Expression, purification, and characterization of TylM1, an N,N- dimethyltransferase involved in the biosynthesis of mycaminose [6]

Full text of DOI:10.1021/ja9815881

J. Am. Chem. Soc. 1998, 120, 9951-9952
9951
Scheme 1
Expression, Purification, and Characterization of
TylM1, an N,N-Dimethyltransferase Involved in the
Biosynthesis of Mycaminose
Huawei Chen, Zhihong Guo,† and Hung-wen Liu*
Department of Chemistry, UniVersity of Minnesota
Minneapolis, Minnesota 55455
ReceiVed May 7, 1998
Tylosin (1), produced by Streptomyces fradiae, is an important
drug used to treat veterinary Gram-positive and mycoplasma
infections as well as to promote livestock growth.1 This macrolide
antibiotic is composed of a polyketide aglycon (tylactone) and
three unusual sugarssmycaminose (2), mycarose (3), and myci-
nose (4). Extensive genetic and phenotypic complementation
studies have led to several classes of blocked mutants, revealing
the genetic organization of the tylosin biosynthetic gene (tyl)
cluster.2 As illustrated in Scheme 1, the tylG region harbors the
polyketide synthase (PKS) genes, while the tylLM, tylIBA, and
tylCK regions contain genes for mycaminose and mycarose
formation.2 In an effort to study the biosynthesis of unusual sugars
found in antibiotics, we have sequenced the extended segments
flanking the PKS genes from which 17 open reading frames
(ORFs) in the tylCK, tylLM, and tylIBA regions have been
identified.3 Further analysis based on sequence similarities to
other sugar biosynthetic genes,4,5 especially those reported by
Cundliffe and co-workers, who have also sequenced the tylIBA
and tylLM segments of the tyl cluster,6 has allowed most of these
ORFs to be tentatively assigned. Three genes in the tylIBA region
are believed to be involved in mycaminose biosynthesis6astylA1
is for R-D-glucose 1-phosphate thymidylyltransferase, tylA2 is for
TDP-D-glucose 4,6-dehydratase, and tylB is likely the gene for a
pyridoxal 5′-phosphate-dependent aminotransferase.7 Other genes
that have been assigned to encode enzymes involved in the
mycaminose pathway reside in the tylLM regionstylM1 is likely
for a S-adenosylmethionine (AdoMet)-dependent methyltrans-
ferase, tylM2 is for the glycosyltransferase, and tylM3 may encode
a tautomerase which displays sequence similarity to some P-450
enzymes but lacks the conserved cysteine residue that coordinates
the heme iron.6b,8 While all of these assignments are based solely
on sequence analysis, the tentative identification of these genes
has allowed initial speculation of their roles, leading to a possible
route for mycaminose biosynthesis shown in Scheme 1.6 To
verify the proposed pathway, it is important to experimentally
determine the function of each gene product. Thus, we have
expressed several of the aforementioned genes and examined the
catalytic properties of the purified enzymes. In this paper, we
report the initial characterization of TylM1, and our results
provide, for the first time, biochemical evidence establishing the
role of TylM1 as the methyltransferase catalyzing N,N-dimethy-
lation of the 3-amino group.
To study the function of TylM1, the tylM1 gene was amplified
by polymerase chain reaction (PCR) and cloned into the pET-
17b expression vector, and the ensuing plasmid was used to
transform Escherichia coli BL21(DE3)pLysS host cells. Growth
of the resulting construct at 30 °C and induction using isopropyl
â-D-thiogalactoside (IPTG) allowed highly efficient expression
of tylM1, in which greater than 20% of the soluble protein in the
crude extract was found to be TylM1. This enzyme was purified
to near homogeneity by a protocol consisting of ammonium sulfate
fractionation and DEAE Sepharose and FPLC MonoQ chromato-
graphic steps. AdoMet (0.1 mM) was included in all buffers
throughout the purification to prevent the precipitation of TylM1
from the solution. Judging from a Mr of 55.2 K estimated by gel
filtration and a calculated mass of 27 427 Da based on the
translated sequence for each subunit, TylM1 exists as a ho-
modimer.9 The electronic absorption spectrum of the purified
enzyme shows no absorbance above 300 nm.
The predicted substrate 5 was synthesized by the reactions
delineated in Scheme 2.10 To test whether TylM1 is the desired
methyltransferase, a mixture of TylM1 (0.05 µmol), 5 (7.0 µmol),
and AdoMet (40 µmol) in 1.5 mL of 50 mM potassium phosphate
buffer (KPi, pH 7.5) was incubated at 23 °C for 4 h. The enzyme
† Current address: Department of Chemistry, University of California,
Berkeley, CA 94720.
(1) (a) McGuire, J. M.; Boniece, W. S.; Higgens, C. E.; Hoehn, M. M.;
Stark, W. M.; Westhead, J.; Wolfe, R. N. Antibiot. Chemother. 1961, 11, 320-
327. (b) Corcoran, J. W.; Huber, M. L. B.; Huber, F. M. J. Antibiot. 1977,
30, 1012-1014.
(2) (a) Baltz, R. H.; Seno, E. T. Antimicrob. Agents Chemother. 1981, 20,
214-225. (b) Fisherman, S. E.; Cox, K.; Larson, J. L.; Reynolds, P. A.; Seno,
E. T.; Yeh, W.-K.; Van Frank, R.; Hershberger, C. L. Proc. Natl. Acad. Sci.
U.S.A. 1987, 84, 8248-8252. (c) Baltz, R. H.; Seno, E. T. Annu. ReV.
Microbiol. 1988, 42, 547-574.
(3) Nine other ORFs have also been identified downstream from the tylCK
region.
(4) (a) Liu, H.-w.; Thorson, J. S. Annu. ReV. Microbiol. 1994, 48, 223-
256. (b) Kirschning, A.; Bechthold, A. F.-W.; Rohr, J. In Bioorganic Chemistry
Deoxysugars, Polyketides & Related Classes: Synthesis, Biosynthesis, En-
zymes; Rohr, J., Ed.; Springer: Berlin, 1997; pp 1-84. (c) Johnson, D. A.;
Liu, H.-w. In ComprehensiVe Chemistry of Natural Products Chemistry; Bar-
ton, D., Nakanishi, K., Meth-Cohn, O., Eds.; Pergamon: New York, in press.
(5) (a) Gaisser, S.; Bohm, G. A.; Corte´s, J.; Leadlay, P. F. Mol. Gen. Genet.
1997, 256, 239-251. (b) Summers, R. G.; Donadio, S.; Staver, M. J.; Wendt-
Pienkowski, E.; Hutchinson. C. R.; Katz, L. Microbiology 1997, 143, 3251-
3262.
(6) (a) Merson-Davies, L. A.; Cundliffe, E. Mol. Microbiol. 1994, 13, 349-
355. (b) Gandecha, A. R.; Large, S. L.; Cundliffe, E. Gene 1997, 184, 197-
203. (c) Fish, S. A.; Cundliffe, E. Microbiol. 1997, 143, 3871-3876.
(7) Thorson, J. S.; Lo, S. F.; Liu, H.-w.; Hutchinson, R. C. J. Am. Chem.
Soc. 1993, 115, 6993-6994.
(8) The other two genes found in the tylLM region are6b tlrD, a resistant
gene that ecodes a rRNA methyltransferase (Gandecha, A. R.; Cundliffe, E.
Gene 1996, 180, 173-176), and ORF4, which is highly homologous to the
crotonyl-CoA reductase gene (Wallace, K. K.; Bao, Z.; Hong, D.; Digate, R.;
Schuler, G.; Speedie, M. K.; Reynolds, K. A. Eur. J. Biochem. 1995, 233,
954-962).
(9) N-terminal amino acid sequencing confirmed that the first 10 residues
(AHSSATAGPQ) of this protein are identical to the translated tylM1 sequence
except for the deletion of the first methionine residue.
(10) Starting from 1,2:5,6-di-O-isopropylidene-R-D-glucose (7), the key
intermediate 8 was prepared in five steps, which involved inversion of the
C-3 hydroxyl group, nucleophilic displacement with sodium azide, and hydride
reduction of the azido group, with an overall yield of 30%. After carboben-
zyloxy chloride treatment to protect the C-3 amino functionality, the 5,6-
isopropylidene group was selectively cleaved to give 9 (83% yield for two
steps).11 Subsequent C-6 iodination followed by NaBH4 reduction gave
compound 10, which was converted to 11 via acid treatment, peracetylation,
and selective removal of the 1-O-acetyl group with hydrazine acetate in DMF12
(45% overall yield from 9). Phosphate 12 was obtained from 11 in pure R-form
by m-CPBA oxidation of the corresponding phosphite13 followed by hydro-
genation and treatment with K2CO3 in methanol (50% yield from 11). After
passing through an Amberlite IR-120 cation exchange column (Et3NH+, 1 ×
30 cm), the triethylammonium salt of 12 was reacted with thymidine
5′-monophosphomorpholidate in the presence of 1H-tetrazole in pyridine to
give 5,13 which was purified by Bio-Gel P-2 chromatography with 25 mM
NH4HCO3 (60% yield).14
S0002-7863(98)01588-1 CCC: $15.00 © 1998 American Chemical Society
Published on Web 09/11/1998

9952 J. Am. Chem. Soc., Vol. 120, No. 38, 1998
Communications to the Editor
Scheme 2
the SAX anion exchange column. The percent conversion of 5
to mono- and dimethylated products was estimated on the basis
of the integration of the corresponding peaks. By fitting the data
to the rate laws for an irreversible unimolecular consecutive
reaction (5 f 13 f 6),18 rate constants of 0.020 and 0.062 min-1
for the mono- and dimethylation step, respectively, were deduced
with good correlation coefficients.19 Our data clearly demonstrate
that TylM1 is indeed the required methyltransferase in the
biosynthesis of mycaminose, and it alone catalyzes the N,N-dimeth-
ylation, via a monomethylated intermediate, in a stepwise manner.
The above results introduce TylM1 as a new member of a small
family of enzymes that are capable of catalyzing N,N-dimethy-
lation. A few representatives of this class include: PEM-2, a
phospholipid methyltransferase that catalyzes the methylation of
phosphatidylethanolamine, preferentially the second and third
methyl transfer steps, to form phosphatidylcholine;20 RMT1, a
protein-arginine methyltransferase that catalyzes both the NG-
mono- and NG,NG-asymmetric dimethylation of protein arginine
residues, an important process in modulating mRNA splicing as
well as growth factor localization and function;21 and TlrA, the
product of a resistance gene isolated from the tylosin producer
S. fradiae, that catalyzes the dimethylation of a single base (A-
2058) to N6,N6-dimethyladenine within 23S rRNA rendering the
bacterial strain resistant to macrolide, lincosamide, and strepto-
gramin B-type (MLS) antibiotics.22 Thus, TylM1, acting on the
amino group of a sugar substrate, is a distinct member of this
AdoMet-dependent N,N-dimethyltransferase family.
It should be pointed out that the deduced sequence of tylM1
reveals significant similarity to those of eryCVI from the
erythromycin cluster of Saccharopolyspora erythraea (60%
identity),5 desVI from the methymycin cluster of Streptomyces
Venezuelae (60% identity),23 snoX from the nogalamycin cluster
of Streptomyces nogalater (54% identity),24 rdmD from the
rhodomycin cluster of Streptomyces purpurascens (50% iden-
tity),25 and srmX from the spiramycin cluster of Streptomyces
ambofaciens (47% identity).26 All of them contain a short
consensus sequence near the N-terminus, LLDV(I)ACGTG, a
conserved motif for many AdoMet binding proteins.5a,27 Although
such a sequence analysis has allowed speculation of their catalytic
roles as methyltransfereases, their actual functions have never been
verified biochemically. The fact that TylM1 has now been fully
established as an AdoMet-dependent N,N-dimethyltransferase
furnishes compelling evidence suggesting a similar role for these
proteins in their respective biosynthetic pathways.
was removed using a Centricon-10, and the product was isolated
by FPLC MonoQ using a linear gradient of 0-0.4 M ammonium
bicarbonate buffer over 25 min. Spectral characterization of the
purified product confirmed that it is indeed the dimethylated
hexose 6.15 Interestingly, when the reaction was quenched at an
earlier time, two products could be detected by HPLC using an
Adsorbosphere SAX column (5 µm, 4.6 × 250 mm) and 50 mM
KPi buffer (pH 3.6). The product with a retention time of 8.79
min is the dimethylated hexose 6, and the other, at 7.16 min, is
the monomethylated species 13.16 The time courses of mono-
and dimethylation were determined by HPLC (at 279 nm)
following the consumption of substrate 5 and the formation of
13 and 6. A typical assay mixture consisted of 10 mM AdoMet,17
0.5 mM 5, 10.9 µM TylM1, and 200 µL of 50 mM KPi (pH 7.5).
Aliquots of 5 µL were withdrawn at appropriate time intervals,
boiled, and subjected to the aforementioned HPLC analysis using
Acknowledgment. We express our great appreciation to Dr. Eugene
Seno and the Lilly Research Laboratories for their generous gifts of the
plasmids pSET552, pHJL309, pHJL311, and POJ190. This work is
supported in part by the National Institutes of Health Grant GM54346.
JA9815881
(11) Alais, J.; David, S. Carbohydr. Res. 1992, 230, 79-87.
(12) Excoffier, G.; Gagnaire, D.; Utille, J.-P. Carbohydr. Res. 1975, 39,
368-373.
(18) The rate laws for an irreversible unimolecular consecutive reaction
of (A f B f C) are υA ) -dx/dt ) k1x, υB ) k1x - k2y, and υC ) k2y, in
which x, y, and z are the respective concentrations of A, B, and C at time t.
Assuming the initial concentration (at time 0) of A is a and those of B and C
are nil, data fitting to any two of the following three equations could solve k1
and k2: x ) a exp(-k1t), y ) ak1[exp(-k1t) - exp(-k2t)]/(k2 - k1), and z )
a[1 - k2 exp(-k1t)/(k2 - k1) + k1 exp(-k2t)/(k2 - k1)]. Since such a treatment
holds only under subsaturation conditions, the fact that our kinetic data fit
well to the above equations strongly suggests a large Km for 5 (much greater
than 0.5 mM).
(19) The correlation coefficients are greater than 0.995 for all data fittings.
(20) Kodaki, T.; Yamashita, S. Eur. J. Biochem. 1989, 185, 243-251.
(21) (a) Gary, J. D.; Lin, W.-J.; Yang, M. C.; Herschman, H. R.; Clarke,
S. J. Biol. Chem. 1996, 271, 12585-12594. (b) Lin, W.-J.; Gary, J. D.; Yang,
M. E.; Clarke, S.; Herschman, H. R. J. Biol. Chem. 1996, 271, 15034-15044.
(22) Zalacain, M.; Cundliffe, E. J. Bacteriol. 1989, 171, 4254-4260.
(23) Xue, Y.; Zhao, L.; Liu, H.-w.; Sherman, D. H. Proc. Natl. Acad. Sci.
U.S.A., in press.
(13) Wittmann, V.; Wong, C.-H. J. Org. Chem. 1997, 62, 2144-2147.
(14) Spectral data of 5: 1H NMR (D2O) δ 1.10 (3H, d, J ) 6.3 Hz, 5-Me),
1.74 (3H, s, 5′′-Me), 2.17-2.22 (2H, m, 2′-Hs), 3.18 (1H, t, J ) 10.2, 4-H),
3.24 (1H, t, J ) 10.2, 3-H), 3.61 (1H, ddd, J ) 10.2, 3.3, 3.1, 2-H), 3.77-
3.87 (1H, m, 5-H), 3.99-4.01 (3H, m, 4′-H, 5′- Hs), 4.41-4.44 (1H, m, 3′-
H), 5.40 (1H, dd, J ) 7.2, 3.3, 1-H), 6.17 (1H, dd, J ) 7.2, 6.9, 1′-H), 7.55
(1H, s, 6′′-H); 13C NMR (D2O) δ 12.8, 17.6, 39.7, 55.9, 66.6 (d, J ) 6.0 Hz),
70.1, 70.6 (d, J ) 8.4), 72.1, 73.3, 86.2, 86.4 (d, J ) 9.2), 95.6 (d, J ) 6.1),
112.8, 138.5, 153.0, 167.9; 31P NMR (D2O) δ -13.47 (d, J ) 20.7), -11.75
(d, J ) 20.7); high-resolution MALDI-MS calcd for C16H25N3O14P2 (M +
H)+ 546.0890, found 546.0894.
(15) Spectral data of 6: 1H NMR (D2O) δ 1.11 (3H, d, J ) 6.3 Hz, 5-Me),
1.75 (3H, d, J ) 1.2, 5′′-Me), 2.18-2.22 (2H, m, 2′-Hs), 2.85 (6H, s, NMe2),
3.36-3.46 (2H, m, 3-H, 4-H), 3.82-3.87 (2H, m, 2-H, 5-H), 4.01-4.02 (3H,
m, 4′-H, 5′-Hs), 4.42-4.46 (1H, m, 3′-H), 5.42 (1H, dd, J ) 7.2, 3.3, 1-H),
6.18 (1H, dd, J ) 6.9, 6.7, 1′-H), 7.56 (1H, d, J ) 1.2, 6′′-H); high-resolution
MALDI-MS calcd for C18H29N3O14P2 (M + H)+ 574.1203, found 574.1196.
(16) The identity of 13 was confirmed by its incorporation of radioactivity
when [3H-CH3]AdoMet was used as the methyl donor and by determining its
exact mass using high-resolution MALDI-MS: calcd for C17H27N3O14P2 (M
+ H)+ 560.1047, found 560.1023.
(24) Ylihonko, K. Tuikkanen, J.; Jussila, S.; Cong, L.; Ma¨ntsa¨la¨, P. Mol.
Gen. Genet. 1996, 251, 113-120.
(25) Niemi, J.; Ma¨ntsa¨la¨, P. J. Bacteriol. 1995, 177, 2942-2945.
(26) Geistlich, M.; Losick, R.; Turner, J. R.; Rao, R. N. Mol. Microbiol.
1992, 6, 2019-2029.
(27) (a) Kagan, R. M.; Clarke, S. Arch. Biochem. Biophys. 1994, 310, 417-
427. (b) Schluckebier, G.; O’Gara, M.; Saenger, W.; Cheng, X. J. Mol. Biol.
1995, 247, 16-20.
(17) Under this condition, the reaction rate is no longer dependent on the
concentration of AdoMet.