH 50), 4.57 (t, J = 8 Hz, H30), 3.95 (m, H40), 3.60–3.72
(m, H50 H500), 2.75 (m, 1H (ibu)), 1.13 (d, J = 7 Hz, 6H (ibu)), 0.87
(s, 9H, tBu–Si), 0.03 (s, 6H, (CH3)2–Si).
31P NMR (163 MHz): d 18.5, 18.2, 16.1, 15.8.
4.13.4. N2-Isobutyryl-guanosine-30-50-di(methylene
bisphosphonate) (5b)
4.12.4.2. Compound (4b): N2-Isobutyryl-20-O-tert-butyldimeth-
ylsilyl guanosine. White powder. Yield: 92%.
Starting material: (4b), 44% yield.
HRMS: Calcd: 468.2272. Found: 468.2272.
HRMS: Calcd: 670.0476. Found: 670.0476.
1H NMR (300 MHz, DMSO-d): 8.10 (s, H8), 5.85 (d, J = 5.8 Hz,
H10), 5.15 (t, J = 11.5 Hz, OH 50), 5.11 (d, J = 6.3 Hz, OH 30), 4.72
(m, H20), 4.2 (m, H30), 4.09 (m, H40), 3.2–3.4 (m, H50 H500), 2.78
(m, 1H (ibu)), 1.10 (d, J = 7 Hz, 6H (ibu)), 0.83 (s, 9H, tBu–Si), 0.02
(s, 6H, (CH3)2–Si).
1H NMR (300 MHz, D2O): 8.10 (s, H8), 5.85 (d, J = 5.8 Hz, H10),
4.72 (t, J = 8 Hz, H20), 4.2 (m, H30), 4.09 (m, H40), 3.5 (m, H50 H500),
2.78 (m, 1H (ibu)), 2.1 (t, J = 19.9 Hz, P–CH2–P), 1.10 (d, 6H (ibu)).
31P NMR (163 MHz): d 19.1, 18.0, 16.1, 15.5.
4.13.5. N2-Isobutyryl-30-O-tert-butyldimethylsilyl-guanosine–
20-50-di(methylene bisphosphonate) (5c)
4.12.5. 20,30-Isopropylideneguanosine (8)
To a suspension of guanosine (5 g, 17 mmol) in 300 mL of ace-
tone 70% perchloric acid (2 mL, 24 mmol) was added. After
70 min, concentrated ammonium hydroxide (3 mL, 25 mmol) was
added to the reaction mixture and cooled down with an ice bath.
The precipitate was filtered out and dried over vacuum. An off-
white powder was obtained. Yield: 80%.
Starting material: (4a), 9% yield.
HRMS: Calcd: 784.1340. Found: 784.1342.
1H NMR (300 MHz, D2O): 8.12 (s, H8), 5.92 (d, J = 5.8 Hz, H10),
5.30 (m, H20), 4.57 (t, J = 8 Hz, H30), 3.95 (m, H40), 3.60–3.72 (m,
H50 H500), 2.75 (m, 1H (ibu)), 2.1–2.2 (t, J = 19.7 Hz, P–CH2–P),
1.13 (d, 6H (ibu)), 0.87 (s, 9H, tBu–Si), 0.03 (s, 6H, (CH3)2–Si).
31P NMR (163 MHz): d 18.8, 18.5, 16.2, 15.9.
HRMS: Calcd: 324.1302. Found: 324.1304.
1H NMR (300 MHz, DMSO-d6): 10.67 (s, 1H, NH), 7.89 (s, H8),
6.50 (s, 2H, NH2), 5.90 (d, J = 5.8 Hz, H10), 5.17 (t, J = 8.2 Hz, H20),
5.03 (t, J = 11.5 Hz, OH 50), 4.94 (t, J = 8 Hz, H30), 4.09 (m, H40),
3.50 (m, H50 H500), 1.49 (s, 3H, CH3), 1.29 (s, 3H, CH3).
4.13.6. N2-Isobutyryl-20-O-tert-butyldimethylsilyl guanosine-30-
50-di(methylene bisphosphonate) (5d)
Starting material: (4b), 8% yield.
HRMS: Calcd: 784.1340. Found: 784.1336.
4.13. General procedure for phosphonylation
1H NMR (300 MHz, D2O): 8.15 (s, H8), 5.87 (d, J = 5.8 Hz, H10),
4.75 (t, J = 8 Hz, H20), 4.2 (m, H30), 4.09 (m, H40), 3.4 (m, H50 H500),
2.78 (m, 1H (ibu)), 2.0–2.2 (t, J = 19.9 Hz, P–CH2–P), 1.10 (d, 6H
(ibu)), 0.83 (s, 9H, tBu–Si), 0.02 (s, 6H, (CH3)2–Si).
31P NMR (163 MHz): d 19.2, 18.6, 16.5, 16.1.
The procedure was adapted from Kalek et al.,29 as follows:
A solution of methylene bis-(phosphonic dichloride) (1.2 equiv
for compounds (7) and (9) and 2.4 equiv for compounds (5a–d)

4496
E. Wexselblatt et al. / Bioorg. Med. Chem. 18 (2010) 4485–4497
4.13.7. 20-Deoxyguanosine-30-50-di(methylene bisphosphonate)
(10)
4.17. 50-O-Acetyl-30-methylene bisphosphonate-20-
deoxyguanosine (13)
Same procedure as used for preparing (6a) using 20-deoxygua-
nosine as starting material. Yield: 73%.
A solution of methylene bis-(phosphonic dichloride) (1.2 equiv)
in trimethyl phosphate (10 mL) cooled to 0 °C was added to a sus-
pension of 1 equiv of compound (12) in trimethyl phosphate
(10 mL) at 0 °C. The reaction mixture was stirred at 0 °C. After
20 h, 0.5 M aqueous ammonium bicarbonate (pH 7.0) was added.
Purification was conducted on HPLC using a RP-C18 column, apply-
ing 5 mM ammonium acetate in water as the mobile phase at a
flow rate of 4 mL/min resulting in white powder after
lyophilization.
HRMS: Calcd: 584.0108. Found: 584.0108.
1H NMR (300 MHz, D2O): 8.15 (s, H8), 5.80 (t, J = 13.4 Hz, H10),
4.7 (m, H20 H200), 4.2 (m, H30), 4.09 (m, H40), 3.6 (m, H50 H500), 2.2
(t, J = 19.7 Hz, P–CH2–P).
31P NMR (163 MHz): d 19.2, 18.3, 16.3, 15.9.
4.14. Guanosine-20(30)-50-di(methylene bisphosphonate) (6a,b)
4.14.1. Guanosine-20-50-di(methylene bisphosphonate) (6a)
Compound (5a) was subjected to hydrolysis overnight at 60 °C
in 25% ammonium hydroxide and was then lyophilized. The crude
material was applied to a RP-C18 preparative column, applying
5 mM ammonium acetate in water as the mobile phase at a flow
rate of 4 mL/min.
Yield: 67%.
HRMS: Calcd: 468.0680. Found: 468.0680.
1H NMR (300 MHz, D2O): 7.85 (s, 1H, H-8), 6.11 (t, J = 14.5 Hz,
H-10), 5.21 (m, H-30), 4.2 (m, 2H, H-50), 2.64 (m, 1H, H-20), 2.32
(m, 1H, H-20), 2.21 (t, J = 19.7 Hz, P–CH2–P), 2.01 (s, 3H, COCH3).
31P NMR (163 Hz, D2O): 18.4; 15.2.
The desired product was obtained as a white powder. 88% yield.
HRMS: Calcd: 600.0057. Found: 600.0057.
4.18. 30-Methylene bisphosphonate-20-deoxyguanosine (14)
1H NMR (300 MHz, D2O): 8.12 (s, H8), 5.88 (d, J = 5.8 Hz, H10),
5.31 (m, H20), 4.57 (t, J = 8.6 Hz, H30), 3.95 (m, H40), 3.72 (m, H50
H500), 2.2 (t, J = 19.8 Hz, P–CH2–P).
Compound (13) was dissolved in concentrated ammonium
hydroxide and heated to 60 °C for 15 h. The ammonium hydroxide
was removed in vacuum and the solid residue was re-dissolved in
water and lyophilized. The resulting solid was purified by HPLC
employing a preparative RP-C18 column, applying 5 mM ammo-
nium acetate in water as the mobile phase at a flow rate of 4 mL/min.
White powder, yield: 91%.
31P NMR (163 MHz): d 18.9, 18.4, 16.0, 15.8.
4.14.2. Guanosine-30-50-di(methylene bisphosphonate) (6b)
Same procedure as used for preparing (6a) using (5b) as the
starting material. White powder. 90% yield.
HRMS: Calcd: 600.0057. Found: 600.0058.
HRMS: Calcd: 426.0574. Found: 426.0572.
1H NMR (300 MHz, D2O): 8.0 (s, H8), 5.9 (d, J = 5.8 Hz, H10), 4.72
(t, J = 8.6 Hz, H20), 4.2 (m, H30), 4.09 (m, H40), 3.5 (m, H50 H500), 2.0
(t, J = 19.8 Hz, P–CH2–P).
1H NMR (300 MHz, D2O): 7.85 (s, 1H, H-8), 6.11 (t, J = 14.5 Hz,
H-10), 5.21 (m, H-30), 4.2 (m, 2H, H-50), 2.64 (m, 1H, H-20), 2.32
(m, 1H, H-20), 2.21 (t, J = 19.7 Hz, P–CH2–P).
31P NMR (163 MHz): d 19.2, 18.3, 16.3, 15.9.
31P NMR (163 Hz, D2O): 18.5; 15.4.
4.15. 30-O-50-O-Diacetyl-20-deoxyguanosine (11)
Acknowledgments
20-Deoxyguanosine hydrate was repeatedly co-evaporated with
pyridine and suspended in dry pyridine (100 mL). Acetic anhydride
(50 mL) was added and the mixture was stirred at room tempera-
ture for 45 min. The solution was evaporated under reduced pres-
sure. Fifty milliliters of water were added to the oily residue. The
white solid formed was filtered and dried over phosphorus pentox-
ide. Yield: 90%.
This work was supported by grants from the German Israel
Foundation (835/2004 and 835/2007) to J.K., R.H., and G.G. R.H.
thanks the Fonds der Chemischen Industrie and the DFG Cluster of
Excellence ‘Inflammation at Interfaces’ for continuous support.
References and notes
HRMS: Calcd: 352.1251. Found: 352.1253.
1. Lock, R. L.; Harry, E. J. Nat. Rev. Drug Disc. 2008, 7, 324.
2. Walsh, C. Nat. Rev. Microbiol. 2003, 1, 65.
3. Cashel, M.; Gentry, D. R.; Hernandez, V. H.; Vinella, D. The stringent response.
In Escherichia coli and Salmonella typhimurium Cellular and molecular biology;
ASM Press: Washington, DC, 1996.
1H NMR (300 MHz, DMSO-d6): 10.67 (s, N1), 7.91 (s, H8), 6.50 (s,
2H, N2), 6.13 (t, J = 14.5 Hz, H-10), 5.30 (m, H-30), 4.29 (m, H-40),
4.16 (m, 2H, H-50), 2.95 (m, 1H, H-20), 2.48 (m, 1H, H-20), 2.08
and 2.04 (two s, 6H, 2COCH3).
4. Cashel, M.; Gallant, J. Nature 1969, 221, 838.
5. Cashel, M.; Lazzarini, R. A.; Kalbacher, B. J. Chromatogr. 1969, 40, 103.
6. Gentry, D. R.; Hernandez, V. J.; Nguyen, L. H.; Jensen, D. B.; Cashel, M. J.
Bacteriol. 1993, 175, 7982.
4.16. 50-O-Acetyl-20-deoxyguanosine (12)
7. Hammer, B. K.; Tateda, E. S.; Swanson, M. S. Mol. Microbiol. 2002, 44, 107.
8. Haralalka, S.; Nandi, S.; Bhadra, R. K. J. Bacteriol. 2003, 185, 4672.
9. Na, H. S.; Kim, H. J.; Lee, H. C.; Hong, Y.; Rhee, J. H.; Choy, H. E. Vaccine 2006, 24,
2027.
10. Nakanishi, N.; Abe, H.; Ogura, Y.; Hayashi, T.; Tashiro, K.; Kuhara, S.; Sugimoto,
N.; Tobe, T. Mol. Microbiol. 2006, 61, 194.
11. Pizarro-Cerda, J.; Tedin, K. Mol. Microbiol. 2004, 52, 1827.
12. Taylor, C. M.; Beresford, M.; Epton, H. A.; Sigee, D. C.; Shama, G.; Andrew, P. W.;
Roberts, I. S. J. Bacteriol. 2002, 184, 621.
13. Hesketh, A.; Chen, W. J.; Ryding, J.; Chang, S.; Bibb, M. Genome Biol. 2007, 8,
R161.
14. Dahl, J. L.; Kraus, C. N.; Boshoff, H. I.; Doan, B.; Foley, K.; Avarbock, D.; Kaplan,
G.; Mizrahi, V.; Rubin, H.; Barry, C. E., III Proc. Natl. Acad. Sci. U.S.A. 2003, 100,
10026.
15. Metzger, S.; Dror, I. B.; Aizenman, E.; Schreiber, G.; Toone, M.; Friesen, J. D.;
Cashel, M.; Glaser, G. J. Biol. Chem. 1988, 263, 15699.
16. Pedersen, F. S.; Kjeldgaard, N. O. Eur. J. Biochem. 1977, 76, 91.
17. Wendrich, T. M.; Blaha, G.; Wilson, D. N.; Marahiel, M. A.; Nierhaus, K. H. Mol.
Cell. 2002, 10, 779.
Compound (11) (0.2 g, 0.57 mmol) was dissolved in 5 mL pyri-
dine and treated with hydrazine hydrate (55 lL, 1.1 mmol) at room
temperature for 20 h. Acetone (1 mL) was added to the reaction
mixture and stirred for 30 min. Then the solvents were removed
under vacuum and the residue was co-evaporated with acetone.
The desired compound was isolated on a preparative RP-C18 HPLC
column using a gradient of 0–90% acetonitrile in water over 50 min
at a flow rate of 5 mL/min. 2D NMR was used to identify the result-
ing compounds.
White powder, yield: 47%.
HRMS: Calcd: 310.1146. Found: 310.1148.
1H NMR (300 MHz, DMSO-d6): 10.67 (s, N1), 7.91 (s, H8), 6.50 (s,
2H, N2), 6.13 (t, J = 14.5 Hz, H-10), 5.44 (br s, 1H, 30OH), 4.39 (m, 1H,
H-30), 4.29 (m, H-40), 4.16 (m, 2H, H-50), 2.95 (m, 1H, H-20), 2.48 (m,
1H, H-20), 2.08 and 2.04 (two s, 6H, 2COCH3).
18. Battesti, A.; Bouveret, E. Mol. Microbiol. 2006, 62, 1048.

E. Wexselblatt et al. / Bioorg. Med. Chem. 18 (2010) 4485–4497
4497
19. Spira, B.; Silberstein, N.; Yagil, E. J. Bacteriol. 1995, 177, 4053.
20. Vinella, D.; Albrecht, C.; Cashel, M.; D’Ari, R. Mol. Microbiol. 2005, 56, 958.
21. Metzger, S.; Sarubbi, E.; Glaser, G.; Cashel, M. J. Biol. Chem. 1989, 264, 9122.
22. Xiao, H.; Kalman, M.; Ikehara, K.; Zemel, S.; Glaser, G.; Cashel, M. J. Biol. Chem.
1991, 266, 5980.
28. Sung, W. L.; Narang, S. A. Can. J. Chem.-Rev. Can. Chim. 1982, 60, 111.
29. Kalek, M.; Jemielity, J.; Stepinski, J.; Stolarski, R.; Darzynkiewicz, E. Tetrahedron
Lett. 2005, 46, 2417.
30. Yoshikawa, M.; Kato, T.; Takenishi, T. Bull. Chem. Soc. Jpn. 1969, 42, 3505.
31. Ishido, Y.; Nakazaki, N.; Sakairi, N. J. Chem. Soc., Perkin Trans.
1 1979,
23. Mechold, U.; Cashel, M.; Steiner, K.; Gentry, D.; Malke, H. J. Bacteriol. 1996, 178,
1401.
2088.
32. Gropp, M.; Strausz, Y.; Gross, M.; Glaser, G. J. Bacteriol. 2001, 183, 570.
33. Yang, X.; Ishiguro, E. E. J. Bacteriol. 2001, 183, 6532.
34. Jones, G.; Willett, P.; Glen, R. C.; Leach, A. R.; Taylor, R. J. Mol. Biol. 1997, 267,
727.
35. Mechold, U.; Murphy, H.; Brown, L.; Cashel, M. J. Bacteriol. 2002, 184, 2878.
36. Block, R.; Haseltine, A. W. J. Biol. Chem. 1975, 250, 1212.
24. Mittenhuber, G. J. Mol. Microbiol. Biotechnol. 2001, 3, 585.
25. Wendrich, T. M.; Marahiel, M. A. Mol. Microbiol. 1997, 26, 65.
26. Nanamiya, H.; Kasai, K.; Nozawa, A.; Yun, C. S.; Narisawa, T.; Murakami, K.;
Natori, Y.; Kawamura, F.; Tozawa, Y. Mol. Microbiol. 2008, 67, 291.
27. Hogg, T.; Mechold, U.; Malke, H.; Cashel, M.; Hilgenfeld, R. Cell 2004, 117, 57.