Regioselective phosphorylation of carbohydrates and various alcohols by bacterial acid phosphatases; probing the substrate specificity of the enzyme from Shigella flexneri

Article abstract of DOI:10.1002/adsc.200505072

Bacterial non-specific acid phosphatases normally catalyze the dephosphorylation of a variety of substrates. As shown previously the enzymes from Shigella flexneri and Salmonella enterica are also able to catalyze the phosphorylation of inosine to inosine monophosphate and D-glucose to D-glucose 6-phosphate (D-G6P) using cheap pyrophosphate as the phosphate donor. After optimization high yields (95%) are achieved in the latter reaction and we show here that it is possible to use these enzymes in a preparative manner. This prompted us to investigate by using ³¹P NMR and HPLC also the phosphorylation of a broad range of carbohydrates and alcohols. Many cyclic carbohydrates are phosphorylated in a regioselective manner. Non-cyclic carbohydrates are phosphorylated as well. Phosphorylation of linear alcohols, cyclic and aromatic alcohols is also possible. In all cases the acid phosphatase from Shigella prefers a primary alcohol function above a secondary one. We conclude that these enzymes are an attractive alternative to existing chemical and enzymatic methods in the phosphorylation of a broad range of compounds.

Full text of DOI:10.1002/adsc.200505072

FULL PAPERS
Regioselective Phosphorylation of Carbohydrates and Various
Alcohols by Bacterial Acid Phosphatases; Probing the Substrate
Specificity of the Enzyme from Shigella flexneri
Teunie van Herk, Aloysius F. Hartog, Alida M. van der Burg, Ron Wever*
Vanꢀt Hoff Institute for Molecular Sciences, University of Amsterdam, Nieuwe Achtergracht 129, 1018 WS Amsterdam,
The Netherlands
Fax: (þ31)-20-525-5670, e-mail: rwever@science.uva.nl
Received: February 2, 2005; Accepted: March 31, 2005
Supporting Information for this article is available on the WWW under http://asc.wiley-vch.de/home/.
Abstract: Bacterial non-specific acid phosphatases Many cyclic carbohydrates are phosphorylated in a
normally catalyze the dephosphorylation of a variety regioselective manner. Non-cyclic carbohydrates are
of substrates. As shown previously the enzymes phosphorylated as well. Phosphorylation of linear al-
from Shigella flexneri and Salmonella enterica are cohols, cyclic and aromatic alcohols is also possible.
also able to catalyze the phosphorylation of inosine In all cases the acid phosphatase from Shigella prefers
to inosine monophosphate and d-glucose to d-glucose a primary alcohol function above a secondary one. We
6-phosphate (D-G6P) using cheap pyrophosphate as conclude that these enzymes are an attractive alterna-
the phosphate donor. After optimization high yields tive to existing chemical and enzymatic methods in
(95%) are achieved in the latter reaction and we the phosphorylation of a broad range of compounds.
show here that it is possible to use these enzymes in
a preparative manner. This prompted us to investigate Keywords: enzymatic phosphorylation; glucose 6-
by using 31P NMR and HPLC also the phosphoryla- phosphate; monophosphate; non-specific acid phos-
tion of a broad range of carbohydrates and alcohols. phatase; primary alcohol; regioselectivity.
Introduction
vantages thus making these syntheses more efficient.
Enzymatic phosphorylation ensures mild reaction con-
Phosphate esters play an important role in a wide variety ditions and, furthermore, protection of functional
of structurally diverse natural and biologically active groups is not necessary. The usefulness of existing enzy-
compounds such as glycolipids, nucleic acids, nucleoti- matic methods, usually kinases, however, is limited by
des, proteins, coenzymes, steroids and in particular car- their narrow substrate specificity. These kinases also re-
bohydrates.[1,2] Given the importance of this functional quire ATP as phosphate donor. Due to its high cost ATP
group it is not surprising that many methods have been cannot be employed in stoichiometric amounts and thus
developed for the phosphorylation of alcohol functions. ATP needs to be regenerated. It is clear that enzymes
Among the more widely used methods are those using with broader substrate specificity, which can use other
phosphoramidites and chlorophosphates as phosphate phosphate donors than ATP, are desired. The synthesis
donors.[1–4] Although these are methods with moderate of organic phosphates by enzymic transfer of the phos-
to high efficiency, they all require elaborate manipula- phate group from organic ꢁdonorꢀ compounds to a suit-
tion of protecting groups to achieve regioselectivity able alcohol function was first demonstrated by Axel-
with, for example, carbohydrate substrates. Potential rod[7] and Appleyard.[8] Phosphatases hydrolyzing
disadvantages of these chemical methods[4–6] are the monophosphate esters are expected to yield equal
low overall yields of phosphorylated product, formation amounts of inorganic phosphate (Pi) and an alcohol.
of oligophosphate esters as undesired side products aris- However, Axelrod found that some phosphatase prepa-
ing from overphosphorylation, the need for multi-step rations (e.g., orange juice phosphatase) acting on nitro-
procedures (e.g., phosphitylation, ligand exchange, oxi- phenyl phosphate in the presence of certain alcohols
dation and deprotection), and the use of toxic reagents gave lower amounts of Pi compared to the nitrophenol
and organic solvents.
formed.[7] It was then proposed that the preparation con-
Employing enzymes for the regioselective formation tained both a phosphatase and a transphosphatase,
of phosphate esters will eliminate many of these disad- which was also the conclusion of Appleyard. Later it
Adv. Synth. Catal. 2005, 347, 1155–1162
DOI: 10.1002/adsc.200505072
ꢂ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1155

Teunie van Herk et al.
FULL PAPERS
was found that phosphate transferase activity is a prop- arative phosphorylation of glucose and related com-
erty of the phosphatase itself.[9]
pounds, but also in the phosphorylation of structurally
The term non-specific acid phosphatases (NSAPs) re- different alcoholic substrates using PPi as a cheap phos-
fers to a group of secreted bacterial enzymes which do phate donor [Equation (1)].
not exhibit a marked substrate specificity, and show hy-
drolytic activity towards several different and structur-
ally unrelated organic phosphate monoesters such as nu-
cleotides, sugar phosphates, phosphorylated proteins
ð1Þ
and lipids, phytic acid, etc.[10,11] These enzymes are be-
lieved to function essentially in scavenging organic
phosphoesters that cannot cross the cytoplasmic mem-
brane. Inorganic phosphate (Pi) and organic products
are released, that can be transported across the mem-
brane, thus providing the cell with essential nutrients.
Some phosphohydrolases have evolved specialized
Results and Discussion
Efficient Phosphorylation of Glucose
functions in microbial virulence, signal transduction as As published previously, PhoN-Sf mediates the phos-
well as in several metabolic pathways. On the basis of phorylation of d-glucose (D-1) to d-glucose 6-phos-
amino acid sequences, NSAPs are categorized into three phate (D-G6P) using pyrophosphate (PPi) as the phos-
classes, designated as class A, B and C.[10] The active site phate donor.[21] A maximum concentration of D-G6P
of class A acid phosphatases [KX6RP-(X12–54)-PSGH- of 35–40 mM at pH 4 after 3 hours was reported. Our
(X31–54)-SRX5HX3D[10,11]] is also conserved in mammali- 31P NMR experiments reveal a maximal concentration
an glucose 6-phosphatases (G6Pase), lipid phosphatases of 65 mM at pH 4 after 200 minutes (Figure 1), which
and vanadium-containing haloperoxidases.[12,13] The ac- was confirmed by HPLC measurements.
tive site residues participate in the binding of phosphate,
The yield of D-G6P could be further increased by in-
compensate the negative charges of the oxygen atoms, creasing the concentration of D-1 in the reaction mix-
act as nucleophiles, stabilize the pentacoordinated tran- ture to 400 mM and 1 M. As a result maximal D-G6P
sition state and play a role in leaving group protona- concentrations were found of 80 and 95 mM, respective-
tion.[14,15] G6Pase is a key enzyme in glucose homeosta- ly, formed from 100 mM PPi. Thus we are dealing with a
sis, catalyzing the hydrolysis of G6P to glucose and phos- very efficient transphosphorylation process. Apparently
phate; the terminal steps in gluconeogenesis and glyco- hydrolysis of the activated phosphorylated enzyme (or
genolysis. G6Pase catalyses not only the hydrolysis of indirectly the hydrolysis of PPi) by competing water is
G6P but it also has phosphotransferase activity, being limited and transphosphorylation of a phosphoryl group
able to synthesize G6P from glucose and a wide variety to D-1 is favored over transphosphorylation to water
of phosphate donors including pyrophosphate (hydrolysis). As previously described,[21] D-1 inhibits
(PPi).[16,17]
the D-G6P dephosphorylation. The mechanism of in-
Recently, Asano et al. reported a new enzymatic hibition is complex and indicates a mixed type. Compet-
method of phosphorylation of inosine to produce ino- itive inhibition with a Kic of 50 mM and uncompetitive in-
sine 5’-monophosphate (5’IMP) by the recombinant hibition with a Kiu of 150 mM were observed. Effectively
PhoC from Morganellamorganii, a class A1 non-specific
acid phosphatase, using PPi as a phosphate donor. They
investigated the phosphotransferase activity of a num-
ber of enterobacteria, and using PPi as the phosphate do-
nor they showed that especially class A1 acid phospha-
tases, exhibit high regioselective transphosphorylation
activity.[18–20] As we have previously shown,[21] acid phos-
phatases are able to phosphorylate and dephosphory-
late glucose and G6P and based on a steady-state kinetic
analysis an overall mechanism of phosphorylation and
dephosphorylation was derived. The key intermediate
in the mechanism is an activated phosphorylated en-
zyme intermediate that may react with a phosphate ac-
ceptor to yield a phosphorylated product. The inter-
mediate may also react with water, resulting in hydroly-
Figure 1. Time course of the synthesis of G6P from glucose
sis and formation of free phosphate.[21] This prompted us
to see if we can use the acid phosphatases from Shigella
flexneri (PhoN-Sf, a class A1 enzyme) and Salmonella
and PPi by PhoN-Sf at 308C as followed by 31P NMR. The re-
action mixture contains 1 mM of phosphatase, 100 mM PPi,
100 mM sodium acetate buffer (pH 4.0) and different concen-
enterica (PhoN-Se, a class A2 enzyme) not only in prep- trations of d-glucose.
1156
ꢂ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
Adv. Synth. Catal. 2005, 347, 1155–1162

Regioselective Phosphorylation of Carbohydrates and Various Alcohols
FULL PAPERS
as outlined[21] high concentrations of D-1 will drive the
equilibrium in the direction of D-G6P. Thus the high
D-1 concentration present in our reaction mixture inhib-
its the hydrolysis of D-G6P. This explains the very slow
decrease of D-G6P under these reaction conditions.
The D-G6P concentration could also be increased by ad-
dition of more PPi. When a reaction mixture of 100 mM
PPi and 100 mM D-1 in 100 mM sodium acetate (pH 4.0)
containing 1 mM PhoN-Sf was allowed to react for 3
hours at 308C the addition of another aliquot of
100 mM PPi and subsequent incubation for 3 hours in-
creased the D-G6P concentration from 65 to 90 mM.
A third aliquot of PPi resulted in 95 mM of D-G6P. A
further increase in D-G6P is hard to reach since the
Km for D-1 at pH 4.0 is 8.2 mM[21] and hydrolysis of the
formed D-G6P (Km of 0.021 mM at pH 6.0) takes over
when the concentration of D-1 drops.
To assess the potential usefulness of the enzymatic
method to produce D-G6P, a small-scale pilot experi-
ment was carried out. A 5 mL reaction mixture contain-
ing 100 mM of PPi, 1 M of d-glucose in 100 mM acetate
buffer pH 4.0 was used for preparative phosphorylation
by 1 mM PhoN-Sf. Based on PPi, conversion reached
97.3% after 16 hours. Ba(CH3COO)2 was used to precip-
itate the D-G6P. This yielded 76.5% D-G6P in a >95%
purity. This shows that the enzyme can indeed be used in
preparative methods.
Figure 2. Time course of D- and L-G6P synthesis and Pi for-
mation by PhoN-Sf. The reaction mixture contains 1 mM of
PhoN-Sf, 100 mM glucose, 100 mM PPi, 100 mM sodium ace-
tate buffer (pH 5.3) at 308C. The reactions were monitored
by HPLC. Panel A: G6P synthesis. Panel B: Pi formation dur-
ing G6P production.
Stereospecificity of the Phosphorylation
A clear difference was observed when D-1 and L-1 were
phosphorylated by PhoN-Sf with PPi as phosphate do-
nor (Figure 2, panel A). Whereas D-G6P formation
continues for over 360 minutes to a concentration of
50 mM, the L-G6P concentration reaches only a maxi-
mal concentration of 15 mM. After 150 minutes the L-
G6P formed has been hydrolyzed completely. Clearly
D-1 has a much higher affinity for the substrate site
than L-1. The amount of inorganic phosphate (Pi)
formed in the process was also measured. When 1 mM
PhoN-Sf was added to 100 mM PPi in the absence of ac-
ceptor alcohol, PPi is hydrolyzed completely to 200 mM
Pi within 150 minutes. In the presence of L-1 (Figure 2,
panel B) also 200 mM of Pi is formed within 150 minutes.
In contrast when D-1 is present as an acceptor about
60 mM of Pi is formed after 150 minutes. These observa-
tions confirm that D-1 is a much better substrate for the
enzyme than L-1
Figure 3. Time course of G6P and inorganic phosphate for-
mation from d-glucose and PPi by PhoN-Sf as determined
by HPLC. The reaction mixture contains 100 mM D-glucose,
100 mM PPi, 100 mM sodium acetate buffer (pH 4.0), 1 mM
PhoN-Sf at 308C.
When the phosphorylation of D-1 is studied in more
detail (Figure 3) it becomes clear that the phosphoryla-
tion is efficient. At 100 minutes of incubation, 60 mM of
D-G6P and 80 mM of Pi are produced from 100 mM of
PPi. The 20 mM difference between the two values re- to hydrolysis is 6:1 and the transphosphorylation cata-
flects Pi that is enzymatically formed from either PPi lyzed by the acid phosphatase is indeed a very efficient
or D-G6P with water asacceptor. Thus during phosphor- process. The known enzymatic method for D-G6P pro-
ylation of D-1 by PPi the ratio of transphosphorylation duction with yeast hexokinase needs an ATP regenera-
Adv. Synth. Catal. 2005, 347, 1155–1162
asc.wiley-vch.de
ꢂ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1157

Teunie van Herk et al.
FULL PAPERS
tion system.[5] This method using either phosphoenol
pyruvate/pyruvate kinase or acetylphosphate/acetate
kinase has major disadvantages, e.g., the kinases are
strongly inhibited by the phosphate donors. Thus the
phosphorylation system described here is an attractive
alternative to produce D-G6P.
Phosphorylation of Other Carbohydrates.
Because of these findings, a variety of polyhydroxy com-
pounds were investigated to evaluate the synthetic ap-
plicability of the phosphatase. Figure 5 shows the phos-
phorylation of d-glucose (D-1) and its derivatives d-
mannose (2), d-galactose (3), and d-allose (4) at a start-
ing concentration of 400 mM. These compounds were
all phosphorylated, but compared to D-1 the dephos-
Figure 5. Time course synthesis of different d-hexose 6-phos-
phates from the corresponding d-hexoses by PhoN-Sf as de-
termined by 31P NMR. The reaction mixtures contain 1 mM
of phosphatase, 400 mM d-hexose, 100 mM PPi, and
phorylation is much faster. Dephosphorylation of D- 100 mM sodium acetate buffer (pH 4.0), 308C.
G6P is not observed before 15 hours of incubation (Fig-
ure 1). However, as seen in Figure 5, dephosphorylation
of phosphorylated 2, 3 and 4 is observed after 60 to 120 to a lesser extent compared to the other sugars. This
minutes. The rapid hydrolysis was also seen with phos- may be related to the fact that the orientation of the hy-
phorylated L-1 (Figure 2). Hexose 3 is phosphorylated droxy group closest to the phosphorylation site differs
Figure 4. Carbohydrate substrates phosphorylated by PhoN-Sf. The values between brackets represent maximal concentrations
of phosphate ester reached as determined by time course HPLC experiments (1–16 h). The reaction mixtures contain 1 mM of
phosphatase, 100 mM of alcohol, 100 mM PPi, and 100 mM sodium acetate buffer (pH 5.3) at 308C.
[a]
Not quantified.
1158
ꢂ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
Adv. Synth. Catal. 2005, 347, 1155–1162

Regioselective Phosphorylation of Carbohydrates and Various Alcohols
FULL PAPERS
from that in D-1. Dephosphorylation of 4 is slower com- er, d-glucosamine 6-phosphate is not stable in solu-
pared to 2 and 3. So, subtle structural differences in the tion[22] and breakdown products were seen. 5-Thio-d-
acceptor compounds give rise to a different behavior in glucose (13), which differs from D-1 in substitution of
the phosphorylation and dephosphorylation by PhoN- the ring oxygen by sulfur is phosphorylated as well,
Sf, but the phosphorylation activity is not abolished. while 1-deoxynojirimycin (14) that differs from D-1 by
Phosphorylation of these sugars by yeast hexokinase substitution of the ring oxygen by nitrogen and absence
has also been performed.[5] The same order of phosphor- of the anomeric hydroxy group, is not phosphorylated.
ylation activity as in the present study was found, name- Apparently, the enzyme does not accept NH on this po-
ly: D-1>4>2>3. The fact that hexokinase is not able to sition. Variation at position C-6, the phosphorylation
produce d-galactose 6-phosphate while it is a substrate site, is not allowed as expected. d-Glucuronic acid (15)
for PhoN-Sf is striking. Thus, hexokinase is tolerant of and 6-deoxy-d-glucose (16) were not phosphorylated
modifications at the C-2 and C-3 positions, but is intoler- at all, showing that the enzyme is selective for the 6-po-
ant of structural changes at C-4, while PhoN-Sf accepts a sition. Interestingly, 15 and 16 inhibit the hydrolysis of
more broad substrate range.
PPi by PhoN-Sf (not shown).
Because the phosphatase phosphorylates also inosine
As reported before,[21] the phosphorylation of D-1 is
pH dependent. Table 1 shows that this is also the case to inosine monophosphate[21] in which the d-ribose ring
for the d-glucose epimers. The yield of phosphomo- acts as an acceptor instead of d-glucose, a few riboses
noesters at pH 5.3 is lower compared to the yield from were investigated. In our hands a maximal concentra-
reactions carried out at pH 4.0. At higher substrate con- tion of 6.8 mM IMP at pH 5.0 is formed from 40 mM in-
centrations (0.4 M vs. 0.1 M) the equilibrium is driven in osine and 100 mM PPi. d-Ribose (17), the sugar part of
the direction of the phosphorylated product and hydrol- IMP, is also phosphorylated, although the conversion
ysis is inhibited. At 1 M of d-glucose 96.6 mM of D-G6P is low and only 2.5 mM of product is formed starting
is found (Figure 1).
from 100 mM 17 and 100 mM PPi. The difference in
Substituted sugars are substrates for PhoN-Sf as well. yield between phosphorylation of the ribose structure
Glycosides, where an aglycone is bonded to the anome- in inosine and the molecule as such can be explained
ric carbon atom (in these cases by an oxygen atom) are by the fact that ribose in inosine exists as a ribofuranose,
efficiently phosphorylated. When the anomeric hydroxy in which a primary hydroxy group is present. In solution
group is replaced by other substituents like a methyl 17 itself exists mostly in the pyranose form and no pri-
group as in a- and b-methyl d-glucopyranoside (a-5 mary hydroxy group is present.[23] Both l-arabinose
and b-5), a 2-(hydroxymethyl)-phenyl group as in salicin (L-18) and d-arabinose (D-18) are phosphorylated,
(6) or 4-hydroxyphenyl as in arbutin (7), 20 mM of phos- however, d-xylose (19) is not phosphorylated and inhib-
phorylated product are formed at pH 5.3. The amounts its the hydrolysis of PPi (not shown). When 19 is in its
of a-5 and b-5 phosphorylated are equal. This means pyranose form (>99.5% in solution),[23] the geometry
that the phosphatase does not recognize the difference of the hydroxy substituents is the same as in D-1 and,
between the two anomeric forms. Not only monosac- as shown already, D-1 inhibits also the hydrolysis of
charides, but also disaccharides [d-sucrose (8)] and tri- PPi. d-Glycero-d-gulo-heptose (20), a seven-carbon
saccharides [d-raffinose (9)] are substrates for PhoN- sugar, is converted to 13 mM of phosphomonoester.
Sf. Variation at the C-2 position is also allowed. Phos-
l-Sorbose (21), a ketose, existing mainly as a cyclic
phorylation of d-glucosamine (10), N-acetyl-d-glucosa- hemiacetal, was phosphorylated to 14 mM of product.
mine (11) and 2-deoxy-d-glucose (12) resulted also in d-Fructose (22), a ketose as well, is phosphorylated to
20 mM of phosphorylated product (at pH 5.3). Howev- a considerably lesser extent. The difference in phosphor-
ylation between 21 and 22 probably relates to the spatial
projection of the hydroxy groups in the ring with respect
Table 1. Phosphorylation of different d-hexoses with PPi by
to the primary hydroxy functions. In 21 these groups
have the same orientation as in D-1. This is not the
case in 22 and this might explain why 21 is phosphorylat-
ed almost three times better than 22. Yeast hexokinase
fails to react with 21.[5]
PhoN-Sf.[a]
pH 5.3
pH 4.0
pH 4.0
(0.1 M)
(0.1 M)
(0.4 M)
Glucose
Mannose
Galactose
Allose
(1)
(2)
(3)
(4)
47.3
14.3
6.5
52.8
20.9
8.5
80.4
63.9
30.4
69.0
For phosphorylation it is not necessary for the sub-
strates to be in a cyclic hemiacetal form, as is shown
for d-sorbitol (23) and d-mannitol (24). These com-
pounds lack the carbonyl group that is necessary to
form a hemiacetal and hence are not cyclic. Both com-
12.0
43.3
[a]
The concentrations of the d-hexose 6-phosphates given are
the maximal values observed by HPLC and 31P NMR in pounds contain two primary alcohols as potential phos-
time course studies. The reaction mixtures contain 1 mM
of PhoN-Sf, d-hexose (0.1 or 0.4 M), 100 mM PPi, and
100 mM sodium acetate buffer (pH 5.3 or pH 4.0) at 308C.
phorylation sites. Phosphorylation of either one of the
terminal hydroxy groups in 24 results in the same prod-
uct, whereas phosphorylation of 23 may result in two
Adv. Synth. Catal. 2005, 347, 1155–1162
asc.wiley-vch.de
ꢂ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1159

Teunie van Herk et al.
FULL PAPERS
Table 2. Phosphorylation of some simple alcohols by PhoN-
16.6 mM for 29. Secondary alcohols like 2-butanol (31)
and 2-propanol (28) are hardly accepted by PhoN-Sf
since only about 3.4 mM of phosphorylated product is
formed. This is in agreement with our previous find-
ings[21] that PhoN-Sf only produces 5’-IMP (primary al-
cohol function) and not 3’-IMP (secondary alcohol func-
tion) upon phosphorylation of inosine. When PhoN-Se
was used in the phosphorylation of 31, the yield is in-
creased to 13.3 mM. This is in line with the observation
that PhoN-Se is able to phosphorylate a secondary alco-
hol function in inosine to produce 3’-IMP.[21] No signifi-
Sf.[a]
Compound
Maximal concentration (mM)
(25)
(26)
14.6
8.9
(27)
(28)
11.0
3.5
(29)
(30)
16.6
16.7
R-(31)
S-(31)
1.8 (13.2[b]
3.4 (13.3[b]
)
)
Table 3. Phosphorylation of polyalcohols by PhoN-Sf.[a]
Compound
Maximal concentration (mM)
7.3
(32)
20.3
(33)
[a]
The concentrations of the phosphate esters given are the
maximal values observed by HPLC and 31P NMR in
time course studies. The reaction mixtures contain 1 mM
of PhoN-Sf, 100 mM of alcohol, 100 mM PPi, and
100 mM sodium acetate buffer (pH 5.3) at 308C.
PhoN-Se was used.
(34)
(35)
10.7
9.1 (11.8[b]
)
[b]
[a]
The concentrations of the phosphate esters given are the
maximal values observed by HPLC and 31P NMR in
time course studies. The reaction mixtures contain 1 mM
of PhoN-Sf, 100 mM of alcohol, 100 mM PPi, and
100 mM sodium acetate buffer (pH 5.3) at 308C.
PhoN-Se was used.
products. By 31P NMR it was confirmed that phosphory-
lation of 24 produced only one product peak, whereas
phosphorylation of 23 resulted in two peaks with a ratio
of 15:3 mM.
[b]
From these findings we conclude that PhoN-Sf is capa-
ble of phosphorylating a broad variety of carbohydrates.
A range of hexoses [d-glucose epimers, C-1 derivatives
(glycosides) and C-2 derivatives], pentoses, heptoses,
ketoses and non-cyclic carbohydrates are allowed as
substrates. Higher yields can be obtained by variation
of pH and addition of excess PPi. In addition to being re-
gio-selective, PhoN-Sf is also capable of recognizing the
difference between d- and l-sugars, therefore the en-
zyme is also stereoselective.
Table 4. Phosphorylation of some cyclic and aromatic alco-
hols by PhoN-Sf.[a]
Compound
Maximal concentration (mM)
39.0
(36)
(37)
3.1
(38)
(39)
13.4
3.7
Phosphorylation of Non-Sugar (Poly)hydroxy
Compounds
The observation that non-cyclic sugars 23 and 24 are
phosphorylated by the phosphatase, suggests that other
simple linear alcohols may also serve as substrates. As
Table 2 shows, indeed simple alcohols are phosphorylat-
ed and increasing chain length resulted in increased
yields of phosphorylated products. From ethanol (26)
via 1-propanol (27), 1-butanol (29) to 1-pentanol (32),
the yield increases from 8.9 to 20.3 mM starting from
both 100 mM substrate and PPi. Only methanol (25)
does not fit in this series. Branched carbon chains are al-
lowed as is seen with 3-methyl-1-butanol (30), which
gives 16.7 mM phosphorylated product, compared to
(40)
0.2 (9.1[b]
)
[a]
The concentrations of the phosphate esters given are the
maximal values observed by HPLC and 31P NMR in time
course studies. The reaction mixtures contain 1 mM of
PhoN-Sf, 100 mM of alcohol, 100 mM PPi, and 100 mM so-
dium acetate buffer (pH 5.3) at 308C.
[b]
PhoN-Se was used.
1160
ꢂ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
Adv. Synth. Catal. 2005, 347, 1155–1162

Regioselective Phosphorylation of Carbohydrates and Various Alcohols
FULL PAPERS
cant difference is observed between phosphorylation of from Escherichia blattae (EB-NSAP), a class A1 en-
R-31 and S-31 by PhoN-Sf or by PhoN-Se.
zyme, has been reported,[24] but the similarity in se-
Table 3 shows the phosphorylation of polyalcohols. quence is only 84% with PhoN-Sf and 62% with
1,3-Propanediol (34) is a better substrate compared to PhoN-Se, respectively. Although the similarity between
1,2-propanediol (33). This is in line with the observation EB-NSAP and PhoN-Sf is quite high, we chose not to
that the secondary alcohol functions in 28 and 31 are perform the modelling since small differences can cause
hardly phosphorylated as shown in Table 2 and that 34 major differences in the 3D structure near or at the ac-
has 2 primary alcohol functions. Glycerol (35) is phos- tive site.
phorylated to almost the same extent as 34 and there is
hardly any difference between the two acid phosphatas-
es.
Experimental Section
Not only linear alcohols are substrate for acid phos-
Expression and purification of Salmonella enterica PhoN
(PhoN-Se) and Shigella flexneri PhoN (PhoN-Sf) were as de-
scribed elsewhere.[21,25] All chemicals and other enzymes
were purchased from commercial suppliers and used without
purification.
phatases, cyclic and aromatic alcohols are phosphorylat-
ed as well as is illustrated in Table 4. Again, as we have
seen before, a primary alcohol function is phosphorylat-
ed much easier by PhoN-Sf compared tosecondary ones.
Furthermore, it is obvious that 38 is a less suitable sub-
strate compared to its non-aromatic counterpart 36.
An explanation for this observation might be that 36
adopts a chair conformation as does glucose and this ap-
parentlyresults in a more efficient transphosphorylation
to the corresponding monophosphate ester. The last
substrate, 1-phenylethanol (40) was chosen because of
its chirality. PhoN-Sf only phosphorylates 0.2 mM of
the substrate whereas PhoN-Se produces 9.1 mM of
phosphorylated product. For this reason, PhoN-Se was
chosen to carry out further experiments. Unfortunately,
the enzyme was not able to recognize the difference be-
tween R-40 and S-40.
Assay for Phosphotransferase Activity.
A standard reaction mixture contained 100 mM of substrate,
100 mM disodium pyrophosphate (PPi), 100 mM acetic acid
buffer (final pH 5.3 unless otherwise mentioned) and 1 mM of
PhoN-Sf or PhoN-Se in a total volume of 0.5 mL at 308C.
The amounts of phosphorylated products were determined
by HPLC on a OA-column. The enantiomeric excess (ee) of
the phosphorylation of 1-phenylethanol was determined as fol-
lows: a 50-mL sample of a standard reaction mixture was ex-
tracted with 200 mL n-octane and 20 mL of the extract were
used for HPLC analysis on a Chiralcel OD column. For
In conclusion the bacterial non-specific acid phospha- HPLC experiments, the decrease in substrate peak area was
taken as a measure for conversion.
tases from Shigella flexneri (PhoN-Sf) and Salmonella
enterica (PhoN-Se) are convenient and selective tools
in the preparation of a variety of phosphorylated prod-
ucts. These enzymes are able to phosphorylate a broad
array of structurally different alcoholic compounds in
a regiospecific manner. In addition, the NSAPs use
(cheap) PPi as phosphoryl donor. This is in contrast to
kinases which use ATP and for which regeneration sys-
tems have to be used. Furthermore, kinases exhibit a
more narrow substrate specificity meaning that for
each class of substrates another kinase has to be used.
Employing NSAPs eliminates many of the disadvantag-
es associated with classical chemical methods requiring
protection and deprotection of groups under harsh reac-
tion conditions. The preparative phosphorylation of d-
glucose, resulting in D-G6P is also reported. The enzy-
matic method is easy and efficient and thus it may be-
come a useful alternative to the existing chemical and
enzymatic methods of producing phosphorylated com-
pounds. It probably also allows the phosphorylation of
more labile molecules of which the synthesis up to
now has not been possible.
31P NMR
Phosphorylated products, PPi and Pi were quantified by phos-
phorus nuclear magnetic resonance (31P NMR). Spectra were
determined on a Varian Unity Inova at 202 MHz using a 10-
mm 15N-31P probe. Chemical shifts (d) are expressed in ppm rel-
ative to 85% phosphoric acid. At zero time a spectrum was tak-
en from the reaction mixture containing 100 mM PPi, 100 mM
substrate (or 400 mM and 1 M), 100 mM sodium acetate buffer
(final pH 5.3 or 4.0) in a 10-mm NMR tube. The reaction was
initiated by addition of 1 mM PhoN-Sf or PhoN-Se, and was car-
ried out in a 10-mm NMR tube at 308C. Concentrations of
product and reactants were determined using 50 mM dimethyl
methanephosphonate in deuterated water as an external stand-
ard, which was coaxially inserted in the NMR tube.
Preparative Synthesis of G6P
A 5-mL reaction mixture containing 100 mM of PPi, 1 M of d-
glucose (1) in 100 mM acetate buffer pH 4.0 was stirred and
kept at 308C. The reaction was started by the addition of
1 mM PhoN-Sf. Conversion reached 97.3% based on PPi after
16 hours. Isolation of G6P was based on the method described
by Pollak et al.[26] and Wong et al.[27] The pH was adjusted to 9.8
with NaOH and Pi was precipitated by the addition of 2 mL of
To explain the difference in reactivity towards the dif-
ferent substrates, docking of the substrates into the ac-
tive site of the enzymes would be an option. However,
crystal structures of PhoN-Sf and PhoN-Se are not
(yet) available. The crystal structure of the enzyme 0.5 M Ba(CH3COO)2. The precipitate was washed with 2 times
Adv. Synth. Catal. 2005, 347, 1155–1162
asc.wiley-vch.de
ꢂ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1161

Teunie van Herk et al.
FULL PAPERS
with 2 mL of water. Cold ethanol (final concentration 75%)
was added to the filtrate to precipitate the barium salt of
G6P, which was washed with 2 mL of 75% ethanol after filtra-
tion. Drying over KOH resulted in 0.237 g of a white solid
(>95% pure, determined by HPLC). The precipitate was dis-
solved using ion exchange resin (Dowex 50x8) in Hþ form. This
isolation resulted in 78.6% recovery of the formed D-G6P
(coupled enzymatic assay based on the method of Noltman
et al.[28]). This quantity corresponds to a 76.5% yield based
[7] B. Axelrod, J. Biol. Chem. 1948, 172, 1–13.
[8] J. Appleyard, Biochem. J. 1948, 42, 596–597.
[9] R. K. Morton, Biochem. J. 1958, 70, 150–155.
[10] G. M. Rossolini, S. Schippa, M. L. Riccio, F. Berlutti,
L. E. Macaskie, M. C. Thaller, Cell. Mol. Life Sci. 1998,
54, 833–850.
[11] J. Stukey, G. M. Carman, Prot. Sci. 1997, 6, 469–472.
[12] W. Hemrika, R. Renirie, H. L. Dekker, P. Barnett, R.
Wever, Proc. Natl. Acad. Sci. USA 1997, 94, 2145–2149.
[13] A. F. Neuwald, Prot. Sci. 1997, 6, 1764–1767.
[14] W. Hemrika, R. Renirie, S. Macedo-Ribeiro, A. Mes-
serschmidt, R. Wever, J. Biol. Chem. 1999, 274, 23820–
23827.
1
on PPi. H NMR (500 MHz, D2O): d¼5.22 (d, J¼3.8 Hz, H-
1a), 4.63 (d, J¼8.0 Hz, H-1b), 4.06–3.84 (m), 3.74–3.44 (m),
3.27 (t, J¼8.9 Hz). This is in agreement with reported values.[5]
To determine the amount of G6P, 50 mL of a 10 or 20ꢀdiluted
fraction were added to 2 mL of a G6P assay mixture. This assay
mixture contains 0.01 mg mLꢁ1 glucose 6-phosphate dehydro-
genase, 1 mM of NADPþ and 10 mM MgCl2 in 100 mM Tris/
acetate (pH 7.5). The formed NADPH can be monitored at
340 nm (extinction coefficient 6.22 mMꢁ1 cmꢁ1).
[15] R. Renirie, W. Hemrika, R. Wever, J. Biol. Chem. 2000,
275, 11650–11657.
[16] W. J. Arion, B. K. Wallin, P. W. Carlson, A. J. Lange, J.
Biol. Chem. 1972, 247, 2558–2565.
[17] R. C. Nordlie, in: The Enzymes, Vol. 4, 3rd edn., (Ed.:
P. D. Boyer), Academic Press, New York, 1972, pp.
543–609.
Acknowledgements
[18] K. Ishikawa, Y. Mihara, N. Shimba, N. Ohtsu, H. Kawa-
saki, E.-I. Suzuki, Y. Asano, Protein Eng. 2002, 15, 539–
543.
[19] Y. Asano, Y. Mihara, H. Yamada, J. Mol. Catal. B: Enzy-
matic 1999, 6, 271–277.
We gratefully acknowledge Dr. W. Hemrika and Dr. V. Dumay
for providing us the plasmid harbouring the S. enterica ser. Ty-
phimurium PhoN gene and Prof. A. J. Lange, University of
Minnesota USA, for providing us the plasmid pET3a PhoN-
Sf. 2-Deoxynojirimicin was kindly provided by Prof. G. J. Koo-
men. Valuable suggestions and advice from Dr. J. van Maar-
seveen are appreciated. This work was supported by the Dutch
National Research School Combination (NRSC-Catalysis).
[20] Y. Mihara, T. Utagawa, H. Yamada, Y. Asano, Appl. En-
viron. Microbiol. 2000, 66, 2811–2816.
[21] N. Tanaka, Z. Hasan, A. F. Hartog, T. van Herk, R. We-
ver, Org. Biomol. Chem. 2003, 1, 2833–2889.
[22] D. H. Brown, Biochim. Biophys. Acta 1951, 7, 487–493.
[23] P. M. Collins, R. J. Ferrier, Monosaccharides. Their chem-
istry and their roles in natural products, John Wiley and
Sons, Chichester, 1995.
References
[1] S. Jones, D. Selitsianos, K. J. Thomson, S. M. Toms, J.
Org. Chem. 2003, 68, 5211–5216.
[24] K. Ishikawa, Y. Mihara, K. Gondoh, E.-I. Suzuki, Y. Asa-
no, EMBO J. 2000, 19, 2412–2423.
[2] P. Lemmen, W. Richter, B. Werner, R. Karl, R. Strumpf,
I. Ugi, Synthesis 1993, 1–10.
[25] N. Tanaka, V. Dumay, Q. Liao, A. J. Lange, R. Wever,
Eur. J. Biochem. 2002, 269, 2162–2167.
[3] L. A. Slotin, Synthesis 1977, 737–752.
[4] D. V. Yashunsky, A. V. Nikolaev, J. Chem. Soc. Perkin
Trans. 1 2000, 1195–1198.
[5] H. K. Chenault, R. F. Mandes, K. R. Hornberger, J. Org.
Chem. 1997, 62, 331–336.
[26] A. Pollak, R. L. Baughn, G. M. Whitesides, J. Am. Chem.
Soc. 1977, 99, 2366–2367.
[27] C.-H. Wong, G. M. Whitesides, J. Am. Chem. Soc. 1981,
103, 4890–4899.
[28] E. A. Noltmann, C. J. Gubler, S. A. Kuby, J. Biol. Chem.
1961, 236, 1225–1230.
[6] E. E. Nifantiev, M. K. Grachev, S. Y. Burmistrov, Chem.
Rev. 2000, 100, 3755–3799.
1162
ꢂ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
Adv. Synth. Catal. 2005, 347, 1155–1162