Synthesis of a non-natural glucose-2-phosphate ester able to dupe the acc system of: Agrobacterium fabrum

Article abstract of DOI:10.1039/c8ob03086c

The first non-natural derivative of the rare d-glucose-2-phosphate (G2P), namely glucose-2-(O-lactic acid phosphate) (G2LP), has been synthesized. When used as sole carbon source, G2LP enables bacterial growth of the plant pathogenic strain Agrobacterium fabrum C58 (formerly referred to as Agrobacterium tumefaciens). X-ray crystallography and affinity measurements investigations reveal that G2LP binds the periplasmic binding protein (PBP) AccA similarly to the natural compounds and with the same affinity. Moreover, enzymatic assays show that it is able to serve as substrate of the phosphodiesterase AccF. The properties found for G2LP demonstrate that the very unusual glucose-2-phosphoryl residue, present in G2LP, can be used as structural feature for designing non-natural systems fully compatible with the Acc cascade of A. fabrum.

Full text of DOI:10.1039/c8ob03086c

Organic &
Biomolecular Chemistry
View Article Online
View Journal
PAPER
Synthesis of a non-natural glucose-2-phosphate
ester able to dupe the acc system of
Agrobacterium fabrum†
Cite this: DOI: 10.1039/c8ob03086c
Si-Zhe Li,‡a Armelle Vigouroux,‡b Mohammed Ahmar, a Abbas El Sahili,b
Laurent Soulère, a Laïla Sago,b David Cornu,b Solange Moréra *b and
a
Yves Queneau
*
The first non-natural derivative of the rare D-glucose-2-phosphate (G2P), namely glucose-2-(O-lactic acid
phosphate) (G2LP), has been synthesized. When used as sole carbon source, G2LP enables bacterial growth
of the plant pathogenic strain Agrobacterium fabrum C58 (formerly referred to as Agrobacterium tumefa-
ciens). X-ray crystallography and affinity measurements investigations reveal that G2LP binds the periplasmic
binding protein (PBP) AccA similarly to the natural compounds and with the same affinity. Moreover, enzy-
matic assays show that it is able to serve as substrate of the phosphodiesterase AccF. The properties found
for G2LP demonstrate that the very unusual glucose-2-phosphoryl residue, present in G2LP, can be used as
structural feature for designing non-natural systems fully compatible with the Acc cascade of A. fabrum.
Received 12th December 2018,
Accepted 4th January 2019
DOI: 10.1039/c8ob03086c
rsc.li/obc
abbreviations for glucose 1 and 6 phosphates, G2P or G3P
more often refer to glycerol or glyceraldehyde derivatives. As a
Introduction
Carbohydrate phosphates and phosphodiesters are essential building block present in more complex systems, the glucose-
metabolites, being substrates, intermediates or products 2-phosphoryl residue was identified or suggested in some
involved in a huge number of biological processes. With phosphorylated starches and glycogens,1–3 and with respect to
respect to glucose phosphates, the most common ones are smaller systems, it was found only in a few biological mole-
glucose-6-phosphate (G6P) and glucose-1-phosphate (G1P), cules, namely the phosphodiesters agrocinopines C and D and
involved in the conversion of glucose to glycogen or starch, or the phosphoramidate agrocin 84 (Fig. 1), produced by various
their metabolism in most multicellular animals and plants. agrobacteria stains.4–9 Agrocin 84 is a natural antibiotic pro-
Comparing bibliographic databases using glucose-X-phosphate duced by the non-pathogenic bacterial strain Agrobacterium
keywords is a sign of how much phosphorylation at sites other radiobacter K84 used as a biocontrol agent, being also
than O-1 or O-6 are far much less encountered (see for imported by Agrobacterium fabrum C58 by the periplasmic
example hits in database searches for references mentioning
glucose-X-phosphate).§ Whereas G1P and G6P are accepted
aUniv Lyon, Institut de Chimie et Biochimie Moléculaires et Supramoléculaires,
CNRS, Université Lyon 1, INSA Lyon, CPE Lyon, ICBMS, UMR 5246, Université
Claude Bernard, Bâtiment Lederer, 1 Rue Victor Grignard, 69622 Villeurbanne
Cedex, France. E-mail: Yves.queneau@insa-lyon.fr
bInstitute for Integrative Biology of the Cell (I2BC), CNRS CEA Université Paris-Sud,
Université Paris-Saclay, Avenue de la Terrasse, Gif-sur-Yvette 91198, France.
E-mail: Solange.morera@i2bc.paris-saclay.fr
§Hits in database searches for glucose-X-phosphate, (December 2018 numbers)
in SciFinder: X = 1: 7066; X = 2: 19; X = 3: 93; X = 4: 13; X = 5: 14; X = 6: 59332;
and in web of knowledge: X = 1: 3509; X = 2: 9; X = 3: 38; X = 4: 7; X = 5: 14; X = 6:
34 806.
†Electronic supplementary information (ESI) available: NMR spectra for new
compounds, 2D NMR structural elucidation of G2LP and crystallographic data
and refinement parameters for the structure of AccA in complex with G2LP. See
DOI: 10.1039/c8ob03086c
Fig. 1 Structures of D-glucose-2-phosphate (G2P), its esters agrocino-
pines C and D, its amide agrocin 84, and D-arabinose-2-phosphate (A2P)
and its ester agrocinopine A.
‡Equal contribution.
This journal is © The Royal Society of Chemistry 2019
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
binding protein (PBP) AccA associated to its ABC transporter.
As Agrobacterium tumefaciens is now recognized as a complex
of several species including A. fabrum to which belongs
A. fabrum C58, here we use its new name although the most
abundant literature is associated with the A. tumefaciens
species complex, and especially with the strain C58.
Agrocin 84 structure is a phosphoramidate connecting a
glucose moiety with the toxic moiety TM84, a di-substituted
adenine nucleotide analog which inhibits leucine synthase
once imported and delivered inside the cell, thus killing the
bacteria. Initially proposed with a glucofuranosyl-1-phosphoryl
moiety,10,11 its structure was then suggested to be possibly a
pyranose system though with imprecise connection position,11
and recently fully ascertained with the help of the X-ray struc-
ture of agrocin 84 bound to AccA.12 Assessing this structure
was essential for explaining how the same PBP AccA is able to
bind several types of ligands. A few other G2P derivatives,
namely the phosphodiester connecting two glucoses at O-2
and O-6, and a phenyl glucoside derivative were reported.13–15
The quite rare D-glucose-2-phosphate, which we will refer to
as G2P in this paper, is known since the late 1940s from the
work by Fleury et al. and by Farrar,16,17 and has been observed
as a product of the opening of the cyclic glucose-1,2-phosphate
(GCP).18 G2P was included in several studies on phosphate
hydrolysis rates and phosphatases characterizations,19 and used
for analytical purposes for assessing starch or glycogen phos-
phorylation positions.3,20 As an isolated molecule, apart from
the studies mentioned above, only one specific biological prop-
erty has been established for G2P by El Sahili et al.,12 being able
to activate quorum sensing (QS) and Ti plasmid transfer respon-
sible for the virulence of the pathogenic bacteria Agrobacterium
fabrum causing the crown gall. This property is associated with
its ability to bind the AccR transcriptional repressor just like the
natural ligand for this process, L-arabinose-2-phosphate (A2P)
(Fig. 1), does.12 This common property of G2P and A2P was one
of several clues unveiling the complete role of agrocinopine A,
the 2-O-L-arabinose-4′-O-sucrose phosphodiester produced in
the tumors which develop in plants infected by Agrobacterium
fabrum and re-imported in the bacteria by the PBP AccA to serve
as nutrient and precursor of signals of QS regulated virulence.12
Overall, the ability of the rarely encountered glucosyl-2-O-
phosphoryl moiety to bind AccA suggested that the two first
steps in the acc cascade (AccA-AccF, import then degradation)
could be used as a strategy for delivering molecules inside
Agrobacterium fabrum. The work presented herein aims at
probing this strategy, by preparing the first, non-natural,
glucose-2-O-phosphodiester, namely D-glucose-2-(O-L-lactic
acid phosphate) (G2LP), and by studying its ability to be used
as carbon source by A. fabrum, its binding with AccA and its
ability to be hydrolyzed by the phosphodiesterase AccF.
Scheme 1 Preparation of 2-O-D-glucose-2-O-L-lactic acid phosphate.
lactic acid benzyl ester residue and one benzyl residue
(Scheme 1). Thus, benzyl (2S)-2-(((benzyloxy)(diisopropyl-
amino)-phosphanyl)oxy)propanoate (3) was prepared from bis
(diisopropylaminochlorophosphine) (1) and benzyl L-lactate,
leading to the intermediate bisaminated phosphoramidite 2
which gave the monoaminated phosphoramidite 3 after substi-
tution of only one diisopropylamino group by benzyl alcohol,
thanks to the use of diisopropylammonium tetrazolide as mild
and hindered base.21
Coupling of phosphoramidite 3 with the known benzyl
2,3,4,6-tetra-O-benzyl-α-D-glucopyranoside
4
(obtained by
TIBAL-mediated selective 2-O-debenzylation22 of perbenzylated
glucose) proceeded in 84% yield in the presence of tetrazole as
activating agent, giving the desired phosphite
5 in a
1 : 1 mixture of the two diastereoisomers at the chiral phos-
phorus atom. The phosphite 5 showed limited thermal stabi-
lity and should be used rapidly for the subsequent oxidation
step performed at 0 °C using tert-BuOOH, affording the fully
protected phosphate 6 in 80% yield. The alternative order for
constructing the phosphite, namely coupling the glucose with
the bisaminated phosphoramidite 2 followed by substitution
with benzyl alcohol was found less efficient overall.
Palladium catalyzed hydrogenolysis of all 5 benzyl groups
towards the target compound G2LP was found problematic
due to the formation of the free lactic carboxylic acid which
could be either hydrolyzed or transformed to an undesired
methyl ester when the reaction was performed in methanol.
Several conditions were tested (Table 1), showing that saliniza-
tion of the medium from the beginning using sodium or pot-
assium carbonate was advantageous, as previously reported in
the case of unstable substrates,23 however in our case the reac-
tion became incomplete. Finally the most favorable conditions
were to use neutral aqueous ethanol as solvent then to salinize
the medium with potassium carbonate after filtration of the
catalyst and before any evaporation of the solvent.
Further ion exchange chromatography on a DEAE-Sephadex
column led to D-glucose-2-(O-L-lactic acid phosphate)
ammonium salt in 63% yield. In addition to mass spec-
troscopy characterization of G2LP, detailed NMR investigations
ascertained the position of the phosphodiester on the 2-O-
glucose and 2-O-lactic acid moieties (see 2D NMR in ESI†).
Results and discussion
G2LP synthesis relied on the preparation of an intermediate
phosphite 5 bearing one benzylated glucosyl residue, one
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Organic & Biomolecular Chemistry
Paper
Table 1 Optimization of hydrogenolysis conditions leading to G2LP
Entry Conditions
Results
a
1
2
3
CH3OH
R = CH3
THF/H2O (1/1, v/v)
Na2CO3/methanol/H2O (1 eq., 1/1,
v/v)
M = 390b; and M = 332b
Incomplete
debenzylation
Incomplete
4
K2CO3/THF/H2O (1 eq., 1/1, v/v)
debenzylation
5
6
Ethanol/H2O (1/1, v/v)
Ethanol/H2O (1/1, v/v), K2CO3
(1 eq.)
M = 332c
+ d
R = K, and R = NH4
a Transesterification product. b The mass was measured in both posi-
tive and negative ionization mode. c Decomposition after purification.
d Purification by ion exchange chromatography on a DEAE-Sephadex
column eluted with NH4CO3.
Fig. 3 G2PL bound to the ligand-binding site of AccA. (A) Two views of
G2PL shown in sticks in its annealing Fo-Fc omit map contoured at 3σ.
(B) Superposition of the bound G2PL and G2P shown in yellow and
green sticks, respectively, in the binding site of AccA (C) G2PL, (D) G2P
bound to the binding site of AccA are shown in the same code colour as
in B. Hydrogen bonds between AccA (labelled amino acids shown in
pink and slate) and each ligand are shown as dashed lines in black (dis-
tances are up to 3.2 Å).
Having G2LP in hands, several biological investigations were
conducted with respect to its uptake as nutrient by
Agrobacterium fabrum, its ability to bind the PBP AccA and its
ability to be hydrolyzed by the phosphodiesterase AccF, which
are reported below. When present as the sole carbon source for
A. fabrum C58 strain, G2LP was used as nutrient by the bacteria
as indicated by bacterial growth (Fig. 2). A similar behaviour
was observed with L-arabinose-2-phosphate used as a sole
carbon source meaning that G2LP was degraded in the bacterial
cytoplasm once imported by AccA and its ABC transporter.
For assessing the ability of G2LP to bind AccA, an X-ray
diffraction investigation was then performed and the structure
of the PBP AccA in complex with G2LP at 1.8 Å was solved by
molecular replacement using the structure of AccA in complex
with G2P (PDB code 4RA1).12 Both structures are very similar dis-
playing an average root mean square deviation of 0.38 Å over all
Cα atoms. As expected, G2LP is bound between the two closed
lobes of AccA. The G2LP is well defined in its electron density
maps and can present an α or β-conformation for the O1 atom
of the 4C1 “chair” glucose moiety (Fig. 3A). This was also
observed for the bound G2P (PDB code 4RA1). Notably, the
β-conformation O1 atom makes protein interactions while the
α-conformation none. The G2P part of the bound G2LP overlaps
the bound G2P when the two structures are superimposed
making very numerous similar protein interactions (Fig. 3B–D).
In contrast, the lactate part of the G2LP interacts with AccA
through water molecules. G2LP binding to AccA was investigated
using tryptophan fluorescence spectroscopy. Titration experi-
ments yielded an apparent KD value of 2 0.2 μM (Fig. 4) which
is very similar to the reported KD value of 2.5 0.5 μM for G2P.12
Finally, the ability of G2LP to serve as a substrate for the
phosphodiesterase AccF was investigated.
A mass spec-
trometry-based enzyme assay was performed by measuring
substrate and product after mixing AccF and G2LP. A close
comparison of mass spectra showed that G2LP peak at 331.04
m/z was not detected in the presence of AccF and revealed a
hydrolysis product at 281.24 m/z corresponding to a sodium
adduct of G2P (Fig. 5).
Overall, a clear proof of the hydrolysis of the phosphodie-
sters linkage of G2LP by the phosphodiesterase AccF is
Fig. 2 G2LP is used as a carbon source. OD monitoring (600 nm) of
cultures in presence of G2LP (black squares) L-arabinose-2-phosphate
(black triangles) as a carbon source and in absence of any carbon source
(black circles) in AB minimum media.
Fig. 4 AccA fluorescence monitoring upon titration with G2PL and fit
(solid line) to a single binding model using Origin. Measures were done
in triplicates.
This journal is © The Royal Society of Chemistry 2019
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
Bruker 400 MHz or Bruker 500 MHz spectrometers. The
chemical shifts (ppm) are referenced to the solvent residual
peak and coupling constants are reported in Hz. The following
abbreviations are used to explain the multiplicities: s = singlet,
d = doublet, t = triplet, q = quartet, m = multiplet, br = broad.
Electrospray ionization (ESI) mass spectrometry (MS) experi-
ments were performed on a Thermo Finnigan LCQ Advantage
mass. High-resolution mass spectra (HRMS) were recorded on
a Finnigan Mat 151xL mass spectrometer using electrospray.
Analytical thin-layer chromatography was carried out on silica
gel Merck 60 D254 (0.25 mm). Flash chromatography was per-
formed on Merck Si 60 silica gel (40–63 μm). Optical rotations
were measured on a PerkinElmer 241 or Jasco P1010 polari-
meter with a 10 cm cell (concentration c expressed as
g/100 mL).
Benzyl (S)-2-((bis(diisopropylamino)phosphanyl)oxy)
propanoate (2)
To a suspension of bis(diisopropylaminochlorophosphine) (1)
(1.0 g, 3.76 mmol) in dry Et2O (40 mL), a solution of benzyl
L-lactate (812 mg, 4.512 mmol) and Et3N (1.05 mL, 7.52 mmol)
in dry Et2O (10 mL) was added dropwise at 0 °C and then the
white solid were removed by filtration, the solvent of the fil-
trate was evaporated, and the residue was further purified by
chromatography column (pentane/Et3N = 30 : 1) to yield benzyl
Fig. 5 Enzymatic activity of AccF monitored by mass spectrometry:
mass spectra of G2LP (upper panel) and G2LP mixed with AccF (lower
panel) are shown. Peaks marked with filled and empty circles corres-
pond to G2LP and a G2LP hydrolysis product, respectively. In lower
spectrum, G2LP peak at 331.04 m/z is not detected while a peak at
281.24 is exclusively observed. This latter peak derived from the hydro-
lysis of G2LP corresponds to a sodium adduct of G2P (G2PNa).
(S)-2-((bis(diisopropylamino)phosphanyl)oxy)propano-ate
2
(1.21 g, yield 79%) as a colorless oil.
1H NMR (300 MHz, chloroform-d) δ 7.39–7.03 (m, 5H, ArH),
5.07 (s, 2H, PhCH2), 4.40–4.04 (m, 1H, CH–O), 3.63–3.16 (m,
4H, 4CH), 1.36 (d, J = 6.8 Hz, 3H, CH3), 1.19–0.80 (m, 24H,
8CH3). 31P NMR (122 MHz, chloroform-d) δ 118.79. 13C NMR
(75 MHz, chloroform-d) δ: 173.6 (d, J = 3.9 Hz, CvO), 136.0
(ArC), 128.4 (2ArC), 128.1 (2ArC), 69.0 (PO–C̲H̲–CH3), 66.2
provided. This, combined with the ability to bind AccA, is in
adequation with the ability of A. fabrum to grow when G2LP is
the unique source of carbon.
(PhCH2), 44.7 (CH–N–CH), 44.5 (CH–N–CH), 24.4 (CH3), 24.3
(CH3), 24.24 (CH3), 24.19 (CH3), 24.1 (2CH3), 24.0 (2CH3),
20.23 (d, J = 4.5 Hz, CH3). HRMS m/z (ESI): calcd for
C22H39NaN2O3P (M + Na)+: 433.2698. Found: 433.2616.
Conclusions
Benzyl (2S)-2-(((benzyloxy)(diisopropylamino)phosphanyl)oxy)
propanoate (3)
The newly synthesized D-glucose-2-(O-L-lactic acid phosphate)
(G2LP) was found to enable Agrobacterium fabrum C58 growth,
bind AccA and serve as substrate of AccF. This supports the
idea that, mimicking Agrobacterium radiobacter which uses
Agrocin 84 for competing with Agrobacterium fabrum, other
derivatives possessing the glucose-2-phosphate moiety can
mystify the PBP AccA, based on the unusual glucose-2-phos-
phoryl key for entering Agrobacterium fabrum.
To a solution of 1H-tetrazole (3.88 mL 0.45 M CH3CN solution,
1.746 mmol) in 5 mL of anhydrous CH2Cl2 was added diiso-
propylamine (0.286 mL, 2.037 mmol) under argon at room
temperature, the mixture was allowed to stir at the same temp-
erature for 30 min, and the solvent was removed under
reduced procedure to afford diisopropylammonium tetrazolide
as a white solid. Then a solution of benzyl (S)-2-((bis(diiso-
propylamino)phosphanyl)oxy)propanoate
(2)
(1.21
g,
2.91 mmol) in dry CH2Cl2 (10.0 mL) was added dropwise to a
stirred solution of diisopropylammonium tetrazolide (0.380 g,
2.22 mmol), which was dried by repeated co-evaporation with
dry MeCN prior to use, and benzyl alcohol (0.318 mL,
3.054 mmol) in dry CH2Cl2 (20.0 mL) at 0 °C under argon.
Experimental
Chemistry – general information and synthesis of G2LP and
intermediates 2, 3, 5 and -6
Reagents and solvents were purchased from Aldrich, Acros, After being stirred for 4 h at 0 °C, the mixture was diluted with
Lancaster, Alfa Aesar, Fluka or TCI to be used without further CH2Cl2 (30 mL) and washed with a saturated NaHCO3 aqueous
purification. NMR spectra were recorded on a Bruker 300 MHz, solution (30 mL). The aqueous layer was separated and
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Organic & Biomolecular Chemistry
Paper
extracted with CH2Cl2 (10 mL). The organic layers were com- (4ArC), 127.7 (ArC), 127.6 (2ArC), 127.5 (2ArC), 127.16 (ArC),
bined, dried over Na2SO4, filtered and concentrated under 97.8 (d, J = 2.0 Hz, C-1), 97.6 (d, J = 2.0 Hz, C-1), 81.4 (t, J =
reduced pressure. The residue was purified by silica gel 4.0 Hz, 2C-3), 78.0 (C-4), 77.9 (C-4), 75.5 (2OBn), 75.1 (2OBn),
column chromatography (hexane/AcOEt = 30/1, v/v with 2% 73.9 (d, J = 10.0 Hz, C-2), 73.6 (d, J = 9.0 Hz, C-2), 73.5 (2OBn),
TEA) to afford benzyl (2S)-2-(((benzyloxy)(diisopropylamino) 70.6 (d, J = 6.5 Hz, 2C-5), 69.6 (OBn), 69.5 (OBn), 68.5 (C-6),
phosphanyl)oxy)propanoate as a colorless oil. The spectrum 68.4 (C-6), 67.9 (d, J = 11.6 Hz, CH–O–P), 67.3 (d, J = 11.6 Hz,
showed that the product is a pair of enantiomers (ratio 1/1).
1H NMR (300 MHz, chloroform-d) δ: 7.40–7.25 (m, 10H, P–OBn), 64.0 (d, J = 13.0 Hz, P–OBn), 20.0 (t, J = 3.8 Hz, 2CH3).
ArH), 5.24–5.11 (m, 2H, ArCH2), 4.79–4.64 (m, 2H, 31P NMR (122 MHz, chloroform-d) δ: 139.12, 138.88. [α]D
CH–O–P), 66.8 (d, J = 4.0 Hz, 2OBn), 64.1 (d, J = 10.0 Hz,
=
PhCH2OOC–), 4.54–4.37 (m, 1H, –CH), 3.68 (m, 2H, N–CH), +22.5 (c 3.0, CHCl3). HRMS m/z (ESI): calcd for C51H54O10P
1.49 (d, J = 6.9 Hz, 3H, CH3), 1.25–1.10 (m, 12H, 4CH3). 13C (M + H)+: 857.3376. Found: 857.3449.
NMR (75 MHz, chloroform-d) δ: 173.1 (d, J = 2.8 Hz, CvO),
172.7 (d, J = 2.8 Hz, CvO), 139.5 (d, J = 7.8 Hz, ArC), 139.4 (d,
J = 7.7 Hz, ArC), 135.7 (ArC), 128.5 (ArC), 128.2 (ArC), 128.1
(ArC), 127.3 (ArC), 127.1 (ArC), 127.0 (ArC), 126.9 (ArC), 68.9
(d, CH–O), 68.2 (d, CH–O), 66.6 (d, J = 7.0 Hz, COOCH̲ ̲2Ph),
Benzyl (2S)-2-(((benzyloxy)(((2S,3R,4S,5R,6R)-2,4,5-tris
(benzyloxy)-6-((benzyloxy)methyl)tetrahydro-2H-pyran-3-yl)oxy)
phosphoryl)oxy)propanoate (6)
65.7 (d, PhCH2), 65.4 (d, PhCH2), 43.4 (CH), 43.2 (2CH), 43.0 To a solution of compound 5 (220 mg, 0.257 mmol) in 10 mL
(CH), 24.7 (CH3), 24.6 (CH3), 24.5 (CH3), 24.4 (CH3), 20.1 (d, J = anhydrous DCM was added t-BuOOH (47 µL, 6 mol L−1 in
3.6 Hz, CH3), 20.0 (d, J = 4.8 Hz, CH3). 31P NMR (122 MHz, decane, 0.283 mmol) at 0 °C, the reaction was allowed to stir at
chloroform-d) δ 148.63, 148.47. HRMS m/z (ESI): calcd for 0 °C for 2 h, then room for 1 h until the TLC (pentane/AcOEt =
C23H32NaNO4P (M + Na)+: 440.2069. Found: 440.2026.
3 : 1) showed the starting material was consumed completely,
then the reaction solvent was evaporated and the residue was
purified by chromatography column (pentane/ether = 1 : 1) to
afford phosphate 6 (180 mg, yield = 80%) as colorless syrup.
The spectrum showed that the product is a pair of diastereo-
Benzyl (2S)-2-(((benzyloxy)(((2S,3R,4S,5R,6R)-2,4,5-tris
(benzyloxy)-6-((benzyloxy)methyl)tetrahydro-2H-pyran-3-yl)oxy)
phosphanyl)oxy)propanoate (5)
Benzyl (2S)-2-(((benzyloxy)(diisopropylamino)phosphanyl)oxy)- mers with ratio 3/2.
propanoate (3) (520 mg, 1.247 mmol) was stirred with 1H-tetra-
1H NMR (400 MHz, chloroform-d) δ: 7.34–7.14 (m, 28H,
zole (0.45 mol L−1 in acetonitrile, 2.8 mL, 1.25 mmol) in dry ArH), 7.11–7.04 (m, 2H, ArH), 5.36 (d, J = 3.7 Hz, 0.6H, H-1),
CH2Cl2 (20 mL) under argon for 30 min at room temperature. 5.22 (d, J = 3.7 Hz, 0.4H, H-1), 5.11 (m, 1H, ArCH2), 5.06–4.88
Compound 4 (225 mg, 0.416 mmol) dissolved in dry CH2Cl2 (m, 3H, ArCH2), 4.83 (d, J = 11.3 Hz, 1H, CHO–P), 4.87–4.81
(5 mL) was added and the resulting mixture was stirred for 2 h. (m, 0.5H, ArCH2), 4.77–4.70 (m, 2H, ArCH2), 4.70–4.59 (m,
TLC (petroleum ether/ethyl acetate 10 : 1) showed that the reac- 2H, ArCH2), 4.58–4.44 (m, 3.5H, ArCH2), 4.43 (m, 0.6H, H-2),
tion was completed. The reaction was diluted in 50 mL DCM 4.35 (m, 0.4H, H-2), 4.03–3.93 (m, 1H, H-3), 3.84–3.79 (m, 1H,
and washed with 50 mL of saturated NaHCO3 solution, the H-5), 3.68 (dd, J = 4.0, 4.0 Hz, 1H, H-6), 3.65–3.60 (m, 1H,
aqueous layer was washed with DCM (2 × 20 mL), the com- H-4), 3.59–3.55 (m, 1H, H-6), 1.38 (d, J = 6.9 Hz, 2H, CH3),
bined organic layer was washed with brine (50 mL) and dried 1.34 (d, J = 6.8 Hz, 1H, CH3). 13C NMR (101 MHz, chloroform-
over Na2SO4, filtered, and concentrated under reduced d) δ: 170.3 (d, J = 4.6 Hz, CvO), 170.1 (d, J = 5.8 Hz, CvO),
pressure. Purification by silica gel column chromatography 138.5 (d, J = 2.0 Hz, 2Glu-OCH2–C
̲
Ph), 138.1 (2Glu-OCH2–
Ph), 138.0 (d, J = 5.0 Hz, 2Glu-OCH2–CPh), 137.4 (Glu-
Ph), 135.8 (d, J = 8.0 Hz, P–
Ph), 135.7 (d, J = 7.6 Hz, P–OCH2–CPh), 135.3
Ph), 135.2 (COOCH2–CPh), 128.7 (2ArC), 128.50
(petroleum ether/ethyl acetate 13 : 1) yield 5 (300 mg, 84%) as
C
̲
̲
a colorless oil.
OCH2–C
̲
̲
Ph), 137.2 (Glu-OCH2–C
̲
1H NMR (400 MHz, chloroform-d) δ: 7.45–7.28 (m, 27H, OCH2–C
̲
ArH), 7.24 (dd, J = 2.0, 1.6 Hz, 1H, ArH), 7.18 (m, 2H, ArH), (COOCH2–C
̲
̲
5.19 (d, J = 4.0 Hz, 0.5H, H-1), 5.19–5.16 (d, 1H, ArCH2), (2ArC), 128.4 (d, J = 2.3 Hz, 6ArC), 128.3 (d, J = 2.1 Hz, 6ArC),
5.12–5.07 (m, 1H, ArCH2), 5.09 (d, J = 4 Hz, 0.5H, H-1), 4.98 (d, 128.2 (ArC), 128.1 (ArC), 128.0 (d, J = 2.1 Hz, 4ArC), 127.9
J = 8.0 Hz, 0.5H, ArCH2), 4.92–4.82 (m, 3.5H, ArCH2), 4.81–4.67 (ArC), 127.8 (d, J = 2.4 Hz, 4ArC), 127.7 (ArC), 127.6 (d, J = 5.1
(m, 3H, ArCH2), 4.78 (dd, 0.5H, J = 2 Hz, CH3CH–O), 4.75 (dd, Hz, 2ArC), 96.6 (C-1), 96.5 (C-1), 80.7 (C-3), 80.6 (C-3), 77.73
0.5H, J = 2 Hz, CH3CH–O), 4.61–4.51 (m, 3H, ArCH2), 4.35–4.28 (C-4), 77.67 (C-4), 77.5 (d, J = 6.2 Hz, C-2), 77.30 (d, J = 6.2 Hz,
(m, 1H, H-2), 4.08 (m, 1H, H-3), 3.92 (m, 1H, H-5), 3.79 (m, 1H, C-2), 75.5 (OCH2Ph), 75.4 (OCH2Ph), 75.21 (d, J = 2.0 Hz,
H-6), 3.76–3.71 (m, 1H, H-4), 3.70–3.64 (m, 1H, H-6), 1.39 (m, 2OCH2Ph), 73.52 (d, J = 2.9 Hz, 2OCH2Ph), 72.26 (d, J = 5.1
3H, CH3). 13C NMR (101 MHz, chloroform-d) δ: 172.0 (d, J = Hz, P–OCH), 72.04 (d, J = 5.2 Hz, P–OCH), 70.6 (C-5), 70.5
4.0 Hz, CvO), 171.9 (d, J = 2.0 Hz, CvO), 138.8 (2Glu-OCH2– (C-5), 70.1 (OCH2Ph), 70.0 (OCH2Ph), 69.6 (d, J = 5.8 Hz, P–
C̲
̲Ph), 138.2 (P–OCH2–C̲
̲
̲
̲
̲
=
P
(ArC), 128.6 (2ArC), 128.5 (ArC), 128.4 (6ArC), 128.3 (2ArC), +40.4 (c 5.5, CHCl3). HRMS m/z (ESI): calcd for C51H53ClO11
128.2 (2ArC), 128.1 (ArC), 128.0 (3ArC), 127.9 (2ArC), 127.8 (M + Cl)−: 907.3020. Found: 907.3017.
This journal is © The Royal Society of Chemistry 2019
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
D-Glucose-2-(O-L-lactic acid phosphate) ammonium salt (G2LP) manual rebuilding were performed using COOT.28 The three
dimensional model of G2LP was generated using the ProDRG
Pd/C (70 mg) was added to a solution of compound 6 (70 mg,
0.082 mmol) in ethanol and water (10 mL, v/v) and the mixture
is stirred under H2 atmosphere for 24 h at room temperature.
The solution was then filtered through Celite and K2CO3 was
webserver.29 Molecular graphics images were generated using
PYMOL software (http://www.pymol.org).
Measurement of G2LP affinity with AccA. G2LP bound to
AccA was monitored by autofluorescence by excitating the
added to the filtrate, the mixture was stirred for 30 min, then
protein at a wavelength of 295 nm and monitoring the quench-
the solvent was evaporated to give a residue as white solid,
ing of fluorescence emission of tryptophans at 335 nm. The
which was further purified by Chromabond C18 column to
experiment was performed at 22 °C in microplates using Cary
afford 21 mg desired compound as a potassium salt. The solid
Eclypse spectrofluorometer (Varian) in 25 mM Tris-HCl pH 8.0
was dissolved in water and further purified by ion exchange
and 150 mM NaCl with a fixed amount of proteins (2 µM) and
chromatography on a DEAE-Sephadex column eluted with
increasing concentrations of G2LP. G2LP has no emission
signal at 335 nm. The data were analysed using Origin 7 soft-
ware and fitted to the equation f = ΔFluorescencemax × abs(x)/
(KD + abs(x)).
Investigation of G2LP activity with AccF by mass
spectrometry
NH4CO3, to give G2LP (19 mg) in 63% yield (α/β = 5/3).
1H NMR (400 MHz, deuterium oxide) δ: 5.36 (d, J = 3.6 Hz,
1H, H-1α), 4.68 (d, J = 7.9 Hz, 0.6H, H-1β), 4.60–4.46 (m, 1.6H,
P–OCHαβ), 3.95 (td, J = 4.0, 8.0, 4.0 Hz, 1H, H-2α), 3.89–3.82
(m, 1.6H, H-6αβ), 3.81–3.75 (m, 2.6H, H-5α, H-3α, H-2β),
3.73–3.65 (m, 1.6H, H-6αβ), 3.63–3.56 (m, 0.6H, H-3β),
AccF purification. AccF expression plasmid was chemically
3.47–3.42 (m, 2.2H, H-4αβ, H-5β), 1.40 (dd, J = 6.9, 2.8 Hz, 5H,
synthesized using codon optimization for the expression in
CH3αβ). 13C NMR (101 MHz, deuterium oxide) δ: 180.4 (d, J =
E. coli and inserted into pET-21b plasmid using NdeI and
4.7 Hz, CvO), 180.1 (d, J = 5.8 Hz, CvO), 95.0 (d, J = 5.1 Hz,
BamHI restriction enzyme (Genscript, Piscataway, NJ). E. coli
C-1β), 90.9 (d, J = 2.0 Hz, C-1α), 78.9 (d, J = 6.6 Hz, C-2β), 75.9
(C-5β), 75.7 (d, J = 6.8 Hz, C-2α), 75.0 (d, J = 2.7 Hz, C-3β), 73.1
(t, J = 5.7 Hz, P–OCH), 71.5 (d, J = 5.6 Hz, C-3α), 71.2 (C-5α),
BL21 competent cells transformed with pET21b-AccF were
grown in LB media at 37 °C until OD600 of 0.6. 0.5 mM of iso-
propyl β-D-1-thiogalactopyranoside (IPTG) was added to the
69.2 (C-4α), 69.0 (C-4β), 60.6 (d, J = 5.0 Hz, C-6β), 60.5 (C-6α),
culture for overnight expression at 25 °C. The cells were pel-
20.2 (d, J = 2.3 Hz, CH3β), 19.8 (d, J = 3.2 Hz, CH3α). 31P NMR
leted by centrifugation at 8000g for 15 min at 4 °C, resus-
(122 MHz, deuterium oxide) δ: −0.86(β), −1.16 (α). HRMS m/z
pended in 50 mM Tris-HCl pH 8, 300 mM NaCl and 20 mM
(ESI): Calcd for C9H16O11P (M − H)−: 331.0436. Found:
331.0441. [α]D = +45.0 (c 1.0, H2O).
imidazole and disrupted by sonication. After centrifugation at
25 000g for 30 minutes, the filtered supernatant is injected on
a nickel affinity column (HiTrap 5 ml, GE Healthcare). After a
Biology – materials and methods
washing step of 6% of 50 mM Tris-HCl pH 8, 150 mM NaCl
G2LP uptake assay. A single colony of the plant pathogen and 300 mM imidazole (Buffer B), the protein is eluted with
Agrobacterium fabrum C58 derivative pTi::Gm24 was grown over- Buffer B and injected on a gel filtration Superdex 200 26/60
night at 28 °C in AB media supplemented with mannitol (GE Healthcare) using 50 mM Tris-HCl pH 8 and 150 mM
(2 g L−1) and gentamicin 25 µg ml−1. The bacteria pellet was NaCl. The protein fractions are pooled, concentrated at 5 mg
washed twice with 0.8% NaCl. 500 μl of AB media in presence ml−1 and stored at −80 °C.
of a carbon source (G2LP or L-arabinose-2-phosphate at a final
Mass spectrometry experiments. Mass spectrometry measure-
concentration of 1 mM) or in absence of any carbon source ments were performed with an electrospray Triple-TOF
and supplemented with gentamicin (25 µg ml−1) were inocu- 4600 mass spectrometer (ABSciex) used in negative polarity
lated at an initial OD600 of 0.1 and the OD was monitored for mode. For ESI-MS measurements, the Triple-TOF instrument
56 hours.
was operated in RF quadrupole mode with the TOF data being
Structure of AccA in complex with G2LP. The AccA protein at collected between m/z 50–1000. Collision energy was set to 10
14 mg ml−1, purified as described,12 with 5 mM of G2LP eV, and nitrogen was used as collision gas. External calibration
cocrystallized in 25% PEG 4000, 0.2 M sodium acetate and 0.1 was performed with PPG 300 μM in the m/z 100–1000 mass
M sodium citrate pH 5.6. Crystals were transfered to a cryo-pro- range. The analyst 1.6 and PeakView 1.2 softwares were used
tectant solution containing the mother solution supplemented for acquisition and data processing, respectively. The monoiso-
with 22% (w/v) PEG 400 and flash frozen in liquid nitrogen. A topic masses are annotated in the spectrum, and the estimated
1.8 Å resolution diffraction data was collected at 100 K on the mass accuracy is 0.01 Da. For mass spectra acquisitions,
PROXIMA I beamline at SOLEIL synchrotron (Saint-Aubin, direct analyzes were performed after sample dilution in 50%
France). Data collection and processing statistics are given in acetonitrile and 1% formic acid.
Table 1 of ESI.† Data processing was performed using the XDS
package.25 The structure determination of AccA-G2LP complex
was performed by molecular replacement with PHASER26
Accession numbers
using the structure of AccA in complex with G2P (PDB code
4RA1)12 as a search model. The structure was refined with Coordinates and structure factors have been deposited at the
BUSTER-2.1027 and inspection of the density maps and Protein Data Bank (PDB) under accession code 6I7W.
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Organic & Biomolecular Chemistry
Paper
11 R. J. Thompson, R. H. Hamilton and C. F. Pootjes,
Antimicrob. Agents Chemother., 1979, 16, 293–296.
12 A. El Sahili, S. Z. Li, J. Lang, C. Virus, S. Planamente,
M. Ahmar, B. G. Guimaraes, M. Aumont-Nicaise,
A. Vigouroux, L. Soulere, J. Reader, Y. Queneau, D. Faure
and S. Moréra, PLoS Pathog., 2015, 11, e1005071.
13 O. Holst, H. Moll, H. Brade and J. Thiem, Carbohydr. Res.,
1989, 188, 219–221.
14 M. Franzkowiak, J. Thiem and C. Demoulin, Carbohydr.
Res., 1986, 158, 13–35.
15 P. Rivaill and L. Szabo, Bull. Soc. Chim. Fr., 1963, 716–721.
16 P. Fleury, J. Courtois and A. Desjobert, Bull. Soc. Chim. Fr.,
1948, 694–700.
Conflicts of interest
There are no conflicts to declare.
Acknowledgements
We acknowledge SOLEIL for provision of synchrotron radiation
facilities (proposal ID 20160782) in using Proxima 1 beamline.
This work was supported by ANR-Blanc SENSOR (ANR-12-BSV8-
0003-01/02/03), and benefited from the I2BC crystallization plat-
form supported by ANR FRISBI (ANR-10-INSB-05-01) and the
I2BC mass spectrometry SICaPS platform. A PhD grant to SZL
from the China Scholarship Council is gratefully acknowledged,
as well as financial support from CNRS and MESR.
17 K. R. Farrar, J. Chem. Soc., 1949, 3131–3135.
18 F. C. Kokesh, D. A. Cameron, Y. Kakuda and P. V. Kuras,
Carbohydr. Res., 1978, 62, 289–300.
19 F. C. Kokesh, R. K. Stephenson and Y. Kakuda, Biochim.
Biophys. Acta, 1977, 483, 258–262.
20 T. Whiteside, L. Carreira and S. Hilal, QSAR Comb. Sci.,
2007, 26, 587–595.
21 N. Oka, Y. Morita, Y. Itakura and K. Ando, Chem. Commun.,
2013, 49, 11503–11505.
22 T. Lecourt, A. Herault, A. J. Pearce, M. Sollogoub and
P. Sinay, Chemistry, 2004, 10, 2960–2971.
Notes and references
1 V. S. Tagliabracci, C. Heiss, C. Karthik, C. J. Contreras,
J. Glushka, M. Ishihara, P. Azadi, T. D. Hurley,
A. A. DePaoli-Roach and P. J. Roach, Cell Metab., 2011, 13,
274–282.
2 V.
M.
Chikwana,
M.
Khanna,
S.
Baskaran,
V. S. Tagliabracci, C. J. Contreras, A. DePaoli-Roach,
P. J. Roach and T. D. Hurley, Proc. Natl. Acad. Sci. U. S. A.,
2013, 110, 20976–20981.
3 S. Lim and P. A. Seib, Cereal Chem., 1993, 70, 145–152.
4 J. G. Ellis and P. J. Murphy, Mol. Gen. Genet., 1981, 181, 36–
43.
23 L. Y. Wu, M. O. Anderson, Y. Toriyabe, J. Maung,
T. Y. Campbell, C. Tajon, M. Kazak, J. Moser and
C. E. Berkman, Bioorg. Med. Chem., 2007, 15, 7434–7443.
24 E. Haudecoeur, S. Planamente, A. Cirou, M. Tannieres,
B. J. Shelp, S. Moréra and D. Faure, Proc. Natl. Acad.
Sci. U. S. A., 2009, 106, 14587–14592.
5 J. G. Ellis, A. Kerr, A. Petit and J. Tempe, Mol. Gen. Genet.,
1982, 186, 269–274.
25 W. Kabsch, Acta Crystallogr., Sect. D: Biol. Crystallogr., 2010,
66, 125–132.
6 G. T. Hayman and S. K. Farrand, Mol. Gen. Genet., 1990,
223, 465–473.
7 P. Oger and S. K. Farrand, Mol. Microbiol., 2001, 41, 1173–
1185.
8 M. Lindberg, T. Norberg and S. Oscarson, J. Carbohydr.
Chem., 1992, 11, 243–253.
26 A. J. McCoy, R. W. Grosse-Kunstleve, P. D. Adams,
M. D. Winn, L. C. Storoni and R. J. Read, J. Appl.
Crystallogr., 2007, 40, 658–674.
27 E. Blanc, P. Roversi, C. Vonrhein, C. Flensburg, S. M. Lea
and G. Bricogne, Acta Crystallogr., Sect. D: Biol. Crystallogr.,
2004, 60, 2210–2221.
9 M. Lindberg and S. Oscarson, J. Carbohydr. Chem., 1993,
12, 1139–1147.
28 P. Emsley and K. Cowtan, Acta Crystallogr., Sect. D: Biol.
Crystallogr., 2004, 60, 2126–2132.
10 W. P. Roberts, M. E. Tate and A. Kerr, Nature, 1977, 265,
379–381.
29 A. W. Schuttelkopf and D. M. van Aalten, Acta Crystallogr.,
Sect. D: Biol. Crystallogr., 2004, 60, 1355–1363.
This journal is © The Royal Society of Chemistry 2019
Org. Biomol. Chem.