1354
Biochemical Society Transactions (2003) Volume 31, part 6
Fructosamine 3-kinase, an enzyme involved
in protein deglycation
G. Delpierre and E. Van Schaftingen1
Laboratoire de Chimie Physiologique, UCL and ICP, Avenue Hippocrate 75, B-1200 Brussels, Belgium
Abstract
The in vivo formation of fructosamines following non-enzymatic reaction of proteins with glucose (i.e.
glycation) was first described almost 30 years ago. Until recently, the only known fate of fructosamines in
mammalian cells was their spontaneous conversion into advanced glycation end products. The identification
in human erythrocytes of a new enzyme, fructosamine 3-kinase, disclosed the existence of a so-far
unsuspected intracellular metabolism of these compounds. Fructosamine 3-kinase phosphorylates with
high affinity both low-molecular-mass and protein-bound fructosamines on the third carbon of their
deoxyfructose moiety, leading to the formation of fructosamine 3-phosphates. The latter are unstable and
spontaneously decompose into inorganic phosphate and 3-deoxyglucosone, with concomitant regeneration
of the unglycated amine. The presence of proteins related to fructosamine 3-kinase in many prokaryotic
and eukaryotic genomes suggests that this ‘deglycation’ process is not restricted to mammals.
Fructosamines: formation, role in diabetes
and metabolism
Fructosamines are the result of a spontaneous condensation
between glucose and primary amines, followed by an
Amadori rearrangement [1]. The most famous glycated protein is HbA1c, a form of haemoglobin bearing fructosamine
residues at the N-terminal valine of its β-chains, and which
represents approx. 5% of total haemoglobin in normal
individuals [2]. It is noteworthy that protein modification
by glucose usually mostly affects the ε-amino group of
lysyl residues, since the N-terminal end of many eukaryotic
cytosolic proteins is blocked by acetylation [3,4]. Glycated
haemoglobin may therefore be considered as an exception
to this rule, since approximately 50% of the fructosamine
residues are contributed by modified β-N-terminal ends,
the other 50% being bound to lysyl side chains [5]. Molecules
other than proteins, such as aminophospholipids, can also
undergo glycation [6].
Protein glycation is of importance with respect to diabetes
for two reasons. First, since the reaction rate is low and
directly proportional to glucose concentration [7,8], the levels
of glycated haemoglobin and serum fructosamines are an
integrated measure of the blood glucose concentration over
time, and are therefore commonly assayed to assess efficacy of
treatment [9–11]. Secondly, fructosamines may undergo
further spontaneous reactions, yielding AGEs (advanced
glycation end products). Both fructosamines [12] and AGEs
Key words: Amadori compound, complications, diabetes, glycation, fructose 3-phosphate,
protein repair.
Abbreviations used: AGE, advanced glycation end product; DMF, 1-deoxy-1-morpholinofructose; FN3K, fructosamine 3-kinase; FN3P-Hb, haemoglobin bound to a fructosamine
3-phosphate group; fructoseglycine, Nε -(1-deoxyfructosyl)glycine; fructoselysine, Nε -(1deoxyfructosyl)lysine; fructosevaline, Nε -(1-deoxyfructosyl)valine.
1
To whom correspondence should be addressed (e-mail [email protected]).
C 2003
Biochemical Society
(reviewed in [13–15]) are probably involved in the development of diabetic complications.
The glycation process has long been considered as irreversible, although, retrospectively, some early findings may
have suggested the existence of an intracellular deglycation
mechanism. Schleicher and Wieland [16] predicted the halflife of proteins by comparing their rate of in vitro glycation
with their extent of in vivo glycation. This method should
provide correct values provided that glycation is a purely
spontaneous (chemical) process and there is no enzymatic
deglycation. The predicted half-lives fitted with accepted
values in the case of extracellular proteins such as albumin,
but not for haemoglobin, for which the glycation-derived
value was much shorter than the average life-span of human
erythrocytes (13.6 compared with 120 days). Along the same
lines, Shapiro et al. [5] demonstrated that the prevalent sites of
glycation of haemoglobin in vivo differ from those obtained
after incubation of this protein with glucose in vitro. The
authors hypothesized that this discrepancy could be due
to the fact that glycation is affected by the composition of
the intracellular milieu, or that fructosamine residues can
undergo ‘enzymatic modifications’.
Different kinds of enzymes enable some micro-organisms
to metabolize fructosamines. Amadoriases are oxidases that
cleave either the keto–amino or the alkyl–amino bond of these
compounds, producing respectively glucosone and a free
amine [17,18], or free fructosamine and an aldehyde [19]. In
Escherichia coli, a specific kinase has recently been
identified and shown to convert free fructoselysine [Nε (1-deoxyfructosyl)lysine] into fructoselysine 6-phosphate,
which is then cleaved by a specific ‘deglycase’ to yield lysine
and glucose 6-phosphate [20]. All of these microbial enzymes
act only on low-molecular-mass fructosamines and not on
protein-bound residues, and therefore do not participate
The Enzymatic Defence against Glycation in Health, Disease and Therapeutics
Figure 1 Effect of DMF on the formation of fructose 3-phosphate
in intact human erythrocytes
Suspensions of erythrocytes were incubated at 37◦ C without or with
50 mM fructose, together with the indicated concentrations of DMF.
At different times, the cells were extensively washed and extracted
with perchloric acid. Fructose 3-phosphate concentrations correspond to
fructose liberated by alkaline phosphatase. Modified from [23].
NMR analysis showed not only that DMF inhibited fructose
phosphorylation, but that it was itself phosphorylated. The
attachment of the phosphate group to the third carbon of
the deoxyfructose moiety of DMF was demonstrated by tandem MS and NMR analysis of the phosphorylated compound
isolated from erythrocytes incubated in the presence of this
fructosamine [23].
Synthesis of radiolabelled [14 C]DMF permitted us to
calculate an apparent K m of approx. 100 µM for DMF 3phosphate formation. This value is identical to the K i of the
inhibition exerted by this fructosamine on fructose phosphorylation. Since the kinase acting on fructose and DMF
has a much higher affinity (approx. 300-fold) for the model
fructosamine than for the ketose, its designation as ‘fructosamine 3-kinase’ (FN3K) appeared to be fully justified.
Cloning and heterologous expression of
human and mouse FN3Ks
in protein deglycation. Furthermore, no such enzymes are
found in mammalian cells.
A model fructosamine inhibits the
formation of fructose 3-phosphate in
human erythrocytes
The starting point for the identification of a deglycation
pathway in mammalian cells was the discovery of fructose
3-phosphate in human erythrocytes [21]. This unusual
phosphate ester is formed from fructose in erythrocytes by
a ‘fructose 3-kinase’, which displays a very low affinity for
its substrate (apparent K m of approx. 30 mM) [22]. The V max
of fructose 3-phosphate formation in erythrocytes is approx.
0.25 µmol · h−1 · ml−1 of packed cells [23], which is at least
two orders of magnitude lower than the rate of glycolysis
in human erythrocytes. Another point is that no enzyme is
known to metabolize fructose 3-phosphate, accounting for its
accumulation in cells incubated with high concentrations of
fructose [22]. These considerations argued against a role
of ‘fructose 3-kinase’ in fructose metabolism, and led us
to postulate that the physiological substrate of this kinase
was not fructose itself, but compounds with a closely related
structure, possibly fructosamines. This was further supported
by an abstract suggesting the presence in human erythrocytes
of a kinase acting on fructose and glycated histones [24].
We tested this hypothesis by investigating the effect of a
model fructosamine, DMF (1-deoxy-1-morpholinofructose),
on the formation of fructose 3-phosphate in intact human
erythrocytes. Figure 1 shows that these cells accumulate
large amounts of fructose 3-phosphate when incubated
with 50 mM fructose, and that DMF strongly inhibits this
accumulation, with an apparent K i of approx. 100 µM. 31 P-
FN3K was purified to near homogeneity from human
erythrocytes, taking advantage of its strong affinity for Blue
Sepharose. The activity co-purified with a polypeptide of
approx. 35 kDa on SDS/polyacrylamide gels. This candidate
was isolated, digested with trypsin and submitted to tandem
MS. The sequence of one tryptic peptide obtained in this
way allowed the identification of human and mouse ESTs
(expressed sequence tags) encoding proteins of unknown
function, and enabled us to reconstruct and clone the cDNAs
of putative human and mouse FN3Ks [25].
The corresponding proteins were expressed in Escherichia
coli and purified, and their kinetic properties were compared
with those of the purified ‘natural’ enzyme. The three
enzymes phosphorylate DMF and the glycated ε-amino
group of free lysine or of lysyl residues within lysozyme, with
K m values in the micromolar range. Their affinity for N-αglycated amino acids such as fructoseglycine [Nε -(1-deoxyfructosyl)glycine] [25] or fructosevaline [Nε -(1-deoxyfructosyl)valine] [26] is much lower (K m values in the millimolar
range), and fructose is only poorly phosphorylated.
Proteins sharing approx. 30% identity with FN3K are
present in several prokaryotic and eukaryotic genomes,
although not in those of Drosophila melanogaster and
Saccharomyces cerevisiae. An alignment of some of these
proteins with human and mouse FN3Ks is shown in Figure 2.
A conserved motif located towards the C-terminal end,
i.e. PXLXHGDLWSXN, appears to be shared by all these
proteins. Interestingly, some residues (HGDXXXXN) of this
motif are also highly conserved in aminoglycoside kinases
[27], which, like FN3K, transfer a phosphoryl group on to a
secondary alcohol nearby an amino group.
Role of fructosamine phosphorylation
Intracellular glycated proteins are physiological substrates
of FN3K, as demonstrated by the identification of fructoselysine 3-phosphate bound to haemoglobin using 31 P-NMR
C 2003
Biochemical Society
1355
1356
Biochemical Society Transactions (2003) Volume 31, part 6
Figure 2 Alignment of human and mouse FN3Ks with related
Figure 3 Effect of DMF on the sum of glycated haemoglobin and
proteins of unknown function present in different genomes
Residues conserved with respect to the human sequence are shaded.
FN3P-Hb measured in intact human erythrocytes incubated ex vivo
Erythrocytes were incubated at 37◦ C in the presence of 200 mM glucose,
Abbreviations: HS, Homo sapiens; MM, Mus musculus; NC, Neurospora
crassa; Nsp, Nostoc sp.; EC, Escherichia coli.
with (䊏) or without (䊊) 5 mM DMF. After 48 h, the cells were washed
and resuspended in a medium containing 5 mM glucose in the presence
(䊏, 䊉) or in the absence (䊐, 䊊) of 5 mM DMF. At the indicated times,
incubations were stopped and glycated haemoglobin was measured
by affinity chromatography on boronate columns. FN3P-Hb corresponds
to the alkali-labile phosphate released from a form of haemoglobin
retained on anion-exchangers. Modified, with permission, from [28].
c The Biochemical Society.
[26]. The significance of this phosphorylation was elucidated
by studying the effect of DMF, a substrate and competitive
inhibitor of FN3K, on the level of glycated haemoglobin
in human erythrocytes incubated ex vivo in the presence of
200 mM glucose [28]. Such a high concentration was needed
to reach a level of glycated haemoglobin of approx. 13% in
2 days (compared with 5% in control cells incubated with
5 mM glucose), since the rate of fructosamine formation
is first-order with respect to sugar concentration. Under
these conditions of heavy glycation, a form of haemoglobin
containing alkali-labile phosphate could be detected that
accounted for approx. 5% of total haemoglobin. This amount
was much lower if DMF was present in the incubation
medium, indicating that alkali-labile phosphate corresponds
to fructosamine 3-phosphate residues (FN3P-Hb, i.e.
haemoglobin bound to a fructosamine 3-phosphate group).
DMF in the medium also almost doubled the rate of
accumulation of glycated haemoglobin, confirming that the
latter is a substrate for FN3K in intact erythrocytes. When
cells incubated for 48 h with 200 mM glucose were returned
to a medium containing 5 mM glucose, the level of total
glycated haemoglobin (i.e. the sum of glycated haemoglobin
and FN3P-Hb) decreased only if DMF was omitted from the
medium (Figure 3). This net decrease confirmed the presence
of a FN3K-dependent deglycation process, rather than a mere
phosphorylation/dephosphorylation cycle between glycated
haemoglobin and FN3P-Hb.
It can be seen in Figure 3 that the level of glycated haemoglobin in the absence of DMF did not return to the expected
C 2003
Biochemical Society
background level of 5% during the second phase of incubation (after 48 h), indicating that some of the fructosamine
residues are resistant to deglycation. A likely explanation
is that some glycated residues, for instance the N-terminal
valine of the β-chains, are poor substrates of FN3K and
undergo only slow deglycation, while others (the ε-amine
of lysines) are readily phosphorylated by this enzyme. This
would be in agreement with the kinetic properties of the
enzyme mentioned above.
In vitro, free fructoselysine 3-phosphate spontaneously
decomposes into inorganic phosphate and 3-deoxyglucosone
[26], and this is most probably accompanied by regeneration
of the unglycated ε-amine. [14 C]Fructoselysine 3-phosphate
labelled on its amino acid moiety spontaneously converts
to a radioactive compound with the same chromatographic
behaviour as lysine (G. Delpierre and E. Van Schaftingen,
unpublished work). In the experiment shown in Figure 3,
the oxidation product of 3-deoxyglucosone, i.e. 2-keto-3deoxygluconate, was assayed using a specific kinase from
Escherichia coli [28]. The amount of 2-keto-3-deoxygluconate
measured in the cells incubated for 48 h with 200 mM
glucose and without DMF was found to match the amount
of haemoglobin that underwent deglycation over the same
period. Furthermore, the estimated half-life of FN3P-Hb in
this experiment was close to the value determined in vitro
for free fructoselysine 3-phosphate ([26]; G. Delpierre and E.
Van Schaftingen, unpublished work), suggesting an identical
degradation mechanism for both protein-bound and lowmolecular-mass fructosamine 3-phosphates.
The Enzymatic Defence against Glycation in Health, Disease and Therapeutics
Conclusion
Phosphorylation of fructosamine residues on their third
carbon appears to be a means by which these residues are
removed from protein. FN3K appears therefore to catalyse a
new protein repair mechanism. Other protein repair mechanisms allow, e.g., the reduction of methionine sulphoxide
residues to methionine, or the conversion of L-isoaspartyl
residues to L-aspartyl, and their importance has been
underlined by mouse knock-out experiments [29,30]. Further
work is needed to evaluate the impact of the absence of FN3K.
It will also be most interesting to know if the related genes
found in different species exert a similar activity to strive
against ‘sugar stress’.
This work was supported by the Concerted Research Action Program
of the Communauté Française de Belgique, the Belgian Federal
Service for Scientific, Technical and Cultural Affairs, and the European
Foundation for the Study of Diabetes, and by a grant from the
European Community (Euroglycan network, QLG1-CT-2000-00047).
G.D. is Chargé de Recherches of the Belgian FNRS
References
1 Amadori, M. (1931) Atti. R. Acad. Natl. Linnai 13, 72
2 Bunn, H.F., Haney, D.N., Gabbay, K.H. and Gallop, P.M. (1975) Biochem.
Biophys. Res. Commun. 67, 103–109
3 Brown, J.L. and Roberts, W.K. (1976) J. Biol. Chem. 251, 1009–1014
4 Persson, B., Flinta, C., von Heijne, G. and Jornvall, H. (1985) Eur. J.
Biochem. 152, 523–527
5 Shapiro, R., McManus, M.J., Zalut, C. and Bunn, H.F. (1980) J. Biol. Chem.
255, 3120–3127
6 Ravandi, A., Kuksis, A., Marai, L., Myher, J.J., Steiner, G., Lewisa, G. and
Kamido, H. (1996) FEBS Lett. 381, 77–81
7 Higgins, P.J. and Bunn, H.F. (1981) J. Biol. Chem. 256, 5204–5208
8 Day, J.F., Thornburg, R.W., Thorpe, S.R. and Baynes, J.W. (1979) J. Biol.
Chem. 254, 9394–9400
9 Goldstein, D.E., Little, R.R., Wiedmeyer, H.M., England, J.D. and McKenzie,
E.M. (1986) Clin. Chem. 32, B64–B70
10 Johnson, R.N., Metcalf, P.A. and Baker, J.R. (1983) Clin. Chim. Acta 127,
87–95
11 Armbruster, D.A. (1987) Clin. Chem. 33, 2153–2163
12 Cohen, M.P., Clements, R.S., Cohen, J.A. and Shearman, C.W. (1996)
Biochem. Biophys. Res. Commun. 218, 72–75
13 Baynes, J.W. (1991) Diabetes 40, 405–412
14 Vlassara, H., Bucala, R. and Striker, L. (1994) Lab. Invest. 70, 138–151
15 Brownlee, M. (2001) Nature (London) 414, 813–820
16 Schleicher, E. and Wieland, O.H. (1986) Biochim. Biophys. Acta 884,
199–205
17 Horiuchi, T., Kurokawa, T. and Saito, N. (1989) Agric. Biol. Chem. 53,
103–110
18 Takahashi, M., Pischetsrieder, M. and Monnier, V.M. (1997) J. Biol. Chem.
272, 12505–12507
19 Saxena, A.K., Saxena, P. and Monnier, V.M. (1996) J. Biol. Chem. 271,
32803–32809
20 Wiame, E., Delpierre, G., Collard, F. and Van Schaftingen, E. (2002)
J. Biol. Chem. 277, 42523–42529
21 Petersen, A., Szwergold, B.S., Kappler, F., Weingarten, M. and Brown,
T.R. (1990) J. Biol. Chem. 265, 17424–17427
22 Petersen, A., Kappler, F., Szwergold, B.S. and Brown, T.R. (1992)
Biochem. J. 284, 363–366
23 Delpierre, G., Vanstapel, F., Stroobant, V. and Van Schaftingen, E. (2000)
Biochem. J. 352, 835–839
24 Szwergold, B.S., Taylor, K., Lal, S., Su, B., Kappler, F. and Brown, T.R.
(1997) Diabetes 46 (Suppl. 1), 108A
25 Delpierre, G., Rider, M.H., Collard, F., Stroobant, V., Vanstapel, F., Santos,
H. and Van Schaftingen, E. (2000) Diabetes 49, 1627–1634
26 Szwergold, B.S., Howell, S. and Beisswenger, P.J. (2001) Diabetes 50,
2139–2147
27 Perlin, M.H., Brown, S.A. and Dholakia, J.N. (1999) Front. Biosci. 4, 63–71
28 Delpierre, G., Collard, F., Fortpied, J. and Van Schaftingen, E. (2002)
Biochem. J. 365, 801–808
29 Moskovitz, J., Bar-Noy, S., Williams, W.M., Requena, J., Berlett, B.S. and
Stadtman, E.R. (2001) Proc. Natl. Acad. Sci. U.S.A. 98, 12920–12925
30 Kim, E., Lowenson, J.D., Clarke, S. and Young, S.G. (1999) J. Biol. Chem.
274, 20671–20678
Received 6 June 2003
C 2003
Biochemical Society
1357