Delineating the biosynthesis of gentamicin X2, the common precursor of the gentamicin C antibiotic complex

Article abstract of DOI:10.1016/j.chembiol.2014.12.012

Gentamicin C complex is a mixture of aminoglycoside antibiotics used worldwide to treat severe Gram-negative bacterial infections. Despite its clinical importance, the enzymology of its biosynthetic pathway has remained obscure. We report here insights into the four enzyme-catalyzed steps that lead from the first-formed pseudotrisaccharide gentamicin A2 to gentamicin X2, the last common intermediate for all components of the C complex. We have used both targeted mutations of individual genes and reconstitution of portions of the pathway in vitro to show that the secondary alcohol function at C-3″ of A2 is first converted to an amine, catalyzed by the tandem operation of oxidoreductase GenD2 and transaminase GenS2. The amine is then specifically methylated by the S-adenosyl-l-methionine (SAM)-dependent N-methyltransferase GenN to form gentamicin A. Finally, C-methylation at C-4″ to form gentamicin X2 is catalyzed by the radical SAM-dependent and cobalamin-dependent enzyme GenD1.

Full text of DOI:10.1016/j.chembiol.2014.12.012

Article
Delineating the Biosynthesis of Gentamicin X2, the
Common Precursor of the Gentamicin C Antibiotic
Complex
Highlights
Authors
d
Key portion of the pathway to a globally used antibiotic is
defined
Chuan Huang, Fanglu Huang, ...,
Peter F. Leadlay, Yuhui Sun
d
d
Four enzymes convert gentamicin A2 into gentamicin X2
Correspondence
pfl10@cam.ac.uk (P.F.L.),
yhsun@whu.edu.cn (Y.S.)
GenD1 is an unusual cobalamin- and radical SAM-dependent
methyltransferase
In Brief
All five components of the clinically
valuable antibiotic gentamicin C complex
derive from the pseudotrisaccharide
gentamicin X2. Huang et al. use gene
knockouts and in vitro reconstitution to
identify the four pathway enzymes
catalyzing the final steps leading to this
key intermediate.
Huang et al., 2015, Chemistry & Biology 22, 251–261
February 19, 2015 ª2015 The Authors
http://dx.doi.org/10.1016/j.chembiol.2014.12.012

Chemistry & Biology
Article
Delineating the Biosynthesis of Gentamicin X2,
the Common Precursor of the
Gentamicin C Antibiotic Complex
Chuan Huang,1,4 Fanglu Huang,2,4 Eileen Moison,2 Junhong Guo,1 Xinyun Jian,1 Xiaobo Duan,1 Zixin Deng,1,3
1,
Peter F. Leadlay,2, and Yuhui Sun
*
*
1Key Laboratory of Combinatorial Biosynthesis and Drug Discovery, Ministry of Education, and Wuhan University School of Pharmaceutical
Sciences, Wuhan University, Wuhan 430071, People’s Republic of China
2Department of Biochemistry, University of Cambridge, Cambridge CB2 1GA, UK
3Hubei Engineering Laboratory for Synthetic Microbiology, Wuhan Institute of Biotechnology, Wuhan 430075, People’s Republic of China
4Co-first author
*Correspondence: pfl10@cam.ac.uk (P.F.L.), yhsun@whu.edu.cn (Y.S.)
http://dx.doi.org/10.1016/j.chembiol.2014.12.012
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
SUMMARY
gentamicin biosynthetic pathway an attractive target for reengin-
eering to favor a specific component.
Gentamicin C complex is a mixture of aminoglycoside
antibiotics used worldwide to treat severe Gram-
negative bacterial infections. Despite its clinical
importance, the enzymology of its biosynthetic
pathway has remained obscure. We report here
insights into the four enzyme-catalyzed steps that
leadfromthe first-formedpseudotrisaccharidegenta-
micin A2 to gentamicin X2, the last common interme-
diate for all components of the C complex. We have
used both targeted mutations of individual genes
and reconstitution of portions of the pathway in vitro
Gentamicins are modified sugars, characteristically contain-
ing an unusual aminocyclitol ring (2-deoxystreptamine, 2-DOS
[1]) (Houghton et al., 2010; Park et al., 2013). The outlines of
the gentamicin pathway have been known for some time (Testa
and Tilley, 1975, 1976; Kase et al., 1982a, 1982b), and in the
wake of the comprehensive sequencing of aminoglycoside
biosynthetic gene clusters (Ota et al., 2000; Unwin et al., 2004;
Kharel et al., 2004a, 2004b, 2004c; Huang et al., 2005; Subba
et al., 2005; Aboshanab, 2005; Hong et al., 2009) rapid progress
has been made in identifying the enzymatic steps that lead to the
2-DOS scaffold, and thence to the pseudodisaccharide parom-
amine (2) (Llewellyn and Spencer, 2006; Thibodeaux et al.,
to show that the secondary alcohol function at C-300 2008; Kudo and Eguchi, 2009; Kudo et al., 2009) and the pseu-
dotrisaccharide gentamicin A2 (3) (Park et al., 2008) (Figure 1).
This has enabled a sharper focus on those genes likely to govern
of A2 is first converted to an amine, catalyzed by the
tandem operation of oxidoreductase GenD2 and
transaminase GenS2. The amine is then specifically
methylated by the S-adenosyl-L-methionine (SAM)-
dependent N-methyltransferase GenN to form genta-
micin A. Finally, C-methylation at C-400 to form
gentamicin X2 is catalyzed by the radical SAM-depen-
dent and cobalamin-dependent enzyme GenD1.
the later steps of gentamicin biosynthesis.
We (Guo et al., 2014) and others (Karki et al., 2012; Hong and
Yan, 2012; Li et al., 2013) have used specific gene deletions
to probe the identity of the enzymes that act on gentamicin X2
(4), partitioning this intermediate between a methylated branch
that gives rise via G418 (5) to gentamicin components C2a, C2,
and C1 and an unmethylated branch that gives rise to C1a
and C2b (Figure 1). The cluster contains multiple, mutually
homologous genes for methyltransferases, oxidoreductases,
INTRODUCTION
and aminotransferases with potential for overlapping substrate
specificities, but by combining analysis of specific gene dele-
Gentamicins are clinically valuable aminoglycoside antibiotics tions with in vitro reconstitution of the enzymes involved in dehy-
isolated as gentamicin C complex, a mixture of five components drogenation/transamination of G418 (5) (Guo et al., 2014) it has
(Figure 1), from the filamentous bacterium Micromonospora been possible to deconvolute the specific roles of some of the
echinospora. Gentamicins are protein synthesis inhibitors used individual genes in this final stage of the pathway. Also, Liu
to combat Gram-negative bacterial infections. They are also be- and colleagues have demonstrated the activity in vitro of
ing explored in other therapeutic areas (Hainrichson et al., 2008; GenK, which methylates 4 to produce 5, and have confirmed
Linde and Kerem, 2008; Cuccarese et al., 2013). However, gen- that it is a cobalamin-dependent radical S-adenosyl-L-methio-
tamicins carry a serious risk of kidney damage and hearing loss nine (SAM) enzyme (Kim et al., 2013).
(Bockenhauer et al., 2009) which limits their utility, and it is there-
Here we present a similar in vivo and in vitro analysis to define
fore encouraging that evidence is available that an individual the enzymology of the pathway from the first pseudotrisacchar-
component of the gentamicin mixture may have lower toxicity ide 3 to the last common precursor of the gentamicin C complex,
(Sandoval et al., 2006; Kobayashi et al., 2008). This makes the gentamicin X2 (4). It has already been established that disruption
Chemistry & Biology 22, 251–261, February 19, 2015 ª2015 The Authors 251

Figure 1. The Pathway to Gentamicin X2
The roles revealed in this study of GenD2, GenS2, GenN, and GenD1 are highlighted in orange. Shunt products of GenK are shown in blue. For details, see the
text. A2 (3): gentamicin A2; DOA2 (8): 300-dehydro-300-oxo-gentamicin A2; DAA2 (9): 300-dehydro-300-amino-gentamicin A2; X2 (4): gentamicin X2; A2e (7):
60-methylgentamicin A2; Ae (6): 60-methylgentamicin A. Carbon numbering is shown with the structure of 3.
252 Chemistry & Biology 22, 251–261, February 19, 2015 ª2015 The Authors

of the gene genD1, also known as gntE (Unwin et al., 2004) or frame in both genD2 and genK was constructed (Figure S1C).
gtmI (Kharel et al., 2004c), by a thiostrepton resistance gene LC-ESI-HRMS analysis confirmed that this mutant accumulated
(tsr) leads to accumulation of 3 (Figure 1) (Kim et al., 2008). These 3 but not 7 (Figure 2E). Complementation of the DgenD2DgenK
authors interpreted their result as meaning that GenD1 catalyzes strain with plasmid pWHU67 containing genK under the control
N-methylation of the 300-amino group in 3; but several alternative of the PermE* promoter (Figure S1K), restored the coproduction
explanations of this result are possible, especially as polar ef- of the shunt product 7 (Figure 2I).
fects on the downstream genS2 gene were not ruled out. By
GenS2, the product of the unassigned gene genS2 (also
analysis of the gentamicin-related metabolites accumulated in known as gntF [Unwin et al., 2004] or gtmD [Kharel et al.,
strains bearing single and multiple specific gene deletions, by 2004c]) is a plausible candidate to partner GenD2 in amination
following the bioconversion of specific intermediates fed to at C-300 because it is predicted to encode a pyridoxal phos-
such mutants, and by reconstituting all of the steps using purified phate-dependent aminotransferase with significant sequence
recombinant enzymes, we demonstrate here that in fact the identity to Sis15, KanS2, and TobS2, enzymes catalyzing similar
respective (and essential) catalytic contributions of the dehydro- reactions in other aminoglycoside pathways. Accordingly, this
genase GenD2, the pyridoxal phosphate-dependent amino- gene was knocked out by targeted in-frame deletion and the
transferase GenS2, the SAM-dependent N-methyltransferase mutant was confirmed by PCR and Southern hybridization (Fig-
GenN, and the radical SAM-dependent C-methyltransferase ure S1B). LC-ESI-HRMS analysis confirmed that fermentation of
GenD1 are as shown in Figure 1. We also provide evidence for this mutant only produced 3 and 7 (Figure 2D). Complementation
the ability of GenK to act precociously on 3 and gentamicin A of the DgenS2 mutant, carried out by using plasmid pWHU115
(6) (Figure 1) to generate shunt metabolites.
containing genS2 under the control of the PermE* promoter (Fig-
ure S1H), restored the production of gentamicin C complex and
of various intermediates to wild-type levels (Figure 2H). Chemical
complementation of the DgenS2 mutant by feeding 6 also simi-
larly restored production of gentamicin C complex (Figure S3D),
confirming that the genS2 gene, like genD2, is essential and acts
RESULTS
GenD2, GenS2, and GenN Are Essential for Conversion
of Gentamicin A2 into Gentamicin A
The gene genD2 (also known as gntC [Unwin et al., 2004] or gtmC exclusively in the section of the pathway between 3 and 6,
[Kharel et al., 2004c]) is a candidate to act on 3 because it is pre- consistent with a specific role in partnering GenD2 to accomplish
dicted to encode an NAD(P)H-linked dehydrogenase with signif- amination at C-300. Also, the shunt product gentamicin 7 was not
icant sequence identity to Sis12, KanD2, and TobD2, enzymes present in a DgenS2DgenK double mutant (Figures 2F and S1D)
catalyzing similar reactions in other aminoglycoside pathways. but its appearance was restored by complementation in trans by
To investigate the proposed role of the GenD2 dehydrogenase constitutive expression of genK (Figures 2J and S1L). The pre-
in the specific oxidation of 3 at the C-300-hydroxyl, this gene sumed intermediate in amination at C-300 is 300-dehydro-300-
was knocked out by targeted in-frame deletion of a 930 bp inter- oxo-gentamicin A2 (DOA2 [8]; Figure 1), and this was not
nal fragment (Figure S1A available online). The mutant was detected in the DgenS2 or DgenS2DgenK mutants. Either the
confirmed by PCR and Southern hybridization (Figure S1A). compound is unstable, or the GenD2-catalyzed oxidation does
Liquid chromatography/electrospray ionization/high-resolution not proceed if the ketone product is not continuously removed
mass spectrometry (LC-ESI-HRMS) analysis confirmed that (see below).
fermentation of this mutant produces elevated levels of 3, but
The third and final step between 3 and the known intermediate
no other gentamicins normally seen in the wild-type (Figures 6 involves N-methylation of the newly introduced 300-amino
2C and S2A). A second new species was also observed, which group in the garosamine ring (Figure 1). Of the several predicted
from LC-ESI-HRMS and tandem mass spectrometry (MS/MS) methyltransferase genes in the gentamicin gene cluster, genN
analysis (Figure S2B) appears to represent a derivative of 3 meth- (GI: 85814038; Aboshanab, 2005) is an attractive candidate for
ylated at the C-60 position. This species was first detected by Kim this role. It is predicted to encode a conventional SAM-depen-
et al. (2008) as a by-product of their genD1 disruptant and was dent methyltransferase, and its closest known homolog (90%
named by them gentamicin A2e (7; Figure 1) but not character- sequence identity) is Sis30 in the cluster for sisomicin ([19];
ized. Complementation of the DgenD2 mutant, carried out by Hong et al., 2009), a closely related aminoglycoside which re-
using plasmid pWHU184 containing genD2 under the control of quires a similar N-methylation (Supplemental Experimental Pro-
the PermE* promoter (Figure S1G), restored the production of cedures). The gene clusters for the aminoglycosides kanamycin
gentamicin C complex and of various intermediates to wild- (11) and tobramycin (12), which do not contain a 300-N-methyl
type levels (Figure 2G). Chemical complementation of the group (Supplemental Experimental Procedures), contain no
DgenD2 mutant by feeding 6 also similarly restored production gene homologous to GenN.
of gentamicin C complex (Figure S3C), confirming the primary
role of GenD2 in the section of the pathway between 3 and 6.
GenN was knocked out by targeted in-frame deletion of a
933 bp internal fragment (Figure S1E). LC-ESI-HRMS analysis
Recent in vivo (Kim et al., 2008; Hong and Yan, 2012; Guo revealed that fermentation of this mutant produced 3 and 7 (Fig-
et al., 2014) and in vitro (Kim et al., 2013) work has unequivocally ure 2L), together with small amounts of a compound with the pre-
established that methylation at the C-60 position is normally car- dicted mass of 300-dehydro-300-amino-gentamicin A2 (DAA2 [9];
ried out on 4 by the SAM-dependent methyltransferase GenK. Figures 1 and S2C), but none of the gentamicin C components
We suspected therefore that the appearance of 7 in our DgenD2 was detected. Trace amounts of demethylgentamicin C1, deme-
mutant was a consequence of the action of GenK in the absence thylgentamicin C2, and demethylgentamicin C1a were observed.
of its normal substrate. To test this, a double mutant deleted in- The fragmentation profiles of these demethylgentamicins
Chemistry & Biology 22, 251–261, February 19, 2015 ª2015 The Authors 253

(legend on next page)
254 Chemistry & Biology 22, 251–261, February 19, 2015 ª2015 The Authors

indicate that they lack a methyl group in the garosamine ring. LC-HRMS. GenD2, GenS2, and GenN were each expressed in
Complementation of the DgenN mutant, carried out by using Escherichia coli as soluble N-His6-tagged proteins and purified
plasmid pWHU68 containing genN under the control of the to near homogeneity (Figure S4A). The molecular weights of
PermE* promoter (Figure S1I), restored the production of genta- the proteins as analyzed by LC-MS were in agreement with the
micin C complex and of various intermediates to wild-type levels calculated values assuming loss of the N-terminal methionine:
(Figure 2N). Chemical complementation of the DgenN mutant by 38,533 Da for GenD2 (calculated 38,539 Da); 48,820 Da
feeding gentamicin A restored production of gentamicin C com- for GenS2 (calculated 48,826.5 Da); and 37,218 Da for GenN
plex (Figure S3E), confirming that the genN gene, like genD2 (calculated 37,221.8 Da), respectively. Adventitious N-gluconyl
and genS2, is essential and acts exclusively in the section of modification (+178 Da) (Geoghegan et al., 1999) was also seen
the pathway between 3 and 6, consistent with a specific role in with these proteins.
N-methylation of the C-300 amino group.
First, the predicted dehydrogenase GenD2 was incubated
The remaining step in the conversion of 3 to 4 is the C-methyl- with 3, and either NAD+ or NADP+ as redox cofactor, at 30ꢀC.
ation of 3 at the C-400 position (Figure 1). The gene genD1 (also Samples were analyzed using LC-ESI-HRMS after 10 min,
known as gntE (Unwin et al., 2004) or gtmI (Kharel et al., 60 min, 90 min, and overnight incubation. However, no DOA2
2004c), previously assigned as an oxidoreductase/methyltrans- (8) was detected, even after overnight incubation. Addition of
ferase (Kim et al., 2008), is a strong candidate to encode this the aminotransferase GenS2, and L-glutamate as amino donor,
methyltransferase, because it shows significant sequence iden- also failed to reveal any of the expected DAA2 (9). In contrast,
tity to authentic cobalamin-dependent and radical SAM-depen- when all three enzymes (GenD2, GenS2, and GenN) were pre-
dent methyltransferases (Kim et al., 2013). Accordingly, the sent, as well as NAD+, L-glutamate, and SAM (as a methyl
genD1 gene was knocked out by in-frame deletion of an donor), essentially complete (>96%) conversion of 3 (Figure 3A)
1,851 bp internal fragment (Figure S1F). In contrast to previous to 6 (Figure 3B) was achieved within 10 min at 30ꢀC. The identity
findings based on disruption of the genD1 gene (Kim et al., of the product was confirmed by LC-ESI-HRMS analysis, and
2008), LC-ESI-HRMS analysis showed that fermentation of this comparison with authentic 6: [M+H]+ m/z = 469.2498 (ꢁ1.28
mutant produced not only 3 and 7 but also 6 (Figures 2K and ppm) and [M + Na]+ m/z = 491.2319 (ꢁ1.02 ppm) (Figure 3B).
S2D). The most abundant peak was a new species with the Reactions lacking any one of the three enzymes did not yield
mass of 4 but with a different retention time and MS/MS frag- any product. These results strongly support the hypothesis
mentation pattern (Figure S2E). The appearance of this species that the 300-modifications on 3 leading to 6 formation are indeed
(called here gentamicin Ae [10]) is consistent with GenK-cata- catalyzed by the coupled activities of GenD2, GenS2, and
lyzed methylation of 6 at C-60 in the absence of its normal genta- GenN. It is reasonable to speculate that the activities of
micin X2 (4) substrate. Complementation of the DgenD1 mutant GenD2 and GenS2 may be inhibited by very low levels of their
in trans, using a plasmid housing genD1 under a constitutive pro- products and that subsequent 300-methylation by GenN allevi-
moter (Figure S1J), restored wild-type levels of gentamicins (Fig- ates this product inhibition.
ure 2M). Chemical complementation of the DgenD1 mutant by
feeding 6 did not restore production of gentamicin C complex The Aminoglycosides Kanamycin B and Tobramycin
(Figure S3F), but complementation by feeding with 5 did restore Are Deaminated by GenS2 and GenD2 and Methylated
the production of gentamicin C2a, C2, and C1, the products in by GenN
the ‘‘methylated branch’’ (Figure S3G). This confirmed that the To help confirm the respective roles of GenD2 and GenS2, kana-
genD1 gene is essential and acts exclusively in the conversion mycin B (11) and tobramycin (12) were used. These aminoglyco-
of 6 to 4, consistent with a specific role in C-methylation at the side antibiotics are close structural analogs of DAA2 (9), and both
C-400 position. It appears likely that in the disruption mutant pre- possess an unmethylated 300-amino group, so they could be
viously studied (Kim et al., 2008), there was a deleterious polar tested as surrogate substrates for 300-deamination, the reverse
effect on the downstream genS2 gene, which would account of the reaction normally catalyzed by GenS2/GenD2. In these
for the lack of accumulation of 6 in that study.
assays, using 2-oxoglutarate as the amino acceptor, LC-ESI-
HRMS analysis showed that GenS2 catalyzed low-level deami-
nation of both compounds. The change could be localized to
the kanosamine ring, consistent with the production, respec-
tively, of trace amounts of 300-deamino-300-oxo-kanamycin B
(13; [M + H]+ m/z 483.2287, [M + Na]+ m/z 505.2113) (Figure S5B)
Purified Recombinant GenD2, GenS2, and GenN
Together Catalyze Successive 300-Oxidation,
Transamination, and N-Methylation of Gentamicin A2 to
Form Gentamicin A
To obtain direct evidence for the catalytic roles of GenD2, and 300-deamino-300-oxo-tobramycin (14; [M + H]+ m/z 467.2342)
GenS2, and GenN, we turned to in vitro reconstitution experi- (Figure S5F). In contrast, in the presence of GenS2 and GenD2
ments, starting with gentamicin A2 (3). This substrate was puri- (and either NADH or NADPH) essentially quantitative yields
fied from fermentation extracts of the DgenS2DgenK strain of were obtained of, respectively, 300-deamino-300-hydroxy-kana-
M. echinospora (Figure 2F), and its identity was confirmed by mycin B (15; [M + H]+ m/z 485.2443, [M + Na]+ m/z 507.2263)
Figure 2. Production of Gentamicins by Micromonospora echinospora Mutants
LC-ESI-HRMS total ion current traces of (A) gentamicin standard; and of mutant fermentation culture extracts from (B) wild-type; (C) DgenD2 mutant; (D) DgenS2
mutant; (E) DgenD2DgenK mutant; (F) DgenS2DgenK mutant; (G) DgenD2::genD2 mutant; (H) DgenS2::genS2 mutant; (I) DgenD2DgenK::genK mutant; (J)
DgenS2DgenK::genK mutant; (K) DgenD1 mutant; (L) DgenN mutant; (M) DgenD1::genD1 mutant; (N) DgenN::genN mutant. For the structure of metabolites, see
Figure 1.
Chemistry & Biology 22, 251–261, February 19, 2015 ª2015 The Authors 255

295.1146
A
MS/MS
m/z 456.22
456-162+H=295
6.9 min
456.2180
100
163.0630
80
60
40
456-133+H=324
324.1371
m/z 456
(3)
160 200 240 280 320
m/z
478.1999
I% 20
0
0
5
10
15
20
440
480
520
m/z
560
600
Time (min)
B
324.1690
MS/MS
m/z 469.25
469-162+H=308
308.1361
469.2498
9.1 min
100
80
163.0485
469-146+H=324
60
m/z 469
160 200 240 280 320
40
(6)
m/z
20
I%
491.2320
0
0
5
10
15
20
440
480
520
m/z
560
600
Time (min)
Figure 3. Enzymatic One-Pot Conversion of Gentamicin A2 to Gentamicin A
LC-ESI-HRMS selective ion monitoring was carried out on (A) [M + H]+ (m/z 456) and [M + Na]+ (m/z 478) ions of gentamicin A2 (3); (B) [M + H]+ (m/z 469) and
[M + Na]+ (m/z 491) ions of gentamicin A (6), produced by the coupled action of GenD2, GenS2, and GenN on gentamicin A2 (3). MS/MS fragments from the
[M + H]+ (m/z 456) and [M + H]+ (m/z 469) ions are shown as insets.
(Figure S5C) and 300-deamino-300-hydroxy-tobramycin (16; [M + ever, 15 was smoothly converted by GenS2, GenD2, and GenN
H]+ m/z 469.2496, [M + Na]+ m/z 491.2314) (Figure S5G).
11 and 12 were excellent substrates for SAM-dependent
into 17.
methylation by GenN, the reaction being complete after 1 hr at The Radical SAM Methyltransferase GenD1 Catalyzes
30ꢀC. For kanamycin B GenN showed kcat = 1.93 0.1 sꢁ1 and Cobalamin-Dependent Methylation of Gentamicin A to
Km = 34 3.8 mM (see Supplemental Experimental Procedures Produce Gentamicin X2
for assay details). LC-ESI-HRMS analysis confirmed that methyl- The putative radical SAM enzyme GenD1 is one of two such
ation in both cases occurred specifically on the kanosamine intriguing enzymes in the gentamicin pathway, the other being
ring, consistent with the formation, respectively, of 300-N- GenK, which catalyzes C-methylation of gentamicin X2 (4) at the
methyl-kanamycin B (17; [M + H]+ m/z 498.2761 (ꢁ1.81 ppm), C-60 position (Kim et al., 2013). Like GenK, a putative cobal-
[M + Na]+ m/z 520.2582 (ꢁ1.35 ppm)) (Figure S5D) and 300-N- amin-dependent radical SAM C-methyltransferase, and like
methyltobramycin (18; [M + H]+ m/z 482.2814 (ꢁ1.45 ppm), Sis14 from the gene cluster for the closely related aminoglycoside
[M + Na]+ m/z 504.2633 (ꢁ1.39 ppm)) (Figure S5H). In contrast, sisomicin (19; Supplemental Experimental Procedures), GenD1
GenN had no effect on 6, 4, or 5, all of which are gentamicin in- houses a sequence motif (Cx4Cx2C) highly similar to the
termediates already methylated at the 300-N position. These re- conserved Cx3Cx2C binding site for the [4Fe-4S] iron-sulfur clus-
sults further confirm the 300-oxidoreductase activity of GenD2, ter found in the radical SAM superfamily (Kim et al., 2013). The
the 300-aminotransferase activity of GenS2, and the 300-N-methyl- iron-sulfur cluster would mediate homolysis of SAM, triggering
ation activity of GenN. Purified 15, when used as substrate for formation of a gentamicin substrate radical which in turn recruits
the forward reaction, gave neither detectable 13 with GenS2 a methyl group from methylcobalamin. Important preliminary evi-
alone nor detectable 11 with GenS2 and GenD2 together. How- dence has recently been presented for the in vitro activity of GenK,
256 Chemistry & Biology 22, 251–261, February 19, 2015 ª2015 The Authors

Figure 4. C-Methylation of Gentamicin A to
Gentamicin X2 Catalyzed by GenD1
B
A
LC-ESI-HRMS selective ion monitoring was car-
ried out on (A) [M + H]+ m/z 483 of the product of
GenD1-catalyzed methylation of gentamicin A (6);
(B) [M + H]+ m/z 252 of coproduced 50-dAdo (20);
(C) MS and MS/MS spectra of gentamicin A (6); (D)
MS and MS/MS spectra of gentamicin X2 (4).
Different high-performance liquid chromatography
conditions were used for detection of gentamicin
X2 (4) and 50-dAdo (20), respectively, as described
in Supplemental Experimental Procedures. 50-
dAdo, 50-deoxyadenosine; BV, benzyl viologen;
MV, methyl viologen.
X2
5'-dAdo
No enzyme (control)
MV + NADPH
MV + NADPH
BV + NADPH
MV + dithionite
BV + dithionite
0
4
8
12
16
20
15
20
30
0
5
25
10
radical SAM-dependent methyltrans-
ferases, and showed a UV-visible ab-
sorbance peak centered on 415 nm (Fig-
Time (min)
Time (min)
C
ure S4B) as expected for
a protein
containing an iron-sulfur cluster (Duin
et al., 1997; Pierrel et al., 2002; Kim et al.,
2013). The color of the protein faded grad-
ually within hours upon exposure to air.
Freshly purified and reconstituted
GenD1 was assayed for its ability to cata-
lyze C-methylation of 6 at the C-400 posi-
tion to yield 4, using conditions previously
found successful for GenK (Kim et al.,
2013). Incubations (12 hr) were carried
out in an anaerobic chamber at 30ꢀC
and the reaction mixtures were analyzed
by using LC-ESI-HRMS. Reaction mix-
tures contained commercially available 6
(0.4 mM), SAM (4 mM), dithiothreitol
(DTT) (10 mM) and methylcobalamin
(1 mM) in 50 mM Tris-HCl buffer (pH
8.0). The reaction was initiated by addition
of GenD1 (50 mM). With sodium dithionite
(4 mM) as the source of electrons, and
with either methyl viologen (MV) or benzyl
viologen (BV) (each at 1 mM) present to
mediate reductive activation of the iron-
sulfur center, only modest (4%–5%)
conversion of 6 to a methylated product
was seen. However, conversion was
increased to 84%–88% when the source
of electrons was NADPH (4 mM) (Fig-
324.1396
MS/MS
m/z 469.25
469.2495
469-162+H=308
100
80
308.2100
163.0801
469-146+H=324
m/z 469
60
40
(6)
150
200
250
300
350
400
491.2313
507.2053
m/z
I%
20
0
400
440
480
520
560
600
m/z
D
324.1305
MS/MS
m/z 483.26
483-162+H=322
483.2652
100
80
322.1928
163.0460
483-160+H=324
60 m/z 483
40
20
0
150
200
250
300
350
400
(4)
m/z
I%
505.2472
520
400
440
480
560
600
m/z
including the involvement of cobalamin (Kim et al., 2013). GenD1 ure 4A). The methylated product had a mass consistent with
was accordingly expressed in E. coli as a soluble N-His6-tagged that of 4: [M + H]+ m/z 483.2650 (ꢁ2.28 ppm); [M + Na]+ m/z
protein and purified to near homogeneity (Figure S5A). The molec- 505.2468 (ꢁ2.37 ppm). Its MS-MS fragmentation pattern, iden-
ular weight of purified recombinant GenD1 as determined by LC- tical to that of authentic 4 (Figure 4B), confirmed that the methyl-
MS analysis was 77,410 Da, consistent with the calculated mass ation had occurred as expected on the C-400 position of 6. As
(77,417.5 Da) of the protein lacking the first methionine. Affinity found for GenK, methylcobalamin could be replaced by hydrox-
purification, buffer exchange to remove imidazole, and reconstitu- ocobalamin (14%–18% conversion), suggesting that methylco-
tion of the iron-sulfur cluster were carried out in an anaerobic balamin can be regenerated by methyltransfer from SAM during
chamber to minimize oxidative damage. Purified GenD1 had a catalysis. 4 was not formed in the absence of either cobalamin or
brownish color. After reconstitution of the iron-sulfur cluster in viologen. Omission of the reconstitution step, or carrying out as-
GenD1 using ferrous ammonium sulfate and sodium sulfide, the says under aerobic conditions, gave little or no activity. The for-
protein solution assumed a gray-black color, as reported for other mation of 50-deoxyadenosine (20, 50-dAdo) as a reaction product
Chemistry & Biology 22, 251–261, February 19, 2015 ª2015 The Authors 257

Figure 5. Production of 50-Deoxyadenosine
during GenD1-Catalyzed Methylation of
Gentamicin A
LC-ESI-HRMS selective ion monitoring and MS/
MS fragmentation of 50-deoxyadenosine (50-dAdo)
(20), [M + H]+ m/z 252.
gentamicin A2 (3) and gentamicin A (6)
could nevertheless be obtained by study-
ing the reactions in reverse, using as sub-
strate analogs of 6 the structurally related
was also confirmed by LC-ESI-HRMS ([M+H]+ m/z 252.1089 aminoglycosides kanamycin B (11) and tobramycin (12), which
(ꢁ1.19 ppm)) (Figures 4A and 5), suggesting a similar catalytic each have a free 300-amino group. GenS2 alone did not catalyze
mechanism to that proposed for GenK, in which a 50-dAdo transamination of either substrate but each was efficiently con-
radical generated via the reductive cleavage of SAM abstracts verted into the respective 300-hydroxy-derivative by the joint
a hydrogen from the substrate, leading to the formation of a sub- action of GenS2 and GenD2. The 300-hydroxy-kanamycin B
strate radical (Kim et al., 2013).
(15), when used as a substrate for the forward reaction, behaved
exactly like gentamicin A2 (3): no reaction with GenD2 and
GenS2 unless all three enzymes (GenD2, GenS2, and GenN)
were present, strongly suggesting that the presence of GenN
DISCUSSION
The confident application of synthetic biology to reconfigure relieves product inhibition by 13 and 11.
gene sets and thus redirect the course of an antibiotic biosyn-
It has been previously suggested (Kim et al., 2008) that GenD1
thetic pathway requires an excellent understanding of the under- catalyzes the N-methylation at C-300 of gentamicin A (6), but it is
lying enzymology. For the biosynthetic pathway to the genta- clear from our present work that GenN governs this step. Rather,
micin C complex, an established drug to treat life-threatening GenD1 can be confidently assigned, on the basis of both in vivo
infections caused by Gram-negative bacteria, even the task of gene disruption and direct in vitro assay, as the C-methyltrans-
assigning functions to individual gene products involved in the ferase that transforms gentamicin A (6) into gentamicin X2 (4).
pathway is not trivial because the cluster contains multiple, After GenK (Kim et al., 2013), GenD1 is the second representa-
mutually homologous genes for methyltransferases, oxidore- tive of this mechanistically intriguing class of cobalamin-depen-
ductases, and aminotransferases with potential for overlapping dent radical SAM methyltransferases to be characterized in the
substrate specificities. There are numerous precedents from gentamicin pathway. Both these enzymes activate and achieve
other aminoglycoside pathways for enzyme promiscuity or substitution at unactivated sp3 C centers. Unlike GenK, which
even dual function (Huang et al., 2007; Fan et al., 2008; Yo- has required refolding from inclusion bodies after heterologous
koyama et al., 2008; Park et al., 2011). From our analysis of the expression (Kim et al., 2013), GenD1 is expressed in E. coli as
DgenD2, DgenS2, and DgenN mutants of M. echinospora it is a soluble protein in excellent yield, which should be helpful in
clear that GenD2 (dehydrogenase), GenS2 (aminotransferase), future mechanistic study of this enzyme.
and GenN (N-methyltransferase) uniquely govern the replace-
ment of the C-300-hydroxyl group in gentamicin A2 (3) by the SIGNIFICANCE
methylamino group of gentamicin X2 (4). No other enzymes in
M. echinospora were able to take over these roles under the Aminoglycosides, mainly produced by actinobacteria,
fermentation conditions used, and these enzymes are appar- constitute a vital clinical asset. Gentamicin in particular is
ently not required elsewhere in the pathway. The appearance of continuing interest because of its remarkable potency in
in these mutants of the shunt metabolite gentamicin A2e (7) is treating systemic Gram-negative infections, and yetbecause
readily explained by the action of the methyltransferase GenK of the significant known toxicity of the gentamicin complex it
on the earlier precursors gentamicin A2 (3) and gentamicin A requires constant and expensive individual monitoring of pa-
(6) in the absence of its normal substrate.
tients. The perspective of a safer and less expensive genta-
The selective production of gentamicin A2 (3) from the micin through administration of a single component, or using
DgenD2DgenK mutant provided the substrate to study the re- a single component for semisynthesis of a novel derivative, is
combinant enzymes in vitro. The in vitro work, in turn, confirmed therefore very attractive. The work we report here has used
the conclusions from the initial analysis of specific gene knock- complementary in vivo and in vitro approaches to identify
outs but importantly also revealed that the GenD2/GenS2- the four key enzymes that lead from the first-formed pseudo-
catalyzed dehydrogenation/amination only proceeds when the trisaccharide to gentamicin X2, the most advanced common
gentamicin A (6) product is removed (by GenN-catalyzed methyl- precursor of all the components of gentamicin C complex. It
ation) as fast as it is formed. The simplest explanation for this hasconfirmed the activity of soluble recombinant GenD1 as a
observation is the relief of product inhibition, a neat way to avoid cobalamin-dependent radical SAM methyltransferase in this
unwanted buildup of reactive pathway intermediates. Indirect pathway, paving the way for detailed mechanistic study of
confirmation of the likely nature of those intermediates between this intriguing enzyme; and, by more closely defining the
258 Chemistry & Biology 22, 251–261, February 19, 2015 ª2015 The Authors

these plasmids were introduced individually into DgenD2 (pWHU184),
DgenS2 (pWHU115), DgenN (pWHU68), DgenD1 (pWHU66), DgenD2DgenK
(pWHU67), and DgenS2DgenK (pWHU67) by conjugation. Complemented
exconjugants were verified based on thiostrepton resistance and confirmed
by PCR (Figure S1).
molecular enzymology of the pathway in M. echinospora, has
brought closer the goal of assembling a defined set of en-
zymes to deliver single gentamicin C components.
EXPERIMENTAL PROCEDURES
Targeted In-Frame Gene Deletion
Bacterial Strains and Plasmids, Chemicals, and Culture Conditions
E. coli strains NovaBlue and BL21(DE3) (Novagen) were used as cloning and
expression hosts, respectively. For routine cloning E. coli strains were main-
tained in 23 TY medium (tryptone 1.6%, yeast extract 1%, NaCl 0.5%) at
37ꢀC with appropriate antibiotic selection at the indicated final concentrations:
ampicillin (100 mg/ml), apramycin (50 mg/ml), kanamycin (25 mg/ml), or chloram-
phenicol (25 mg/ml). For E. coli protein expression, Luria-Bertani (LB) medium
(1% tryptone, 0.5% yeast extract, 1% NaCl) was used at 37ꢀC with appropriate
antibiotic selection at the indicated final concentrations: tetracycline (15 mg/ml)
for NovaBlue cells, kanamycin (50 mg/ml) for cells harboring the recombinant
pET28a(+) plasmids, and carbenicillin (100 mg/ml) for plasmid pDB1282,
containing essential genes for biosynthesis of the iron-sulfur cluster (Zheng
et al., 1998). Ferrous ammonium sulfate, LB medium, imidazole, NaCl, and
Tris base were purchased from Fisher Scientific. Restriction endonucleases,
Pfu DNA polymerases, and T4 DNA ligase were from Fermentas (Thermo Sci-
entific). Oligonucleotide primers were synthesized by Invitrogen Life Technolo-
gies. Gentamicin A (6), gentamicin X2 (4), and kanamycin B (11) were products
from Toku-E. G418 used for feeding studies was from Gibco. Amino acids, hy-
droxocobalamine, methylcobalamine (MeCbl), MV, BV, SAM, sodium dithion-
ite, sodium sulfide, and tobramycin (12) were obtained from Sigma-Aldrich.
M. echinospora ATCC15835 wild-type and mutants were grown on ATCC
172 medium (glucose 1%, yeast extract 0.5%, soluble starch 2%, N-Z amine
0.5%, CaCO3 0.2%) at 28ꢀC for genomic DNA isolation and cultivation. For
aminoglycoside production, the seed culture was shaken at 220 rpm in liquid
ATCC 172 medium at 28ꢀC for 2 days and then used to initiate the fermentation
culture (5% inoculum) in fermentation medium (Soybean powder 2.0%,
peptone 0.1%, glucose 0.3%, (NH4)2SO4 0.03%, CaCO3 0.3%, KNO3
0.03%, CoCl2 5 ppm) shaken at 220 rpm and at 28ꢀC for 5 days. For feeding
experiments, filter-sterilized compounds (150 mg/ml) were added to the
fermentation medium before inoculation.
To create individual in-frame deletion mutants of genD2, genS2, genN,
and genD1, the corresponding plasmid pWHU6, pWHU21, pYH289, and
pYH287 was introduced into the wild-type strain by conjugation from E. coli
ET12567/pUZ8002 (MacNeil et al., 1992) on ABB medium (soytone 0.5%, sol-
uble starch 0.5%, CaCO3 0.3%, MOPS 0.21%, FeSO4 0.0012%, thiamine-HCl
0.001%, agar 3%) with addition of 10 mM MgCl2 solution. After 10 hr incuba-
tion at 28ꢀC, the exconjugants were selected with nalidixic acid (12.5 mg/ml)
and apramycin (12.5 mg/ml). Exconjugants were transferred onto ABB medium
containing nalidixic acid (25 mg/ml) and apramycin (25 mg/ml). To promote a
second crossover, these mutants were propagated on A medium (soluble
starch 1%, corn steep powder 0.25%, yeast extract 0.3%, CaCO3 0.3%,
FeSO4 0.0012%, agar 3% [pH 7.0], adjusted with KOH) with addition of
10 mM MgCl2 solution. To select the recombinant progeny by their apramy-
cin-sensitive phenotype, single colonies on A plates were patched in parallel
onto antibiotic (25 mg/ml apramycin)-containing A medium and antibiotic-
free A medium. The desired in-frame deletion mutants were identified by
PCR using the checking primers and further confirmed by Southern blot anal-
ysis (Figures S1A, S1B, S1E, and S1F). Double in-frame deletion mutants,
DgenD2DgenK and DgenS2DgenK, were prepared using the same protocol
with DgenD2 and plasmids pWHU1 (Figures S1C and S1D).
Gene Complementation of the DgenD2, DgenS2, DgenN, DgenD1,
DgenD2DgenK, and DgenS2DgenK Mutants
The complementation plasmids were introduced individually into DgenD2
(pWHU184), DgenS2 (pWHU115), DgenN (pWHU68), DgenD1 (pWHU66),
DgenD2DgenK (pWHU67), and DgenS2DgenK (pWHU67) by conjugation.
Complemented exconjugants were identified based on thiostrepton resis-
tance and confirmed by PCR (Figures S1G–S1L).
Cloning of genD1, genD2, genN, and genS2 Genes for Expression in
E. coli
Isolation and Analysis of Gentamicin and Intermediates
The genD2, genS2, genN, and genD1 genes were each amplified from the
genomic DNA of M. echinospora ATCC15835 by PCR using Pfu DNA polymer-
ase with 25 cycles of denaturing at 94ꢀC for 1 min, annealing at 60ꢀC for 1 min,
and extension at 72ꢀC for 2 min plus a final extension at 72ꢀC for 10 min. The
PCR products were digested with appropriate restriction enzymes, purified
by gel extraction (Fermentas), and inserted into a pET28a(+) plasmid. The
resulting constructs were verified by DNA sequencing.
The fermentation broth was adjusted to pH 2.0 with concentrated HCl. The
cultures were agitated for 2 hr on a laboratory rocker and then centrifuged at
5000 g for 10 min at 4ꢀC. Each supernatant was applied to a column of
1.5 g DOWEX 50WX8-200 ion-exchange resin preconditioned with 50 ml
acetonitrile followed by 50 ml 3 4 distilled-deionized water. The column was
washed with 15 ml of water and then aminoglycosides were eluted with
15 ml of 1 M ammonium hydroxide. The eluate was freeze-dried, the residue
was dissolved in 1 ml of water, and a sample was subjected to high-perfor-
mance liquid chromatography-HRMS analysis.
Overexpression and Purification of Recombinant Proteins
E. coli BL21(DE3) cells bearing the recombinant plasmids were cultured in LB
broth containing kanamycin (50 mg/ml) at 37ꢀC until the cell density reached
0.5–1.0 at 600 nm. Overexpression of the proteins was induced by isopropylth-
iogalactoside (IPTG) (0.1 mM) at 20ꢀC with shaking at 180 rpm overnight. For
overexpression of GenD1, E. coli BL21(DE3) cells harboring both genD1 gene
in pET28a(+) plasmid and pDB1282 plasmid containing the iron-sulfur cluster
biosynthetic genes (Zheng et al., 1998) were incubated in 1.5 l of LB broth con-
taining kanamycin and carbenicillin in a 2 l flask at 37ꢀC, 150 rpm until the cell
density reached A600 = 0.8 – 1.0. The iron-sulfur cluster protein expression was
induced with 20 mM L-(+)-arabinose and the culture incubated for a further 45–
60 min. The culture was then cooled to room temperature before addition of
0.2 mM IPTG to induce GenD1 protein expression at 20ꢀC, 150 rpm overnight;
2 mM Fe(II)SO4 was also added at this point. Cells were harvested by centrifu-
gation and resuspended in Binding Buffer (0.5 M NaCl, 20 mM Tris-HCl
[pH 7.9]). The Binding Buffer for GenD1 resuspension was purged with N2
for at least 30 min before use. The recombinant protein was released by son-
ication for 4 min using a 2 s on/6 s off cycle and the recombinant protein in the
clarified cell lysate was purified using Co2+ or Ni2+ ion-charged His-Bind metal
chelating resin (Novagen) according to the manufacturer’s instructions. Imid-
azole in the eluted GenD2, GenS2, and GenN solutions was removed by buffer
exchange using Amicon Ultra centrifugal filters (Millipore). The purification
Construction of Gene Disruption Plasmids
For in-frame deletion, DNA fragments flanking each target gene were amplified
from the genomic DNA of M. echinospora ATCC15835 using Phusion DNA
polymerase (New England Biolabs). The PCR products were each cloned
into pUC18, then cut out and cloned together into the Streptomyces-E. coli
shuttle vector pYH7 (Sun et al., 2008) to obtain the following gene disruption
plasmids: pWHU6 (for DgenD2), pWHU21 (for DgenS2), pYH289 (for DgenN),
and pYH287 (for DgenD1). For construction of the DgenD2DgenK and
DgenS2DgenK double mutants, pWHU1 (Guo et al., 2014) was employed for
genK in-frame deletion. All plasmids were verified by sequencing.
Construction of Gene Complementation Plasmids
Complementation plasmids were prepared by cloning genD2, genS2, genN,
genD1, and genK, respectively, into pWHU77 (a plasmid derived from
pIB139 [Wilkinson et al., 2002; Del Vecchio et al., 2003] with the apramycin
resistance gene replaced by a thiostrepton resistance gene) under the control
of the PermE* promoter (Figure S1). The PCR products were inserted into
pWHU77 between the NdeI and EcoRI sites to generate pWHU184,
pWHU115, pWHU68, pWHU66, and pWHU67. After sequence confirmation,
Chemistry & Biology 22, 251–261, February 19, 2015 ª2015 The Authors 259

procedure for GenD1 from the His-Bind purification and onward was carried
out in a DAB-10S anaerobic chamber (Saffron Scientific). Imidazole was
removed from the purified GenD1 protein solution using a PD10 column
(GE Healthcare) with a buffer containing 0.25 M NaCl and 20 mM Tris-HCl
(pH 7.9). Anaerobic buffers were prepared outside the anaerobic chamber
and autoclaved. Autoclaved buffers were then purged with nitrogen for at least
30 min with stirring before being transferred to the anaerobic chamber, and
allowed to stir uncapped at least overnight. Chemical solutions to be stored
at ꢁ20ꢀC were prepared within the anaerobic chamber and transferred
immediately to ꢁ20ꢀC after removal from the anaerobic chamber. Chemical
solutions freshly prepared on the day of use were dissolved in anaerobic water
inside the anaerobic chamber. Tips, tubes, and other consumables were
autoclaved and stored in the anaerobic chamber.
The purified proteins were stored at ꢁ20ꢀC in a buffer containing 250 mM
NaCl, 10 mM Tris-HCl (pH 7.9), and 33% glycerol. The identities of the purified
recombinant proteins were confirmed by SDS-PAGE, UV-visible absorbance
analysis (Cary 100 Bio UV-visible spectrophotometer), and LC-ESI-MS
(ThermoFinnigan). Protein concentrations were determined using Bradford
protein dye reagent (Sigma).
LC-ESI-MS Analyses
Conditions for LC-ESI-MS analyses were as described in Supplemental Exper-
imental Procedures.
SUPPLEMENTAL INFORMATION
Supplemental Information including one table, five figures, and Supplemental
Experimental Procedures can be found with this article online at http://dx.doi.
org/10.1016/j.chembiol.2014.12.012.
AUTHOR CONTRIBUTIONS
F.H., C. H., P.F.L., and Y.S. conceived the experiments; C. H., J.G., X.J., and
X.D. constructed and analyzed mutants; F.H. and E.M. carried out in vitro anal-
ysis; C.H., F.H., E.M., J.G., Z.D., P.F.L., and Y.S. analyzed the results; and
F.H., C. H., P.F.L., and Y.S. wrote the paper.
ACKNOWLEDGMENTS
This work was supported by a project grant from the Medical Research
Council, UK (G1001687) to P.F.L.; and by the 973 and 863 programs from
the Ministry of Science and Technology of China, National Science Founda-
tion of China, and the Translational Medical Research Fund of Wuhan Univer-
sity School of Medicine to Y.S.; E.M. thanks the Gates Cambridge Trust for a
scholarship. We also gratefully acknowledge Dr. Xinzhou Yang, South-
Central University for Nationalities, for his assistance in separation of genta-
micin A2. We thank Dr. Andrew Truman (John Innes Institute) for helpful
discussions.
Reconstitution of the Iron-Sulfur Cluster in GenD1
The reconstitution of the iron-sulfur cluster in GenD1 protein was conducted
under strictly anaerobic conditions immediately after elution from PD-10 col-
umns. The average temperature in the glove box was 30ꢀC. The protein solu-
tion (2.5 ml) was incubated with 5 mM DTT for 15 min followed by addition of
1 mM Fe(II)(NH4)2(SO4)2 and 1 mM Na2S, both were freshly prepared, yielding a
translucent, blackish solution. The solution was left at room temperature for
1 hr, and was then exchanged into a buffer containing 0.25 M NaCl and
20 mM Tris-HCl (pH 7.9) using a PD-10 column. The eluted protein solution re-
mained grayish black, indicating a successful reconstitution. A portion of the
reconstituted protein was transferred to a quartz cuvette with a lid and sealed
with Parafilm, and was immediately subjected to UV-vis scanning. The recon-
stituted protein was stored at ꢁ80ꢀC, in Eppendorf tubes wrapped in Parafilm,
after addition of glycerol to 33% (v/v).
Received: September 22, 2014
Revised: December 3, 2014
Accepted: December 6, 2014
Published: January 29, 2015
REFERENCES
Activity Assays of GenD2, GenS2, and GenN
Gentamicin A2 (3) purified from the DgenS2DgenK mutant of M. echinospora
ATCC15835 (see Supplemental Experimental Procedures, Method S1, for
the protocol of purification) was used as a substrate in assays testing the
activity of GenD2 as a 300-dehydrogenase, GenS2 as a 300-transaminase, and
GenN as a 300-N-methyltransferase. A typical reaction (100 ml) contained sub-
strate (400 mM), NAD+ or NADP+ (2.5 mM), L-glutamate (1 mM, when GenS2
was present), SAM (2 mM, when GenN was present), and purified enzyme(s)
(30 mM each) in Tris-HCl buffer (50 mM, pH 7.5). To test the deamination activ-
ity of GenS2 and the subsequent ketoreductase activity of GenD2 (the reverse
reactions), kanamycin B (11) and tobramycin (12) were used as substrates. A
typical reaction mixture consisted of substrate (400 mM), NADH or NADPH
(2.5 mM), 2-oxoglutarate (1 mM, when GenS2 was present), and purified en-
zyme(s) (30 mM) in Tris-HCl buffer (50 mM, pH 7.5). Kanamycin and tobramycin
were used as substrate analogs to test the putative 300-N-methyltransferase
activity of GenN. A typical reaction mixture contained substrate (400 mM),
SAM (2 mM), and purified GenN (30 mM) in Tris-HCl buffer (50 mM, pH 7.5).
Reaction mixtures were incubated at 30ꢀC and quenched by addition of
chloroform (100 ml) followed by vigorous vortexing. Precipitated protein was
removed by centrifugation. Ten to twenty microliters of the aqueous superna-
tant were analyzed by either LC-ESI-MS or LC-ESI-HRMS.
Aboshanab, K.M.A. (2005). Genetic Studies on the Biosynthesis of the Major
Aminoglycoside Antibiotics: Isolation, Analysis and Comparison of the
Biosynthetic Gene Clusters for 12 Aminoglycoside Antibiotics, PhD Thesis.
¨
(Wuppertal: Bergische Universitat Wuppertal).
Bockenhauer, D., Hug, M.J., and Kleta, R. (2009). Cystic fibrosis, aminogly-
coside treatment and acute renal failure: the not so gentle micin. Pediatr.
Nephrol. 24, 925–928.
Cuccarese, M.F., Singh, A., Amiji, M., and O’Doherty, G.A. (2013).
A
novel use of gentamicin in the ROS-mediated sensitization of NCI-H460
lung cancer cells to various anticancer agents. ACS Chem. Biol. 8,
2771–2777.
Del Vecchio, F., Petkovic, H., Kendrew, S.G., Low, L., Wilkinson, B., Lill, R.,
Corte´ s, J., Rudd, B.A.M., Staunton, J., and Leadlay, P.F. (2003). Active-site
residue, domain and module swaps in modular polyketide synthases. J. Ind.
Microbiol. Biotechnol. 30, 489–494.
Duin, E.C., Lafferty, M.E., Crouse, B.R., Allen, R.M., Sanyal, I., Flint, D.H., and
Johnson, M.K. (1997). [2Fe-2S] to [4Fe-4S] cluster conversion in Escherichia
coli biotin synthase. Biochemistry 36, 11811–11820.
Fan, Q.Z., Huang, F.L., Leadlay, P.F., and Spencer, J.B. (2008). The neomycin
biosynthetic gene cluster of Streptomyces fradiae NCIMB 8233: genetic and
biochemical evidence for the roles of two glycosyltransferases and a deacety-
lase. Org. Biomol. Chem. 6, 3306–3314.
GenD1 Assay
Assays were carried out in an anaerobic chamber (ꢂ30ꢀC) overnight. Standard
reaction mixtures (50 ml) contained 1 mM MeCbl, 10 mM DTT, 4 mM SAM,
100 mM purified reconstituted GenD1, 400 mM substrate, 4 mM NADPH, and
1 mM MV in 50 mM Tris-HCl (pH 8). Alternative reducing agents tested were
combinations of 4 mM NADPH or Na2S2O4 and 1 mM MV or BV. After incuba-
tion, assays were removed from the anaerobic chamber and protein was
precipitated by addition of 50 ml of chloroform. The assays were vortexed
and then centrifuged at 13,000 rpm for 5 min. Ten to twenty microliters of
the aqueous supernatant were analyzed by either LC-ESI-MS or LC-ESI-
HRMS.
Geoghegan, K.F., Dixon, H.B., Rosner, P.J., Hoth, L.R., Lanzetti, A.J.,
Borzilleri, K.A., Marr, E.S., Pezzullo, L.H., Martin, L.B., LeMotte, P.K., et al.
(1999). Spontaneous alpha-N-6-phosphogluconoylation of a ‘‘His tag’’ in
Escherichia coli: the cause of extra mass of 258 or 178 Da in fusion proteins.
Anal. Biochem. 26, 169–184.
Guo, J., Huang, F., Huang, C., Duan, X., Jian, X., Leeper, F., Deng, Z., Leadlay,
P.F., and Sun, Y. (2014). Specificity and promiscuity at the branch point in
gentamicin biosynthesis. Chem. Biol. 21, 608–618.
260 Chemistry & Biology 22, 251–261, February 19, 2015 ª2015 The Authors

Hainrichson, M., Nudelman, I., and Baasov, T. (2008). Designer aminoglyco-
sides: the race to develop improved antibiotics and compounds for the treat-
ment of human genetic diseases. Org. Biomol. Chem. 6, 227–239.
Linde, L., and Kerem, B. (2008). Introducing sense into nonsense in treatments
of human genetic diseases. Trends Genet. 24, 552–563.
Llewellyn, N.M., and Spencer, J.B. (2006). Biosynthesis of 2-deoxystrept-
Hong, W.R., and Yan, L.B. (2012). Identification of gntK, a gene required for the
methylation of purpurosamine C-60 in gentamicin biosynthesis. J. Gen. Appl.
Microbiol. 58, 349–356.
amine-containing aminoglycoside antibiotics. Nat. Prod. Rep. 23, 864–874.
MacNeil, D.J., Occi, J.L., Gewain, K.M., MacNeil, T., Gibbons, P.H., Ruby,
C.L., and Danis, S.J. (1992). Complex organization of the Streptomyces
avermitilis genes encoding the avermectin polyketide synthase. Gene 115,
119–125.
Hong, W.R., Ge, M., Zeng, Z.H., Zhu, L., Luo, M.Y., Shao, L., and Chen, D.J.
(2009). Molecular cloning and sequence analysis of the sisomicin biosynthetic
gene cluster from Micromonospora inyoensis. Biotechnol. Lett. 31, 449–455.
Ota, Y., Tamegai, H., Kudo, F., Kuriki, H., Koike-Takeshita, A., Eguchi, T., and
Kakinuma, K. (2000). Butirosin biosynthetic gene cluster from Bacillus
circulans. J. Antibiot. 53, 1158–1167.
Houghton, J.L., Green, K.D., Chen, W., and Garneau-Tsodikova, S. (2010). The
future of aminoglycosides: the end or renaissance? Chembiochem 11,
880–902.
Park, J.W., Hong, J.S., Parajuli, N., Jung, W.S., Park, S.R., Lim, S.K., Sohng,
J.K., and Yoon, Y.J. (2008). Genetic dissection of the biosynthetic route to
gentamicin A2 by heterologous expression of its minimal gene set. Proc.
Natl. Acad. Sci. USA 105, 8399–8404.
Huang, F.L., Haydock, S.F., Mironenko, T., Spiteller, D., Li, Y., and Spencer,
J.B. (2005). The neomycin biosynthetic gene cluster of Streptomyces fradiae
NCIMB8233: characterisation of an aminotransferase involved in the formation
of 2-deoxystreptamine. Org. Biomol. Chem. 3, 1410–1419.
Huang, F.L., Spiteller, D., Koorbanally, N.A., Li, Y., Llewellyn, N.M., and
Spencer, J.B. (2007). Elaboration of neosamine rings in the biosynthesis of
neomycin and butirosin. Chembiochem 8, 283–288.
Park, J.W., Park, S.R., Nepal, K.K., Han, A.R., Ban, Y.H., Yoo, Y.J., Kim, E.J.,
Kim, E.M., Kim, D., Sohng, J.K., and Yoon, Y.J. (2011). Discovery of parallel
pathways of kanamycin biosynthesis allows antibiotic manipulation. Nat.
Chem. Biol. 7, 843–852.
Karki, S., Kim, J.Y., Park, S.H., and Kwon, H.J. (2012). Gene inactivation study
on gntK, a putative C-methyltransferase gene in gentamicin biosynthesis from
Micromonospora echinospora. J. Kor. Soc. Appl. Biol. Chem. 55, 439–442.
Park, S.R., Park, J.W., Ban, Y.H., Sohng, J.K., and Yoon, Y.J. (2013). 2-deox-
ystreptamine-containing aminoglycoside antibiotics: recent advances in the
characterization and manipulation of their biosynthetic pathways. Nat. Prod.
Rep. 30, 11–20.
Kase, H., Odakura, Y., and Nakayama, K. (1982a). Sagamicin and the related
aminoglycosides: fermentation and biosynthesis I. Biosynthetic studies with
the blocked mutants of Micromonospora sagamiensis. J. Antibiot. 35, 1–9.
¨
Pierrel, F., Bjork, G.R., Fontecave, M., and Atta, M. (2002). Enzymatic modifi-
Kase, H., Shimura, G., Iida, T., and Nakayama, K. (1982b). Biotransformation
of sisomicin and verdamicin by Micromonospora sagamiensis. Agric. Biol.
Chem. 46, 515–522.
cation of tRNAs: MiaB is an iron-sulfur protein. J. Biol. Chem. 277, 13367–
13370.
Sandoval, R.M., Reilly, J.P., Running, W., Campos, S.B., Santos, J.R., Phillips,
C.L., and Molitoris, B.A. (2006). A non-nephrotoxic gentamicin congener that
retains antimicrobial efficacy. J. Am. Soc. Nephrol. 17, 2697–2705.
Kharel, M.K., Basnet, D.B., Lee, H.C., Liou, K., Woo, S.W., Kim, B.G., and
Sohng, J.K. (2004a). Isolation and characterization of the tobramycin biosyn-
thetic gene cluster from Streptomyces tenebrarius. FEMS Microbiol. Lett.
230, 185–190.
Subba, B., Kahrel, M.K., Lee, H.C., Liou, K., Kim, B.G., and Sohng, J.K. (2005).
The ribostamycin biosynthetic gene cluster in Streptomyces ribosidificus:
comparison with butirosin biosynthesis. Mol. Cells 20, 90–96.
Kharel, M.K., Basnet, D.B., Lee, H.C., Liou, K., Moon, Y.H., Kim, J.J., Woo,
J.S., and Sohng, J.K. (2004b). Molecular cloning and characterization of a
2-deoxystreptamine biosynthetic gene cluster in gentamicin-producing
Micromonospora echinospora ATCC15835. Mol. Cells 18, 71–78.
Sun, Y.H., He, X.Y., Liang, J.D., Zhou, X.F., and Deng, Z.X. (2008). Analysis of
functions in plasmid pHZ1358 influencing its genetic and structural stability in
Streptomyces lividans1326. Appl. Microbiol. Biotechnol. 82, 303–310.
Kharel, M.K., Subba, B., Basnet, D.B., Woo, J.S., Lee, H.C., Liou, K., and
Sohng, J.K. (2004c). A gene cluster for biosynthesis of kanamycin from
Streptomyces kanamyceticus: comparison with gentamicin biosynthetic
gene cluster. Arch. Biochem. Biophys. 429, 204–214.
Testa, R.T., and Tilley, B.C. (1975). Biotransformation, a new approach to
aminoglycoside biosynthesis. I. Sisomicin. J. Antibiot. 28, 573–579.
Testa, R.T., and Tilley, B.C. (1976). Biotransformation, a new approach to
Kim, J.Y., Suh, J.W., Kang, S.H., Phan, T.H., Park, S.H., and Kwon, H.J. (2008).
Gene inactivation study of gntE reveals its role in the first step of pseudotrisac-
charide modifications in gentamicin biosynthesis. Biochem. Biophys. Res.
Commun. 372, 730–734.
aminoglycoside biosynthesis. II. Gentamicin. J. Antibiot. 29, 140–146.
Thibodeaux, C.J., Melancon, C.E., III, and Liu, H.W. (2008). Natural-product
sugar biosynthesis and enzymatic glycodiversification. Angew. Chem. Int.
Ed. Engl. 47, 9814–9859.
Kim, H.J., McCarty, R.M., Ogasawara, Y., Liu, Y.N., Mansoorabadi, S.O.,
Levieux, J., and Liu, H.W. (2013). GenK-catalyzed C-60 methylation in the
biosynthesis of gentamicin: isolation and characterization of a cobalamin-
dependent radical SAM enzyme. J. Am. Chem. Soc. 135, 8093–8096.
Unwin, J., Standage, S., Alexander, D., Hosted, T., Jr., Horan, A.C., and
Wellington, E.M. (2004). Gene cluster in Micromonospora echinospora
ATCC15835 for the biosynthesis of the gentamicin C complex. J. Antibiot.
57, 436–445.
Kobayashi, M., Sone, M., Umemura, M., Nabeshima, T., Nakashima, T., and
Hellstro¨ m, S. (2008). Comparisons of cochleotoxicity among three gentamicin
compounds following intratympanic application. Acta Otolaryngol. 128,
245–249.
¨
Wilkinson, C.J., Hughes-Thomas, Z.A., Martin, C.J., Bohm, I., Mironenko, T.,
Deacon, M., Wheatcroft, M., Wirtz, G., Staunton, J., and Leadlay, P.F.
(2002). Increasing the efficiency of heterologous promoters in actinomycetes.
J. Mol. Microbiol. Biotechnol. 4, 417–426.
Kudo, F., and Eguchi, T. (2009). Biosynthetic genes for aminoglycoside antibi-
otics. J. Antibiot. 62, 471–481.
Yokoyama, K., Yamamoto, Y., Kudo, F., and Eguchi, T. (2008). Involvement of
two distinct N-acetylglucosaminyltransferases and a dual-function deacety-
lase in neomycin biosynthesis. Chembiochem 9, 865–869.
Kudo, F., Sucipto, H., and Eguchi, T. (2009). Enzymatic activity of a glycosyl-
transferase KanM2 encoded in the kanamycin biosynthetic gene cluster.
J. Antibiot. 62, 707–710.
Li, D., Li, H., Ni, X., Zhang, H., and Xia, H.Z. (2013). Construction of a genta-
micin C1a-overproducing strain of Micromonospora purpurea by inactivation
of the gacD gene. Microbiol. Res. 168, 263–267.
Zheng, L., Cash, V.L., Flint, D.H., and Dean, D.R. (1998). Assembly of iron-
sulfur clusters. Identification of an iscSUA-hscBA-fdx gene cluster from
Azotobacter vinelandii. J. Biol. Chem. 273, 13264–13272.
Chemistry & Biology 22, 251–261, February 19, 2015 ª2015 The Authors 261