there are known NPs produced by the respective organisms (e.g., Macheleidt et al. 2016;
Rutledge and Challis 2015; Sanchez et al.
2012), representing an underexploited reservoir
of novel chemistries. As such, the genomic
approaches used in natural product discovery
are a valuable toolkit for capturing the general
biosynthetic potential of fungi.
Targeted genome mining approaches routinely used to identify BGCs are based primarily
on domain-dependent protocols, which identify conserved elements within the constituent
genes of a cluster. The antiSMASH framework
(Medema et al. 2011), with its latest version,
antiSMASH 4.0 (Blin et al. 2017), represents
one of the most comprehensive algorithms,
capable of recognizing 44 different classes of
biosynthetic enzymes. Others, like SMURF
(Khaldi et al. 2010), ClustScan (Starcevic et al.
2008), or PRISM (Skinnider et al. 2015), recognize core enzymes typical for fungal secondary
metabolite clusters, including PKSs, NRPSs,
DMATSs, and/or hybrid PKS-NRPS entities.
For a recent overview of software tools and databases
for analysis of biosynthetic gene clusters in genomic
data, the reader is directed to a review by Weber (2014)
and references therein.
In contrast to the motif-dependent tools,
protocols have been designed to identify clusters lacking a core enzyme. MIPS-CG, an algorithm developed by Takeda et al. (2014), utilizes
comparative genomics of two species, recognizing gene order within a nonsyntenic genomic
background, rather than conserved motifs.
Similarly, comparative phylogenomics uses
the distribution and evolution of gene clusters
to predict the roles of genes, potentially linking
genetic elements and SMs (Brown and Proctor
2016), identifying novel clusters from phylogenetic alignment outliers (Kang 2017), or revealing an evolutionary plasticity in SM profiles
(Mattern et al. 2017). Comparative genomics
approaches are also highly suited to the identification of BGCs in a scenario where the product is known, but the encoding cluster remains
to be discovered. Using the product structure,
information can be gained about the types of
enzymes involved in its production. This strategy led to the identification of the gene cluster
of griseofulvin, an antifungal polyketide
produced by several Penicillium species. Comparing the genome of the producing P. aethiopicum with that of P. chrysogenum, a nonproducer, facilitated the elimination of orthologous clusters. From the remaining ones, only
one contained methyltransferases and a halogenase necessary for the formation of griseofulvin. The deletion of the PKS confirmed the
prediction (Chooi et al. 2010).
The SM cluster producing echinocandin B1, a novel
antifungal from Emericella rugulosa, was identified in
a similar fashion (Cacho et al. 2012).
While mining of genome data using bioinformatic tools is valuable due to its speed of
processing and high throughput potential
(e.g., Cacho et al. 2015), it should be noted
that this approach is invariably predictive and
the derived hypothesis needs to be validated
experimentally. That links the computational
prediction with wet lab techniques, based
either on the OSMAC strategy and a relatively
inefficient screening of modified physicochemical cultivation conditions [e.g., isolation of
new antibiotics from Aspergillus parasiticus
(Bracarense and Takahashi 2014)], or one of
the molecular techniques discussed in the present work.
Traditionally, an SM cluster identified by
genomic approaches is validated by gene
knockout or overexpression to establish a link
between the genetic prediction and SM production. Deletion of two genes in A. nidulans led to
the identification of the four emericellamides
C-F evident from the comparison between the
metabolic profiles of the wild type and deletion
strains (Chiang et al. 2009). Six azaphilones,
azanigerones A-F, were also discovered in the
same manner in A. niger and further validated
by overexpression and gene deletion mutants
(Zabala et al. 2012). Alternatively, a cluster can
be heterologously expressed, as was the case of
didymellamide B, whose cluster was identified
in A. solani and transferred to A. oryzae for
expression (Ugai et al. 2016).
282
M. Flak et al.
Rutledge and Challis 2015; Sanchez et al.
2012), representing an underexploited reservoir
of novel chemistries. As such, the genomic
approaches used in natural product discovery
are a valuable toolkit for capturing the general
biosynthetic potential of fungi.
Targeted genome mining approaches routinely used to identify BGCs are based primarily
on domain-dependent protocols, which identify conserved elements within the constituent
genes of a cluster. The antiSMASH framework
(Medema et al. 2011), with its latest version,
antiSMASH 4.0 (Blin et al. 2017), represents
one of the most comprehensive algorithms,
capable of recognizing 44 different classes of
biosynthetic enzymes. Others, like SMURF
(Khaldi et al. 2010), ClustScan (Starcevic et al.
2008), or PRISM (Skinnider et al. 2015), recognize core enzymes typical for fungal secondary
metabolite clusters, including PKSs, NRPSs,
DMATSs, and/or hybrid PKS-NRPS entities.
For a recent overview of software tools and databases
for analysis of biosynthetic gene clusters in genomic
data, the reader is directed to a review by Weber (2014)
and references therein.
In contrast to the motif-dependent tools,
protocols have been designed to identify clusters lacking a core enzyme. MIPS-CG, an algorithm developed by Takeda et al. (2014), utilizes
comparative genomics of two species, recognizing gene order within a nonsyntenic genomic
background, rather than conserved motifs.
Similarly, comparative phylogenomics uses
the distribution and evolution of gene clusters
to predict the roles of genes, potentially linking
genetic elements and SMs (Brown and Proctor
2016), identifying novel clusters from phylogenetic alignment outliers (Kang 2017), or revealing an evolutionary plasticity in SM profiles
(Mattern et al. 2017). Comparative genomics
approaches are also highly suited to the identification of BGCs in a scenario where the product is known, but the encoding cluster remains
to be discovered. Using the product structure,
information can be gained about the types of
enzymes involved in its production. This strategy led to the identification of the gene cluster
of griseofulvin, an antifungal polyketide
produced by several Penicillium species. Comparing the genome of the producing P. aethiopicum with that of P. chrysogenum, a nonproducer, facilitated the elimination of orthologous clusters. From the remaining ones, only
one contained methyltransferases and a halogenase necessary for the formation of griseofulvin. The deletion of the PKS confirmed the
prediction (Chooi et al. 2010).
The SM cluster producing echinocandin B1, a novel
antifungal from Emericella rugulosa, was identified in
a similar fashion (Cacho et al. 2012).
While mining of genome data using bioinformatic tools is valuable due to its speed of
processing and high throughput potential
(e.g., Cacho et al. 2015), it should be noted
that this approach is invariably predictive and
the derived hypothesis needs to be validated
experimentally. That links the computational
prediction with wet lab techniques, based
either on the OSMAC strategy and a relatively
inefficient screening of modified physicochemical cultivation conditions [e.g., isolation of
new antibiotics from Aspergillus parasiticus
(Bracarense and Takahashi 2014)], or one of
the molecular techniques discussed in the present work.
Traditionally, an SM cluster identified by
genomic approaches is validated by gene
knockout or overexpression to establish a link
between the genetic prediction and SM production. Deletion of two genes in A. nidulans led to
the identification of the four emericellamides
C-F evident from the comparison between the
metabolic profiles of the wild type and deletion
strains (Chiang et al. 2009). Six azaphilones,
azanigerones A-F, were also discovered in the
same manner in A. niger and further validated
by overexpression and gene deletion mutants
(Zabala et al. 2012). Alternatively, a cluster can
be heterologously expressed, as was the case of
didymellamide B, whose cluster was identified
in A. solani and transferred to A. oryzae for
expression (Ugai et al. 2016).
282
M. Flak et al.