Identification of novel citramalate biosynthesis pathways in Aspergillus niger

Background The filamentous fungus Aspergillus niger is frequently used for industrial production of fermentative products such as enzymes, proteins and biochemicals. Notable examples of industrially produced A. niger fermentation products are glucoamylase and citric acid. Most notably, the industrial production of citric acid achieves high titers, yield and productivities, a feat that has prompted researchers to propose A. niger to serve as heterologous production host for the industrial production of itaconic acid (IA), a promising sustainable chemical building-block for the fabrication of various synthetic resins, coatings, and polymers. Heterologous production of IA in A. niger has resulted in unexpected levels of metabolic rewiring that has led us to the identification of IA biodegradation pathway in A. niger. In this study we have attempted to identify the final product of the IA biodegradation pathway and analyzed the effect of metabolic rewiring on the bioproduction of 9 industrially relevant organic acids. Results IA biodegradation manifests in diminishing titers of IA and the occurrence of an unidentified compound in the HPLC profile. Based on published results on the IA biodegradation pathway, we hypothesized that the final product of IA biodegradation in A. niger may be citramalic acid (CM). Based on detailed HPLC analysis, we concluded that the unidentified compound is indeed CM. Furthermore, by transcriptome analysis we explored the effect of metabolic rewiring on the production of 9 industrially relevant organic acids by transcriptome analysis of IA producing and WT A. niger strains. Interestingly, this analysis led to the identification of a previously unknown biosynthetic cluster that is proposed to be involved in the biosynthesis of CM. Upon overexpression of the putative citramalate synthase and a genomically clustered organic acid transporter, we have observed CM bioproduction by A. niger. Conclusion In this study, we have shown that the end product of IA biodegradation pathway in A. niger is CM. Knock-out of the IA biodegradation pathway results in the cessation of CM production. Furthermore, in this study we have identified a citramalate biosynthesis pathway, which upon overexpression drives citramalate bioproduction in A. niger.


Background
The filamentous fungus Aspergillus niger is widely known for its secretion capacity of metabolites, proteins and enzymes. Due to the species robust nature, and range of interesting compounds with generally regarded as safe (GRAS) status that it can produce, A. niger is a frequently used work-horse in industrial biotechnology [1]. Notable examples of industrial compounds produced by A. niger are citric acid, oxalic acid, gluconic acid, amylase and glucoamylase [2][3][4][5]. A. niger has also been proposed for the production of heterologous products, such as cyclodepsipeptides, a class of secondary metabolites that exhibit a variety of pharmaceutically relevant bioactivities and itaconic acid (IA), a promising sustainable chemical building-block for the fabrication of various synthetic resins, coatings, and polymers [6,7].
The production of IA with A. niger reaches industrially relevant titers due to an rewired pathway involving the endogenous cytosolic citrate synthase CitB and ATP-citrate lyase [8,9]. Together with an improved  6:19 fermentation protocol this resulted in the highest IA titer reported for A. niger (56.5 g/l) [9]. However, this metabolic rewiring towards IA proved more intricate as we have also observed induction of genes that are responsible for IA bioconversion and degradation in high IA producing A. niger strains [10]. The gene products itaconyl-CoA transferase (IctA) and itaconyl-CoA hydratase (IchA) together constitute a pathway that bears much similarity with IA degrading pathways reported in A. terreus and the pathogenic bacteria Pseudomonas aeruginosa and Yersinia pestis [11,12]. However, whereas the IA degrading pathways in aforementioned species convert IA into the cellular building-blocks pyruvate and acetyl-CoA, the end product of the pathway in A. niger is unknown, as the gene encoding the enzyme that facilitates the final step in the conversion of citramalyl-CoA into pyruvate and acetyl-CoA, cclA, although present, is not induced in A. niger under IA degrading conditions. We have previously reported that an unknown peak was detected during HPLC analysis in samples where extracellular IA titers were diminishing [10] Deletion of the pathway specific genes ictA and ichA results in cessation of IA bioconversion and concomitantly the unknown peak is also no longer detected [10]. In this study, we have focused on identifying the gene pathways related to this unknown compound, identified as citramalate (CM), and its link with IA production in A. niger. To further explore the unexpected level of metabolic rewiring in IA producing A. niger strains, we have analyzed the transcriptome of high and low IA producing strains for genes that are related to biosynthesis and transport of the industrially relevant metabolites citric acid, succinic acid, fumaric acid, malic acid, lactic acid, gluconic acid, oxalic acid, itaconic acid and citraconic acid to see the effects on these genes, which interestingly have led to the identification of another completely unknown CM biosynthesis route [13].

Strains and culture conditions
All A. niger strains used in this study are listed in Table 1.

Auxotrophic mutant (pyrE) selection
CimA #B3 was cultivated on MM agar plates in the presence of 5-Fluoroorotic acid (5-FOA) to generate pyrE mutant strains, resulting in uridine auxotrophy. Spores of colonies were transferred to MM + 5-FOA agar supplemented with uridine and uracil in 48 well plates, using sterilized toothpicks, for an additional selection round. Growing strains of the second selection round were transferred to MM without uridine and uracil to check if the 5-FOA resistant mutants were indeed uridine auxotroph. DNA was isolated from uridine auxotrophic transformants, as described in "Vector construction and transformation", and PCR with primers 98 + 99 (Additional file 1: Table S1) was performed to confirm pyrE mutant strains.

Vector construction and transformation
Restriction digestion, ligation and other standard molecular biological techniques were performed using common procedures [17]. All primers were obtained from Eurogentec and are listed in Additional file 1: Table S1. PCR reactions were performed with the Alpha Cycler 4 (PCRmax). All enzymes were purchased from Ther-moFisher and used following the manufacturer's protocols. Fungal DNA isolations for colony PCR were performed on mycelia grown in 1 ml LCM in a 2 ml round well 96-well microtiter plate (MTP) (Axygen) sealed with semi permeable film at 33 °C, 850 rpm, overnight in a rotary shaker. DNA was isolated from the mycelia using the DNA isolation from Plant kit and protocol (Nexttec GmbH). This included homogenization with 300 μl acid washed 0.1 mm Zirconium beads (Biospec Products) and 2 × 1 min bead-beating with cooling on ice in between (Mini-Beadbeater-96). The supernatant was directly used as template for PCR.
To create overexpression construct of cimA the An09g00170 gene was in vitro synthesized at GeneArt (Waltham, MA) and subsequently digested with Hin-dIII. The digested cimA fragment was ligated in HindIII digested pABgpdI vector containing the A. niger gpdA expression signals, thereby establishing the pABgpdI-cimA expression vector.
For the construction of an mfsB (An09g00190) expression vector, mfsB was PCR amplified from AB1.13 genomic DNA with Phusion HF Master Mix and primer pair 432 + 433 (Additional file 1: Table S1) following standard protocols. The resulting fragment was purified, using the QIAquick PCR purification kit and protocol (Qiagen). 2.5 μg of the purified fragment was digested with NcoI/BpiI in one reaction, and BpiI/ BglII in a second reaction. The 1123 bp generated fragment from the first reaction and 623 bp generated fragment from the second reaction were excised from gel and purified using the QIAquick gel extraction kit and protocol (Qiagen). These two fragments were inserted into an NcoI/BglII opened pAB-gpdI backbone carrying the gpdA expression signals, originating from pABgpdI-citC [10], establishing the mfsB expression vector pABgpdI-mfsB. This was done in a ligation reaction consisting of T4 DNA ligase and buffer, and a total of 140 μg DNA with a vector:insert ratio of 1:3. The mixture was incubated at room temperature for 30 min. 4 μl of the ligation mixture was transformed into Escherichia coli JM109 (Promega) according to the manufacturer's standard heat shock protocol. Presence of pABgpdI-mfsB in colonies was checked with colony PCR, using DreamTaq Green PCR Master Mix. Several positive transformants were miniprepped according to the GeneJET Plasmid Miniprep kit and protocol (ThermoFisher). Restriction analysis with BpiI was performed to validate the plasmids identity, followed by maxiprep of designated transformants using the Plasmid Plus Maxi kit and protocol (Qiagen). The identity of the purified plasmid was verified by Sanger sequencing using primers 143, 329, 430 and 431 (Additional file 1: Table S1) (Baseclear).
Transformants were plated on MM + 1.2 M sorbitol as osmotic agent and selected based on the reestablishment of uracil prototrophy due to integration of the functional pyr expression cassette. Individual colonies were transferred to 48-well plates containing MM agar. These were used to inoculate MTPs, and DNA was isolated as described earlier. Successful integration of the mfsB and cimA expression cassettes was determined by colony PCR, using primers 143 + 433 for PgpdA-mfsB-TgpdA and 143 + 331 for PgpdA-cimA-TgpdA. Positive transformants were streaked on MM plates to obtain pure colonies. DNA isolation and colony PCR was repeated as described above, and positive transformants were used to prepare spore suspensions as described in "Strains and culture conditions".

Metabolite analysis
Extracellular metabolite concentrations were determined by high-performance liquid chromatography (HPLC). A WATERS e2695 separations module equipped with an Aminex HPX-87H column (Bio-Rad) was used in combination with 5 mM H 2 SO 4 as eluent, coupled to a refractive index detector (WATERS 2414) and a dualwavelength detector (WATERS UV/Vis 2489) for peak detection. For identification of various organic acids as described in fungal biosynthetic pathways, reference compounds were analyzed for retention time and UV210 nm /RI area ratios. Empower PDA software was used for data processing.

RNA isolation, transcriptome sequencing and analysis
Biomass samples for RNA isolation were taken at several timepoints during controlled-batch cultivation and washed with distilled water and frozen in liquid N 2 . The controlled batch cultivations were performed using 5 l scale benchtop New Brunswick Scientific fermenters (BioFlo 3000) at 33 °C. Starting pH was 3.5 after inoculation and M12 medium [16] was allowed to naturally acidify till pH 2.3 and then kept at pH 2.3 by addition of 4 M KOH. Dissolved oxygen (DO) tension was 25% at the moment of inoculation and when DO dropped till 20% it was kept at 20%. The system was calibrated with 100% sterile air as 100% DO and 100% N 2 as 0% DO. The fermenter was inoculated by 72 h old 100 ml non-baffled shake flask cultures containing 1.0*10 8 spores. mRNA isolation procedures for transcriptome sequencing and analysis are published in Hossain et al. [10].

IA bioconversion in A. niger revisited
In our previous communication, we have reported the role of ictA and ichA during IA bioconversion in A. niger [10]. We have observed that the expression of ictA and ichA is induced under IA producing conditions in high IA producing strain CitB#99 and knock-out of these genes resulted in abolishment of IA bioconversion [10]. Surprisingly, during IA bioconversion, we have also observed the occurrence of an previously unidentified compound in HPLC analysis (Fig. 1). Upon further analysis we hypothesized that this unidentified peak could be citramalic acid (CM), as based on the identified IA bioconversion pathway, citramalyl-CoA is an intermediate that is formed during IA bioconversion, which could be converted to CM by action of IctA [10,11]. To confirm this, detailed HPLC analysis was carried out. Based on this analysis, the unidentified compound shared a very similar retention time and UV210 nm /RI area ratio, as the CM standard (Additional file 1: Tables S2 and S3). Based on this observation we concluded that the unidentified peak is CM. Upon this observation, we have revisited our previous fermentation data as presented in Hossain et al. [8,10] to identify the presence of CM during conditions of IA bioconversion (Fig. 2). Interestingly, in both cases we have been able to identify the presence of CM in the extracellular medium upon decreasing levels of IA in the medium. The occurrence of CM in both cases coincides with reducing titers of IA, further strengthening the hypothesis that IA is converted into CM in A. niger (Fig. 2).

Alternative organic acid production in A. niger
It is well established in literature that A. niger is a prolific producer of organic acids, more particular gluconic, citric and oxalic acid [19]. The unexpected finding of CM biosynthesis and secretion during IA bioconversion prompted us to search for additional and previously unidentified organic acid biosynthesis pathways in this organism. For this purpose we have explored the genome mining efforts presented by Li et al. [13] in A. niger to identify genes potentially related to the biosynthesis of nine industrially relevant organic acids: citric acid, succinic acid, fumaric acid, malic acid, lactic acid, gluconic acid, oxalic acid, itaconic acid and citraconic acid. In Table 2 we have summarized transcriptome data of these genes. Furthermore, transcriptome data of the putative orthologs and paralogs in A. niger of the well described transporters for malic acid (MaeA) [20], itaconic acid (Itp1 and MfsA) [8,21,22] and hydroxyparaconic acid (Itp1) [22] are presented together with the citrate exporter CexA [23] (Table 3). From this latter comparison it is interesting to note that the two functionally characterized IA transporters have different putative orthologs in A. niger (Table 3), which by itself is a IA nonproducing strain. Interestingly, the expression of the canonical glucose oxidase goxC, that is responsible for gluconic acid formation, is practically absent in the analyzed strains under the cultivation conditions that we applied. Furthermore it is interesting to observe the significant downregulation of oahA, that encodes oxaloacetate hydrolase and is responsible for oxalate production, in high IA producing strain CitB#99 compared with AB1.13. Both results correspond with the consequent absence of gluconic acid and oxalic acid in HPLC analyses of cultivations with IA producing strains [8,10].  Glyoxylate shunt specific genes An01g09270 and An15g01860 that code for isocitrate lyase (acuD) and malate synthase respectively are downregulated in CitB#99. It is well established in literature that itaconate can inhibit the glyoxylate shunt in pathogenic bacteria, however it was not known if this is also the case in fungi [12,24,25]. Our results suggest a relation of IA bioproduction and glyoxylate shunt downregulation.
Another interesting observation is that citB overexpression downregulated expression of another putative cytosolic citrate synthase citC, similar as citB, being part of a secondary metabolite pathway of which all genes are downregulated, including two cadA like genes An0g09950 and An01g09930 [10].
Furthermore, we have observed that the expression of a gene encoding a 2-isopropylmalate synthase (IPMS) like protein (An09g00170), with significant similarity to a bacterial citramalate synthase (cimA), is strongly reduced in CitB#99 [26]. However, the expression of An01g13160, that codes for the canonical IPMS, is not affected. This uncharacterized gene encoding the IPMS like protein is clustered together with an major facilitator superfamily transporter (An09g00190), whose expression is also downregulated significantly in CitB#99. CimA and IPMS, together with homocitrate synthase, belong to the LeuA dimer superfamily [27]. To explore the role of this novel gene cluster, its overexpression was studied.

Overexpression of cimA
To test whether the gene product of An09g00170 is involved in organic acid biosynthesis we have overexpressed the putative cimA gene under control of A. niger gpdA expression signals. Upon transformation, 96 colonies were randomly picked from transformation plates, cultivated in microtiter plates and the supernatant analyzed on HPLC. Out of the tested 96 colonies, the strains CimA A10, CimA B3 and CimA D11 produced a compound with the same HPLC profile as CM and colony PCR confirmed the presence of pABgpdI-cimA (Additional file 1: Tables S2 and S3; PCR data not shown). CM production was further tested in cimA overexpressing strains CimA A10, CimA B3 and CimA D11. Non baffled shake flasks were inoculated and samples taken for HPLC measurement. After 280 h of incubation CimA A10 had accumulated 1.83 g/l CM and 10.01 g/l CA, CimA B3 accumulated 7.03 g/l CM and 6.83 g/l CA, CimA D11 accumulated 5.41 g/l CM and 5.87 g/l CA, whereas the parental AB1.13 strain accumulated 19.55 g/l CA and no detectable CM (Fig. 3). These results indicate that the gene product of An09g00170 is involved in citramalate biosynthesis. To further boost the production of CM we have overexpressed the MFS multidrug transporter that is clustered together with cimA in the A. niger genome.

Overexpression of mfsB
Having established CM production upon cimA overexpression, we subsequently tested the effect of overexpressing mfsB in CM producing strain CimA B3. For this purpose auxotrophic (pyrE−) strain was generated by cultivation on 5-fluoro-orotic acid. Transformation of CimA B3 pyrE− with mfsB expression cassette resulted in 21 transformants that were verified by PCR analysis (data not shown). Ten strains were selected for shake flask cultivation together with the parental CimA B3 strain to evaluate CM production (Table 4). Four transformants performed better in CM yield and titer compared with CimA B3 (CimA + MfsB #17, #27, #28, #85). The two best performing strains CimA + MfsB #27 and #28 were selected for further experiments.

Shake flask cultivation
Overexpression of mfsB in CM producing CimA B3 strain resulted strains with increased CM yield. The CM production performance of two of these strains, CimA + MfsB #27 and #28, was compared with the parental CimA B3 and AB1.13 strains in 500 ml nonbaffled shake flask cultivations. Strain AB1.13 produced no detectable CM, while max. 12.4 g/l CA was produced after 236 h, after which CA titers strongly reduced (Fig. 5a). This effect is caused by the depletion of glucose in the medium after 236 h (Fig. 5b). CM production of strains CimA B3, CimA + MfsB #27 and CimA + MfsB #28 is comparable between the three strains and final titers of 6.6 g/l, 6.4 g/l and 5.9 g/l CM is produced respectively after 333 h of cultivation (Fig. 5a). CA is also produced as side product in cultivations with strain CimA B3 (max. 9.2 g/l), CimA + MfsB #27 (max. 6.1 g/l) and CimA + MfsB #28 (max. 3.7 g/l) after 333 h. Interestingly, CA production in mfsB overexpressing strains only starts after 142 h of cultivation, whereas CA titer of 2.4 g/l is already achieved after 72 h in cultivation with strain CimA B3 (Fig. 5a). Equally interesting is the observation that glucose is only depleted in cultivations with strains AB1.13 and CimA B3 but not in cultivations with strains CimA + MfsB #27 and #28 with 24.9 g/l and 45 g/l glucose left respectively. This observation is also in line with the increased CM yield of strains CimA + MfsB #27 and #28 (Table 5). More detailed analysis of the HPLC results from flask cultivations of strains expressing both cimA and mfsB also identified a compound with HPLC characteristics similar to citraconic acid (Fig. 5).
In conclusion, by overexpressing cimA and mfsB we have converted A. niger into a system that predominantly produces CM (and eventually its degradation product citraconic acid) and reduced CA levels ( Table 5).

Discussion
Heterologous IA bioproduction in A. niger resulted in high levels of unexpected metabolic rewiring, as exemplified by the induction of two genes, ictA and ichA, that are involved in IA degradation upon high IA titers [5]. The proteins encoded by these genes intracellularly convert IA into a previously unknown compound. In this study we identified CM as being the end product of the IA biodegrading pathway in A. niger. We have shown that IA is converted into CM during IA biodegradation, by action of IctA and IchA as the genes encoding these enzymes are strongly induced upon IA bioproduction [10]. This is in contrast with the end products of the IA biodegrading pathways in Y. pestis,   P. aeruginosa and A. terreus, which are pyruvate and acetyl-CoA. Surprisingly, the bacterium Alcaligenes xylosoxidans has also been reported to intracellularly convert IA into CM, indicating that A. niger is not the only organism with this phenotype [29]. IA degradation and concomitant CM bioproduction cessate by knocking out either ictA or ichA [10].
What the role of CM is in A. niger metabolism and why A. niger converts IA into CM is not yet clear. One explanation for the secretion of CM in a IA overproducing strain could be that the gene encoding the last step in the IA biodegrading pathway, cclA which codes for citramalyl-CoA lyase, is not induced in A. niger upon IA biodegradation and the conversion to pyruvate and acetyl-CoA therefore does not occur in A. niger, being a natural non-IA producing host [10]. Interestingly, Meijer et al. [30] have also detected citramalate in A. niger, however, this was in cell lysates of WT A. niger where normally the IA degradation pathway is not induced [10,31]. This suggests that there must be other endogenous biosynthesis pathways for CM whose function is yet unknown. To explore possible novel organic acid biosynthesis pathways in A. niger we have looked into metabolic pathway rewiring in transcriptome data (Table 2). Interestingly, in this dataset we have observed the downregulation of a putative IPMS An09g00170, which upon overexpression drives CM production. This result prompted us to designate An09g00170 as citramalate synthase cimA. Furthermore, in our transporter comparison analysis, we have seen that the ortholog of the functionally characterized IA transporter from U. maydis, Itp1, is the characterized citrate transporter in A. niger, CexA (Table 3), while the putative ortholog to the functionally orthologous A. terreus MfsA is An02g07580. These results suggest that these organic acid transporters show significant redundancy, also explaining that without co-expression of a pathway specific transporter the related organic acid can still be exported albeit at low(er) levels [8,15,22,32].
Citramalate synthase has been described as an enzyme from the archaea M. jannaschii that is a part of the isoleucine biosynthesis pathway and together with IPMS belongs to the LeuA dimer superfamily [26,27]. Whereas IPMS catalyzes the condensation of acetyl-CoA with α-ketoisovalerate to form isopropylmalate in the leucine biosynthesis pathway, citramalate synthase catalyzes the condensation of acetyl-CoA with pyruvate to form citramalate [26,27]. To the best of our knowledge, this is the first example of CimA driven CM production in filamentous fungi. We are still speculating about the role of CM in the metabolism of A. niger, however, it is possible that CM is an intermediate in the isoleucine biosynthesis pathway as is the case in archaea, and is clearly a topic for further research. It is also interesting to note that cimA is clustered together with an major facilitator superfamily transporter An09g00190, which we have termed mfsB. This observation led us to speculate that mfsB is responsible for or involved in the cellular export of citramalate. Previously, it has been shown that the cellular export of metabolites can be the limiting factor resulting in low titers and yields [8,32]. However, upon overexpression of mfsB we have not observed strongly increased titers of CM, but we have observed an increased CM yield and secretion of citraconate (Table 5). Moreover, during IA biodegradation and concomitant CM production, the expression of mfsB is strongly repressed ( Upon mfsB overexpression, the titer and yield of CA dropped, suggesting a change in the metabolism where CA production is reduced to favor CM production. As also citraconate is produced after prolonged cultivation upon mfsB overexpression, other metabolic conversion may take place driven by transporter action. This result further shows the crucial role of these transporters in organic acid production as is also recently shown by Wierckx et al. [33]. It is further interesting to note  that also CA secretion resumed later on during cultivation by strains CimA + MfsB #27 and #28 (Fig. 5a). We speculate that a (nutrient) limitation in the cultivation medium may be causing this phenotype. This result would then indicate that medium optimization towards optimal CM production in A. niger is required. We have recently successfully performed medium optimization towards improved heterologous IA production in our lab [9]. Apart from medium optimization, genetic engineering to further optimize the CM biosynthesis may also be applied. We hypothesize that the overexpression of ATPcitrate lyase would improve the biosynthesis of CM by increasing the precursor pool of acetyl-CoA for CimA, similar as observed for IA bioproduction [9]. It is also relevant to mention that the two CM biosynthesis pathways, as identified in our research, would produce two different enantiomers of CM (Fig. 4). The further elucidation of these pathways and the pathwayspecific enantiomer that is produced is topic for further research, however the fact that in cimA/mfsB overexpression strains the produced CM seems to be converted further into citraconic acid ( Fig. 5; Additional file 1: Tables S2 and S3) suggests that in that case R-citramalate is produced [34], while in the itaconic acid degradation pathway this can only be S-citramalate [35]. Moreover, CM is an interesting compound from industrial perspective, due to its potential to serve as bio-based precursor for methyl methacrylate synthesis, which in turn is the building block for acrylic glass (Plexiglas) [36,37]. This has spurred further research activities into optimizing the bioproduction of CM [38].

Conclusion
We have previously reported the intracellular biodegradation pathway of IA in A. niger. In this study we have identified the end product of this biodegradation pathway as being CM. Knock-out of the biodegradation pathway specific genes ictA or ichA results in the cessation of IA biodegradation and concomitant CM production. Furthermore, in this study we have identified, through transcriptome analysis, an alternative citramalate biosynthesis pathway, which upon overexpression drives bioproduction of citramalate in A. niger. The biosynthetic citramalate synthase is clustered with a putative transporter, which upon overexpression results in almost 2-fold higher citramalate yield on glucose, suggesting it to be a citramalate exporter. However, as also citraconate is secreted, these observations would require additional research similar as was recently done for the IA transporters [22,33].