Báo cáo khoa học: Novel modified version of nonphosphorylated sugar metabolism – an alternative L-rhamnose pathway of Sphingomonas sp. doc

Novel modified version of nonphosphorylated sugar
metabolism – an alternative
L-rhamnose pathway of
Sphingomonas sp.
Seiya Watanabe
1,2,3
and Keisuke Makino
1,2,3,4
1 Institute of Advanced Energy, Kyoto University, Japan
2 New Energy and Industrial Technology Development Organization, Gokasho, Uji, Kyoto, Japan
3 CREST, Japan Science and Technology Agency, Gokasho, Uji, Kyoto, Japan
4 Innovative Collaboration Center, Kyoto University, Japan
Microorganisms can utilize pentoses and deoxyhexoses
as their sole carbon source. There are generally two
pathways for the metabolism of these sugars, one with
phosphorylated intermediates and the other without
such intermediates. The former pathways of bacteria
and ⁄ or fungi have been studied extensively. Many
Keywords
Entner–Doudoroff pathway; gene cluster;
L-rhamnose; metabolic evolution;
Sphingomonas sp.
Correspondence
S. Watanabe, Institute of Advanced Energy,
Kyoto University, Gokasho, Uji, Kyoto
611-0011, Japan
Fax: +81 774 38 3524
Tel: +81 774 38 3596
E-mail:
(Received 30 October 2008, revised 9
December 2008, accepted 5 January 2009)

doi:10.1111/j.1742-4658.2009.06885.x
Several bacteria, including Azotobacter vinelandii, possess an alternative
pathway of l-rhamnose metabolism, which is different from the known
bacterial pathway. In a previous article, a gene cluster related to this path-
way was identified, consisting of the genes encoding the four metabolic
enzymes l-rhamnose-1-dehydrogenase (LRA1), l-rhamnono-c-lactonase
(LRA2), l-rhamnonate dehydratase (LRA3) and l-2-keto-3-deoxyrhamno-
nate (l-KDR) aldolase (LRA4), by which l-rhamnose is converted into
pyruvate and l-lactaldehyde, through analogous reaction steps to the well-
known Entner-Doudoroff (ED) pathway. In this study, bioinformatic
analysis revealed that Sphingomonas sp. possesses a gene cluster consisting
of LRA1–3 and two genes of unknown function, LRA5 and LRA6. LRA5
catalyzed the NAD
+
-dependent dehydrogenation of several l-2-keto-3-de-
oxyacid-sugars, including l-KDR. Furthermore, the reaction product was
converted to pyruvate and l-lactate by LRA6; this is different from the
pathway of Azotobacter vinelandii. Therefore, LRA5 and LRA6 were
assigned as the novel enzymes l-KDR 4-dehydrogenase and l-2,4-diketo-3-
deoxyrhamnonate hydrolase, respectively. Interestingly, both enzymes were
phylogenetically similar to l-rhamnose-1-dehydrogenase and d-2-keto-3-
deoxyarabinonate dehydratase, respectively, and the latter was involved in
the archeal nonphosphorylative d-arabinose pathway, which is partially
analogous to the ED pathway. The introduction of LRA1–4 or LRA1–3,
LRA5 and LAR6 compensated for the l-rhamnose-defective phenotype of
an Escherichia coli mutant. Metabolic evolution and promiscuity between
the alternative l-rhamnose pathway and other sugar pathways analogous
to the ED pathway are discussed.
Abbreviations
COG, cluster of orthologous groups of proteins;

D-KDA, D-2-keto-3-deoxyarabinonate; D-KGD, D-2-keto-3-deoxygluconate; ED, Entner–
Doudoroff; FAH, fumarylacetoacetate hydrolase;
L-DKDR, L-2,4-diketo-3-deoxyrhamnonate; L-KDF, L-2-keto-3-deoxyfuconate; L-KDL, L-2-keto-
3-deoxylyxonate;
L-KDM, L-2-keto-3-deoxymannonate; L-KDR, L-2-keto-3-deoxyrhamnonate; LRA1, L-rhamnose-1-dehydrogenase; LRA2,
L-rhamnono-c-lactonase; LRA3, L-rhamnonate dehydratase; LRA4, L-2-keto-3-deoxyrhamnonate aldolase; LRA5, L-2-keto-3-deoxyrhamnonate-
(4)-dehydrogenase; LRA6,
L-2,4-diketo-3-deoxyrhamnonate hydrolase; MhpD, 2-oxopent-4-enoate hydratase; npED, nonphosphorylative
Entner–Doudoroff; SDR, short-chain dehydrogenase ⁄ reductase; aKGSA, a-ketoglutaric semialdehyde.
1554 FEBS Journal 276 (2009) 1554–1567 ª 2009 The Authors Journal compilation ª 2009 FEBS
bacteria, including Escherichia coli, also metabolize
l-rhamnose (l-6-deoxymannose) through this type of
pathway, using enzymes consisting of l-rhamnose
isomerase (EC 5.3.1.14), rhamnulokinase (EC 2.7.1.5),
and rhamnulose-1-phosphate aldolase (EC 4.1.2.19)
[1]. The l-lactaldehyde obtained, together with
dihydroxyacetone phosphate, is further converted to
l-lactate or 1,2-propanediol by l-lactaldehyde dehydro-
genase (EC 1.2.1.22) [2] and lactaldehyde : propanediol
oxidoreductase [EC 1.1.1.77(55)] [3] under aerobic and
anaerobic conditions, respectively.
The pathways without phosphorylated intermediates
are classified into two groups, in which the sugar is
commonly converted into a dl-2-keto-3-deoxyacid-
sugar through the participation of dehydrogenase,
lactonase and dehydratase enzymes (schematic reac-
tions a–c in Fig. 1) (each enzyme is referred to as ‘E’
below). In the ‘type I pathway’ of d-glucose [4,5],
d-galactose [6], d-fucose [7–10] and l-arabinose [11],
the dl-2-keto-3-deoxyacid-sugar is cleaved through an

aldolase reaction (schematic reaction e) to the appro-
priate aldehyde and pyruvate as well as to the Entner–
Doudoroff (ED) pathway. Although most metabolic
genes have not yet been identified, except for the ‘non-
phosphorylative ED (npED) pathway’ in Archaea, a
previous article [12] recently characterized this type of
l-rhamnose pathway in the fungi Pichia stipitis and
Debaryomyces hansenii (E12–E15) and the bacterium
Azotobacter vinelandii (E16–E19). In this pathway,
l-rhamnose is converted into pyruvate and l-lactalde-
hyde via l-rhamnono-c-lactone and l-rhamnonate, by
the consecutive action of the enzymes l-rhamnose-
D-Galactose
L-Fucose

D-Fucose
L-Arabinose
Pathway
Hexaric acids
L-Arabinose
D-Arabinose
II
D-Xylose
71
COG2706
COG3386
COG3618
COG2220
COG0129
COG4948

COG0800
COG0329
COG3836
COG0179
COG1012COG0364
COG1063
COG4993
COG0673
COG1028
COG0667
Glyceraldehyde 3P
D-Glyceraldehyde
L-Lactate

D-Lactaldehyde
Glycolaldehyde
Final product
α-Ketoglutarate

g
52
55
60
65
70
74
79
84
Hexaric acids
L-Rhamnose

I
Tartronate semialdehyde
L-Lactaldehyde
L-Lactate
Schematic reactions
Pathway Domain
ED

npED

Gluconate
5
Final product
(pyruvate +)
Glyceraldehyde 3P
D-Glyceraldehyde
–
–
–
20
D-Galacturonate
B
E
B
B
B
Domain
B
B
B

B
B
A
B
A
B
E
B
B
A
B
B
B
B
E
b
–
–
31
34
38
43
47
57
62
67
76
81
–
13

17
b
2
10
21
–
c
32
35
39
44
48
50
53
58
63
68
72
77
82
25
14
18
22
c
3
6
28
f
36

41
45
49
51
54
59
64
69
73
78
83
26
15
19
24
e
4
7
29
–
40
–
–
–
–
–
–
–
–
–

–
–
–
–
–
23
d
–
–
–
L-Glyceraldehyde
a
–
–
30
33
37
42
46
56
61
66
75
80
–
12
16
a
1
8

9
11
27
III
III
I
I
PPP
PPP
PPP
ED
Fig. 1. Comparison of sugar metabolic path-
ways analogous to the ED pathway in bacte-
ria (B), eukaryotes (E), and Archaea (A).
Sugar is commonly metabolized through
the participation of sugar dehydrogenase (a),
lactone-sugar hydrolase (lactonase) (b), acid-
sugar dehydratase (c), 2-keto-3-deoxyacid-
sugar dehydrogenase (d), aldolase (e), and
dehydratase (f) for 2-keto-3-deoxyacid-sugar,
and aKGSA dehydrogenase (g). Colored
COGs are homologous to each other, and
white indicates that the metabolic gene has
not yet been identified. Numbers (1–84)
correspond to enzymes catalyzing each
reaction (listed in Table S1), which are
referred to as ‘E’ in the text. In this study,
an alternative
L-rhamnose pathway including
E20–E24 was focused on (indicated by cir-

cles). The reactions of E24 and E41 can be
assigned as equivalent to e and f (see text).
S. Watanabe and K. Makino Novel
L-rhamnose metabolic pathway
FEBS Journal 276 (2009) 1554–1567 ª 2009 The Authors Journal compilation ª 2009 FEBS 1555
1-dehydrogenase (LRA1; EC 1.1.1.173), l-rhamnono-
c-lactonase (LRA2; EC 3.1.1.65), l-rhamnonate
dehydratase (LRA3; EC 4.2.1.90), and l-2-keto-3-
deoxyrhamnonate (l-KDR) aldolase (LRA4) (referred
to as the ‘aldolase pathway’ in this article) (Fig. 2A).
Furthermore, the metabolic fate of l-lactaldehyde is
dehydrogenation by l-lactaldehyde dehydrogenase
(EC 1.2.1.22), which is similar to the known bacterial
pathway as described above [13]. Interestingly, there
is no evolutionary relationship between the l-KDR
aldolases from fungi and bacteria (E15 and E19).
l-Lactaldehyde dehydrogenases from fungi and bacte-
ria belong to different subfamilies in the aldehyde
dehydrogenase superfamily, and show distinct coen-
zyme and substrate specificities. These findings indicate
that the pathways have evolved independently in spite
of the homologous schematic conversion of l-rham-
nose. The ‘type II pathway’ without phosphorylated
intermediates corresponds to an alternative pathway of
d-arabinose [14], l-arabinose [15–17], and d-xylose
[18]. In these pathways, the dl-2-keto-3-deoxypento-
nate intermediate is converted to a-ketoglutarate via
L-Rhamnose:H
+
symporter

EAM07803 EAM07804 EAM07805 EAM07806 EAM07807 EAM07808
EAM07809 EAM07810
Sugar transporterSugar channel
Azotobacter vinelandii
EAT09360 EAT09361 EAT09362 EAT09364
L-Rhamnose:H
+
symporter
EAT09366
β-Xylosidase
EAT09367
EAT09363
EAT09365
Sphingomonas sp. SKA58
(NBRC101715)
Pichia stipitis
ABN68602 ABN68603ABN68405ABN68404
LRA1 LRA3 LRA6 LRA5 LRA2
LRA1 LRA3LRA2 LRA4
LRA4LRA1 LRA2LRA3
Debaryomyces hansenii
CAG87574CAG87575CAG87576CAG87577
LADH
Chr 2
ABN64318
LADH
CAG90160Chr G
LRA4 LRA3
Escherichia coli
yfaU yfaV yfaW

Transporter
Chr8
Chr E
L-Rhamnose
L-Rhamnono-γ-lactone
L-Rhamnonate
L-2-Keto-3-deoxyrhamnonate
(
L-KDR)
L-Rhamnose
1-dehydrogenase
( LRA1, EC 1.1.1.173)
L-Rhamnono-γ-lactonase
( LRA2, EC 3.1.1.65)
L-Rhamnonate dehydratase
( LRA3, EC 4.2.1.90)
NAD(P)
+
NAD(P)H
H2O
H2O
Pyruvate
L-Lactaldehyde
L-KDR aldolase
( LRA4)
NAD
+
NADH
L-2,4-Diketo-3-deoxyrhamnonate
(

L-DKDR)
Pyruvate
L-Lactate
L-DKDR hydrolase
( LRA6)
L-KDR 4-dehydrogenase
( LRA5)
Aldolase pathway Diketo-hydrolase pathway
L-Lactate
L-Lactaldehyde
dehydrogenase
( LADH, EC 1.2.1.22)
NAD
+
NADH
A
B
Xanthomonas campestris
FucDFucB FucCFucA FucE
AAM43286 AAM43287 AAM43288 AAM43289 AAM43290 AAM43291 AAM43292
Sugar transporter
L-Fucopyranoside
mutarotase
C
Fig. 2. The alternative L-rhamnose pathway (A) and schematic gene clusters (B). P. stipitis, D. hansenii and A. vinelandii possess the ‘aldol-
ase pathway’, in which
L-rhamnose is converted into pyruvate and L-lactaldehyde. L-Lactaldehyde dehydrogenase then produces L-lactate
from
L-lactaldehyde in this pathway. In Sphingomonas sp., the L-KDR intermediate is alternatively converted into pyruvate and L-lactate via
L-DKDR (diketo-hydrolase pathway). Chr n indicates chromosome number. Homologous genes are indicated in the same color, and corre-

spond to Fig. 1. Gray putative genes are similar in sequence to other
L-rhamnose-related enzymes involved in sugar uptake. (C) Schematic
gene cluster related to the alternative
L-fucose pathway of bacteria [33].
Novel
L-rhamnose metabolic pathway S. Watanabe and K. Makino
1556 FEBS Journal 276 (2009) 1554–1567 ª 2009 The Authors Journal compilation ª 2009 FEBS
a-ketoglutaric semialdehyde (aKGSA) by dl-2-keto-3-
deoxypentonate dehydratase (EC 4.1.2.18; E59, E64,
E69, E73, E78, and E83) and aKGSA dehydrogenase
(EC 1.2.1.26; E60, E65, E70, E74, E79, and E80) (sche-
matic reactions e and f, respectively, in Fig. 1).
aKGSA is also produced from hexaric acids (d-gluca-
rate and d-galactarate) via d-2-keto-3-deoxyglucarate
by two successive dehydration reactions in bacteria
(E50–E52 and E53–E55) [19].
Most of these alternative pathways of sugar metabo-
lism have been described on the basis only of enzyme
activities in cell-free extracts from experiments
performed several decades ago. It could be useful to
identify a set of the metabolic genes (enzymes) for the
enzymatic synthesis of unavailable specific intermedi-
ates, in particular dl-2-keto-3-deoxyacid-sugars. In this
regard, transcriptomic and ⁄ or proteomic analysis have
significant advantages, as shown in the alternative
d-arabinose pathway of Archaea [14]. We, alternatively,
focused on the following two insights: (a) the equivalent
reaction step-catalyzing metabolic enzymes involved in
sugar pathways analogous to the ED pathway are classi-
fied into limited numbers of the known protein families,

cluster of orthologous groups of proteins (COG)
(Fig. 1); (b) metabolic genes often form a single gene
cluster in the genomes of bacteria and Archaea
(Fig. 2B). Accordingly, homology searches were carried
out using the known metabolic genes (Table S1) against
the genomes of microorganisms, and ‘semi-automati-
cally’ selected a set of four potential metabolic genes,
LRA1–4, involved in the alternative l-rhamnose path-
way described above [12], suggesting that this approach
may be helpful in identifying unknown sugar pathways
analogous to the ED pathway, even if the phenotype of
the microorganism is not available.
This study further developed this possibility and
revealed that in Sphingomonas sp., a phenotype-
unknown but genome sequence-available bacterium,
l-rhamnose is converted into pyruvate and l-lactate
(but not l-lactaldehyde) via four nonphosphorylated
intermediates by five metabolic enzymes (genes), which
differed partially from the aldolase pathway. Compari-
sons between the novel l-rhamnose pathway and other
sugar metabolic pathways and the substrate promiscuity
in the metabolic enzymes are also described.
Results
Gene cluster related to L-rhamnose metabolism
in Sphingomonas sp.
Several significant insights were obtained about sugar
pathways analogous to the ED pathway, including the
alternative l-rhamnose pathways of fungi and bacteria
(Fig. 1). Therefore, an extended bioinformatic analysis
was carried out, and an interesting gene cluster related

to putative sugar metabolism was found in the genome
of Sphingomonas sp. SKA58 (Fig. 2B). In this article,
the prefixes Ps (P. stipitis), Dh (D. hansenii), Av
(A. vinelandii) and Sp (Sphingomonas sp.) have been
added to gene symbols or protein designations when
required for clarity.
When compared with the LRA1–4 gene clusters
of fungi and bacteria, the gene EAT09362 of Sphingo-
monas sp. was homologous to the LRA3 gene encod-
ing l-rhamnonate dehydratase (COG4948, enolase
superfamily); there was 80.2% identity with AvLRA3.
Furthermore, gene EAT09365 belonged to the same
COG group as the LRA2 gene encoding l-rhamnono-
c-lactonase (COG3618, a ⁄ b hydrolase fold enzymes).
On the other hand, the gene cluster of Sphingomonas
sp. SKA58 contained the genes encoding two putative
short-chain dehydrogenase ⁄ reductase (SDR) family
enzymes (COG1028), EAT09360 (SpLRA1) and
EAT09364 (SpLRA5) (Fig. 2B). As the former
showed higher amino acid sequence homology to
other LRA1 proteins than the latter (66.7% and
25.3% identity with AvLRA1, respectively), the
enzyme function of EAT09360 is likely to play a role
as an l-rhamnose-1-dehydrogenase (see below). On
the other hand, there was no homolog to the
EAT09363 gene in the LRA1–4 gene cluster:
COG0179, 2-oxopent-4-enoate hydratase (MhpD)
family. These results indicated that the gene cluster of
Sphingomonas sp. SKA58 may be responsible for non-
phosphorylative sugar metabolism, in which the sche-

matic conversion of the l-2-keto-3-deoxyacid-sugar
intermediate is different from the aldol cleavage by
l-KDR aldolase. In this study, we used Sphingomonas
sp. NBRC 101715 instead of Sphingomonas sp.
SKA58 as a target microorganism; between these,
there is 99.8% identity in 16S rRNA sequence. We
were successful in amplifying SpLRA genes by geno-
mic PCR using oligonucleotide primers designed from
the Sphingomonas sp. SKA58 genome sequence (see
below). Unless otherwise noted, Sphingomonas sp.
hereafter indicates strain NBRC 101715.
Functional expression of SpLRA genes in E. coli
Among the five SpLRA gene products, SpLRA3,
SpLRA5, and SpLRA6 were successfully expressed in
E. coli cells as (His)6-tagged enzymes. The recombi-
nant enzymes were purified to homogeneity using a
nickel-chelating affinity column and then gel filtration
chromatography (Fig. 3B). Western blot analysis with
S. Watanabe and K. Makino Novel L-rhamnose metabolic pathway
FEBS Journal 276 (2009) 1554–1567 ª 2009 The Authors Journal compilation ª 2009 FEBS 1557
an antibody against (His)6-tag confirmed the (His)6-
tag at the N-terminal.
L-Rhamnose-1-dehydrogenase (SpLRA1)
To initially estimate the physiological role of the LRA
gene cluster, we first attempted to biochemically char-
acterize a dehydrogenase with l-rhamnose in Sphingo-
monas sp. When compared with nutrient medium
(0.035 unitÆmg
)1
protein), approximate 9.7-fold higher

activity of NADP
+
-dependent dehydrogenation with
l-rhamnose was found in the cell-free extract from
Sphingomonas sp. cells grown on l-rhamnose (0.34 uni-
tÆmg
)1
protein). Similar induction of NAD
+
-dependent
activity by l-rhamnose was also found, although the
specific activity was slightly lower than the NADP
+
-
dependent activity. The l-rhamnose-1-dehydrogenase
was (partially) purified by four chromatographic steps:
there was NADP
+
-dependent and NAD
+
-dependent
specific activity of 39.8 and 31.7 unitÆmg
)1
protein,
respectively (Fig. 3A). A typical result of purification
is summarized in Table 1A. During the purification
procedure, the ratio of NADP
+
-linked and NAD
+

-
linked activity remained almost constant, suggesting
the presence of only one protein as l-rhamnose-1-
dehydrogenase. Two major protein bands were found
on SDS ⁄ PAGE gel (bands A and B, respectively, in
Fig. 3A). The N-terminal amino acid sequence up to
20 amino acids of band B, MKLLEGKTVLITGAST
GIGR, was completely identical to that of SpLRA1,
and the putative molecular mass of SpLRA1
(26.5 kDa) was similar to that of the native enzyme
( 28 kDa) (band A is described in the next section).
Significant dehydrogenase activity was observed with
l-rhamnose, l-lyxose (75%) and l-mannose (8.4%) in
the presence of NADP
+
[values in parentheses are
activity relative to that with l-rhamnose (100%)]. The
substrate specificity and dual coenzyme specificity
between NAD
+
and NADP
+
(see above) were similar
to the same bacterial AvLRA1 (high concomitant
activity for l-lyxose), compared with fungal PsLRA1
and DhLRA1 (strict NAD
+
-dependence) [12]. These
results indicated that SpLRA1 encodes NAD(P)
+

-
dependent l-rhamnose-1-dehydrogenase, and that the
91
65
48
37
28
M1 2 3kDakDa
195
119
91
65
48
37
28
12 34MM5
AB
20.5
Band A
Band B
Fig. 3. SDS ⁄ PAGE. (A) Purification of native L-rhamnose dehydro-
genase from Sphingomonas sp. Lane 1: cell-free extract (100 lg).
Lane 2: HiPrep 16 ⁄ 10 Q FF (50 lg). Lane 3: HiPrep 16 ⁄ 10 Butyl FF
(20 lg). Lane 4: hydroxyapatite (20 lg). Lane 5: HiLoad 26 ⁄ 60
Superdex 200 pg (20 lg). M is marker protein. Bands A and B
correspond to
L-KDR 4-dehydrogenase and L-rhamnose-1-dehydro-
genase, respectively (see text). (B) Purification of (His)6-tagged
SpLRA3 (lane 1), SpLRA5 (lane 2), and SpLRA6 (lane 3) (each of
5 lg).

Table 1. Summary of concomitant purification of L-rhamnose-1-dehydrogenase (A) and L-KDR 4-dehydrogenase (B) from Sphingomonas sp.
Step Total protein (mg)
Total activity (units)
Specific activity
a
(unitsÆmg
)1
protein)
Yield
(%)
Purification
fold
NAD
+
NADP
+
NADP
+
⁄ NAD
+
A
Cell-free extract 1325 654 821 1.26 0.620 100 1.0
HiPrep 16 ⁄ 10 Q FF 212 544 755 1.39 3.56 92 1.0
HiPrep 16 ⁄ 10 Butyl FF 28.8 437 535 1.22 18.6 65 4.4
CHT ceramic hydroxyapatite 13.8 202 274 1.36 19.9 33 12
HiLoad 26 ⁄ 60 Superdex 200 pg 3.72 118 148 1.25 39.8 18 12
Step
Total protein
(mg)
Total activity

(units)
Specific activity
(unitsÆmg
)1
protein)
Yield
(%)
Purification
fold
B
Cell-free extract 1325 263 0.198 100 1.0
HiPrep 16 ⁄ 10 Q FF 212 118 0.557 45 2.8
HiPrep 16 ⁄ 10 Butyl FF 28.8 49.2 1.71 19 8.6
CHT ceramic hydroxyapatite 13.8 30.5 2.21 12 11
HiLoad 26 ⁄ 60 Superdex 200 pg 3.72 20.4 5.48 8 28
a
NADP
+
-dependent activity.
Novel
L-rhamnose metabolic pathway S. Watanabe and K. Makino
1558 FEBS Journal 276 (2009) 1554–1567 ª 2009 The Authors Journal compilation ª 2009 FEBS
remaining LRA genes may also be related to l-rham-
nose metabolism.
L-2-Keto-3-deoxyrhamnonate-(4)-dehydrogenase
(SpLRA5)
As described above, although SpLRA5 belongs to the
SDR family, together with SpLRA1, the enzyme func-
tion may be different from the dehydrogenation with
l-rhamnose. At present, this protein family contains

approximately 3000 primary structures and consists of
enzymes of several EC classes [20], and relatively high
degrees of similarity to SpLRA5 were found in several
(putative) reductases ⁄ dehydrogenases (30–35% identi-
ties). Significant NAD
+
-dependent dehydrogenation
activity (0.22 unitÆmg
)1
protein) of l-KDR was found
when Sphingomonas sp. was grown on l-rhamnose, but
not when it was grown on nutrient medium. Further-
more, no aldolase activity of l-KDR was induced by
l-rhamnose. Surprisingly, the dehydrogenase for
l-KDR was concomitantly purified with l-rhamnose-
1-dehydrogenase: NAD
+
-dependent specific activity of
5.48 unitÆmg
)1
protein was found (Table 1B). In the
(partially) purified sample, band A should correspond
to this enzyme (Fig. 3A), and the N-terminal amino
acid sequence up to 19 amino acids was significantly
homologous with that of SpLRA5: (M)
SVFAGRYA
GRXAIVTGGAS (underlined letters indicate the same
amino acids as SpLRA5; X, residue was not deter-
mined). On the other hand, the molecular mass of the
native enzyme estimated from SDS ⁄ PAGE ( 29 kDa)

was slightly higher than the value estimated from the
putative amino acid sequence of SpLRA5 (25.7 kDa),
although this is not entirely clear. These results
suggested that SpLRA5 plays a role as a novel
NAD
+
-dependent l-KDR dehydrogenase involved in
l-rhamnose metabolism in Sphingomonas sp.
When the purified (His)6-tagged SpLRA5 was incu-
bated with each l-2-keto-3-deoxyacid-sugar in the
presence of NAD
+
, clear dehydrogenation activity
was detected by a spectrophotometric assay, and the
determined kinetic parameters are shown in Table 2.
The k
cat
⁄ K
m
value with l-KDR was 81 min
)1
Æmm
)1
in the range expected for the physiological substrate;
this was 9.3-fold and 214-fold higher than those
with l-2-keto-3-deoxylyxonate (l-KDL) and l-2-keto-
3-deoxymannonate (l-KDM), respectively, this being
mainly caused by the higher value of k
cat
. l-Rhamnose

was an inactive substrate, and no activity was found
in the presence of NADP
+
when l-KDR was used
as a substrate. To identify the hydrogen absorbed
by SpLRA5, an attempt was made to isolate the
reaction product free of NAD
+
, protein, and buffer,
but this proved to be unsuccessful, probably
because of the unstable nature of the product.
Therefore, the reaction product of SpLRA5 was esti-
mated by a coupling reaction with SpLRA6, as
described below.
The common active center of SDR enzymes consists
of a tightly conserved Ser-Tyr-Lys catalytic triad [20].
Furthermore, the coenzyme-binding mode follows a
classical ‘Rossmann fold’, in which a characteristic
GXXXGX[G ⁄ A] fingerprint motif exists. These motifs
are also conserved in SpLRA5: Ser143–Tyr156–Lys160
and Gly17-Gly-Ala-Ser-Gly-Leu-Gly23 (Fig. 4A).
These findings suggested that the fundamental catalytic
mechanism and coenzyme recognition of SpLRA5 may
be similar to those in known SDR enzymes. The
recombinant SpLRA5 enzyme formed a homotetra-
meric structure by itself, like other SDR proteins (data
not shown). Therefore, it is likely that SpLRA1 is
completely unnecessary to maintain the active form of
SpLRA5, and that concomitant purification of
SpLRA5 and SpLRA1 (Fig. 3A) is due to the similar

properties on the surface rather than the hetero-oligo-
meric structure, although their sequence identity is
only 29%.
L-2,4-Diketo-3-deoxyrhamnonate hydrolase
(SpLRA6)
SpLRA6 is a novel member of the MhpD family
(COG0179), which is different from the protein fami-
lies of LRA1–4 proteins (Fig. 1). In HPLC analysis
(Fig. 5A), the retention time of the reaction product of
SpLRA5 was almost the same as that of l-KDR
Table 2. Kinetic parameters of recombinant SpLRA5 protein. Values are the means ± standard deviation, n =3.
Substrate
Specific activity
a
(unitsÆmg
)1
protein) K
m
b
(mM) k
cat
b
(min
)1
) k
cat
⁄ K
m
b
(min

)1
ÆmM
)1
)
L-KDR 1.95 ± 0.04 0.646 ± 0.086 51.8 ± 2.6 81.0 ± 6.3
L-KDL 0.697 ± 0.025 1.27 ± 0.11 11.0 ± 0.4 8.72 ± 0.42
L-KDM 0.024 ± 0.001 1.90 ± 0.06 0.720 ± 0.024 0.378 ± 0.001
a
Under standard assay conditions as described in Experimental procedures.
b
Ten different concentrations of substrate between 0.2 and
10 m
M were used.
S. Watanabe and K. Makino Novel
L-rhamnose metabolic pathway
FEBS Journal 276 (2009) 1554–1567 ª 2009 The Authors Journal compilation ª 2009 FEBS 1559
( 10 min). Similar results were also observed with
l-KDL and l-KDM. On the other hand, when
l-KDR was incubated with SpLRA5 and SpLRA6 in
the presence of NAD
+
, a novel peak with a later
retention time (13.1 min) appeared, identical to that of
l-lactate. In the case of l-KDL, the peak corresponded
to glycolate. Clearer results were obtained with
l-KDM: two peaks that differed from that of l-KDM
( 10 and  11.4 min) were observed, and were found
to be identical to those of pyruvate and (dl-)glycerate,
respectively. These results indicated that SpLRA6 cata-
lyzes the hydrolysis of l-2,4-diketo-3-deoxyacid-sugar

C O O H
H H
H O H
H O H
C H
3
O
L-KDR L-KDF
C O O H
H H
H O H
H O H
C H
3
O
C O O H
H O H
H O H
H O H
H O H
H H
O H
1
2
3
4
5
6
D-Gluconate
EC 1.1.1.215

D-2-Deoxygluconate
C O O H
H H
H O H
H O H
H O H
H H
O H
C O O H
H H
H O H
H O H
H H
O H
O
D-KDG
EC 1.1.1.126 EC 1.1.1.127
(KduD)
L-Rhamnose
C O O H
H O H
H O H
H O H
H O H
H H
O H
C O O H
H H
H O H
H O H

H H
O H
O
H
O
H
O H
H
O H
H
O H
H
3
C
H O
EC 1.1.1.264
(IdnD)
C O O H
H O H
H O H
H O H
H
H H
O H
H O
L-Idonate
C O O H
H
H H
O

O H
O
H O H
C H
3
C O O H
H H
O
H O H
C H
3
O
H
2
O
L-DKDR
hydrolase
(EC 3.7.1 )
Pyruvate
L-Lactate
L-DKDR
C O O H
H H
O
H H
O
H
C O O H
H
C O O H

H H
O
H H
H
C O O H
H
C O O H
H
H
2
O
FAH
(EC 3.7.1.2)
4-Fumarylacetoacetate
Fumarate
Acetoacetate
C O O H
O
H
H
H H
C O O H
H
H
C O O H
O
H
C O O H
H H
C O O H

H
H
CO
2
HpcE
(EC 4.1.1.68)
5-Oxopent-3-ene-
1,2,5-tricarboxylate
2-Oxohept-3-
ene-1,7-dioate
C O O H
H H
C H
2
O H
H O H
O
C O O H
H H
C H O
H H
O
H
2
O
D-KDA
dehydratase
(EC 4.2.1 )
D-KDA αKGSA
C H

2
H
H
H O
C O O H
H O
H
H O
C O O H
C H
3
H
2
O
MhpD
(EC 4.2.1.80)
4-Hydroxy-2-
oxopentanoate
2-Hydroxy-2,4-
pentadienoate
EC 1.1.1.69
(IdnO/Gno)
EC 1.1.1.125
(KduD)
EC 1.1.1.173
D W E V E L G
D W E V E L G
L P E P E L A
H Y E A E L V
D M E L E M A

R I E A E I A
N D V S E R F N Q
N D L S E R E F Q
D D V S A R D L E
N D Y A I R D Y L
N D W S A R D I Q
L E V V G S - R I
p L R A 6
u c E
d a D
S
F
K
H p c E
A H F
M h p D
W 1 1 7 K
W 1 2 2 K
L 1 4 1 K
N 2 7 4 K
P 1 9 7 K
T 1 0 3 N
W 1 4 9 K
W 1 5 4 K
L 1 6 3 K
N 3 0 6 K
P 2 3 2 K
T 1 3 7 N
K Q - - -
L E - - -

A E N - -
E N - - -
Q W E Y V
R D W S I
W S K G K G H D T
W V K G K S A D T
L P Q S K I Y A G
N L R V K S R D G
P F L G K S F - -
T V A D N A S C G
P G D L M I T G T P P G V G
P G D V I S T G T P P G V G
D G T I L T T G T A I V P G
P G D M I A T G T P K G L S
P G D L L A S G T I S G S D
T G D I I L T G A L G P M V
W 2 3 0 K
W 2 3 5 K
L 2 4 8 K
N 3 8 7 K
P 3 4 2 K
T 2 2 4 N
R G T Q
H G G Q
- P L Y
Y Y R P
- P L G
Q F V D





A v L R A 1
S p L R A 1
S p L R A 5
F u c D
I d n O
K d u D
V S S I
V S S I
L A S V
S A L V G G A M Q T H Y T P T K A G
S A L V G G E Y Q T H Y T P T K A G
A G K E G N P N A S A Y S A S K A G
M S S V A S S I K G V P N R F V Y G V T K A A
I C S V Q
I A S M L
S
S
E L A R P G I A P Y T A T K G A
F Q G G I P V P S Y T A S K K R
-
-
-
-
-
G

G
1 4 4 A
Q 1 4 2 A
D 1 4 1 V
D 1 4 1 T
L 1 4 5 T
R 1 4 3 S
V T G A S R G I G R A
V T G A S T G I G R A
V T G G A S G L G K Q
I T A A G A G I G R E
V T G A S R G I G L T
I T G C D T G L G Q G
0
1 1
1 5
2 0
1 6
1 5
1
1
A
B
C
D
* * * * * * *
Fig. 4. (A) Partial sequence alignment between L-rhamnose-1-dehydrogenases from A. vinelandii (AvLRA1) and Sphingomonas sp. (SpLRA1,
this study),
L-KDR 4-dehydrogenase from Sphingomonas sp. (SpLRA5, this study), L-KDF 4-dehydrogenase from X. campestris (FucD), D-glu-
conate 5-dehydrogenase from Gluconobacter oxydans (IdnO, CAA56322), and

D-KDG 5-dehydrogenase from Erwinia chrysanthemi (KduD,
CAA43989). Open and closed circles indicate NAD(H)
+
-dependent and NADP(H)
+
-dependent enzymes, respectively. Asterisks, GXXXGX(G ⁄ A)
coenzyme-binding motif of Rossmann fold; diamond, Ser-Tyr-Lys catalytic triad. (B) Structurally analogous nonphosphorylated substrate to
L-KDR. Each dehydrogenase acts at the gray-shadowed boxes. Enzymes in boxes belong to the SDR protein family as described in (A). (C)
Partial sequence alignment between
L-DKDR hydrolases from Sphingomonas sp. (SpLRA6, this study) and X. campestris (FucE), D-KDA
dehydratase from S. solfataricus (KdaD; Protein Data Bank ID 2Q18), HpcE from E. coli (1I7O), FAH from mouse (1QQJ), and MhpD from
E. coli (1SV6). Circles, metal ion ligands; triangles, active sites. The full (structure-based) sequence alignment is shown in Fig. S2. (D)
Schematic reactions of MhpD family enzymes. Black triangles indicate cleavage sites of C–C bonds.
Novel
L-rhamnose metabolic pathway S. Watanabe and K. Makino
1560 FEBS Journal 276 (2009) 1554–1567 ª 2009 The Authors Journal compilation ª 2009 FEBS
to pyruvate and hydroxyl acid [physiologically,
l-2,4-diketo-3-deoxyrhamnonate (l-DKDR) hydrolase]
and that SpLRA5 should be assigned as ‘l-KDR
4-dehydrogenase’, which produces l-DKDR from
l-KDR.
The MhpD family contains the archetypal MhpD
(EC 4.2.1.80) [21] fumarylacetoacetate hydrolase (FAH;
EC 3.7.1.2) [22], 5-oxopent-3-ene-1,2,5-tricarboxy-
late decarboxylase ⁄ 2-hydroxyhepta-2,4-diene-1,7-dioate
isomerase (EC 4.1.1.68) [23], and d-2-keto-3-deoxyara-
binonate (d-KDA) dehydratase [14,24] (Fig. 4C; the full
sequence alignment is shown in Fig. S2). Among them,
FAH catalyzes the hydrolytic cleavage of a C–C bond in
fumarylacetoacetate to yield fumarate and acetoacetate,

and the catalytic reaction is partially analogous to that
of l-DKDR hydrolase (Fig. 4D). Furthermore, it is
noteworthy that d-KDA dehydratase is involved in an
archeal d-arabinose pathway that is (partially)
analogous to the npED pathway and similar to the alter-
native l-rhamnose pathway (E73) (Fig. 1). MhpD
family enzymes contain Ca
2+
(5-oxopent-3-ene-1,2,5-
tricarboxylate decarboxylase ⁄ 2-hydroxyhepta-2,4-diene-
1,7-dioate isomerase and FAH) or Mg
2+
(d-KDA
dehydratase) in the active center; these are coordinated
with two highly conserved glutamates and one aspartate
(Fig. 4C). l-DKDR hydrolase also possesses a structur-
ally equivalent metal ion-binding site, Glu119–Glu121–
Asp150, whereas there are several significant variations
C O O H
H H
H O H
H O H
C H
3
O
C O O H
H H
O
H O H
C H

3
O
C O O H
H O H
C H
3
+
NAD
+
NADH H2O
L-Lactate
C O O H
C H
3
O
L-KDR
Pyruvate
L-2,4-Diketo-
3-deoxyrhamnonate
C O O H
O
H H
H O H
C H
2
O H
C O O H
C
+
Glyocolate

NAD
+
NADH H2O
L-KDL
C O O H
C H
3
O
H
2
O H
L-2,4-Diketo-
3-deoxylyxonate
C O O H
H H
C H
2
O H
O
O
C O O H
H H
H O H
H O H
C H
2
O H
O
C O O H
H H

O
H O H
C H
2
O H
O
C O O H
H O H
C H
2
O H
C O O H
C H
3
O
+
Glycerate
NAD
+
NADH H2O
L-KDM
L-2,4-Diketo-
3-deox
y
mannonate
0
200
0
200
0

200
0
200
400
0
200
0
200
0
200
400
8 9 10 11 12 13 14
0
200
0
200
0
200
400
8 9 10 11 12 13 14
L-KDR + SpLRA5 + NAD
+
L-KDR + SpLRA5 + NAD
+
+ SpLRA6
L-KDR
L-KDL + SpLRA5 + NAD
+
L-KDL + SpLRA5 + NAD
+

+ SpLRA6
L-KDM + SpLRA5 + NAD
+
L-KDM + SpLRA5 + NAD
+
+ SpLRA6
0
200
Pyruvate + Glycolate
8 9 10 11 12 13 14
Pyruvate + L-Lactate
Retention time (min)
mV mV mV
Retention time (min) Retention time (min)
mV
L-KDL L-KDM
0
200
Pyruvate + DL-Glycerate
A
B
Fig. 5. (A) HPLC analysis of the reaction products from L-KDR, L-KDL and L-KDM formed by SpLRA5 and SpLRA6. Authentic pyruvate, L-lac-
tate, glycolate and
DL-glycerate were present at a concentration of 10 mM. Arrows indicate the peak corresponding to pyruvate produced
from
L-KDM. (B) Schematic of enzyme reaction products from L-KDR, L-KDL and L-KDM formed by L-KDR 4-dehydrogenase (LRA5) and
L-DKDR hydrolase (LRA6). Pyruvate is shown in the dashed box.
S. Watanabe and K. Makino Novel
L-rhamnose metabolic pathway
FEBS Journal 276 (2009) 1554–1567 ª 2009 The Authors Journal compilation ª 2009 FEBS 1561

in several amino acids in the (putative) active sites,
which may reflect their different catalytic reactions.
L-Rhamnonate dehydratase (SpLRA3)
As AvLRA3 utilizes only l-rhamnonate, l-lyxonate
and l-mannonate as substrates efficiently [13], other
(dl-)2-keto-3-deoxyacid-sugars (and 2-diketo-3-deoxy-
acid-sugars) in addition to l-KDR, l-KDL and
l-KDM are unavailable as substrates for SpLRA5 and
SpLRA6. Therefore, there is still a possibility that the
gene cluster of Sphingomonas sp. is responsible for the
metabolism not only of l-rhamnose but also of other
sugars. In this regard, SpLRA3 is more suitable for
estimating the physiological role of the gene cluster as
well as of SpLRA1, because a library was constructed
previously of 11 acid-sugars as potential substrates for
acid sugar dehydratases [12].
Therefore, recombinant (His)6-tagged SpLRA3 was
prepared using the same procedures as for SpLRA5
and SpLRA6 (Fig. 3B). By semicarbazide endpoint
measurement, significant activity of SpLRA3 was
found with l-rhamnonate, l-lyxonate (128%), and
l-mannonate (89.2%) [values in parentheses are rela-
tive to the activity with l-rhamnonate (100%)].
Furthermore, SpLRA3 showed similar kinetic para-
meters with l-rhamnonate to those of AvLRA3 [12]
(Table 3), and the k
cat
⁄ K
m
value with l-rhamnonate

(177 min
)1
Æmm
)1
) was 26.2- and 59.2-fold higher than
those with l-lyxonate (6.75 min
)1
Æmm
)1
) and l-manno-
nate (2.99 min
)1
Æmm
)1
), respectively, mainly due to
significantly higher K
m
values (13.6- and 30.2-fold,
respectively). These results clearly suggested that
SpLRA3 should be assigned as an ‘l-rhamnonate
dehydratase’ and that the gene cluster of Sphingomonas
sp. is related only to l-rhamnose metabolism.
In vivo expression of LRA genes in an
L-rhamnose-defective E. coli mutant
We here identified an alternative l-rhamnose pathway
in Sphingomonas sp. (referred to as the ‘diketo-hydro-
lase pathway’) that differs from the complete analo-
gous pathway to npED pathway (aldolase pathway)
(Fig. 2A). To estimate the physiological meaning of
both pathways of l-rhamnose metabolism in vivo,

genes LRA1–6 were introduced into plasmid vectors
for multiple gene expression and transformed into an
l-rhamnose-defective E. coli mutant, the KRX strain
(Fig. S1 and Table S1). In this study, the genes for
AvLRA1, DhLRA2, AvLRA3, PsLRA4, AvLRA4,
SpLRA5 and SpLRA6 were used, because of expres-
sion level and ⁄ or solubility problems with other LRA
proteins in E. coli cells (see Experimental procedures).
Western blot analysis using (His)6-tag attached to the
N-terminus of all LRA proteins revealed their func-
tional expression in E. coli cells grown on a nutrient
medium supplemented with 0.2% (w ⁄ v) l-rhamnose
(Fig. 6A). On the other hand, when the recombinant
E. coli strains were cultivated in a minimal medium
containing 2% (w ⁄ v) l-rhamnose, the l-rhamnose-neg-
ative phenotype was compensated for by introduction
of the genes LRA1–4 [Duet-1234(Ps) and Duet-
1234(Av)] or LRA1–3, LRA5, and LRA6 (Duet-12356)
(Fig. 6B), suggesting significant physiological roles of
both of the alternative l-rhamnose pathways in vivo.
Unexpectedly, the introduction of only LRA1 and
LRA2 also led to slow growth of the cells (see Duet-
12). Indeed, E. coli possesses genes homologous to
LRA3 and LRA4 (that encoding yfaW, 63.4% identity
with AvLRA3; that encoding yfaU, 50.2% identity
with AvLRA4) (Fig. 2B), and their enzyme functions
were recently assigned as l-rhamnonate dehydratase
and l-KDR aldolase, respectively [25,26].
Discussion
As illustrated in Fig. 1, there is significant phylogenetic

mosaicism between the metabolic enzymes involved in
sugar pathways analogous to the ED pathway: sugar
dehydrogenase, lactone-sugar hydrolase (lactonase),
acid-sugar dehydratase, and aldolase and dehydratase
for (dl-)2-keto-3-deoxyacid-sugar. One of the most
interesting findings in this study was that two enzymes
belonging to the same protein family (COG1028) are
involved in a single sugar metabolic pathway: LRA1
Table 3. Kinetic parameters of recombinant SpLRA3 protein. Values are the means ± standard deviation, n =3.
Substrate
Specific activity
a
(unitsÆmg
)1
protein) K
m
(mM) k
cat
(min
)1
)
k
cat
⁄ K
m
(min
)1
ÆmM
)1
)

L-Rhamnonate
b
0.401 ± 0.03 (0.891 ± 0.058)
c
0.121 ± 0.006 (0.115 ± 0.001) 21.3 ± 0.6 (43.1 ± 0.2) 177 ± 5 (375 ± 3)
L-Lyxonate
c
0.196 ± 0.02 1.64 ± 0.11 11.0 ± 0.5 6.75 ± 0.14
L-Mannonate
d
0.134 ± 0.01 3.66 ± 0.67 10.8 ± 1.4 2.99 ± 0.16
a
Under standard assay conditions as described in Experimental procedures.
b
Ten different concentrations of substrate between 0.02 and
1m
M were used.
c
Values of AvLRA3 [12].
d
Six different concentrations of substrate between 0.5 and 5 mM were used.
Novel
L-rhamnose metabolic pathway S. Watanabe and K. Makino
1562 FEBS Journal 276 (2009) 1554–1567 ª 2009 The Authors Journal compilation ª 2009 FEBS
and LRA5. In the ‘recruitment model’ of enzyme evo-
lution proposed by Jensen [27], new enzymes evolve by
duplication and mutation of the same enzyme classes
from other pathways, leading to ‘catalytically promis-
cuous’ enzymes and ‘patchwork’-like metabolic path-
ways. In case of the diketo-hydrolase pathway of

Sphingomonas sp., similar evolutionary processes have
occurred more than once in the same pathway.
The ED pathway often plays a role as the metabolic
funnel of the npED pathway. For example, in several
hyperthermophilic Archaea, d-2-keto-3-deoxygluconate
(d-KDG) and glycerate (produced from d-glyceralde-
hyde) are phosphorylated by specific kinases and
subsequently metabolized through the ED pathway: the
so-called ‘semi-phosphorylative ED pathway’ [28].
On the other hand, in several bacterial pathways,
d-gluconate and ⁄ or d-KDG is produced from d-2,5-di-
keto-3-deoxygluconate (in pectin degradation [29]), d-5-
dehydrogluconate (in the l-idonate pathway [30]), and
d-2-deoxygluconate (in the 2-deoxyglucose pathway
[31]), and the (reverse) reactions, in particular the first
reaction, are analogous to that of SpLRA5. Therefore,
it may be reasonable to suppose that l-KDR 4-dehydro-
genase is phylogenetically similar to the enzymes cata-
lyzing these reactions [29,32] (Fig. 4A,B).
To the best of our knowledge, the conversion of
l-KDR into pyruvate and l-lactate via l-DKDR is a
novel metabolic fate in the known sugar pathways.
Gerlt et al. [33] recently identified a gene cluster related
to the alternative l-fucose metabolism of the bacterium
Xanthomonas campestris (E37–E41) (Fig. 2B). Among
the protein products of the five metabolic genes, FucB,
FucC, FucD, and FucE are homologous to SpLRA2
(28% identity), SpLRA3 (30% identity), SpLRA5
(35% identity; see Fig. 4A), and SpLRA6 (56%
identity, see Fig. 4D), respectively, suggesting that

l-DKDR is also produced from l-2-keto-3-deoxyfuco-
nate (l-KDF) in this pathway. On the other hand, as
l-fucose and l-fuconate are inactive substrates for
SpLRA1 and SpLRA3, respectively (see Results), the
LRA gene cluster should be related only to l-rhamnose
metabolism of Sphingomonas sp. but not l-fucose
metabolism; in contrast, the archaeon Sulfolobus
solfataricus metabolizes both d-glucose and d-galactose
promiscuously through the npED pathway [4,5].
l-Lactaldehyde is subsequently converted to l-lactate
by l-lactaldehyde dehydrogenase in the aldolase path-
way [13], whereas the continuous reactions of l-KDR
4-dehydrogenase and l-DKDR hydrolase allow the
metabolism of l-rhamnose into the same pyruvate and
l-lactate products without involvement of the physio-
logically toxic aldehyde (Fig. 2A), by which the diketo-
hydrolase pathway may be more favorable than the
aldolase pathway.
Another interesting insight obtained in this study is
that l-DKDR hydrolase belongs to the same MhpD
protein family as d-KDA dehydratase (Fig. 4C). As
mentioned previously, there is another version of
the npED pathway (type II in Fig. 1), in which the
dl-2-keto-3-pentonate intermediate is converted to
50
30
0 1 12 123 1235 12356 M
1
1
2

1
2
3
1
2
3
4
1
3
4
2
1
3
5
2
3
1
5
2
6
1234(Ps) 1234(Av) kDa
E. coli Duet strains
A
B
0
0.2
0.4
0.6
0.8
1

1.2
1.4
0 2 4 6 8 10 12
A600
Da
y
s
1234(Av)
1234(Ps)
12356
12
0
Fig. 6. Expression of LRA genes in an
L-rhamnose-defective mutant of E. coli. The
constructed recombinant E. coli strains are
summarized in Table S3. (A) Western blot
analysis. All LRA enzymes were overexpres-
sed as (His)6-tagged proteins in E. coli cells
grown in a nutrient medium supplemented
with
L-rhamnose, and purified on an
Ni
2+
-chelating affinity column. One hundred
micrograms of each of the purified proteins
was applied to 11% (w ⁄ v) gel. M indicates
marker protein. (B) Growth in M9 minimal
liquid medium supplemented with 2% (w ⁄ v)
L-rhamnose.
S. Watanabe and K. Makino Novel

L-rhamnose metabolic pathway
FEBS Journal 276 (2009) 1554–1567 ª 2009 The Authors Journal compilation ª 2009 FEBS 1563
a-ketoglutarate via aKGSA by dl-2-keto-3-deoxypent-
onate dehydratase and aKGSA dehydrogenase. d-KDA
dehydratase is involved in this type of archeal d-arabi-
nose metabolism (E73 [14,24]), and the l-arabinose and
d-xylose pathways of bacteria and ⁄ or Archaea also
contain the homolog (E69, E78, and E83). It was found
previously that, in spite of the different pathways of
catalysis, dl-2-keto-3-deoxyacid-sugar aldolases (E7
and E15) and dehydratases (E51, E54, E59, and E64)
belong to the same dihydrodipicolinate synthase ⁄
N-acetylneuraminate lyase protein family (COG0329)
(Fig. 1 [12]). With respect to l-DKDR hydrolase and
d-KDA dehydratase, the former is not an aldolase
enzyme but produces pyruvate by hydrolyzing cleavage
of the C–C bond (Fig. 5D–F). Therefore, not only aldo-
lases and dehydratases with dl-2-keto-3-deoxyacid-
sugar but also l-DKDR hydrolase can be assigned as
equivalent within the analogous schematic conversions
to the ED pathway (Fig. 1).
In this article, we identified a novel metabolic fate of
dl-2-keto-3-deoxyacid-sugar intermediate to pyruvate
and hydroxyl-acid through a continuous reaction of
dehydrogenase and hydrolase, which is referred to as
‘type III of nonphosphorylated sugar pathways’
(Fig. 1). The significant insights about metabolic genes
illustrated in Fig. 1 (modified from a previous study
[12]) will be helpful in identifying unknown sugar path-
ways analogous to the ED and npED pathways.

Experimental procedures
Identification of a putative gene cluster related to
alternative
L-rhamnose metabolism
Amino acid sequence homology searches using the Micro-
bial Genomic blast program distributed by the National
Center for Biotechnology Information (NCBI) was carried
out against bacterial and archeal genome sequences using
the metabolic genes involved in an alternative l-rhamnose
pathway of P. stipitis, D. hansenii and A. vinelandii LRA1–
4. Candidate genes showing significant homology were fur-
ther examined by estimating whether these enzymes belong
to the same protein family as the probe proteins in the
flanking region.
Functional expression and purification of
(His)6-tagged proteins
Sphingomonas sp. NBRC 101715 was purchased from the
National Institute of Technology and Evaluation (Chiba,
Japan). As described in Results, DNA sequences of five
genes from Sphingomonas sp. SKA58 (GenBank accession
numbers EAT09360, EAT09365, EAT09362, EAT09364,
and EAT09363) were used to design genomic PCR primers
for the amplification of a DNA fragment of SpLRA1–3,
SpLRA5, and SpLRA6, respectively (Table S2). Each
amplified DNA fragment was introduced into the BamHI–
PstI sites in pQE-80L (Qiagen, Tokyo, Japan), a plasmid
vector for introducing an N-terminal (His)6-tag to
expressed proteins.
E. coli DH5a harboring the expression plasmid for (His)6-
tagged enzymes (SpLRA3, SpLRA5, and SpLRA6) was

grown at 37 °C to an absorbance of 0.6 at 600 nm in Super
broth medium containing 50 mgÆL
)1
ampicillin. The culture
was rapidly cooled on ice and further incubated for 24 h at
15 °C after the addition of 1 mm isopropyl thio-b-d-galacto-
side. Cells were harvested and resuspended in buffer A
[50 mm sodium phosphate buffer (pH 7.0) containing 1 mm
MgCl
2
, 300 mm NaCl, 10 mm imidazole, and 10% (v ⁄ v)
glycerol]. The cells were disrupted by sonication, and the
solution was centrifuged at 39 120 g for 1 h. The supernatant
was loaded onto an Ni
2+
–nitrilotriacetic acid Superflow col-
umn (Qiagen) equilibrated with buffer A. The column was
washed with buffer B (pH 7.0; buffer A containing 50 mm
imidazole instead of 10 mm imidazole). The enzymes were
then eluted with buffer C (pH 7.0; buffer A containing
250 mm imidazole). The elutant was concentrated by ultrafil-
tration with a Centriplus YM-30 (Millipore, Bedford, MA,
USA) and loaded onto a HiLoad 16 ⁄ 60 Superdex 200 pg
(1.6 · 60 cm; Amersham Biosciences, Uppsala, Sweden) col-
umn equilibrated with buffer D (buffer A without NaCl or
imidazole). Fractions corresponding to the main peak were
pooled, concentrated, and dialyzed against buffer D [con-
taining 50% (v ⁄ v) glycerol], and stored at )35 °C until use.
Enzyme assays
All enzyme assays were performed at 25 °C. Spectrophoto-

metric analysis was carried out using a Jasco spectropho-
tometer model V-550 (JASCO, Tokyo, Japan) linked to an
ETC-505T Temperature Controller (JASCO). The kinetic
parameters K
m
and k
cat
were calculated by Lineweaver–
Burk plot. Protein concentrations were determined by the
method of Lowry et al. [34], with BSA as the standard.
l-Rhamnose-1-dehydrogenase and l-rhamnonate dehydra-
tase assays were performed by the methods described previ-
ously [12]. l-KDR 4-dehydrogenase activity was assayed
routinely in the direction of l-KDR oxidation by measuring
the reduction of NAD
+
at 340 nm. The standard assay
mixture contained 1 mml-KDR in 50 mm Hepes ⁄ NaOH
(pH 8.0) buffer. The reaction was started by the addition of
10 mm NAD
+
solution (100 lL), with a final reaction vol-
ume of 1 mL. l-KDR, l-KDL and l-KDM were synthe-
sized enzymatically from l-rhamnonate, l-lyxonate, and
l-mannonate, respectively, using AvLRA3 (l-rhamnonate
dehydratase), according to the method described previously
[12]. The l-DKDR hydrolase assay was performed in
50 mm Hepes ⁄ NaOH (pH 8.0) buffer containing 10 mm
Novel L-rhamnose metabolic pathway S. Watanabe and K. Makino
1564 FEBS Journal 276 (2009) 1554–1567 ª 2009 The Authors Journal compilation ª 2009 FEBS

l-KDR, 1 unit of l-KDR 4-dehydrogenase (SpLRA5),
10 mm NAD
+
, and purified SpLRA6. Reaction products
were estimated by HPLC as described in the next section
by reference to the authentic reagents.
Reaction product identification by HPLC
The reaction products were identified by HPLC with a
Multi-Station LC-8020 model II system (TOSOH, Tokyo,
Japan). The reaction solution consisted of 50 mm Hepes ⁄
NaOH (pH 8.0), 10 mm substrate (l-KDR, l-KDL, or
l-KDM), and 10 mm NAD
+
. The prepared recombinant
SpLRA5 and SpLRA6 enzymes were gel filtrated using a
HiLoad 16 ⁄ 60 Superdex 200 pg column with 50 mm
Hepes ⁄ NaOH (pH 8.0), and 10 lg of each eluted protein
was added to the reaction solution. Samples were applied
at 35 °C to an Aminex HPX-87H Organic Analysis column
(300 · 7.8 mm; Bio-Rad, Hercules, CA, USA) linked to an
RID-8020 refractive index detector (TOSOH), and eluted
with 5 mm H
2
SO
4
at a flow rate of 0.6 mLÆmin
)1
. If neces-
sary, 12% (w ⁄ v) trichloroacetic acid was added to the sam-
ples (0.1 v) to remove proteins. After filtration, 100 lLof

this solution was analyzed.
Complementation of an L-rhamnose-defective
E. coli mutant by LRA genes
To express multiple LRA genes involved in l-rhamnose
metabolism in E. coli cells, three plasmid vectors, pCO-
LADuet, pETDuet, and pACYCDuet (Novagen, Madison,
WI, USA), and E. coli KRX (Promega, Tokyo, Japan), an
l-rhamnose-defective mutant, were used. We frequently
encounter expression level and ⁄ or solubility problems in
our attempts to express LRA proteins in E. coli (this study
and [12]). For example, among four l-rhamnono-c-lacton-
ase enzymes from P. stipitis, D. hansenii, A. vinelandii and
Sphingomonas sp., only DhLRA2 was successfully expressed
as a soluble form in E. coli cells, despite efforts involving
other plasmid vectors except pQE-80L and growth condi-
tions including protein induction at low temperature.
Therefore, to express multiple LRA genes in a single cell of
E. coli in this study, we used AvLRA1,DhLRA2 ,AvLRA3,
PsLRA4,AvLRA4,SpLRA5, and SpLRA6.
All LRA genes were expressed as N-terminal (His)6-tagged
proteins (Doc. S1, Fig. S1, and Tables S2 and S3). The
purified plasmid constructs and ⁄ or empty vectors were
mixed appropriately and used to transform E. coli
KRX (Table S3). Kanamycin (50 lgÆ mL
)1
), ampicillin
(50 lgÆmL
)1
), chloramphenicol (34 lg Æ mL
)1

) and 1 mm
isopropyl thio-b-d-galactoside were added as necessary. To
estimate the protein expression level, the recombinant E. coli
strains were cultivated overnight in Super broth medium
(100 mL) supplemented with 0.2% (w ⁄ v) l-rhamnose. This
experiment was performed at 20 °C, because of no expres-
sion of SpLRA5 and SpLRA6 at 37 °C (data not shown).
The expressed (His)6-tagged proteins were purified using an
Ni
2+
–nitrilotriacetic acid Spin Column (Qiagen), using the
same buffer system as for the purification of SpLRA pro-
teins, and subjected to western blot analysis (see next sec-
tion). Alternatively, the recombinant E. coli strains were also
grown at 20 °C in M9 minimal liquid medium [6 g of
Na
2
HPO
4
,3gofKH
2
PO
4
, 0.5 g of NaCl, 1 g of NH
4
Cl,
1mm MgSO
4
, 0.1 mm CaCl
2

, and 0.001% (w ⁄ v) thiamine]
supplemented with 2% (w ⁄ v) l-rhamnose.
Western blot analysis
Protein samples were separated by SDS ⁄ PAGE, and the
proteins on the gels were transferred to a nitrocellulose
membrane (Hybond-ECL; Amersham Biosciences). Western
blot analysis was carried out using an ECL Western Blot-
ting Analysis System (Amersham Biosciences), RGSÆHis
horseradish peroxidase antibody [horseradish peroxidase-
fused mouse monoclonal antibody against Arg-Gly-Ser-
(His)6 in the N-terminal additional peptide of the expressed
recombinant proteins (Qiagen)] and 6· His Protein Ladder
(Qiagen) as a marker.
Amino acid sequence alignment
The amino acid sequences were aligned using the program
clustal w, distributed by GenomeNet (Bioinformatics
Center, Kyoto University, Japan) ().
Acknowledgements
This work was supported by Grants-in-aid for Young
Scientists (B), 18760592, from the Ministry of Educa-
tion, Culture, Sports, Science and Technology of Japan
(to S. Watanabe), the Fermentation and Metabolism
Research Foundation, the Japan Bioindustry Associa-
tion (to S. Watanabe), the Research Foundation of the
Association for the Progress of New Chemistry (to
S. Watanabe), the New Energy and Industrial Tech-
nology Development Organization (to S. Watanabe),
and CREST, Japan Science and Technology Agency
(to K. Makino). We thank Y. Takada and T. Hirose,
Hokkaido University, for their help in the determina-

tion of amino acid sequences.
References
1 Egan SM & Schleif RF (1993) A regulatory cascade in
the induction of rhaBAD. J Mol Biol 234, 87–98.
2 Baldoma
´
L & Aguilar J (1987) Involvement of lactalde-
hyde dehydrogenase in several metabolic pathways of
Escherichia coli K12. J Biol Chem 262, 13991–13996.
3 Hacking AJ, Aguilar J & Lin EC (1978) Evolution of
propanediol utilization in Escherichia coli: mutant with
S. Watanabe and K. Makino Novel L-rhamnose metabolic pathway
FEBS Journal 276 (2009) 1554–1567 ª 2009 The Authors Journal compilation ª 2009 FEBS 1565
improved substrate-scavenging power. J Bacteriol 136,
522–530.
4 Lamble HJ, Heyer NI, Bull SD, Hough DW & Danson
MJ (2003) Metabolic pathway promiscuity in the archa-
eon Sulfolobus solfataricus revealed by studies on glu-
cose dehydrogenase and 2-keto-3-deoxygluconate
aldolase. J Biol Chem 278, 34066–34072.
5 Lamble HJ, Milburn CC, Taylor GL, Hough DW &
Danson MJ (2004) Gluconate dehydratase from the
promiscuous Entner–Doudoroff pathway in Sulfolobus
solfataricus. FEBS Lett 576, 133–136.
6 Elshafei AM & Abdel-Fatah OM (2001) Evidence for a
non-phosphorylated route of galactose breakdown in
cell-free extracts of Aspergillus niger. Enzyme Microb
Technol 29, 76–83.
7 Dahms AS & Anderson RL (1972) D-Fucose metabo-
lism in a pseudomonad. I. Oxidation of D-fucose to

D-fucono-d-lactone by a D-aldohexose dehydrogenase.
J Biol Chem 247, 2222–2227.
8 Dahms AS & Anderson RL (1972) D-Fucose metabo-
lism in a pseudomonad. II. Oxidation of D-fucose to
D-fucono-d-lactone by an L-arabino-aldose dehydroge-
nase and hydrolysis of the lactone by a lactonase. J Biol
Chem 247, 2228–2232.
9 Dahms AS & Anderson RL (1972) D-Fucose metabo-
lism in a pseudomonad. III. Conversion of D-fuconate
to 2-keto-3-deoxy-D-fuconate by a dehydratase. J Biol
Chem 247, 2233–2237.
10 Dahms AS & Anderson RL (1972) D-Fucose metabo-
lism in a pseudomonad. IV. Cleavage of 2-keto-3-
deoxy-D-fuconate to pyruvate and D-lactaldehyde by
2-keto-3-deoxy-L-arabonate aldolase. J Biol Chem 247,
2238–2241.
11 Dahms AS & Anderson RL (1969) 2-Keto-3-deoxyl-L-
arabonate aldolase and its role in a new pathway of
L-arabinose degradation. Biochem Biophys Res Commun
36, 809–814.
12 Watanabe S, Saimura M & Makino K (2008) Eukary-
otic and bacterial gene clusters related to an alternative
pathway of non-phosphorylated L-rhamnose metabo-
lism (2008). J Biol Chem 283, 20372–20382.
13 Watanabe S, Piyanart S & Makino K (2008) Metabolic
fate of L-lactaldehyde derived from an alternative
L-rhamnose pathway. FEBS J 275, 5139–5149.
14 Brouns SJ, Walther J, Snijders AP, van de Werken HJ,
Willemen HL, Worm P, de Vos MG, Andersson A,
Lundgren M, Mazon HF et al. (2006) Identification of

the missing links in prokaryotic pentose oxidation path-
ways: evidence for enzyme recruitment. J Biol Chem
281, 27378–27388.
15 Watanabe S, Kodaki T & Makino K (2006) Cloning,
expression and characterization of bacterial L-arabi-
nose 1-dehydrogenase involved in an alternative path-
way of L-arabinose metabolism. J Biol Chem 281,
2612–2623.
16 Watanabe S, Kodaki T & Makino K (2006) A novel
a-ketoglutaric semialdehyde dehydrogenase: evolution-
ary insight into an alternative pathway of bacterial
L-arabinose metabolism. J Biol Chem 281, 28876–
28888.
17 Watanabe S, Shimada N, Tajima K, Kodaki T & Maki-
no K (2006) Identification and characterization of
L-arabonate dehydratase, L-2-keto-3-deoxyarabonate
dehydratase and L-arabinolactonase involved in an
alternative pathway of L-arabinose metabolism: novel
evolutionary insight into sugar metabolism. J Biol Chem
281, 33521–33536.
18 Stephens C, Christen B, Fuchs T, Sundaram V,
Watanabe K & Jenal U (2007) Genetic analysis of a
novel pathway for D-xylose metabolism in Caulobacter
crescentus. J Bacteriol 189, 2181–2185.
19 Watanabe S, Yamada M, Ohtsu I & Makino K (2007)
a-Ketoglutaric semialdehyde dehydrogenase isozymes
involved in metabolic pathways of D-glucarate,
D-galactarate and hydroxy-L-proline: molecular and
metabolic convergent evolution. J Biol Chem 282,
6685–6695.

20 Oppermann U, Filling C, Hult M, Shafqat N, Wu X,
Lindh M, Shafqat J, Nordling E, Kallberg Y, Persson
B et al. (2003) Short-chain dehydrogenases ⁄ reductases
(SDR): the 2002 update. Chem Biol Interact 143–144,
247–253.
21 Pollard JR & Bugg TD (1998) Purification, characteri-
sation and reaction mechanism of monofunctional 2-hy-
droxypentadienoic acid hydratase from Escherichia coli.
Eur J Biochem 251, 98–106.
22 Timm DE, Mueller HA, Bhanumoorthy P, Harp JM
& Bunick GJ (1999) Crystal structure and mechanism
of a carbon–carbon bond hydrolase. Structure 7,
1023–1033.
23 Tame JR, Namba K, Dodson EJ & Roper DI (2002)
The crystal structure of HpcE, a bifunctional decarbox-
ylase ⁄ isomerase with a multifunctional fold. Biochemis-
try 41, 2982–2989.
24 Brouns SJ, Barends TR, Worm P, Akerboom J, Turn-
bull AP, Salmon L & van der Oost J (2008) Structural
insight into substrate binding and catalysis of a novel
2-keto-3-deoxy-D-arabinonate dehydratase illustrates
common mechanistic features of the FAH superfamily.
J Mol Biol 379, 357–371.
25 Rakus JF, Fedorov AA, Fedorov EV, Glasner ME,
Hubbard BK, Delli JD, Babbitt PC, Almo SC & Gerlt
JA (2008) Evolution of enzymatic activities in the
enolase superfamily: L-rhamnonate dehydratase.
Biochemistry 47, 9944–9954.
26 Rea D, Hovington R, Rakus JF, Gerlt JA, Fu
¨

lo
¨
pV,
Bugg TD & Roper DI (2008) Crystal structure and
functional assignment of YfaU, a metal ion dependent
class II aldolase from Escherichia coli K12. Biochemistry
47, 9955–9965.
Novel L-rhamnose metabolic pathway S. Watanabe and K. Makino
1566 FEBS Journal 276 (2009) 1554–1567 ª 2009 The Authors Journal compilation ª 2009 FEBS
27 Jensen RA (1976) Enzyme recruitment in evolution of
new function. Annu Rev Microbiol 30, 372–376.
28 Ahmed H, Ettema TJ, Tjaden B, Geerling AC, van der
Oost J & Siebers B (2005) The semi-phosphorylative
Entner–Doudoroff pathway in hyperthermophilic
archaea: a re-evaluation. Biochem J 390, 529–540.
29 Chatterjee AK, Thurn KK & Tyrell D (1985) Isolation
and characterization of Tn5 insertion mutants of
Erwinia chrysanthemi that are deficient in polygalacturo-
nate catabolic enzymes oligogalacturonate lyase and
3-deoxy-D-glycero-2,5-hexodiulosonate dehydrogenase.
J Bacteriol 162, 708–714.
30 Bausch C, Peekhaus N, Utz C, Blais T, Murray E,
Lowary T & Conway T (1998) Sequence analysis of the
GntII (subsidiary) system for gluconate metabolism
reveals a novel pathway for L-idonic acid catabolism in
Escherichia coli. J Bacteriol 180, 3704–3710.
31 Eichhorn MM & Cynkin MA (1965) Microbial
metabolism of 2-deoxyglucose; 2-deoxygluconic acid
dehydrogenase. Biochemistry 4, 159–165.
32 Klasen R, Bringer-Meyer S & Sahm H (1995)

Biochemical characterization and sequence analysis of
the gluconate:NADP 5-oxidoreductase gene from
Gluconobacter oxydans. J Bacteriol 177, 2637–2643.
33 Yew WS, Fedorov AA, Fedorov EV, Rakus JF, Pierce
RW, Almo SC & Gerlt JA (2006) Evolution of enzy-
matic activities in the enolase superfamily: L-fuconate
dehydratase from Xanthomonas campestris. Biochemistry
45, 14582–14597.
34 Lowry OH, Rosebrough NJ, Farr AL & Randall RJ
(1951) Protein measurement with the folin phenol
reagent. J Biol Chem 193, 265–275.
Supporting information
The following supplementary material is available:
Fig. S1. Schematic construction of plasmids for LRA
gene expression in E. coli.
Fig. S2. Full sequence alignment of MhpD family
enzymes (COG0179), including L-DKDR hydrolase.
Table S1. Metabolic enzymes involved in nonphospho-
rylative sugar pathways.
Table S2. List of primers used in this study.
Table S3. Recombinant E. coli strains expressing LRA
gene(s).
Doc. S1. Supplementary methods.
This supplementary material can be found in the
online version of this article.
Please note: Wiley-Blackwell is not responsible for
the content or functionality of any supplementary
materials supplied by the authors. Any queries (other
than missing material) should be directed to the corre-
sponding author for the article.

S. Watanabe and K. Makino Novel L-rhamnose metabolic pathway
FEBS Journal 276 (2009) 1554–1567 ª 2009 The Authors Journal compilation ª 2009 FEBS 1567