Biologia, Bratislava, 57/Suppl. 11: 93—99, 2002
REVIEW
Structural basis of the synthesis of large cycloamyloses
by amylomaltase
Norbert Sträter1*, Ingo Przylas1, Wolfram Saenger1,
Yoshinobu Terada2, Kazutoshi Fujii2, Takeshi Takaha2
1
Institut für Chemie – Kristallographie, Freie Universität Berlin, Takustraße 6, 14195 Berlin, Germany;
tel.: ++ 49 30 8385 3456, fax: ++ 49 30 8385 6702, e-mail: [email protected]
2
Biochemical Research Laboratory, Ezaki Glico Co. Ltd., 4-6-5 Utajima, Nishiyodogawa-ku, Osaka
555-8502, Japan
STRÄTER, N., PRZYLAS, I., SAENGER, W., TERADA, Y., FUJI, K. & TAKAHA,
T., Structural basis of the synthesis of large cycloamyloses by amylomaltase.
Biologia, Bratislava, 57/Suppl. 11: 93—99, 2002; ISSN 0006-3088.
Large cycloamyloses probably form hydrophobic channels resembling those
of V-type amylose helices. Because these channels can accomodate guest
molecules, cycloamyloses may find technical applications similar to the smaller
cyclodextrins. Cycloamyloses are synthesized by bacterial amylomaltases and
by the related plant D-enzymes. We have analyzed the three-dimensional
structures of amylomaltase from Thermus aquaticus and also of a complex of
this enzyme with the maltotetraose inhibitor acarbose in order to gain insight
into the structural basis of cycloamylose formation. Amylomaltase differs from
α-amylase or CGTase in the absence of a C-terminal domain C and in the
presence of an additional predominantly α-helical insertion between barrel
strands 2 and 3. Two acarbose molecules have bound to the enzyme, one
at the active site and a second molecule ∼14 Å away from the nonreducing
end of the first acarbose. The subdomains B1, B2 and B3, which are located
around the rim of the C-terminal side of the barrel, and in particular the two
loops around residues 460 and 250, probably influence the product specificity.
Key words: acarbose, α-amylase family, glucanotransferase, protein crystallography.
Abbreviations: CA, cycloamylose; CD, cyclodextrin; CGTase, cyclodextrin
glucanotransferase; D-enzyme, starch disproportionating enzyme.
Cyclodextrins and cycloamyloses
Cyclodextrins (CDs) are small cylic maltooligosaccharides with 6 (α-CD), 7 (β-CD) and 8 (γ-CD)
glucose residues. These molecules have an annular
structure with a hydrophilic rim, formed by hydroxyl groups, at both sides of the cone. The cen-
tral hydrophobic cavity with a diameter of 4.7–5.3
Å for α-CD to 7.5–8.3 Å for γ-CD can complex hydrophobic guest molecules (SAENGER et al., 1998).
These inclusion complexes can change the solubility, reactivity and stability of the complexed substances and have found many applications in the
food, chemical and pharmaceutical industries.
* Corresponding author
93
These larger cycloamyloses have also been
reported to be useful in protein refolding. CA
can complex and remove detergents from the refolding buffer and has a chaperone-like activity
(MACHIDA et al., 2000).
Production of CDs and CAs by CGTase and
amylomaltase
Fig. 1. Structure of CA26. The spheres in the channel
along the helix axis are disordered water molecules.
This channel may also accommodate hydrophobic
guest molecules.
Larger cycloamyloses containing 8 and 9 glucoses apparently experience steric strain, which
results in a distortion of the previously annular
molecule into a boat-shaped structure. The X-ray
structures of CA-10 (a cycloamylose with 10 glucose residues) and CA-14 show a butterfly-shaped
molecule. A novel hydrogen-bonding pattern is
present in these molecules, by which the steric
strain is relieved through 1800 rotations about
two diammetrically-opposed glucosidic bonds. The
central cavities of these CA molecules are likely to
be too small to accomodate guest molecules.
The largest cycloamylose that could be characterized by X-ray crystallography is CA-26 (GESSLER et al., 1999). CA-26 folds into two short lefthanded V-amylose helices in an antiparallel arrangement. A hydrophobic channel of 5.0–5.5 Å
diameter runs along the axis of each helix (Fig. 1).
This channel may form complexes with a large
variety of hydrophobic guest molecules, in particular with long, extended molecules. Butanol, octanol and oleic acid can form inclusion complexes
with a CA mixture containing molecules with DP
greater than 50 (TAKAHA et al., 1996; TAKAHA
& SMITH, 1999). On the basis of the high solubility, non-reducing character and the ability to
form inclusion complexes, it therefore appears possible that larger cycloamyloses will find industrial
applications similar to those of the cyclodextrins.
However, it is difficult and costly to prepare large
amounts of pure cycloamylose with a defined ring
size.
94
Cyclodextrins are produced by the action of cyclodextrin glucanotransferase (CGTase) that converts starch into a mixture of α-CD, β-CD and γCD as well as residual dextrin. However, CGTase
initially produces larger cycloamyloses which are
subsequently converted to cyclodextrins (TERADA
et al., 1997). The product specificity (proportions
of α-, β-, and γ-CDs) is different for CGTases from
different sources, but it can be influenced by the
reaction conditions (RENDLEMAN, 1993; MORI et
al., 1995) and by single amino-acid substitutions
(NAKAMURA et al., 1993, 1994; PENNINGA et al.,
1995; WIND et al., 1998).
Larger cycloamyloses are produced by the action of bacterial amylomaltases or the plant Denzymes (starch disproportionating enzymes) on
amylose. It has been demonstrated that in the
early stages of the reaction of potato D-enzyme
with amylose AS-320, CAs with several hundred
glucoses are produced (TAKAHA et al., 1996). As
the reaction progressed, the mean product size
tended towards a DP of about 90 and the yield of
CA reached more than 95 %. The smallest product formed was CA-17. It was later shown that
the action of the amylomaltases from E. coli and
Thermus aquaticus on amylose also produced cycloamyloses (TERADA et al., 1999). The former
enzyme produced CA-17 and the latter CA-22 as
the smallest product.
4-α-Glucanotransferases
CGTase (EC 2.4.1.19) as well as amylomaltase
(EC 2.4.1.25) are 4-α-glucanotransferases. A third
member of this group of enzymes is glycogen debranching enzyme (EC 3.2.1.33 and EC 2.4.1.25).
These enzymes break an α-1,4 link and transfer
the resulting glucan group to an acceptor molecule
through creation of a new α-1,4 link. The intermolecular reaction is often called the “disproportionating reaction”:
(α − 1, 4 − glucan)m + (α − 1, 4 − glucan)n =
(α − 1, 4 − glucan)m−x + (α − 1, 4 − glucan)n+x
An intramolecular glucan transfer yields cyclic
glucan products:
(α − 1, 4 − glucan)m =
= cyclic(α−1, 4−glucan)x +(α−1, 4−glucan)m−x
D-Enzyme and amylomaltase
Amylomaltase was first described by MONOD &
TORRIANI (1948) as a maltose-inducible enzyme.
Since then the amylomaltase gene has been cloned
from several bacteria and homologous genes were
identified in further bacterial genomes. It is a
widely-distributed enzyme, but probably not ubiquitous since some genomes do not contain an obvious amylomaltase gene. A similar enzyme is
present in plants and is termed disproportionating enzyme (D-enzyme).
Early studies on the biochemical properties
focused on D-enzymes from different plant sources.
These studies showed that the different D-enzymes
have common reaction characteristics. The smallest donor molecule is maltotriose, the smallest acceptor is glucose and the smallest transferred glucan is a maltose unit. The major difference between the plant D-enzymes and the bacterial amylomaltases is that amylomaltase can catalyse a glucosyl transfer.
Based on the presence of a few conserved
regions, in particular the four conserved regions
which are characteristic for the α-amylase family,
it has been proposed that these enzymes belong
to the α-amylase family (HEINRICH et al., 1994;
SVENSSON, 1994). It was also noted, that amylomaltase appears to be the most distantly-related
member of this family (JANECEK, 1997) and was
therefore classified as family 77, whereas related
members such as amylase and CGTase are members of the large family 13.
Amylomaltase appears to have different physiological functions in bacteria. In E. coli, amylomaltase is expressed with glucan phosphorylase
from the same operon. It is proposed to be a member of the maltooligosaccharide transport and utilization system (SCHWARTZ, 1987) and apparently
plays a role in converting short maltooligosaccharides into longer chains upon which glucan phosphorylase can act (TAKAHA & SMITH, 1999). However, the genomes of Haemophilus influenzae and
Aquifex aelicus lack the genes coding for maltose
transport proteins. In these organisms, the genes
for amylomaltase are part of the glycogen operon,
indicating that the enzyme might be involved in
glycogen metabolism. It has recently been shown,
that the plastidial plant D-enzyme is required for
malto-oligosaccharide metabolism during starch
degradation (CRITCHLEY et al., 2001).
We have crystallized the amylomaltase from
Thermus aquaticus and analyzed its structure to
a resolution of 2.0 Å resolution by X-ray crystallography (PRZYLAS et al., 2000a). On the basis
of the good resolution of these crystals an atomic
model of all residues of the 57 kDa (500 aa) protein could be built and the conformations of ordered side chains and in particular of the acarbose
inhibitor, which was soaked into the crystals could
be reliably determined (PRZYLAS et al., 2000b).
Protein fold and topography
Amylomaltase contains several insertions between
the strands of the central (βsα)8 -barrel (subdomain A). All insertions are present at the Cterminal side of the barrel, where the substrate
binding site also is located. These insertions have
been subdivided into three subdomains, in order to
facilitate comparison to related enzymes (Fig. 2).
Subdomain B1 corresponds to domain B in the
related family-13 enzymes. The core of this domain is always composed of an insertion between
the third and fourth barrel strand. Whereas in related CGTases and α-amylases two strands from
an insertion between the second and third barrel
strand complete domain B, in amylomaltase two
strands from an insertion between the fourth and
fifth barrel strand are part of this subdomain. In
fact, amylomaltase also contains an insertion between the second and third barrel strand (subdomain B2). This insertion, however, is considerably
larger in amylomaltase and mainly α-helical. Subdomain B2 is unique to amylomaltase.
The most striking difference between amylomaltase and the other members of the α-amylase
superfamily is the absence of the C-terminal domain C, which is present in all other structures of
these enzymes determined so far. The function of
this domain is not clear. When the amylomaltase
structure is superimposed onto CGTase and amylase, subdomain B3 of amylomaltase is located at
approximately the same position as domain C in
the other enzymes and might have a similar function.
No calcium-binding site was located in the
electron density maps. This finding is in agreement
with biochemical data, which indicate that this
enzyme is not influenced by metal ions.
A co-crystal structure with acarbose
In order to study substrate binding to the enzyme,
95
B3
B3
250s loop
250s loop
460s loop
460s loop
B2
B1
B2
crystals of amylomaltase were soaked for 4 hours
in a buffer containing 100 mM acarbose. After
that, the crystals were shock-frozen in liquid nitrogen and data to 1.9 Å resolution were collected
at 100 K.
Acarbose is a maltotetraose derivative that
is a potent inhibitor of many enzymes of the αamylase family. It contains the inhibitory acarviosine group in which a valeinamine ring is linked to
a 4-amino-4,6-dideoxy-D-glucopyranose via an α1,4-linkage. The double bond between C5 and C7
(corresponding to O5 of a glucose unit) promotes a
2
H3 half-chair conformation, which resembles the
presumed transition state anticipated for glycoside
hydrolysis. Although it is not an ideal mimic of
such a transition state, acarbose is a tight-binding
inhibitor.
The active site
A surprising result of the co-crystal structure with
acarbose was that the acarviosine moiety was not
bound to subsites −1 and +1, as in the related
family-13 enzymes. Instead, the inhibitor occupies
subsites −3 to +1 of the active center (Fig. 3).
No salt bridge is present between the amino group
of the valienamine moiety and a protein residue.
This interaction has been proposed to contribute
significantly to the inhibition of these enzymes by
acarbose.
The acarbose inhibitor has also not been processed as in many other cases of reactions of this
inhibitor with enzymes of family 13. In the catalytic subsites −1 and +1 an intact maltose group
is bound, which is nevertheless not turned over.
The high pH of 9.0 of the crystallization condi-
96
B1
Fig. 2. Stereo diagram of
the fold of Thermus aquaticus amylomaltase and acarbose binding sites.Two acarbose molecules are shown
in black. The upper right
molecule is located at the
active site. Subdomain A is
shown in dark grey whereas
subdomains B1–B3 are coloured in light grey.
tions, the low reaction temperature compared to
the optimum of 75–80 ◦C for this thermophilic enzyme and the absence of a suitable acceptor unit
might account for this finding. Indeed, the Xray structure shows that two catalytically important side chains have unexpected conformations.
Firstly, Asp-293, which is the nucleophile that attacks the C1 carbon of the substrate in a putative mechanism involving a covalent intermediate,
is not positioned to interact with the substrate.
Instead, this Asp-293 is hydrogen-bonded to Arg291 and Asp-213. Secondly, Glu-340, which protonates the glycosidic oxygen atom of the scissile
bond (UITDEHAAG et al., 1999), is, in the amylomaltase structure, bound to the oxygens O1, O2
and O3 of the glucose unit at the reducing end.
In structures of α-amylase and CGTase with inhibitor, this residue is bound to O4 of the scissile bond. Thus, the two most important catalytic
residues are not well positioned in the crystal with
respect to their catalytic function.
However, the observed binding mode of acarbose to subsites +1 to −3 is in agreement with
the analysis of the location of the cleavage sites
of potato D-enzyme on maltotetraose substrates
(JONES & WHELAN, 1969).
The catalytic residues
A structural superposition of the catalytic site of
amylomaltase with CGTase and amylase shows
that seven residues are conserved and have a similar orientation in these structures: The three catalytic residues Asp-293, Glu-340 and Asp-395 and
residues Tyr-59, Asp-213, Arg-291 and His-394.
These seven residues build up the core of the cat-
His 294
Glu 340
His 394
Trp 258
+1
Asp 395
D
Gln 256
-1
Tyr 59
Et(OH)2
C
Trp 473
Ser 57
-2
Asn 464
B
Gln 60
-3
Fig. 3. Scheme of the interactions
of acarbose bound to the active
site. Hydrogen bonds of less than
3.5 Å are shown as dashed lines.
A
alytic cleft. Except for Tyr-59, they are part of
the four conserved motifs of family 13. The sequence homology is too weak to detect if Tyr-59
is conserved in all α-amylase family members, as
are the other six residues. As has been previously
noticed on the basis of sequence comparisons, the
histidine residue of the first conserved region (His122 of Taka-amylase), which is conserved in most
other family members, is not present in amylomaltase.
A secondary acarbose binding site
A second acarbose molecule was located in a
groove close to the active center. The distance between the reducing end of this maltotetraose and
the nonreducing end of the substrate analog bound
to the active site is ∼14 Å. Hydrophobic contacts
of Tyr-54 with glucose unit B and of Tyr-101 with
unit C of the inhibitor are probably the most important interactions that determine the conformation and binding of the inhibitor to this site. The
acarbose winds around Tyr-54, which is highly
solvent-exposed in the unliganded structure. Tyr101 is involved in a hydrophobic stacking interaction with glucose unit C. Overall, the second acarbose has significantly fewer interactions with the
Gly 55
A
Asp 56
Gly 98
B
Tyr 54
Gly 53
C
Tyr 101
D
Gly 153
Fig. 4. Scheme of the interactions of acarbose bound
to the secondary binding site.
97
Fig. 5. Molecular surface of amylomaltase. The two
acarbose molecules are shown in addition to two putative binding paths for the smallest cycloamyloses that
are formed by this enzyme.
466, Asn-472 and Trp-473 are strictly conserved
and the first and last of these residues are involved in the binding of acarbose at the active
site (Fig. 3). In addition, a view of the molecular surface suggests that this loop might also sterically influence the size of the smallest cycloamylose formed (Fig. 5).
Thus, one putative binding pathway for the
smallest cycloamylose products of around 20 glucoses is obtained by connecting the two acarbose
molecules bound to amylomaltase, as indicated
in Figure 5 by a straight white line. As outlined
above, it is not clear if the secondary binding
site is conserved among amylomaltases from other
species and if this binding site is occupied for the
formation of smaller cycloamyloses. An alternative pathway appears possible, which goes around
the 250s loop (dotted black line in Figure 5). This
pathway includes a cleft which leads from the reducing end of the acarbose molecule at the active
site to the outside of the barrel. There are indications that this cleft, which is near to the tip of
the 250s loop, is part of the glucan-binding site
and possibly needs to be occupied for the glucantransfer reaction to occur.
The 250s loop
protein compared to the acarbose bound to the
active site (Fig. 4).
In contrast to the residues liganding the acarbose bound to the active site, the residues binding to the second acarbose are not strictly conserved among other amylomaltases or D-enzymes.
However, the amino-acid sequence around Tyr-54
shows a pattern with several glycines and two conserved prolines (PLGPTGYGDSP). Tyr-54 is not conserved. It remains unclear if this binding site is of
low affinity and if it is only occupied because of
the high inhibitor concentration during soaking or
if it plays a role during cycloamylose formation.
Cycloamylose formation
For the formation of cyclic products, the nonreducing end of the glucan chain has to fold back to
the active center. In T. aquaticus amylomaltase,
the secondary binding site around Tyr-54 might
help to form a curved conformation of the amylose
chain in this region, which would favour the formation of cyclic products. Two loops are proposed
to influence the product specificity of amylomaltase: the 460s loop (the loop around residue 460)
and the 250s loop (Fig. 5).
The 460s loop is involved in substrate binding at the active-site cleft. Residues Asn-464, Pro-
98
The 250s loop (residues 247–255), which is part of
subdomain B1, is located on top of a cleft which
leads from the +1 binding site out of the active site
groove. Two solvent-exposed hydrophobic residues
sit at the tip of the loop: Tyr-250 and Phe-251.
Both residues show some disorder (flexibility) in
the unliganded enzyme structure as well as in the
complex with acarbose. In the complex structure a
weak density feature, which might correspond to a
single glucose residue, is present ∼4.5 Å away from
the 250s loop. Furthermore, this loop is highly
conserved among the amylomaltases. These findings suggest that the 250s loop is involved in substrate binding. Possibly, the loop changes conformation or becomes less flexible when larger substrates bind close to the loop.
A comparison of the molecular surfaces of αamylase, CGTase and amylomaltase shows that
α-amylase has a relatively open active-site cleft,
which is formed by residues of domains A and B.
In CGTase, domain B also forms part of the active
site pocket. A tyrosine or phenylalanine side chain
(Tyr-195 in CGTase from Bacillus circulans strain
8), which is derived from this domain, has been
proposed to favour the synthesis of cyclodextrins
by forming a non-polar core around which the αglucan could wrap (NAKAMURA et al., 1994; PEN-
et al., 1995). Trp-258 is near this residue
in amylomaltase and indeed also interacts with
the acarbose bound to the active site by forming hydrophobic contacts to the glucose residue
bound to subsite +1 (Fig. 3). These findings suggest an important role of domain B (subdomain
B1 in amylomaltase) in controlling substrate and
product specificity.
NINGA
Acknowledgements
N.S. thanks the Deutsche Forschungsgemeinschaft for
financial support.
References
CRITCHLEY, J. H., ZEEMAN, S. C., TAKAHA, T.,
SMITH, A. M. & SMITH, S. M. 2001. Plant J. 26:
89–100.
GEßLER, K., USÓN, I., TAKAHA, T., KRAUSS, N.,
SMITH, S. M., OKADA, S., SHELDRICK, G. M. &
SAENGER, W. 1999. Proc. Natl. Acad. Sci. USA
96: 4246–4251.
HEINRICH, P., HUBER, W. & LIEBL, W. 1994. Syst.
Appl. Microbiol. 17: 297–305.
JANECEK, S. 1997. Prog. Biophys. Mol. Biol. 67: 67–
97.
JONES, G. & WHELAN, W. J. 1969. Carbohydr. Res.
9: 483–490.
MACHIDA, S., OGAWA, S., XIAOHUA, S., TAKAHA, T.,
FUJII, K. & HAYASHI, K. 2000. FEBS Lett. 486:
131–135.
MONOD, J. & TORRIANI, A. M. 1948. Compt. rend.
Acad. 227: 240.
MORI, S., GOTO, M., MASE, T., MATSUURA, A., OYA,
T. & KITAHATA, S. 1995. Biosci. Biotech. Biochem.
59: 1012–1015.
NAKAMURA, A., HAGA, K. & YAMANE, K. 1993. Biochemistry 32: 6624–6631.
NAKAMURA, A., HAGA, K. & YAMANE, K. 1994. Biochemistry 33: 9929–9936.
PENNINGA, D., STROKOPYTOV, B., ROZEBOOM, H. J.,
LAWSON, C. L., DIJKSTRA, B. W., BERGSMA, J.
& DIJKHUIZEN, L. 1995. Biochemistry 34: 3368–
3376.
PRZYLAS, I., TOMOO, K., TERADA, Y., TAKAHA, T.,
OKADA, S., SAENGER, W. & STRÄTER, N. 2000a.
J. Mol. Biol. 296: 873–86.
PRZYLAS, I., TERADA, Y., FUJII, K., TAKAHA, T.,
SAENGER, W. & STRÄTER, N. 2000b. Eur. J.
Biochem. 267: 6903–6913.
RENDLEMAN, J. A. 1993. Carbohydr. Res. 247: 223–
237.
SAENGER, W., JACOB, J., GESSLER, K., STEINER, T.,
HOFFMANN, D., SANBE, H., KOIZUMI, K., SMITH,
S. M. & TAKAHA, T. 1998. Chem. Rev. 98: 1787–
1802.
SCHWARTZ, M. 1987. In: Escherichia coli and Salmonella typhimurium: cellular and molecular biology
(Neidhardt, F. D., Ed.), pp. 1482–1502, Am. Soc.
f. Microbiology, Washington, D.C.
SVENSSON, B. 1994. Plant Mol. Biol. 25: 141–157.
TAKAHA, T., YANASE, M., TAKATA, H., OKADA, S. &
SMITH, S. M. 1996. J. Biol. Chem. 271: 2902–2908.
TAKAHA, T. & SMITH, S. M. 1999. Genet. Eng. Rev.
16: 257–280.
TERADA, Y., YANASE, M., TAKATA, H., TAKAHA, T.
& OKADA, S. 1997. J. Biol. Chem. 272: 15729–
15733.
TERADA, Y., FUJI, K., TAKAHA, T. & OKADA, S. 1999.
Appl. Environ. Microbiol. 65: 910–915.
WIND, R. D., BUITELAAR, R. M. & DIJKHUIZEN, L.
1998. Eur. J. Biochem. 253: 598–605.
Received October 22, 2001
Accepted April 18, 2002
99