Phylogenetic and Structural Analysis of Polyketide Synthases in Aspergilli

Introduction

Aspergillus species have recently gained great attention in view of their impact on humans and agriculture and due to the production of bioactive secondary metabolites (SMs). Sequencing of Aspergillus species genome led to the identification of more than 200 SM gene clusters with the potential to produce SMs, which still need to be explored. Many of these clusters include polyketide synthases (PKSs) as the principal enzyme. ¹ The SM compounds presently identified from Aspergillus species under studied culture conditions are only handful, and various research groups are exploring SM compounds using different approaches. Web-based online tools, such as SMURF: genomic mapping of fungal SM clusters, have been developed and are extremely useful as they give the annotation of gene clusters for sequenced fungal genomes. ² Considering the wealth of information provided by whole-genome sequencing and the presence of schematically arranged gene clusters in Aspergilli, researchers are encouraged to explore medically and industrially important compounds.^3–5 Polyketides are industrially well-exploited class of compounds in microbes mainly due to their medicinal importance. ⁶ Several Aspergillus polyketide products have been very well characterized and understood, eg, aflatoxins from Aspergillus flavus,^7,8 melanin pigments from Aspergillus fumigatus, ⁹ and lovastatin from Aspergillus terreus.^10,11 In order to exploit the potential of Aspergillus species for the secretion of useful polyketides, it is necessary to examine and understand the PKS enzyme machinery. ¹²

PKS enzymes of Aspergilli are iterative type I PKSs and are close structural and functional analogs to mammalian fatty acid synthases (FASs). ¹³ This type of enzymes reuses their domains in cyclic fashion. In Aspergilli, PKSs catalyze the condensation of precursor, acetyl-CoA and malonyl-CoA(n), to produce polyketides.^8,12 Three essential domains of PKSs are ketosynthase (KS), acyltransferase (AT), and acyl carrier protein (ACP). The presence of other domains such as β-ketoacyl reductase (KR), enoyl reductase (ER), dehydratase (DH), methyltransferase (MT), and Claisen cyclase/thioesterase (CLC/TE) is variable and depends on the type of PKSs in Aspergillus species. ¹⁴ The functionality of reducing and non-reducing (NR) domains such as MT, DH, ER, and CLC/TE of PKS enzyme directs the types of end compounds produced. PKSs based on the presence of their domains may have differences in end product catalysis. Gene clusters containing PKS and nonribosomal peptide synthases (NRPSs) are putatively identified, and experimental studies are being conducted to link these putative clusters with the secreted compound in various in vitro conditions.^15–17 Selecting the gene clusters with PKS gene based on in silico evidence and short-listing the candidates for further knockdown studies to explore the polyketide compound will narrow down the research efforts for possible SM exploration.

A comparative sequence analysis of the Aspergillus full-length PKS proteins has been carried out with functionally characterized fungal PKSs by a phylogenetic approach, to examine and assign the putative function to unexplored PKSs of Aspergillus species. In the current study, attempts have been made to characterize selected PKS genes from six Aspergillus species and to better understand the gene architecture and protein structures of PKS enzyme.

Methods and Materials

Fungal strains and cultural conditions. Aspergillus isolates were collected from the Indian Type Culture Collection, Indian Agricultural Research Institute, New Delhi, India, in BSL2 facility. These isolates were isolated from various agricultural and animal sources and were morphologically characterized as Aspergillus species and later verified by amplification using Internal transcribed spacer (ITS) region primers. All fungal isolates were handled in biosafety hood cabinet A2 as per the biosafety protocol of the institute. Cultures were grown on YGT media (0.5% yeast extract, 2% dextrose, and 1 mL of trace elements) at 27 °C for 24 hours at 125 rpm. Genomic DNA was extracted using the LETS buffer that contains 0.1 M lithium, 20 mM EDTA, and 0.5% SDS. ¹⁸ DNA from fungal species (Aspergillus, Trichoderma spp., and Fusarium spp.) was isolated, as described earlier.^18,19 The quality of DNA from the isolations was checked by gel electrophoresis, and DNA concentration was determined by NanoDrop.

Database search. Six Aspergillus species, A. fumigatus Af293, A. flavus NRRL3357, Aspergillus niger CBS 513.88, A. terreus NIH2624, Aspergillus oryzae RIB40, and Aspergillus clavatus NRRL1, were selected for this study based on their agricultural, medical, and industrial relevance. Aspergillus genome was searched for putative PKS sequences by subjecting the protein sequence of KS domain of PKSP (XP_756095.1) of A. fumigatus Af293 as a query into BLASTp. Domains in PKS were searched by subjecting each putative amino acid sequence to online tools SEARCHPKS, ²⁰ MapsiDB, ²¹ and CDD (NCBI). A total of 190 Aspergilli PKS sequences were retrieved and analyzed in this study. Iterative type I PKS sequences from other fungi, where polyketide products are characterized, were also retrieved and included for comparison. ²² These PKS proteins are from Penicillium chrysogenum Wisconsin 54-1255, ²³ Penicillium marneffei ATCC 18224, ²⁴ Pyrenophora tritici-repentis Pt-1C-BFP, Gibberella zeae PH-1, ²⁵ Penicillium citrinum, ²⁶ Gibberella moniliformis,^27,28 Emericella nidulans, ²⁹ Monascus purpureus, ³⁰ Gibberella fujikuroi, ²⁷ and Cochliobolus heterostrophus. ³¹ Type III PKSs from bacteria (Streptomyces coelicolor A3(2), ³² Myxococcus xanthus DK 1622, ³³ and Mycobacterium tuberculosis H37Rv) ³⁴ and FASs (from Homo sapiens [AAC50259], Gallus gallus [P12276], and Caenorhabditis elegans [NP_492417]) were used in this study for comparison and outgrouping. Gene accession numbers for these PKSs are given in Supplementary Table 2.

Phylogenetic analysis. Full-length PKS protein sequences were used in multiple sequence alignments using Clustal X (2.0.12). ³⁵ The resulting data were saved as Clustal and PHYLIP format files, and alignments were written as postscript files for further analysis. The phylogenetic analyses were performed in PHYLIP (ver. 3.69) programs, SEQBOOT, PROTDIST, NEIGHBOR, and CONSENSE, in the order as described previously.^36,37 The Jones–Taylor–Thornton amino acid substitution matrix was performed where input order of sequences for phylogenetic analysis was randomized. For generating the KS domain tree, maximum parsimony-based method with a bootstrap value of 1000 was used and the final consensus tree was selected from the 100 MP trees obtained. Alignment gaps were treated as fifth character state, parsimony uninformative characters were excluded, and all the characters were unordered and equal weight. ³⁸ Tree length and consistency index were also calculated. Phylogenetic tree CONSENSE files obtained from PHYLIP were viewed with TreeView and MEGA (5.0). ³⁹

Conserved motifs for KS, sequencing, and phylogeny verification. Sequence alignments were studied for consensus sequences in KS domain across the Aspergillus species. Amino acid change in KS sequences was analyzed for predicting the functionality of model proteins and type of end polyketide product it catalyzes. Multiple alignment files of KS sequences were used to identify homologous regions of amino acid from Aspergillus spp. Amino acid sequences were converted into gene sequences using ExPASy (www.expasy.org/translate/), and forward and reverse primers of length 20-28 bp spanning KS region were designed by using Primer3 software (www.bioinfo.ut.ee/primer3-0.4.0) to check Tm, delta G, and self-hybridization. Degenerate primers were checked for specificity by BLASTn at NCBI before synthesis. Primer set KS_F and KS_R was standardized to amplify the PCR product of same size from three Aspergillus species in one reaction. Degenerate primer sequences are:

KS_F:AGTCTTGCKGCYATYCAYWTGGSCTGCAAYKCSATCTGGAGRA, KS_R: TGAGTWCCTG-KTCCRTGCATTTCMACGTAGC. As positive control, universal fungal primers – ITS1 (based on 18S rRNA) and ITS4 (based on 28S rRNA): TCCGTAGGTGAACCTGCGG and ITS4: TCCTCCGCTTATTGATATG – were used for amplification from Aspergillus DNA (data not shown here). ²⁴ PCR was standardized in a total reaction volume of 50 μL. The mixture contained 10 × reaction buffer (100 mM Tris, 500 mM KCl; pH 8.3), 10 mM dNTP mix, forward and reverse primer (20 μM), five units of Taq DNA Polymerase (Platinum® Taq DNA Polymerase; Invitrogen), 500 ng of template DNA, and autoclaved water in 50 μL of total volume. PCR conditions were initial denaturation at 95 °C for 5 minutes, followed by 35 cycles of denaturation at 95 °C for 30 seconds, annealing of 56 °C for 30 seconds, and extension at 72 °C for 1 minute, followed by a final extension at 72 °C for 10 minutes in a Mastercycler gradient (Eppendorf). Amplified DNA products were separated by electrophoresis in a 1% agarose gel with a 100-bp DNA ladder as a molecular size marker (Promega Corporation). PCR products were purified by Qiagen PCR purification column (Qiagen, Germany) and submitted for sequencing. End PCR product was sequenced by either KS_F or KS_R and searched for sequence homology by BLASTX at NCBI. The deduced amino acid sequence was determined using ExPASy server at EBI (http://web.expasy.org/translate/). Multiple alignments of annotated KS sequences with KS protein from Aspergillus species were carried out using Clustal X. Consensus sequences were observed with respect to earlier bioinformatics analysis of KS for the presence of motifs. Neighbor joining tree was generated using PHYLIP package. Sequences were submitted to NCBI (KT221846-KT221852 and KT213730-KT213740).

Structural modeling of KS domains. KS sequence from UniProt or program database (PDB) was retrieved, and PDB files were generated by subjecting sequences to SBSPKS software (http://www.nii.ac.in/~pksdb/sbspks/master.html). ⁴⁰ KS domain sequences from Aspergillus PKSs were named in this study as PKSP (XP_756095.1/Q4 WZA8), 6-methylsalicylic acid (6MSA; BAE65442.1/Q2PIT2), lovastatin nonaketide synthase (LNKS; Q9Y8A5), Afl25 (XP_002384329.1), Afu1 (XP_753141.2), and Anr (XP_001402408.2), which were modeled using the template structure of beta-ketoacyl-ACP synthase II from Escherichia coli (PDB ID: 1KAS). ⁴¹ Modeller9v7 was used to build protein structures (target-template alignment and model generation), which uses an automated approach for comparative protein structure modeling. ⁴² For each of the target protein sequence, 20 independent structures were generated and the best model was selected on the basis of minimum Discrete Optimized Protein Energy score. Then, the top model was further minimized through 400 steps of the steepest descent, followed by 3000 steps of conjugate gradient using GROMACS 4.0.5. ⁴³ Six conserved motifs identified through sequence alignment of each clade have been mapped on these modeled structures. The area and volume of the active site conformations were measured through CASTp server. ⁴⁴

Results

Modeling of KS domain and mapping of conserved motifs

KS sequences from eight major clades were aligned, and six conserved consensus motifs were identified. ⁵⁵ Particularly, motif – DTACSSSL – in KS carries Cys residue in the active site, which is a signature conserved amino acid across the KS domain in the PKSs from other species. ⁵⁶ The two His residues, known to be in KS catalytic triad, were also found to be conserved in motifs EXHGTGTXXGDP and GSXKX-NXGHXE in the KS sequences. A careful analysis of conserved motifs shows the amino acid changes with respect to the specific type of pk. Figure 3 shows the conserved motifs in KS sequences and variations among them with respect to the type of PKSs. To further verify our grouping of NR, PR, and HR based on phylogeny, we used modeling study for KS proteins. Our hypothesis was that NR, PR, and HR types of KS may have their differences in how they dock the substrate at the active site and thus impacts the catalytic activity of the enzyme. In order to find the structural changes in the three types of PKS enzymes (NR, PR, and HR), KS domain structures have been predicted by homology modeling. Model proteins used for this study were PKSP from A. fumigatus for NR PKS, 6MSA from A. terreus for PR PKS, and LNKS from A. terreus for HR PKS. Three Aspergillus PKSs classified and observed in the current study as NR, PR, and HR were also taken to test our hypothesis. These hypothetical proteins were Afl25 (XP_002384329.1) from A. flavus as NR, Anr (XP_001402408.2) from A. niger as PR, and Afu1 (XP_753141.2) from A. fumigatus as HR type. The percentage identities of the template sequence of 1KAS (used as a standard reference for generating the model) with KS sequences of Afl25, Anr, and Afu1 are 28%, 26%, and 27%, respectively. As reported in the crystal structure of KS domain of FAS of E. coli, the catalytic triad Cys-His-His is observed to be conserved in all these structures. The area and volume of these active site conformations were measured through CASTp server. The active site pocket volume of these different types of proteins varies according to the type of PKS and the nature of chain elongation reactions (NR, PR, and HR; Fig. 4). Chain elongations eventually determine the diversity of chemical compounds in iterative PKSs. The alignments of KS domain sequences from NR type KS, HR type KS, and PR type KS are shown in Figure 5. The conserved motifs are highlighted in yellow color, and two Tyr residues that are found conserved in all three types of PKSs are marked in the box. PR PKS, 6MSA from A. terreus, contains the smallest cavity volume (365.8 Å3), as the iteration reactions in this enzyme are reported to be only 3. Similar results were found with the modeling of protein Anr from A. niger. NR PKS such as PKSP from A. fumigatus shows medium cavity volume (749.8 Å3) and is known to have five iteration cycles. Modeled structure of Afl25 from A. flavus classified in clade I (pigment) is also found to have moderate cavity volume (~750 Å3). HR type of PKS such as LNKS from A. terreus is found to have the largest cavity volume (1408.3 Å3) with the variable (<20) iteration reactions. Similar results were found in modeled protein Afu1 from A. fumigatus. In these proteins, the number of iteration cycles was observed to be dependent on the size of the cavity in the active site of the PKS enzyme in a particular type with products such as pigment, 6MSA, and lovastatin, as described in Figure 5. OPEN IN VIEWER

Figure 2

KS genealogy of Aspergillus PKSs (1000 boot strap value).

OPEN IN VIEWER

Figure 3

Conserved motifs on type I KS sequences.

OPEN IN VIEWER

Figure 4

Three-dimensional modeled structures of NR, PR, and HR type of KS.

OPEN IN VIEWER

Figure 5

Three-dimensional modeled structure from Aspergillus KSs.

The conserved motif amino acid sequences identified as pigment KS were converted to DNA sequences, and degenerate primers were designed for amplification for specific detection of Aspergillus species. Primers based on KS region from PKS genes were used for PCR amplification using the DNA of Aspergillus species, such as A. fumigatus, A. flavus, A. parasiticus, A. niger, and A. oryzae (Supplementary Fig. 6). A PCR product of 450 bp was observed with the DNA template of A. fumigatus (5 isolates), A. flavus (18 isolates), and A. niger (6 isolates) specifically. DNA from Trichoderma, Fusarium, and Bacillus species and human DNA were used for PCR controls, which did not give any amplification in the expected range. Sequences were confirmed with multiple alignments of pksP (gene) from A. fumigatus, A. flavus, and A. niger and submitted to NCBI (KT221846-KT221852 and KT213730-KT213740; Supplementary Fig. 6).

Discussion

PKSs play a major role in contributing to the diversity of polyketide products produced in Aspergilli. Polyketide biosynthesis mainly uses acetyl-CoA and malonyl-CoA as precursors and synthesizes hexanoyl-CoA or pentaketide as the end product. This is used by PKS to generate the stable polyketide precursor with various lengths, which will eventually be processed by the remaining biosynthetic enzymes in the pathway to generate polyketide compound with different applications, such as toxins, antibiotics, and pigment compounds. To gain insight into the synthesis of diverse product by observing the diversity in the PKS enzyme, full-length Aspergillus PKSs and only KS domain from six important Aspergillus species with functionally characterized fungal PKS enzymes were used for comparative sequence analysis. This also correlates the unknown PKSs of Aspergillus with known polyketides they produce. Phylogenetic analysis of Aspergillus PKSs based on the domain architecture facilitated the classification of PKS proteins into three different types of enzymes, ie, NR, PR, and HR PKSs grouped into eight clades.

Earlier, phylogenetic studies were performed for type I PKSs of Ascomycota group of fungi, which divided PKSs into 18 clades, indicating that the grouping was based on the presence/absence of reducing and NR domains in PKS.^22,57 Distribution of PKS enzymes in Aspergilli is studied by phylogeny, ⁵⁷ and we have attempted to assign Aspergilli PKS protein to the probable chemical nature of the compound based on sequence homology. In the current study, Aspergillus PKSs fall into eight clades corresponding to the probable chemistry of the end compound they may synthesize based on the rationale of homology. The NR fungal clade contains PKSs that synthesize unreduced, and usually aromatic, PKs that are precursors to toxins, eg, aflatoxins and pigments.^7,58 All PKSs within this clade lacked ER, DH, and KR domains, which are interpreted as a loss of reducing domains, compared to the domain structure of type I PKSs. The NR fungal PKSs are predicted to synthesize PKs in which the keto groups are either not reduced or reduced by enzymes other than PKS. Unreduced PKs are typically synthesized from acetyl- and malonyl-CoA. PKS proteins with an additional TE/PP domain were scattered throughout the NR PKS clade, as was the case for reducing PKS subclade III. The functional significance of these duplicated PP domains is not known. Aspergillus PKSs are iterative type I class of enzymes, and exceptions to these are recently found type III PKSs in A. flavus and A. oryzae. To outgroup any other type of enzymes, such as closely homologous proteins in FASs, and also type III and modular type I PKSs, representative proteins were also included in our study. M. tuberculosis PKSs are known to have type I modular and type III PKS and are characterized to be a part of gene cluster producing virulent lipids, such as phthiocerol and phenol-phthiocerol.^59,60 Modular type I PKS from S. coelicolor is known to produce antibiotics. ⁶¹ These enzymes were classified into different subclades, and none of Aspergillus PKSs are categorized in this clade. This confirms that Aspergillus species do not have modular PKSs that may produce lipids such as polyketides.

Conserved motifs in the KS domain have been identified that are specific to pk they produce. Three-dimensional structures also reported the changes at the active site confirmations with respect to the type of compound they produce and the cavity volume in their active sites. Change in the cavity groove volume can be linked to the malonyl starter units fitted at the enzymatic site and thus may help predict substrate utilization by the enzyme. In the modeled 6MSAS KS structure, Yadav et al has observed that two tyrosine residues are protruding into the cavity blocking the downward flow of the cavity. ⁶² In the current study, Aspergillus-modeled proteins (annotated as NR, PR, and HR) were also found to have two Tyr residues highly conserved in all KSs and were also aligned on the structure. This finding supports the hypothesis that putative Aspergillus PKS may be functionally active in appropriate in vitro conditions. These three proteins can be selected for further in vitro studies and explored for polyketide exploration studies using advance methods. ⁶³ The presence of certain amino acids at their active site pocket and their alignment in a particular fashion to accommodate the substrate clearly suggest that the diversity in the end product is related to the substrate size and the number of molecules of substrate it can fit for condensation reaction. Recent experiments involving the generation of altered fatty acid-polyketide hybrid products by the rational manipulation of benastatin biosynthetic pathway also suggest that the number of chain elongations is dependent on the size of the PKS enzyme cavity. ⁶⁴ The in silico analysis of the sequence and structural features of iterative KS domains reported in this study may provide logical selection of residues to be mutated and help in exploring the effect of substrate specificity and the end product. Strategic site-directed mutagenesis studies can be planned; knockout and overexpression studies using molecular cassette in reference strains can be performed to identify which products are accumulating in a reasonable amount at certain cultural conditions compared to wild type. Such bioinformatics studies will be helpful in providing pilot results to choose best PKSs candidates for the exploration of novel polyketide compounds for knockout studies. No experimental studies on the modeled Aspergillus proteins analyzed in the current study have yet been reported; therefore, these proteins can be top PKS candidates to explore pk compound. The present in silico analysis gives key leads for such experiments.

We report the unexplored repertoire of PKSs in Aspergillus species. We predicted three proteins of Aspergillus spp. that can be explored for their reducing or NR pk products, namely, XP_002384329.1, XP_753141.2, and XP_001402408.2. Among Aspergillus spp., A. flavus, predominantly an agricultural pathogen and mycotoxin producer, is often reported in immunosuppressed patients. ⁶⁵ A. flavus genome contains 25 PKSs. ⁴⁹ It has been reported by various comparative genomics studies that A. flavus has 55 gene clusters associated with secondary metabolism. However, only handful metabolites with their pathways have been assigned to these clusters. ⁶⁶ Our analysis suggests that A. flavus has the potential to produce numerous polyketides of different natures. This remains to be explored by various expression studies. In the current study, A. flavus protein Afl19 has been linked with pigment production, which has been explored for DHN melanin pathway, while this manuscript was under preparation. ⁶⁷ Aspergillus ochraceus is known to produce ochratoxin A and penicillic acid, which cause significant problems in animal and human health. ⁴⁷ It also produces other metabolites derived from these mycotoxins, such as diaporthin, orthosporin, and asperlactone. The biological activity of these SMs has not been characterized so far, and many PKSs have not been functionally characterized in this species. These molecules may be beneficial (antibiotics) or harmful (mycotoxins) to human health. ⁶⁸ The analysis of this report may find helpful to start with selective pickup for PKs to study.

Utility of KS region diversity can be exploited in the detection of three important Aspergilli, together with one degenerate primer pair designed in this study. Such detection method can be standardized further to identify Aspergillus species, A. flavus, A. fumigatus, and A. niger, in one reaction for screening agricultural and clinical samples.

In conclusion, the sequence-based analysis reported in this study highlights the diversity of the compounds that may probably be produced by Aspergillus species. This study shall serve as a platform to screen the particular type of metabolite synthesized by an Aspergillus species. These in silico analyses of PKSs have facilitated the understanding of the biosynthetic pathways for unknown SMs from Aspergillus species, as well as its utility for unique identification.

Author Contributions

Conceived and designed the experiments: PB, MP, PUS. Work was carried out under the project conceived by PUS. Analyzed the data: PB, AB, RM. Wrote the first draft of the manuscript: PB, PUS. Contributed to the writing of the manuscript: PB, AB, RM. Made critical revisions and approved final version: PB, PUS, TM, YS, AV. All authors reviewed and approved of the final manuscript.

Supplementary Material