Content uploaded by Yuhui Sun
Author content
All content in this area was uploaded by Yuhui Sun on Nov 15, 2024
Content may be subject to copyright.
Content uploaded by Jian Wang
Author content
All content in this area was uploaded by Jian Wang on Jul 10, 2024
Content may be subject to copyright.
Nature Metabolism
nature metabolism
https://doi.org/10.1038/s42255-024-01024-9Article
Elucidation of genes enhancing natural
product biosynthesis through co-evolution
analysis
Xinran Wang1,11, Ningxin Chen1,11, Pablo Cruz-Morales 2,11, Biming Zhong 1,
Yangming Zhang 1, Jian Wang 3, Yifan Xiao4, Xinnan Fu4, Yang Lin5,
Suneil Acharya6, Zhibo Li1, Huaxiang Deng1, Yuhui Sun 3,7, Linquan Bai4,
Xiaoyu Tang 5 , Jay D. Keasling 1,6,8,9 & Xiaozhou Luo 1,10
Streptomyces has the largest repertoire of natural product biosynthetic gene
clusters (BGCs), yet developing a universal engineering strategy for each
Streptomyces species is challenging. Given that some Streptomyces species
have larger BGC repertoires than others, we proposed that a set of genes
co-evolved with BGCs to support biosynthetic prociency must exist in
those strains, and that their identication may provide universal strategies
to improve the productivity of other strains. We show here that genes
co-evolved with natural product BGCs in Streptomyces can be identied
by phylogenomics analysis. Among the 597 genes that co-evolved with
polyketide BGCs, 11 genes in the ‘coenzyme’ category have been examined,
including a gene cluster encoding for the c of ac tor p yr roloquinoline
quinone. When the pqq gene cluster was engineered into 11 Streptomyces
strains, it enhanced production of 16,385 metabolites, including 36 known
natural products with up to 40-fold improvement and several activated silent
gene clusters. This study provides an innovative engineering strategy for
improving polyketide production and nding previously unidentied BGCs.
Microbial-based natural products and their derivatives comprise a
notable portion of the pharmaceuticals in clinical use1, with polyketides
and their derivatives constituting 20% of the top-selling drugs, generat-
ing over US$20 billion in annual revenue worldwide2. In addition to the
need to enhance production of these drugs for cost-effectiveness, the
rise of antibiotic resistance also calls for discovery of new antibiotics.
Streptomyces have the largest repertoire of natural product biosyn-
thetic gene clusters (BGCs) known so far. However, due to the distinct
physiology and metabolic characteristics among different species, few
engineering strategies can be applied to different Streptomyces hosts
Received: 24 March 2023
Accepted: 6 March 2024
Published online: xx xx xxxx
Check for updates
1Shenzhen Key Laboratory for the Intelligent Microbial Manufacturing of Medicines, Key Laboratory of Quantitative Synthetic Biology, Center for
Synthetic Biochemistry, Shenzhen Institute of Synthetic Biology, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen,
China. 2Novo Nordisk Foundation Center for Biosustainability, Technical University of Denmark, Lyngby, Denmark. 3Key Laboratory of Combinatorial
Biosynthesis and Drug Discovery (Ministry of Education), Wuhan University, Wuhan, China. 4State Key Laboratory of Microbial Metabolism and School of
Life Sciences and Biotechnology, Shanghai Jiao Tong University, Shanghai, China. 5Institute of Chemical Biology, Shenzhen Bay Laboratory, Shenzhen,
China. 6Department of Chemical and Biomolecular Engineering and Department of Bioengineering, University of California, Berkeley, CA, USA. 7School
of Pharmacy, Huazhong University of Science and Technology, Wuhan, China. 8Joint BioEnergy Institute, Emeryville, CA, USA. 9Biological Systems and
Engineering Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA. 10Shenzhen Infrastructure for Synthetic Biology, Shenzhen Institute of
Synthetic Biology, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China. 11These authors contributed equally:
Xinran Wang, Ningxin Chen, Pablo Cruz-Morales. e-mail: xtang@szbl.ac.cn; keasling@berkeley.edu; xz.luo@siat.ac.cn
Nature Metabolism
Article https://doi.org/10.1038/s42255-024-01024-9
the ubiquitous enhancement of natural product production through
the metabolic effect of the PQQ cofactor. This study provides innova-
tive engineering strategies and targets for development of platforms
for polyketide biosynthesis production and drug discovery.
Results
Phylogenomic insights on Streptomyces PKS-enriched family
For this analysis, we integrated a database with high-quality, pub-
licly available genomes of strains that were accurately classified as
members of the Streptomyces clade. To avoid variations in quality or
errors in taxonomic classification, we filtered a set of 7,762 bacterial
genomes from a previously integrated database23; from it, we selected
720 actinobacterial genomes assembled in no more than ten contigs
and with a completeness of 85% or more (Supplementary Data 1). This
cut off allowed for comprehensive retrieval of actinobacterial genomes
including those with several plasmids or those that may be minimally
fragmented and for which enough markers for phylogenetic recon-
struction can be found. We extracted 18 conserved protein markers
from the 720 genomes (Supplementary Table 1 and Supplementary
Fig. 1) and used them to obtain a species tree. Using this tree as refer-
ence, we selected 201 genomes found within the Streptomycetaceae
clade including 17 non-Streptomyces genomes and 184 members of the
genus Streptomyces. These genomes have an average completeness
of 98% and share 604 orthologous proteins (Supplementary Data 2).
This set of orthologues was used to construct a Streptomyces species
tree that was rooted using the 17 non-Streptomyces strains. This tree
showed excellent branch support for most nodes. Therefore, we used
it as the basis for further analyses (Fig. 1).
To search for potential PKS-associated genes, we established 14
clades and recorded the G+C content, genome length and the number
and type of BGC for each strain. We observed that the variations in these
traits correlate with the topology of the tree and with our clade classifi-
cation (Fig. 1). This analysis made evident that clade 12 includes all the
strains with the largest genomes among in our set and it also encodes
the largest number of PKS-related domains and BGCs, this result is
consistent with previous reports17. We proposed that, although these
BGCs may be expressed under different conditions, this outstanding
abundance may be associated with other metabolic traits that support
PKS production.
Conserved pqq gene cluster in polyketide-rich Streptomyces
To explore this hypothesis, we calculated the core genomes (genes
exist in all strains of a clade) for each of the 14 clades. Then, to identify
clade-specific conserved genes, we compared the clade-level core
genomes and looked at the genes that are uniquely conserved in each
clade.
Following this approach, we found 597 genes uniquely conserved
in clade 12 (Supplementary Data 3). To establish whether correlations
between these 597 genes and PKS proficiency exist, we recorded the
presence or absence of each of them in the 201 genomes (Supple-
mentary Data 4). Then, the correlation between the 597 genes and
KS domain abundance was evaluated as point biserial correlations
(Supplementary Data 4). There was a positive correlation (P < 0.005,
correlation >0) in most unique genes (545 out of 597, 91.2%) (Sup-
plementary Data 4 and Supplementary Fig. 2). These results are
consistent with our hypothesis that unique genes identified in the
PKS-proficient clade by phylogenomic analysis are correlated with PKS
abundance.
To identify traits and genes that may explain biosynthetic profi-
ciency, the unique genes were sorted according to their functions into
different families. Of them, 57% could be assigned a function, and clas-
sified in one of 12 families, namely regulators, fatty acids, lipids and pol-
yketide related functions, transporters, carbohydrate-active enzymes
(cazymes), sigma factors, coenzymes, cell wall biosynthesis, metal
related functions, cell division, peptidase, amino acid metabolism
to enhance the titre of natural products or to explore a strain for the
discovery of new antibiotics.
The current engineering strategies developed for natural product
production are mostly based on two strategies: (1) metabolic engineer-
ing to enhance precursor biosynthetic pathways
2
or delete competi-
tor pathways3 or (2) gene regulation engineering such as modulating
pathway regulation or global regulation processes4. These strategies
came from numerous investigations of natural product biosynthetic
processes. However, given the wide diversity of natural product pro-
ducers and pathways, we are far from being able to apply a universal
strategy to improve production of all natural products.
The popularization of sequencing technologies has granted public
access to genomes of natural product producers, allowing the perfor-
mance of bottom-up studies aimed at pinpointing the mechanisms
behind complex traits, and paving the way for methods development
in synthetic biology for natural products overproduction as well as
identification of new molecules. Large-scale genomic studies have
shown that some bacterial and fungal taxa are more talented than
others at making a diverse range of natural products5. This observa-
tion indicates the existence of key genes related to high productivity.
Identifying these genes is important to develop metabolic engineer-
ing and synthetic biology strategies to improve production and
accelerate the discovery of these valuable molecules. Pan-genomic
analysis has been developed as a tool to explore the genomes of a spe-
cies for desired traits, and the corresponding ‘gene-trait-application’
strategy has been successfully applied in many species, including
plants
6
, fungi
7
and bacteria
8
, to look for genes correlated to specific
traits and guide further optimization of the species for the desired
applications.
Streptomyces is the producer of most polyketide drugs and was
shown to encode for more natural product BGCs than other bacte-
ria
9
. Streptomyces is a large genus containing around 22,900 known
species
10
. Genomic and pan-genomic analysis addressing the abun-
dance and distribution of natural product BGCs inside Streptomyces
have shown that the potential to produce natural products diverges
at the species level9,11–15, and the relationship between taxonomy
and biosynthetic proficiency has been established, indicating that
some Streptomyces lineages have more BGCs and potentially pro-
duce more natural products than others
16,17
. However, the identity and
function of the genes associated with biosynthetic proficiency are
not known.
Pyrroloquinoline quinone (PQQ) is a redox cofactor involved in
alcohol or glucose dehydrogenase reactions in many Gram-negative
bacteria18. PQQ participates in a variety of intracellular roles, including
the use of carbon sources19 and the respiratory metabolism related to
oxidative stress resistance18. The biosynthesis of PQQ requires specific
precursors and enzymes encoded by the pqq operon, which contains
five (pqqABCDE) or six biosynthetic genes (pqqABCDEF) depending
on the source organism20,21. In Streptomyces, it has been reported that
PQQ-dependent dehydrogenases participated in lankacidin biosyn-
thesis in Streptomyces rochei 7434AN4 (ref. 22). The linkage of PQQ
with other natural products in other Streptomyces species has not been
previously reported.
In this study, we applied pan-genomic analysis to Streptomyces
species to look for genes that co-evolved with polyketide BGCs. We
found that 597 genes were conserved among a closely related group of
Streptomyces that encode for an outstandingly large number of polyke-
tide synthases (PKSs), including the biosynthesis of the cofactor PQQ.
We confirmed the association of PQQ with biosynthetic proficiency
by introduction or overexpression of the PQQ biosynthetic pathway
in 11 Streptomyces strains and two other industrial actinobacterial
strains. PQQ could effectively enhance natural product production
and activation of silent gene clusters in Streptomyces. Besides previ-
ous studies describing PQQ-dependent dehydrogenases involved in
lankacidin biosynthesis22, this is, to our knowledge, the only report of
Nature Metabolism
Article https://doi.org/10.1038/s42255-024-01024-9
0
50
100
150
200
0
5
10
15
BGC abundance
PKS domain abundance
PQQ biosynthetic pathway
Genome length (MB)
G+C content (%)
20
0
7
8
9
10
12
11
13
70
72
71
74
73
75
Present
Absent
Nucleosides
100
13
14
0.05
1
2
3
4
5
6
7
8
9
10
11
12
Phosphonates
Others
KS in type 1 iterative PKS
Type 2 PKS
Type 3 PKS
Trans-AT PKS
Terpenes
Lantipeptides
Siderophore
Aminoglycosides
Arylpolyenes
Beta-lactams
Indoles
Type 1 PKS
NRPS
ACP in any PKS
KR in any PKS
ER in any PKS
DH in any PKS
AT in any PKS
KS in trans-AT PKS
KS in type1 modular PKS
G+C content
PQQ
Length
Phenazines
Fig. 1 | Phylogenomic analysis of the Streptomyces genus and the number of
PKS gene clusters. The phylogenetic tree is divided into 14 clades highlighted
in different colours. Genome length, G+C content, natural product BGCs, PKS-
related domains and pqq BGCs are shown as coloured boxes to the right. The
values are indicated by legends on the left. KS, ketosynthase; AT, acyltransferase;
DH, dehydratase; ER, enoyl reductase; KR, ketoreductase; ACP, acyl-carrier
protein.
Nature Metabolism
Article https://doi.org/10.1038/s42255-024-01024-9
and other functions. The ‘regulators’ category is the most abundant;
however, as regulators are already considered engineering targets, we
did not further explore them (Fig. 2a). The genes within the ‘fatty acids,
lipids and polyketide related functions’ category include enzymes such
as phosphopantetheinyl transferases, methylmalonyl-coenzyme A
(CoA) mutases, 3-oxoacyl-acyl-carrier-protein synthases, acyl-carrier
proteins, long-chain-fatty-acid-CoA ligases and thioesterases (Sup-
plementary Data 3), some of which have been already pinpointed as
engineering targets for enhancing PKS productivity3.
In addition, there are genes that have few reports associated with
polyketide production, such as coenzymes, peptidase and/or pro-
teases or proteins with metal related functions and so on (Fig. 2a).
This discovery suggests connections between these genes and PKS
production, which could be useful for developing tools to enhance
polyketide production. As a proof of concept, we investigated the con-
nection of genes in the ‘coenzyme’ family with PKS production. There
are 11 genes inside the ‘coenzyme’ family, all of which were found to
have a positive correlation with the PKS genes (Supplementary Data 4
and Supplementary Fig. 2). The 11 genes are responsible for biosyn-
thesis of coenzyme B6, coenzyme Q, PQQ and folate (Supplementary
Table 2).
To test the contribution of genes in ‘coenzyme’ family on natural
product production, all 11 orthologous genes were cloned from the
genome of Streptomyces hygroscopicus XM201 (a strain in clade 12),
placed under control of the strong promoter kasOp*, and introduced
into the model strain Streptomyces coelicolor M145. These five PQQ
biosynthetic genes (pqqA–E) were co-expressed as an operon. The
blue-colour actinorhodin, a polyketide naturally produced by S. coeli-
color, was followed as an indicator of titre for this class of compounds.
As shown in Fig. 2b, introduction of shxm_4480, shxm_1296, shxm_1613,
shxm_3774 and shxm1453-57 (the pqq biosynthetic cassette) had a posi-
tive effect on actinorhodin production, which resulted in 62, 189, 51, 20
and 85% increases in actinorhodin titre, respectively.
Given that all the functions needed to produce PQQ were found
among the biosynthetic proficiency-correlated genes, and that their
expression in S. coelicolor induces polyketide production, we reasoned
that PQQ should play a conserved role in enhancing the metabolism
that leads to polyketide production. The positive effect of PQQ on
actinorhodin production was first confirmed by feeding PQQ molecule
into the fermentation medium (Supplementary Fig. 3). For further
confirmation, truncated pqq gene clusters were constructed. The
biosynthesis of PQQ starts with the ribosomal production of a short
peptide encoded in pqqA, which then undergoes a series of modifica-
tions by pqqB–E to form the characteristic pyridine and pyrrole rings
in PQQ
20
(Fig. 2c). The genes pqqA and pqqC are responsible for the
first and final biosynthetic steps, respectively. To confirm that the
increase in actinorhodin production is caused by PQQ and not by the
isolated function of one or more of the pqq genes, two truncated pqq
gene clusters—one lacking pqqA and the other lacking pqqC—were con-
structed and introduced to S. coelicolor. As shown in Supplementary
Fig. 4, their previously observed positive influence on actinorhodin was
distinctly abolished, affirming that it was the mature PQQ molecule
that enhanced actinorhodin biosynthesis.
Mechanism of pqq operon for enhanced actinorhodin yield
To understand why the presence of PQQ improves production of acti-
norhodin in S. coelicolor, and indeed other natural products in a wide
range of species, as demonstrated later in this work (sections ʻEffects
SHXM_1453-57
pSET152
0
SHXM_4480
SHXM_1574
SHXM_1296
SHXM_1613
SHXM_3774
SHXM_4530
0.5
Actinorhodin (A640 nm)
1.0
1.5
2.0
a
PQQ
pqqEpqqDpqqCpqqBpqqA
**
***
***
***
***
***
NS
b
c
Peptidase and/or proteases
Cell division, cell wall biosynthesis,
replication, translation
Metal related functions
Cell wall biosynthesis,
division and replication
Coenzymes
Sigma factors and
related functions
Cazymes
Transporters
FAs, lipids and
polyketide related
functions
Others
Regulators
300
200
100
Proteins
0
Annotated
functions
(343)
Unknown
functions
(254)
Amino acid metabolism
Fig. 2 | Exploration and evaluation of unique genes in PKS rich clade.
a, Functional classification of the genes uniquely conserved in clade 12. Among
the 597 co-evolved genes, 254 encoded for proteins with unknown functions. The
rest of the 343 genes encoding for proteins with specific functions were mainly
distributed in the family (from large to small) of other functions, regulators,
FAs, lipids and polyketide related functions, transporters, cazymes, sigma
factors, coenzymes, cell wall biosynthesis, metal related functions, cell division,
peptidase and amino acid metabolism. FAs, fatty acids. b, Evaluation of genes
in the ‘coenzyme clade’ by heterologous expression in S. coelicolor. The titre of
actinorhodin was determined by the absorbance at 640 nm (A640). SHXM_4480,
S. coelicolor M145 expressing gene SHXM_4480; SHXM_1574, S. coelicolor M145
expressing gene SHXM_1574; SHXM_1296, S. coelicolor M145 expressing gene
SHXM_1296; SHXM_1613, S. coelicolor M145 expressing gene SHXM_1613; SHXM_3774,
S. coelicolor M145 expressing gene SHXM_3774; SHXM_4530, S. coelicolor M145
expressing gene SHXM_4530; SHXM_1453-57, S. coelicolor M145 expressing gene
SHXM_1453-57 and pSET152, control strain, S. coelicolor M145 introduced with the
backbone vector pSET152. Mean concentrations with error bars showing s.d. are
plotted (n = 3, three biologically independent samples). Multiple comparison
significance was tested to **P < 0.05, ***P < 0.01 by an unpaired two-sided Student’s
t-test. NS, not significant. c, Organization of the pqq operon.
Nature Metabolism
Article https://doi.org/10.1038/s42255-024-01024-9
of the pqq operon on several Streptomycesʼ, ʻpqq operon boosts natural
product biosynthesisʼ and ʻpqq operon improves natural products in
industrial strainsʼ), we then investigated the mechanism in the model
strain S. coelicolor M145.
We first examined the changes in the host proteome in the pres-
ence of the pqq gene cluster by performing a fractionated proteomic
analysis. The results revealed that 299 proteins were upregulated.
We classified these proteins following functional categories defined
by the Kyoto Encyclopedia of Genes and Genomes. The upregulated
proteins (Supplementary Data 5) spanned 12 distinct pathways (ratio
>0.08). Upregulation of bacteria secretion system (ko03070) and
protein exports (ko03060) indicate more active transport of metabolic
substrates and products (Fig. 3a). Notably, several pathways related
to cofactor regeneration—including glycan degradation (ko00511),
oxidative phosphorylation (ko00190) and pyrimidine metabolism
(ko00240)—were upregulated, indicating the potential increase in
intracellular cofactor levels. Moreover, the enhancement of glycer-
ophospholipid metabolism (ko00564) indicated a more robust inter-
conversion between lipids and glycerophospholipid or triacylglyerol
(TAG), a potential carbon source for secondary metabolite production,
especially for polyketides.
In Streptomyces, polyketides are commonly produced during
the stationary phase, and it has been established that TAG and fatty
acids are the direct carbon sources for actinorhodin
3
. Consistently, we
observed production of actinorhodin in our strains with and without
the pqq operon after the glucose was totally consumed (Fig. 3b). Given
these results, we reasoned that the expression of the pqq operon should
cause an increase in the production of cofactors such as ATP, NADH and
NADPH via glycan degradation and oxidative phosphorylation, and that
the mobilization of carbon stored as lipids should lead to an increase
in the acyl-CoA pools. That extra ATP, NADH, NADPH and acyl-CoA
should then account for the increase in polyketide production.
To test this idea, we examined the changes in the abundance of
ten important cofactors involved in natural product biosynthesis on
expression of the pqq operon. As shown in Fig. 3c–e, the intracellular
level of ATP and the ratios of NADPH to NADP and NADH to NAD were
found to increase by 32, 392 and 253%, respectively. No significant
change was observed for other cofactors (Supplementary Fig. 5). These
results indicate that NADPH/NADH and ATP assisted in improving the
product levels by providing additional reducing power and cellular
energy.
We also assessed the changes in TAG abundance during different
growth phases on expression of the pqq operon. A significant increase
in TAG accumulation was observed in the first 48 hours of growth
with the highest at 48 hours, which was twofold higher than the strain
without the pqq gene cluster. In addition, the consumption of TAG after
48 hours has also been accelerated, potentially providing a larger flux
of acetyl-CoA for actinorhodin biosynthesis (Fig. 3f). Increased indi-
vidual and overall fatty acids pools were higher in the pqq-expressing
strain during actinorhodin production (Supplementary Figs. 6 and 7),
which suggests that PQQ-enhanced accumulation and degradation of
TAG leads to a generalized increase in acyl-CoA-dependent pathways.
0
0.05
With PQQ
Without PQQ
With PQQ
Without PQQ
With PQQ
Without PQQ
0.10
0.15
0.20
0.25
ATP (µmol g–1 DW)
0
0.1
0.2
0.3
0.4
0.5
NADH/NAD+
0
0.02
0.04
0.06
NADPH/NADP
+
Labelling amount per cell
dry weight
Labelled
Ribosome
a
Two-component system
Peptidoglycan biosynthesis
Pyrimidine metabolism
Quorum sensing
Sulfur metabolism
ABC transporters
Oxidative phosphorylation
Glycan degradation
Glycerophospholipid metabolism
Bacterial secretion system
Protein export
00.1 0.2
1.0 6
4
2
0
0.8
0.6
1.5 × 107
1 × 107
5 × 106
48 60 72 84 96
0
With PQQ
With PQQ
With PQQ
With PQQ
With PQQ
Without PQQ
Without PQQ
Without PQQ
Without PQQ
Without PQQ
0.4
0.20
0.15
0.10
0.05
0
0.2
ACT (A640 nm)
TAG (mg mg–1 cdw)
0
050
ACT-with PQQ
With PQQ
Unlabelled
Without PQQ
ACT-without PQQ
Glucose-with PQQ
Glucose-without PQQ
100
ATP
NADH/NAD+ NADPH/NADP+ TAG
***
*** ***
Time (h)
050 100
Time (h)
Time (h)
Glucose (mg ml–1)
Ratio
0.3 0.4
bc
d e f g
Fig. 3 | PQQ increases actinorhodin production by enhancing cofactor
biosynthesis and modulating intracellular TAG level. a, Pathway
enrichment analysis of the upregulated proteins. The ratio indicates the
fraction of upregulated proteins enriched in a pathway versus all the proteins
in the particular pathway. b, Actinorhodin production curve and glucose
consumption curve for S. coelicolor with and without the pqq gene cluster. Mean
concentrations with error bars showing s.d. are plotted (n = 3, three biologically
independent samples). ACT, actinorhodin. c–e, Introduction of the pqq BGC
increased the levels of ATP (c), NADH/NAD+ (d) and NADPH/NADP+ (e). DW, dry
weight. Mean concentrations with error bars showing s.d. are plotted (n = 3, three
biologically independent samples). Multiple comparison significance was tested
to **P < 0.05, ***P < 0.01 by an unpaired two-sided Student’s t-test. f, Introduction
of the pqq BGC increased TAG formation and degradation efficiency. Mean
concentrations with error bars showing s.d. are plotted (n = 3, three biologically
independent samples). g, The labelled and unlabelled amount of actinorhodin at
different time points during fermentation after supplementation of 13C-labelled
glucose. Mean concentrations with error bars showing s.d. are plotted (n = 4,
three biologically independent samples). Multiple comparison significance was
tested to **P < 0.05, ***P < 0.01 by an unpaired two-sided Student’s t-test.
Nature Metabolism
Article https://doi.org/10.1038/s42255-024-01024-9
To further study the link between this TAG reserve and production of
actinorhodin, we performed a glucose feeding experiment, in which
unlabelled glucose was fed during the first 24 hours after which the
carbon source was switched to [13C] glucose. The results indicate that
in the strain harbouring the pqq gene cluster, more than half of the acti-
norhodin originates from unlabelled glucose, indicating the elevated
product may be associated with TAG reservoir generated during the
period where the cells were growing on unlabelled glucose (Fig. 3g).
We also determined the intracellular concentration of 121 com-
pounds, including nucleotides, amino acids and various organic
compounds that are building blocks of various natural products. The
concentrations of compounds were quantified throughout the period
of actinorhodin production aiming to provides a comprehensive and
systematic overview of changes during the entire fermentation period.
As shown in Supplementary Table 3 and Supplementary Data 6, the
levels of 41 compounds changed significantly on PQQ treatment (fold
change >1.0, P < 0.05). These compounds include amino acids, car-
bohydrates and tricarboxylic acid and Embden–Meyerhof–Parnas
(EMP) pathway intermediates, purine and pyrimidine ribonucleotides
(pyruvate, asparagine and UDP-glucose and so on), which are precur-
sors to nucleoside antibiotics, alkaloids, peptides and sugar-derived
natural products. We found the concentration of acetyl-CoA, the direct
precursor to actinorhodin, and all polyketides (Supplementary Table 4)
significantly increased from 103 to 126% between 36 and 60 h (Sup-
plementary Fig. 8). This coincided with the degradation of lipids and
the concurrent augmentation of actinorhodin.
In conclusion, our integrated proteomic and metabolomic inves-
tigation revealed that PQQ enhances the biosynthesis of natural prod-
ucts. This enhancement is achieved by augmenting ATP levels as well as
elevating the ratios of NADPH/NADP and NADH/NAD, which are vital
cofactors in natural product biosynthesis. Additionally, PQQ boosts
the supply of substrates, notably acetyl-CoA—a fundamental precur-
sor for numerous natural products—by promoting the accumulation
of triacylglycerol (TAG) stores within the initial 48 hours, followed by
accelerating their catabolism concurrent with the onset of actinorho-
din production.
Effects of the pqq operon on several Streptomyces
Since PQQ was shown to be effective in increasing actinorhodin produc-
tion in S. coelicolor, we speculated that PQQ production may enhance
natural product biosynthesis in other Streptomyces species. To explore
this idea, we introduced the pqq gene cluster into 11 Streptomyces
strains. These strains were selected from eight different clades, encom
-
passing the taxonomic diversity of the genus (Supplementary Fig. 9).
The changes in growth and metabolites production were measured.
There were no observable changes in growth of the 11 strains in which
the pqq gene cluster was introduced, implying that the pqq gene had
a neutral metabolic burden (Supplementary Fig. 10).
Global changes in metabolism were measured using
high-performance liquid chromatography with quadrupole time of
flight (HPLC–Q-TOF) (Fig. 4a). This analysis showed increases in the
abundance of 16,385 metabolites, which is 54% of all the measured
metabolites in 11 strains (Fig. 4b, Supplementary Fig. 9 and Supplemen-
tary Table 5). More than 50% of the metabolites analysed increased on
introduction of the pqq gene cluster in 8 out of the 11 strains (Fig. 4c and
Supplementary Table 6) and in 6 of the 11 strains, several metabolites
increased more than 100,000-fold after integration of the pqq operon.
To assess the gross metabolomic differences caused by PQQ produc-
tion, we performed a principal component analysis on strains that
showed changes larger than 1,000-fold (Supplementary Fig. 11). This
analysis showed effective clustering of replicates for each strain and
clear segregation of strains with and without the pqq operon. Overall,
these results showed that introduction of the pqq operon or the addi-
tion of PQQ molecules could induce significant metabolic changes in
a wide range of Streptomyces strains.
pqq operon boosts natural product biosynthesis
The general effect of PQQ on known natural products was also investi-
gated. For this purpose, the metabolites whose production increased
on integration of the pqq gene cluster were annotated with the Strep-
tomeDB chemical dictionary, as in ref. 24. The annotated metabolites
were further confirmed by manual inspection of the corresponding
spectra (Fig. 4a).
For the 11 strains, introduction of the pqq operon successfully
improved production of 34 known natural products including antibacte-
rials, antifungals, anticancer, antiparasitic and immunosuppressive mol-
ecules (Fig. 5a and Supplementary Table 4). We also noted that 70% of the
natural products whose production increased were PKS or PKS-hybrid
products, and that 85% of them use acyl-CoAs as precursors. This is
consistent with our observation that the acyl-CoA pools are increased
on incorporation of pqq operon in the model strain S. coelicolor.
In the following section, we elaborate on two examples that illus-
trate the effects of PQQ in natural product biosynthesis. Detailed infor-
mation on the upregulated natural products found in this study is
available in Supplementary Table 4 and Supplementary Data 7.
(1) Streptomyces rapamycinicus is a member of clade 12. This clade in-
cludes the largest genomes and biosynthetic repertoires among
Streptomyces, and all its members have endogenous pqq operons.
In this strain, all known polyketides detected were increased after
integration of an extra copy of the pqq operon (Fig. 5a and Sup-
plementary Table 4). For instance, rapamycin, a macrolide drug
used as immunosuppressive agent after organ transplantation,
increased by 12.5-fold (Fig. 5a); this is the greatest single increase
in production of rapamycin reported in the literature (Supple-
mentary Table 7). Nigericin, a potassium ionophore that triggers
the NLRP3 inammasome and is used in inammation research,
increased by 31.5-fold.
(2) In Streptomyces albus, introduction of the pqq operon signi-
cantly enhanced production of seven antifungal compounds
derived from three independent endogenous BGCs, namely al-
teramide A (11.5-fold) and B (2.5-fold), antimycin A1 (16.9-fold),
A2 (11.4-fold) and A3 (2.1-fold), and surugamide A (38.4-fold)
and E (36.7-fold) (Fig. 5a and Supplementary Table 4). S. albus is
a member of clade 8, which includes the smallest genomes within
the genus and cannot produce PQQ naturally. S. albus is a popular
host for heterologous expression of BGCs and only basal expres-
sion of its endogenous products has been reported25,26.
In summary, our results showed that the introduction of the pqq
operon can significantly enhance endogenous natural product pro-
duction across the Streptomyces genus. In addition, supplementation
of PQQ increased biosynthesis of natural products in ten out of 11
strains (Supplementary Figs. 3 and 12), indicating exogenous sup-
plementation of PQQ can also be an effective way to natural product
titre improvement.
pqq operon improves natural products in industrial strains
Industrial strains are usually heavily modified for optimal perfor-
mance27, thus additional modifications usually lead to only small
increases in productivity28. Given the successful improvement of
natural product production in various Streptomyces strains through
the introduction of the pqq operon, we sought to explore whether PQQ
could similarly enhance production in industrial antibiotic producer
strains. Our investigation extended to two industrial Actinobacte-
ria: Actinosynnema sp. XYF21, an ansamitocin-P3 overproducer, and
Micromonospora echinospora Q11, an overproducer of gentamicin
aminoglycosides. As show in Fig. 5b, expression of the pqq gene clus-
ter significantly increased the efficiency of gentamicin production
(measured as the mix of the four main gentamicins), resulting in a 58%
increase. While introduction of the pqq operon in Actinosynnema pre-
tiosum XYF21 resulted in a 40% increase in ansamitocin-P3 production
Nature Metabolism
Article https://doi.org/10.1038/s42255-024-01024-9
(Fig. 5c). These results further expand the potential application of
PQQ as an enhancer of natural product biosynthesis, beyond polyke-
tides, even in strains from other actinobacterial genera that are already
optimized.
pqq operon induces the expression of silent BGCs
Since the introduction of the pqq operon enhanced production of
known natural products, we wondered whether it can also activate
silent BGCs. To explore this idea, we analysed the metabolomes of
S. rapamycinicus and S. griseus in the search for unknown molecules. In
S. rapamycinicus, integration of the pqq operon resulted in the appear-
ance of a peak at a retention time of 6.65 min in the total ion chroma-
togram (Fig. 6a). The induced metabolite has a mass to charge ratio
of 1,175.647 m/z, which corresponds to the formula C
62
H
97
NO
18
S. This
formula matches that of the polyketide mediomycin-A (mass error
3.2 ppm), a linear polyene polyketide with potent antifungal activ-
ity. Consistently, we then identified a BGC in the S. rapamycinicus
genome matching the previously reported BGC for mediomycins in
S. mediocidicus ATCC (American Type Culture Collection) 23936 (ref. 29)
(Supplementary Fig. 13). We also found that this BGC is conserved
among other strains in clade 12 (Supplementary Fig. 14).
In S. griseus, introduction of the pqq gene cluster resulted in the
appearance of a peak with retention time of 7.33 min and an m/z of
303.1794 (Fig. 6b and Supplementary Fig. 15), which is consistent with
the m/z value of the γ-butyrolactone SCB3. We confirmed this match
using its MS2 fragmentation patterns (Supplementary Fig. 16). SCB3
is a variant of the γ-A-factor, a butyrolactone previously described
in S. griseus featuring a ketone instead of a hydroxyl moiety at C-6.
Until now, no reduced form of A-factor analogue (that is, SCB-like
γ-butyrolactones) has been detected in S. griseus. The enzyme respon-
sible for the reduction of the ketone group (scbB) in S. coelicolor,
a producer of SCB3, is missing in the gene cluster of S. griseus (Sup-
plementary Fig. 17). However, we found homologues of ScbB that may
be involved in the generation of the SCB series molecules in S. griseus
(Supplementary Table 8).
To obtain a wider perspective on the changes that may account for
unidentified products in these two strains, we ranked their metabolites
according to fold changes. We found that 71 and 226 metabolites in
Percentage (%)
10,000–100,000
1,000–10,000
bc
a
Deconvolution and peak
alignment
Streptomyces
fermentation culture
LC–MS for metabolite
detection
Non-targeted data
collection
Grouped dierential
analysis
Unannotated
Compounds
Compound annotation
according to StreptomeDB
Annotation filter by
references or BGC prediction
Data confirmation by wrong
peak remove
Known natural
products
√
×
×
×
0–2
2–10
10–100
>100,000
100–1,000
√
×
×
0
50
Streptomyces coelicolor
Streptomyces avermitilis
Streptomyces albus
Streptomyces griseus
Streptomyces cinnamonensis
Streptomyces sp NP10
Streptomyces venezuelae
Streptomyces clavuligerus
Streptomyces noursei
Streptomyces rapamycinicus
Streptomyces cattleya
100
150
0
50
100
150
Percentage (%)
0–2
2–10
10–100
100–1,000
1,000–10,000
10,000–100,000
>100,000
Increased after
PQQ introduction
Retention time (min)
Relative abundance
Retention time (min)
×
100
80
60
40
20
0
2 4 6 8 10 12 10 20 30 40
Compound A
Compound B
Fig. 4 | Introduction of the pqq gene cluster results in a positive impact on
the metabolome of 11 Streptomyces strains. a, Overall experimental flow chart
of metabolomics analysis. The upper panel is the process from cultivation to
determining differentially changed metabolites. The lower panel is the process
of compound annotation and confirmation. b, The fraction of compounds that
showed various fold increases in concentrations. 1–2 indicates a fold increase
of 1–2; 2–10, a fold increase of 2–10 and so on. c, The impact of incorporating
the pqq gene cluster into 11 Streptomyces. The upper panel is the percentage
of compounds that increased in concentration, whereas the lower panel is the
percentage of compounds in each strain by the fold increase in concentration.
Nature Metabolism
Article https://doi.org/10.1038/s42255-024-01024-9
S. griseus and S. rapamycinicus went from not having any signal to inten-
sities higher than 10
5
(Supplementary Tables 9 and 10), respectively.
These results showed that introduction of the pqq gene cluster leads
to the production of many compounds that could not be detected
otherwise. There are 46 (65% S. griseus) and 132 (58% S. rapamycini-
cus) of these induced compounds that cannot be matched to those
in current databases (Supplementary Tables 9 and 10), indicating the
possibility of unidentified antibiotics generated by introducing the pqq
gene cluster.
Besides the known compounds in S. rapamycinicus, we found
increased transcription of key structural genes in unknown BGCs 4, 6,
23, 26, 27, 29, 35, 36, 38, 41, 46 and 47 by two- to 32-fold after pqq gene
cluster introduction (Fig. 6c). These BGCs encode for PKSs, nonribo-
somal peptide synthetases (NRPSs) or terpenes that have not been
previously characterized (Supplementary Table 11). We also found
increased transcription of key structural genes of BGCs 5, 14, 16 and 24
in S. griseus (Fig. 6d), and these BGCs encode for melanin, siderophore,
lanthipeptide and thiopeptide products (Supplementary Table 12).
Upregulation of these BGCs is very likely to result in generation of previ-
ously uncharacterized natural products. Additionally, we conducted
a bioactivity assay of the 11 strains with and without pqq cluster to
assess the potential generation of unidentified bioactive compounds.
As shown in Fig. 6e and Supplementary Table 13, introduction of the
pqq gene cluster into S. venezuelae, S. albus, S. rapamycinicus, Strep-
tomyces sp. NP10 and S. avermitilis all resulted in strains exhibiting
distinguished bioactivities. Importantly, introducing pqq gene cluster
to S. venezuelae results in bioactivities against A. baumannii ATCC
19606 and K. pneumoniae ATCC 13883, indicating the possibility of
developing antibiotics and therapies against severe clinic infection
caused by these bacteria.
Nigericin
a
Rapamycin
Actinoplanic acid A
Elaiophylin
Efomycin A
Efomycin G
Monensin A
Monensin B
Premonensin
Alteramide A
Alteramide B
Antimycin A1
Antimycin A2
Antimycin A3
Surugamide A
Surugamide E
THN
Actinorhodin
Actinorhodinic acid
epsilon-Actinorhodin
CDA
CD3A
Germicidin L
Germicidin N
Nocardamin
Oligomycin A
Oligomycin D
A factor
Clavulanic acid
Isocycloheximide
Ectoine
Hydroxyectoine
1-Deoxychloramphenicol
L-681217
Fold change (with/without PQQ)
S. rapamycinicus
S. cinnamonensis
S. albus
S. coelicolor
S. avermitilis
S. griseus
Streptomyces sp. NP10
S. venezuelae
S. cattleya
S. clavuligerus
S. noursei
0
100
200
300
400
500
Antibacterial
Antifungal
Anticancer
Antioxidant
Osmoprotectant
Immunosuppressant
Type1 PKS
Type2 PKS
Type3 PKS
NRPS
PKS–NRPS hybrid
Others
AP-3 titre (mg l–1)
0
With PQQ
Without PQQ
With PQQ
Without PQQ
100
100101
200
300
400
500 **
***
Gentamicins titre (U g–1)
b
c
Fig. 5 | Introduction of pqq gene cluster resulted in increased production
of known natural products. a, Introduction of pqq gene cluster increased
production of 34 known natural products in 11 Streptomyces strains. The
bioactivities and types of natural product are shown in the right panel next
to the compound name (n = 3, three independent cultures analysed). THN,
1,3,6,8-tetrahydroxynaphthalene. b, Introduction of pqq gene cluster increased
gentamicin production in gentamicin industrial strain; gentamicin production
(U g−1) was calculated as: peak area/dry weight (in g). Mean concentrations
with error bars showing s.d. are plotted (n = 3, three biologically independent
samples). Multiple comparison significance was tested to **P < 0.05, ***P < 0.01
by an unpaired two-sided Student’s t-test. c, Introduction of pqq gene cluster
increased ansamitocin-P3 (AP-3) production in an ansamitocin industrial strain.
The ansamitocin titre (mg l−1) was determined by comparing peak area with the
standard curve. Mean concentrations with error bars showing s.d. are plotted
(n = 3, three biologically independent samples). Multiple comparison significance
was tested to **P < 0.05, ***P < 0.01 by an unpaired two-sided Student’s t-test.
Nature Metabolism
Article https://doi.org/10.1038/s42255-024-01024-9
Discussion
Streptomyces encodes the largest repertoire of natural product BGCs
in bacteria5. Efforts have been made to engineer Streptomyces, but
because of the complexity of natural product biosynthesis, only a hand-
ful of strategies have been developed30.
Currently, only a few engineering strategies are commonly used
in Streptomyces, and they are typically identified in one strain and then
applied to others. For example, Chen et al. established a strategy using
transfer RNA-Asp-AUC to circumvent inefficient wobble base-pairing
to enhance actinorhodin and four other antibiotics
31
. Gessner et al.
a b
RT 7.331 min
m/z 303.1794
Without PQQ
With PQQ
Acquisition time (min)
5 7 9 11 13
0
1.0
2.0
2.5
1.5
0.5
Extracted ion
chromatography
SCB3-like compound
×104×104
0
2
4
6
8
10
Acquisition time (min)
2 4 6 8 10 12
RT 6.648 min
m/z 1,174.6396
Without PQQ
With PQQ
Extracted ion
chromatography
Mediomycin-A-like compound
cd
e
A. baumannii
ATCC 19606
K. pneumoniae
ATCC 13883
S. venezuelae
With PQQWithout PQQ
E. coli
ATCC 25922
With PQQWithout PQQ
With PQQWithout PQQ
With PQQWithout PQQ
With PQQ
Without PQQ
S. albus
S. rapamycinicus
Streptomyces sp.
NP10
S. avermitilis
S. pneumoniae
ATCC 49619
S. pneumoniae
ATCC 49619
S. aureus
ATCC 29213
S. pneumoniae
ATCC 49619
S. aureus
ATCC 29213
S. pneumoniae
ATCC 49619
S. aureus
ATCC 29213
S. pneumoniae
ATCC 49619
LIV37_RS21850
LIV37_RS30480
LIV37_RS31180
Transcripts per million
150 100
10
8
6
4
2
0
SGR_RS02500
SGR_RS11805|lanKC
80
60
40
20
0
Without PQQ
With PQQ
100
50
Transcripts per million
Transcripts per million
Transcripts per million
0
LIV37_RS04730
LIV37_RS04735
LIV37_RS04740
LIV37_RS04745
LIV37_RS04750
LIV37_RS04755
Cluster 4
Cluster 36
Cluster 26
Cluster 23
Cluster 27
Cluster 29
Cluster 35
Cluster 38
Cluster 41
Cluster 46
Cluster 47
Cluster 6
Cluster 27
Cluster 16
Cluster 14
Cluster 24
LIV37_RS04760
LIV37_RS04765
LIV37_RS04770
LIV37_RS04775
LIV37_RS40725
LIV37_RS21825
LIV37_RS29460
LIV37_RS21830
LIV37_RS21835
LIV37_RS21840
LIV37_RS21845
LIV37_RS21850
LIV37_RS21855
LIV37_RS30480
LIV37_RS31180
LIV37_RS40085
LIV37_RS40150
LIV37_RS40155
LIV37_RS40160
LIV37_RS40165
LIV37_RS40170
LIV37_RS41815
LIV37_RS44460
LIV37_RS47245
LIV37_RS47250
LIV37_RS47615
LIV37_RS05985
SGR_RS11805
SGR_RS02500
SGR_RS08405
SGR_RS08410
SGR_RS08415
SGR_RS08420
SGR_RS21905
SGR_RS21910
SGR_RS21915
SGR_RS21920
Without PQQ
With PQQ
20
15
10
5
0
LIV37_RS21825
LIV37_RS40165
LIV37_RS40170
LIV37_RS41815
LIV37_RS44460
LIV37_RS47245
LIV37_RS47250
LIV37_RS47615
LIV37_RS05985
LIV37_RS40160
LIV37_RS40155
LIV37_RS40150
LIV37_RS40085
LIV37_RS21855
LIV37_RS21845
LIV37_RS21840
LIV37_RS21835
LIV37_RS21830
Fig. 6 | Natural products activated by introduction of the pqq BGC.
a, Extracted ion chromatography showed a PQQ activated mediomycin-A-like
compound in crude extracts of S. rapamycinicus before and after pqq cluster
introduction. b, Extracted ion chromatography showed a PQQ activated
SCB3-like compound from crude extracts of cultures of S. griseus before and
after pqq cluster introduction. c,d, Changes in transcription of unknown
BGCs increased with pqq gene cluster in S. rapamycinicus (c) and S. griseus (d).
The cluster number was called according to antiSMASH cluster prediction
with corresponding genome sequences (Supplementary Tables 9 and 10).
Mean concentrations with error bars showing s.d. are plotted (n = 3, three
biologically independent samples). e, Comparative bioactivity assays are shown
for Streptomyces with and without pqq gene cluster. Without PQQ refers to a
bioactivity assay using fermentation crude extracts by Streptomyces introducing
the empty vector. With PQQ refers to a bioactivity assay using fermentation
crude extracts by Streptomyces introducing the vector containing pqq gene
cluster. RT, retention time.
Nature Metabolism
Article https://doi.org/10.1038/s42255-024-01024-9
expressed bldA in 15 Streptomyces strains, and seven showed changed
HPLC profiles
32
. Introduction of bldA activated potential products and
resulted in overproduction of eight known natural products. In this
study, we used phylogenomic and pan-genomic approaches to identify
functions that co-evolved with polyketide BGCs. We also developed a
universal engineering strategy by using the pqq gene cluster to increase
natural product biosynthesis. This strategy successfully improved the
production of 34 natural products in 11 Streptomyces families, making
it the most widely applicable engineering strategy developed in Strep-
tomyces so far. It is to be noted that current approach diverges from
previous strategies that focused on singular genetic modifications.
Instead, this method simultaneously influences multiple pathways,
which is a distinctive feature enabling the universal enhancement of
natural product biosynthesis
Our study presents an approach for developing engineering
strategies for Streptomyces and other species, as well as for other
natural products such as non-ribosomal peptides or terpenes. Spe-
cifically, we identified Streptomyces strains such as Streptomyces sila-
ceus ACCC 40021, Streptomyces sp. CFMR 7, Kitasatospora albolonga
YIM 101047, Streptomyces bingchenggensis BCW1, Streptomyces sp.
MUM8645 and Streptomyces sp. TLi053 as having more NRPS gene
clusters than other strains, which could lead to strategies for NRPS
engineering by identifying co-evolved genes. Additionally, Strepto-
myces clavuligerus ATCC 27064, Streptomyces sp. TLi053 and S. hygro-
scopicus XM201 encode more terpene gene clusters, which could also
benefit from this approach. Our strategy can be extended to other
natural product-producing bacteria and fungi with the help of massive
sequencing data.
Furthermore, our study identified 597 genes from different meta-
bolic pathways that co-evolved with polyketide BGCs, and we tested the
pqq BGC as a proof of concept. Besides PQQ, other functions encoded
within these genes may also be beneficial for polyketide or natural
product production. For example, co-evolution of peptidase and/or
protease may be involved in the degradation of misfolded proteins
caused by ribosomal stress due to production of natural products;
co-evolution of enzymes with metal related functions may be driven
by the need for metal ligands needed by natural product biosynthetic
enzymes. These observations open the door for future research.
In our study, phylogenomic analysis indicated that the pqq gene
cluster evolved in parallel with PKS gene clusters, and the introduction
of the pqq gene cluster increased biosynthesis of several polyketides in
the 11 strains. We found that biosynthesis of other types of compound
were also elevated. We reasoned that this effect is due to the intercon-
nection between the central metabolic pathways that feed cofactors
and precursors to the pathways for various natural products
33
. First,
many natural products share common cofactors and precursors such
as ATP, NADH, NADPH and acetyl-CoA. Second, the biosynthetic precur-
sors from central metabolism are interconnected. For example, alanine,
a precursor to peptides, and acetyl-CoA, a precursor to polyketides,
are derived from pyruvate. Their pools may be affected simultaneously
through the change in pyruvate, which will, in turn, alter the biosynthe-
sis of downstream natural products. Finally, the biosynthetic pathways
for various natural products may be controlled by global regulators,
such that a change in one pathway may influence the expression of
global regulators that alter production of other natural products
4
.
Therefore, although the prevalence of the pqq gene cluster correlates
directly with abundance of PKS gene clusters, the contribution of the
pqq gene cluster to polyketide production can be extended to other
natural products through the shared and interconnected precursors
and cofactors, as well as the influence of global regulators.
The PQQ-induced changes that we found in this study are con-
sistent with effects on mitochondrial respiration caused by PQQ in
mammalian cells that have been previously reported34. However, the
mechanistic details behind these effects on natural product biosynthe-
sis require further investigation. We found that the pqq gene cluster in
Streptomyces is often associated with carbohydrate-active enzymes
such as chitinases and with some dehydrogenases and catalases; we
speculate that PQQ may also play a role in these catabolic pathways.
Before this study, the only case in which PQQ has been linked to
natural products is in lankacidin biosynthesis. In this case the pqq
gene cluster is located on the chromosome near the lankacidin BGC
in in S. rochei 7434AN4. PQQ works as a cofactor for the quinoprotein
dehydrogenase (orf23), which catalysers the dehydrogenation of the
C23-C25 lactate moiety to pyruvate during lankacidin biosynthesis22.
However, there are no quinoprotein dehydrogenase homologues that
could use PQQ as a cofactor in S. coelicolor and other Streptomyces. The
mechanism by which PQQ affects the intracellular redox state, whether
through specific binding of unknown quinoproteins or by the means
of its electron transfer nature, will be investigated in future.
In conclusion, we combined genomics, proteomics and metabo-
lomics to evaluate the impact of the PKS co-evolved pqq gene cluster
on natural product biosynthesis in Streptomyces genus. Our results
indicate presence of the pqq operon correlates with biosynthetic pro-
ficiency in the Streptomyces genus and we showed that PQQ enhances
the biosynthesis of natural products and activates BGCs for unidenti-
fied products across the genus Streptomyces and beyond. We believe
that this effect can be exploited to discover bioactive natural products
in the future.
Methods
Genomes database integration
We collected the genomes from a previously compiled database with
7,762 genomes retrieved from the GenBank. These sequences were
processed with ad-hoc annotation and formatting tools23. To cre-
ate a high-quality Streptomyces genome database, we first selected
sequences assigned to the Actinomycetota phylum (formerly Actino-
bacteria). For this step, we parsed the taxonomic labels in our database
against the genera listed in the entry for the Actinobacteria phylum at
the National Center for Biotechnology Information (NCBI) taxonomy
database (ID 201174). This selection step resulted in more than 5,000
genomes. Then, we selected for high-quality genomes by filtering-out
assemblies with more than ten contigs. We reasoned that this cut off
would leave high-quality genomes that may have one or more plasmids
or chromosomes. Finally, we assessed the completeness of the selected
genomes with BUSCO using the actinobacteria_class_odb10 reference
lineage dataset35. We kept genomes with a minimum completeness of
85%. The final dataset included 720 genomes (Supplementary Data 1).
Taxonomic classification of Streptomyces strains
We calculated the core genome for the 720 selected strains. For this,
we used the predicted proteomes of the 720 genomes to identify sets
of conserved orthologues at different sequence identity cut offs using
the BPGA pipeline36. We identified a set of 18 Actinomycetota-wide
conserved protein sequences at a minimum sequence identity cut
off of 20% (Supplementary Table 1). The set of 18 orthologues was
then collected from each genome, and the amino acid sequences
were aligned and concatenated to generate a super matrix with 6,517
characters for phylogenetic analysis using IQtree 2 (ref. 37) with the
following parameters -m TEST, -bb 10000, alrt 10000. A substitution
model was calculated for each partition in the super matrix and then
a phylogenetic tree was calculated (Supplementary Fig. 1). Branch
support was obtained from 10,000 ultrafast bootstrap replicates
and SH-like approximate likelihood ratio test (SH-aLRT), which was
consistent with bootstrap-based support, thus, only bootstrap sup-
port is reported herein. The tree showed excellent support for most
nodes, and it is consistent with the known general taxonomy of the
phylum. The tree features a clade with 201 taxa that we could safely
assign to the Streptomycetaceae family (NCBI taxID 2062). This clade
includes strains assigned to the genera Streptomyces (190 strains),
Kitasatospora (eight strains) and Streptacidiphilus (three strains).
Nature Metabolism
Article https://doi.org/10.1038/s42255-024-01024-9
Within this group only a few genomes were misclassified: one Strepto-
myces strain classified as Kitasatospora and six Kitasatospora strains
that were classified as Streptomyces; finally, one strain at the base
of this clade (classified as Streptomyces scabrisporus NF3) did not
group with the other genera and may belong to a different genus
within the Streptomycetaceae family. We used the 201 genomes to
construct a species tree for the Streptomyces genus using the same
approach described above for the Actinomycetota species tree. For
the Streptomyces tree, we identified a set of 619 conserved ortho-
logues across the 201 genomes at a sequence identity cut off of 0.4.
This set of orthologues was processed leading to a super matrix with
604 partitions (Supplementary Data 2) that was used to calculate a
species tree that was rooted using the 17 non-Streptomyces strains
as outgroup. This improved version is available at https://github.
com/WeMakeMolecules/myCORASON. The Actinomycetota and
Streptomyces species tree were constructed using code available at
https://github.com/WeMakeMolecules/Core-to-Tree.
Functional analysis of conserved traits in Streptomyces
The number and type of BGCs for each strain was obtained using Ant-
iSMASH
38
. To calculate the conserved set of orthologues of each of
the 14 clades we used BPGA
36
. To identify the clade-specific conserved
genes, we compared the clade-level cores with BPGA and extracted the
genes that are uniquely conserved in clade 12 and their annotation. For
the identification of pqq BGCs, we used PqqB (UniProt ID A0A0C6FBE1)
as query for a search using Corason3, which has been modified from
https://doi.org/10.1038/s41589-019-0400-9.
Point biserial correlations between KS abundance (continuous
data) and genes uniquely conserved in clade 12 (dichotomous data)
were calculated using a matrix of presence (recorded as 1) or absence
(recorded as 0) of at least one orthologue of the 597 genes. For this,
the amino acid sequences encoded in the 597 genes were used as query
for a blastP search across the 203 genomes dataset. The search was
performed with a cut off of 1 × 10−6, the score cut off was adjusted
depending on the length of the query automatically using script avail-
able at https://github.com/WeMakeMolecules/Core-to-Tree/raw/main/
dichotomizer.pl. Point biserial correlations (r) were calculated using
Microsoft Excel using the function =CORREL (dichotomous data,
continuous data); t-test statistics were calculated using the formula
=r*SQRT(sample size-2/SQRT(1−r
2
) and P values were obtained with
the function =T.DIST.2 T(T, sample size −2), in all cases the simple size
was 202 genomes.
Strains and growth conditions
Strains and plasmids used in this study are listed in Supplementary
Table 14. Escherichia coli DH5α was used for plasmids construction, and
ET12567 (pUZ8002) was used for conjugation. All E. coli strains was cul-
tivated in Luria-Bertani medium with appropriate antibiotics at 37 °C.
For cultivation and harvesting spores, S. coelicolor M145, S. albus
J1074, S. cinnamonensis ATCC 15413 (recently reclassified as S. virginiae
ATCC 15413 (ref. 39)), S. sp. NP10 and S. venezuelae ISP5230 (ATCC
10712) were cultivated on SFM medium (20 g l−1 soybean powder, 20 g l−1
mannitol, 2% agar) plates. S. avermitilis NRRL 8165 (MA-4680), S. cla-
vuligerus ATCC 27064, S. noursei ATCC 11455, S. cattleya NRRL 8057 were
cultivated on YMG medium (4 g l−1 yeast extract, 10 g l−1 malt extract,
4 gl−1 glucose, 2% agar); S. rapamycinicus NRRL 5491 was cultivated on
oatmeal medium (20 g l−1 oatmeal, 2% agar) and S. griseus DSM 40236
was cultivated on ISP-4 medium (37 g l
−1
ISP-4 medium powder (BD
DIFCO), 2% agar).
For conjugation, S. coelicolor, S. albus, S. cinnamonensis, Strep-
tomyces sp. NP10, S. venezuelae, S. avermitilis and S. rapamycinicus
were performed on SFM medium supplemented with 10 mM MgCl
2
.
S. clavuligerus, S. noursei and S. cattleya were performed on YMG
medium supplement of 10 mM MgCl
2
. S. griseus was performed on
ISP-4 medium supplement of 10 mM MgCl2.
Unless stated, cultivation of Streptomyces strains were performed
according to the publications, with only a few modifications to accom-
modate to the present media component stock in our laboratory. For
fermentation, strains were first inoculated on corresponding plates at
30 °C for 2 days, then 1 cm
2
medium was inoculated into seed medium
for growth of mycelium, and then 10% seed medium was transferred
into fermentation medium for another round of cultivation, accumula
-
tion for secondary metabolites. Both seed cultivation and production
cultures were performed in 250 ml flask filled with spring steels circle.
All medium was adjusted to pH 7.0–7.5 before sterilization.
Construction of plasmids and strains
The primers used in this study are listed in Supplementary Table 14.
For genes in the ‘cofactor’ category and the truncated pqq cluster
without pqqA, the open reading frames of corresponding genes were
amplified from genome sequence of S. hygroscopicus XM201 and were
introduced between the NdeI/EcoRI restriction sites of pLQ646 by
Gibson Assembly. For the truncated pqq gene cluster without pqqC,
the whole cassette was directly synthesized and inserted between the
NdeI/EcoRI restriction sites at GenScript Co., Ltd. The plasmid pLQ646
was constructed previously by introduction of kasOp* promoter in the
multiple cloning site of pSET152 (ref. 40). The resulting plasmid from
pXR-1 to pXR-7 was then introduced into E. coli ET12567 (pUZ8002)
for conjugation with different Streptomyces. After conjugation, the
exo-conjugants were confirmed both by apramycin resistance and
PCR verification.
Metabolomic analysis
After fermentation, the culture from three biological replicates was
diluted by mixing with two volumes of methanol and sonicated for
30 min. The samples were centrifuged at 13,800g for 5 min to remove
insoluble particles and then filtered using a 0.22 μm filter membrane.
The filtrate was injected to Agilent 1290 series HPLC (Agilent Tech-
nologies) equipped with an Agilent 6546 Accurate-Mass Q-TOF liquid
chromatography with mass spectrometry (LC–MS) (Agilent Tech-
nologies) via an electrospray ionization source. LC conditions were as
follows: A, water + 0.1% formic acid; B, acetonitrile + 0.1% formic acid;
0.4 ml min
−1
. The data were collected using Agilent Data Acquisition
software (v.10.1) (Agilent Technologies). Then the collected raw data
were further processed by Agilent Profinder (v.10) (Agilent Technolo-
gies) for peak alignment and Agilent Mass Profiler Professional (v.15.0)
(Agilent Technologies) for grouped differential analysis and compound
annotation.
For compound annotation, the compounds were annotated using
the dictionary of the StreptomeDB database covering most natural
products identified in Streptomyces (so far) in Agilent Mass Profiler
Professional (v.15.0) according to the molecular weight and predicted
addiction. The annotated compound was embedded into Agilent Quali-
tative Analysis (v.0.0) (Agilent Technologies) for peak confirmation
to make sure the difference from the target molecule is <5 ppm, and
the shapeless peaks were also removed. The detailed information for
known natural products is shown in Supplementary Fig. 18.
Proteomic analysis
Protein extracts were prepared as previously described with minor
modifications
41
. To prepare protein extracts, strains were grown in
YMG medium for 24 h and cells were collected. Cells were then resus-
pended in a solution containing 50 mM NH4HCO3 and 8 M urea, along
with 0.5 mm glass beads (Biospec), and mechanically disrupted using
a tissue grinding machine (ten times for 30 s each). The resulting cell
debris was pelleted by centrifugation at 12,000g for 10 min at 4 °C.
Protein concentration was determined using a BCA protein assay kit
(Solarbio) and 100 μg of protein was used for digestion. The samples
were first treated with 5 mM reducing dithiothreitol for 30 min at
56 °C, followed by 25 mM iodoacetamide for alkylation in the dark at
Nature Metabolism
Article https://doi.org/10.1038/s42255-024-01024-9
room temperature for 1 h. The samples were then diluted with 50 mM
NH4HCO3 to a final urea concentration of 2 M and digested overnight
with trypsin (Promega) at an enzyme-to-protein ratio of 1:50 (w/w) at
37 °C. The resulting tryptic peptides were desalted using C18 Sep-Pak
cartridges (Welch), vacuum dried and stored at −80 °C before LC with
tandem MS (LC–MS/MS) analysis.
To prepare for LC–MS/MS analysis, the vacuum dried samples
were reconstituted in an aqueous solution containing 0.1% formic acid.
Peptides were then loaded onto a 3 μm × 0.2 cm precolumn (P/N 164535,
Thermo Scientific) and separated on a 2 μm × 15 cm capillary column
(P/N 164534, Thermo Scientific) using the following HPLC gradient:
3–8% B in 5 min, 8–20% B in 65 min, 20–30% B in 20 min, 30–40% B in
10 min, 40–99% B in 5 min, held at 99% B for 5 min and then held at 3%
B for 10 min (A, 0.1% formic acid in water; B, 0.1% formic acid in 80%
acetonitrile) at a flow rate of 300 nl min
−1
. The peptide samples were
analysed using an Ultimate 3000 RSLC nano system (Thermo Scientific)
coupled to a Q-Exactive HF Orbitrap mass spectrometer (Thermo Fisher
Scientific). The mass spectrometer was operated in data-dependent
mode with a full MS scan (350–2,000 m/z) at a resolution of 60,000
followed by higher-energy collisional dissociation fragmentation at a
normalized collision energy of 28%. The MS2 automatic gain control
target was set to 45 s.
The raw MS files were processed by MaxQuant (http://maxquant.
org/, v.1.6.10.43). MS/MS spectra were searched against the S. coelicolor
(strain ATCC BAA-471/A3(2)/M145) protein database (downloaded
from UniProt, v.2002) using the Andromeda search engine embedded
in MaxQuant. Both peptide and protein assignments were filtered to
achieve a false discovery rate <1%. After searching, the reverse hits,
contaminants and proteins only identified by one site were removed.
Transcriptome analysis
To perform transcriptome analysis, strains of S. rapamycinicus and
S. griseus introduced pqq gene cluster as well as the empty vectors were
cultured under normal fermentation condition for 12 h. The total RNA
of each sample was extracted using Total RNA Extractor kit (catalogue
no. B511311, Sangon) according to the manufacturer’s protocol. The
extracted RNA was then sent to the Sangon Biotech Co., Ltd for library
construction and sequencing. Gene expression level was calculated
by the DESeq2 R package (v.1.20.0), showing average values of three
clones from each group.
Growth determination
Unless noted, the strains were cultivated in 24-well plates filled with
2.5 ml of corresponding fermentation medium and the cultivation was
carried out in 30 °C at 800 rpm. For sampling, 50 μl of fermentation
culture was collected every 12 h and used for growth determination.
For S. albus J1074, S. virginiae ATCC 15413, S. sp. NP10, S. avermitilis
NRRL 8165 (MA-4680), S. clavuligerus ATCC 27064, S. noursei ATCC
11455, S. cattleya NRRL 8057 and S. rapamycinicus NRRL 5491, growth
was determined by the diphenylamine colorimetric method following
the protocol in ref. 42.
For S. venezuelae ISP5230 (ATCC 10712), growth was determined
by absorption at A600 nm using a microplate reader.
For S. coelicolor M145, growth was determined by weighing the
cell dry weight. The strains were cultivated in 250 ml flasks, and 1 ml
of fermentation culture was collected every 12 h and used for growth
determination.
PQQ addition experiment
Streptomyces strains were cultivated in 24-well plates and each well
contained 2.5 ml of the corresponding fermentation medium. The cul-
tivation temperature was maintained at 30 °C at 800 rpm (ZQZY-85CH,
Zhichu). PQQ (Macklin, P863945) was added to the fermentation
cultures at different final concentrations. The concentrations were
as follows: 0 nM, 0.4 nM, 400 nM, 40 μM, 200 μM and 400 μM. The
supplementation time for S. coelicolor, S. albus, S. avermitilis, Strep-
tomyces sp. NP10, S. cattleya, S. clavuligerus and S. griseus was 0 h of
fermentation, for S. noursei was 24 h, for S. cinnamonensis was 48 h
and for S. venezuela was 72 h. Each concentration was added to four
replicates. For sampling, 300 μl of fermentation culture was collected
corresponding to fermentation time and analysed by LC–MS (Q-TOF,
Agilent Technologies).
Key metabolites determination
For determination of precursors and fatty acids, the mycelia were col-
lected at different time points by centrifuge at 4 °C, 13,800g for 2 min
followed by removal of the supernatant medium. To extract metabo-
lites from samples, 400 μl of cold methanol and acetonitrile (1:1, v:v)
extraction solvent was added to remove the protein and extract the
metabolites, then adequately vortexed. For absolute quantification of
the metabolites, stock solutions of stable-isotope internal standards
were added to the extraction solvent simultaneously. The mixture was
collected into a new centrifuge tube, and centrifuged at 13,800g for
20 min at 4 °C to collect the supernatant. The supernatant was dried
in a vacuum centrifuge. For LC–MS analysis, the samples were redis-
solved in 100 μl of acetonitrile:water (1:1, v:v) solvent and centrifuged at
13,800g at 4 °C for 20 min, then the supernatant was injected. Analyses
were performed using an UHPLC (1290 Infinity LC, Agilent Technolo-
gies) coupled to a QTRAP MS (AB 6500+, AB Sciex) in Shanghai Applied
Protein Technology Co., Ltd. The metabolites were separated using
hydrophilic interaction chromatography (Waters UPLC BEH Amide
column, 2.1 × 100 mm, 1.7 μm) and C18 columns (Waters UPLC BEH
C18x2.1 ×100 mm, 1.7 μm), and were quantified on a 6500+ QTRAP
(AB SCIEX) in MRM mode, and in positive and negative switch mode.
For determination of TAG, mycelium of S. coelicolor M145 was col-
lected at different time points from 12 to 96 h for the intracellular TAG
analysis. Intracellular TAG levels were analysed using a commercialized
kit (Solarbio, catalogue no. BC0625) following the manufacturer’s
instructions.
For determination of intracellular cofactors, mycelium of S. coeli-
color M145 was collected at 12 h as this time point had shown the most
enrichment concentration of these molecules. The mycelia were collected
by centrifuge at 4 °C, 13,800g for 2 min followed by removal of the super-
natant medium. The levels of ATP (ab113849), NADH/NAD
+
(ab65348) and
NADPH/NADP
+
(ab65349) were analysed by commercialized kits (Abcam)
following the manufacturer’s instructions. For other cofactors, analyses
were performed using an ultra-HPLC (UHPLC) (1290 Infinity LC, Agilent
Technologies) coupled to a QTRAP MS (AB 6500+, AB Sciex) in Shanghai
Applied Protein Technology Co., Ltd. The cofactors were separated using
hydrophilic interaction chromatography (Waters UPLC BEH Amide col-
umn, 2.1 × 100 mm, 1.7 μm) and C18 columns (Waters UPLC BEH C18-2.1
×100 mm, 1.7 μm), and were quantified on a 6500+ QTRAP (AB SCIEX) in
MRM mode, and in positive and negative switch mode.
C13-labelling experiments
During cultivation, the stable uniformly
13
C-labelled glucose (U-
13
C6,
MCE) was fed into the fermentation culture of S. coelicolor M145 at 24 h.
Samples were collected from 48 to 96 h to detect the labelling fraction
of actinorhodin. To prepare samples, the culture from three biological
replicates was diluted by mixing with two volumes of methanol and
sonicated for 30 min. The samples were centrifuged at 13,800g for
5 min to remove insoluble particles and then filtered using a 0.22 μm
filter membrane. The filtrate was injected to Agilent 1290 series HPLC
(Agilent Technologies) equipped with an Agilent 6546 Accurate-Mass
Q-TOF LC–MS (Agilent Technologies) via an electrospray ionization
source. LC conditions were as follows: A, water + 0.1% formic acid; B,
acetonitrile + 0.1% formic acid; 0.4 ml min
−1
. The data were collected
using Agilent Data Acquisition software (v.10.1) (Agilent Technologies).
Then the collected raw data were further processed by Agilent Qualita-
tive Analysis (v.10) (Agilent Technologies). Actinorhodinic acid was
Nature Metabolism
Article https://doi.org/10.1038/s42255-024-01024-9
chosen as a representative for C
13
-labelling analysis among the different
forms of actinorhodin, as the producing curve was most close to the total
actinorhodin changes determined by colorimetric analysis. The peaks
different isotope forms were extracted by corresponding exact molecu-
lar weight in Agilent Qualitative Analysis (v.10) (Agilent Technologies).
The data were further processed and corrected for natural abun-
dance according to the reported method in ref. 43. The unlabelled
amount was calculated as the amount of the [M+0] fraction, and the
labelled amount was calculated as the sum of all the other isotopic
fractions.
Statistics and reproducibility
Statistical analysis was performed using Microsoft Excel software using
a two-tailed t-test analysis of variance hypothesis. Significant differ-
ences are marked as **P < 0.05 and ***P < 0.01. All data are presented as
mean ± s.d. The number of biologically independent samples for each
panel was three unless otherwise stated in the figure legends.
Reporting summary
Further information on research design is available in the Nature
Portfolio Reporting Summary linked to this article.
Data availability
The transcriptomic data of S. griseus and S. rapamycinicus are available
at NCBI GEO database (GSE256209). The mass spectrometry proteom-
ics data are available at the ProteomeXchange Consortium via the
iProX repository (PXD049454). Other data supporting the findings
of this study are included in the published article and Supplementary
Information. Requests for any additional information can be made to
the corresponding authors. Source data are provided with this paper.
Code availability
The code of pan-genomic analysis was described in the Methods section
of the paper. All the code is openly available in GitHub.
References
1. Newman, D. J. & Cragg, G. M. Natural products as sources of new
drugs over the nearly four decades from 01/1981 to 09/2019.
J. Nat. Prod. 83, 770–803 (2020).
2. Li, S., Li, Z., Pang, S., Xiang, W. & Wang, W. Coordinating
precursor supply for pharmaceutical polyketide production in
Streptomyces. Curr. Opin. Biotechnol. 69, 26–34 (2021).
3. Wang, W. et al. Harnessing the intracellular triacylglycerols
for titer improvement of polyketides in Streptomyces. Nat.
Biotechnol. 38, 76–83 (2020).
4. Liu, G., Chater, K. F., Chandra, G., Niu, G. & Tan, H. Molecular
regulation of antibiotic biosynthesis in Streptomyces. Microbiol.
Mol. Biol. Rev. 77, 112–143 (2013).
5. Gavriilidou, A. et al. Compendium of specialized metabolite
biosynthetic diversity encoded in bacterial genomes. Nat.
Microbiol. 7, 726–735 (2022).
6. Gao, L. et al. The tomato pan-genome uncovers new genes and a
rare allele regulating fruit lavor. Nat. Genet. 51, 1044–1051 (2019).
7. de Vries, R. P. et al. Comparative genomics reveals high biological
diversity and speciic adaptations in the industrially and medically
important fungal genus Aspergillus. Genome Biol. 18, 28 (2017).
8. Cruz-Morales, P. et al. Revisiting the evolution and taxonomy
of Clostridia, a phylogenomic update. Genome Biol. Evol. 11,
2035–2044 (2019).
9. Chandra, G. & Chater, K. F. Developmental biology of
Streptomyces from the perspective of 100 actinobacterial
genome sequences. FEMS Microbiol. Rev. 38, 345–379 (2014).
10. Schoch, C. L. et al. NCBI Taxonomy: a comprehensive update on
curation, resources and tools. Database https://doi.org/10.1093/
database/baaa062 (2020).
11. Bentley, S. D. et al. Complete genome sequence of the model
actinomycete Streptomyces coelicolor A3(2). Nature 417, 141–147
(2002).
12. Kim, J. N. et al. Comparative genomics reveals the core and
accessory genomes of Streptomyces species. J. Microbiol.
Biotechnol. 25, 1599–1605 (2015).
13. Bu, Q. T. et al. Comprehensive dissection of dispensable
genomic regions in Streptomyces based on comparative analysis
approach. Micro. Cell Fact. 19, 99 (2020).
14. Zhou, Z., Gu, J., Li, Y. Q. & Wang, Y. Genome plasticity and
systems evolution in Streptomyces. BMC Bioinf. 13, S8
(2012).
15. Belknap, K. C., Park, C. J., Barth, B. M. & Andam, C. P. Genome
mining of biosynthetic and chemotherapeutic gene clusters in
Streptomyces bacteria. Sci. Rep. 10, 2003 (2020).
16. Zaburannyi, N., Rabyk, M., Ostash, B., Fedorenko, V. & Luzhetskyy,
A. Insights into naturally minimised Streptomyces albus J1074
genome. BMC Genom. 15, 97 (2014).
17. Chung, Y. H. et al. Comparative genomics reveals a remarkable
biosynthetic potential of the Streptomyces phylogenetic lineage
associated with rugose-ornamented spores. mSystems 6,
e0048921 (2021).
18. Jonscher, K. R., Chowanadisai, W. & Rucker, R. B.
Pyrroloquinoline-quinone is more than an antioxidant:
a vitamin-like accessory factor important in health and disease
prevention. Biomolecules 11, 1441 (2021).
19. Wagh, J., Shah, S., Bhandari, P., Archana, G. & Kumar, G. N.
Heterologous expression of pyrroloquinoline quinone (PQQ)
gene cluster confers mineral phosphate solubilization ability to
Herbaspirillum seropedicae Z67. Appl. Microbiol. Biotechnol. 98,
5117–5129 (2014).
20. Zhu, W. & Klinman, J. P. Biogenesis of the peptide-derived redox
cofactor pyrroloquinoline quinone. Curr. Opin. Chem. Biol. 59,
93–103 (2020).
21. Shen, Y. Q. et al. Distribution and properties of the genes
encoding the biosynthesis of the bacterial cofactor,
pyrroloquinoline quinone. Biochemistry 51, 2265–2275
(2012).
22. Yamauchi, Y. et al. Quinoprotein dehydrogenase functions at the
inal oxidation step of lankacidin biosynthesis in Streptomyces
rochei 7434AN4. J. Biosci. Bioeng. 126, 145–152 (2018).
23. Cruz-Morales, P. et al. Biosynthesis of polycyclopropanated high
energy biofuels. Joule 6, 1590–1605 (2022).
24. Moumbock, A. F. A. et al. StreptomeDB 3.0: an updated
compendium of streptomycetes natural products. Nucleic Acids
Res. 49, D600–D604 (2021).
25. Olano, C. et al. Activation and identiication of ive clusters for
secondary metabolites in Streptomyces albus J1074. Microb.
Biotechnol. 7, 242–256 (2014).
26. Xu, F., Nazari, B., Moon, K., Bushin, L. B. & Seyedsayamdost, M. R.
Discovery of a cryptic antifungal compound from Streptomyces
albus J1074 using high-throughput elicitor screens. J. Am. Chem.
Soc. 139, 9203–9212 (2017).
27. Beganovic, S. et al. Systems biology of industrial oxytetracycline
production in Streptomyces rimosus: the secrets of a mutagenized
hyperproducer. Microb. Cell Fact. 22, 222 (2023).
28. Abbate, E. et al. Optimizing the strain engineering process
for industrial-scale production of bio-based molecules. J. Ind.
Microbiol. Biotechnol. 50, kuad025 (2023).
29. Sun, F., Xu, S., Jiang, F. & Liu, W. Genomic-driven discovery of an
amidinohydrolase involved in the biosynthesis of mediomycin A.
Appl. Microbiol. Biotechnol. 102, 2225–2234 (2018).
30. Whitford, C. M., Cruz-Morales, P., Keasling, J. D. & Weber, T.
The design-build-test-learn cycle for metabolic engineering of
Streptomycetes. Essays Biochem. 65, 261–275 (2021).
Nature Metabolism
Article https://doi.org/10.1038/s42255-024-01024-9
31. Chen, X. et al. A new bacterial tRNA enhances antibiotic
production in Streptomyces by circumventing ineicient wobble
base-pairing. Nucleic Acids Res. 50, 7084–7096 (2022).
32. Gessner, A. et al. Changing biosynthetic proiles by expressing
bldA in Streptomyces strains. Chem. Bio. Chem. 16, 2244–2252
(2015).
33. Tsunematsu, Y. et al. Distinct mechanisms for spiro-carbon
formation reveal biosynthetic pathway crosstalk. Nat. Chem. Biol.
9, 818–825 (2013).
34. Saihara, K., Kamikubo, R., Ikemoto, K., Uchida, K. & Akagawa, M.
Pyrroloquinoline quinone, a redox-active o-quinone, stimulates
mitochondrial biogenesis by activating the SIRT1/PGC-1alpha
signaling pathway. Biochemistry 56, 6615–6625 (2017).
35. Simao, F. A., Waterhouse, R. M., Ioannidis, P., Kriventseva,
E. V. & Zdobnov, E. M. BUSCO: assessing genome assembly
and annotation completeness with single-copy orthologs.
Bioinformatics 31, 3210–3212 (2015).
36. Chaudhari, N. M., Gupta, V. K. & Dutta, C. BPGA—an ultra-fast
pan-genome analysis pipeline. Sci. Rep. 6, 24373 (2016).
37. Minh, B. Q. et al. IQ-TREE 2: new models and eicient methods
for phylogenetic inference in the genomic era. Mol. Biol. Evol. 37,
1530–1534 (2020).
38. Blin, K. et al. antiSMASH 6.0: improving cluster detection and
comparison capabilities. Nucleic Acids Res. 49, W29–W35 (2021).
39. Komaki, H. & Tamura, T. Reclassiication of Streptomyces
cinnamonensis as a later heterotypic synonym of Streptomyces
virginiae. Int. J. Syst. Evol. Microbiol. 71, 004813 (2021).
40. Wang, X. R., Wang, R. F., Kang, Q. J. & Bai, L. Q. The antitumor
agent ansamitocin P-3 binds to cell division protein FtsZ in
Actinosynnema pretiosum. Biomolecules 10, 699 (2020).
41. Wu, M. et al. Phosphoproteomics reveals novel targets and
phosphoprotein networks in cell cycle mediated by Dsk1 kinase.
J. Proteome Res. 19, 1776–1787 (2020).
42. Zhao, Y., Xiang, S., Dai, X. & Yang, K. A simpliied diphenylamine
colorimetric method for growth quantiication. Appl. Microbiol.
Biotechnol. 97, 5069–5077 (2013).
43. Nanchen, A., Fuhrer, T. & Sauer, U. Determination of Metabolic
Flux Ratios From 13C-Experiments and Gas Chromatography-Mass
Spectrometry Data (Humana, 2007).
Acknowledgements
We acknowledge the inancial support provided by National Key R&D
Program of China (to X.L., grant no. 2018YFA0903200), National
Natural Science Foundation of China (to X.L., grant no. 32071421 and
to H.D., grant no. 32201203), Shenzhen Science and Technology
Program (to X.L., grant nos. ZDSYS20210623091810032 and
RCYX20200714114736026), China Postdoctoral Science Foundation
(to X.W., grant no. 31800023), Shenzhen Medical Research Fund
(to X.W., grant no. D2301005), Shenzhen Science and Technology
Program (to X.W., grant no. JCYJ20220531100207017), Shenzhen
Bay Scholar Fellowship (to X.L. and X.T., grant no. 229100002),
Novo Nordisk Foundation (to P.C.-M., grant no. CFB 2.0,
NNF20CC0035580), which made this research possible. We thank
J. Nikodinovic-Runic for kindly providing the strain Streptomyces
sp. NP10, G. Liu, F. Ni and L. Zhou for help in industrial strains
experiments, M. Wu for technique support on proteomics analysis,
H. He and G. Zhang for project discussion and Z. Wei for meeting
organization for this project.
Author contributions
X.W., P.C.-M., X.T., J.D.K. and X.L. conceived and supervised the project.
X.W. and N.C. designed and performed the main experiments. P.C.-M.
and S.A. performed the bioinformatics analysis. Y.S., L.B., J.W., Y.X.
and X.F. designed and participated in the industry strain experiments.
X.T. and Y.L. designed and participated in the bioassay test. B.Z.
participated in the metabolomics analysis. Y.Z. participated in RNA
sequencing data analysis. Z.L. and H.D. participated in fermentation
data analysis. X.W., P.C.-M., J.D.K. and X.L. wrote and revised the paper.
Competing interests
J.D.K. has inancial interests in Amyris, Ansa Biotechnologies, Apertor
Pharma, Berkeley Yeast, Cyklos Materials, Demetrix, Lygos, Napigen,
ResVita Bio and Zero Acre Farms. X.L. has inancial interests in
Demetrix and Synceres. The other authors declare no competing
interests.
Additional information
Supplementary information The online version contains supplementary
material available at https://doi.org/10.1038/s42255-024-01024-9.
Correspondence and requests for materials should be addressed to
Xiaoyu Tang, Jay D. Keasling or Xiaozhou Luo.
Peer review information Nature Metabolism thanks Kenji Arakawa,
Hyun Uk Kim and Lixin Zhang for their contribution to the peer review
of this work. Primary Handling Editor: Alfredo Giménez-Cassina,
in collaboration with the Nature Metabolism team.
Reprints and permissions information is available at
www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional ailiations.
Springer Nature or its licensor (e.g. a society or other partner) holds
exclusive rights to this article under a publishing agreement with
the author(s) or other rightsholder(s); author self-archiving of the
accepted manuscript version of this article is solely governed by the
terms of such publishing agreement and applicable law.
© The Author(s), under exclusive licence to Springer Nature Limited
2024
A preview of this full-text is provided by Springer Nature.
Content available from Nature Metabolism
This content is subject to copyright. Terms and conditions apply.
