Identification of the S. bombicola transportome
The importance of transport processes i.e. the uptake of nutrients, flux of signaling molecules or the extrusion of toxic compounds is evident. In industrial biotechnology, tinkering with the main driver can lead to increased production rates [32], the usage of waste feedstocks [33] or alternate product export [34]. Fundamental knowledge about these transporters is essential, yet so far limited. The bioinformatic analysis presented in this article is the first step on the way to functionally characterize the S. bombicola transportome. The transporter percentage and its density is in line with what is expected for small eukaryotic transportomes, with 100–500 members [2]. Furthermore, the distribution among transporter families follows roughly the same subdivision as the three yeasts present in the TransportDB database. The largest (super)family by far is the MFS superfamily of secondary active channels. The primary active pumps, ABC transporters, are the second largest family, making up less than a third in size. This finding follows the evolutionary trend that was observed from prokaryotes to higher eukaryotes; low-energy-demanding transport classes occur more frequently in higher eukaryotes compared to prokaryotes because of an evolutionary selection pressure for lower energy [2].
Arguably the most challenging aspect of fully understanding the transportome in any given organism, is the designation of the substrates and their specificity fitting each transport system. Generally, this knowledge cannot be gathered in a high-throughput setup but has a high technical and resource-intensive demand. Preliminary computational analysis is therefore often valuable to start from. This in silico analysis identified 254 putative transporters, and linked them to their respective orthologous genes in better known yeasts. Curated database TCDB as well as RefSeq were searched for substrate information providing the first clue regarding their function (Table S2). To improve accuracy of substrate prediction, evolutionary information and the AAIndex as implemented in the web server ‘TrSSP: The Transporter Substrate Specificity Prediction Server’ helped to confirm the TransAAP division of the transporters in the substrate groups depicted in Fig. 2 [35]. Knowledge of a protein’s subcellular localization is an additional help in understanding its proper function.
Mapping the expression profile of the S. bombicola transportome
RNA sequencing is a tool that is conceptually unbiased; mapping the reads to the genome sequence helped to identify ORFs and therefore correctly annotate genes, finetuning the results that were obtained from the automated annotation pipeline to 254 transporters. The expression of two transporters, SbStl1.1 and SbVht1.1 could not be detected, which could imply that specific growth conditions are needed for their expression or that they might be candidate pseudogenes. We believe the latter to be the most likely, because both have at least one gene duplicate. The expression of all other transporter genes not only confirmed their presence, but also indicate their potential biological relevance.
The transporters with highest expression in the exponential phase are mostly glucose transporters, transporters involved in adenine and methionine synthesis, sulfur metabolism and cell wall. Also in S. cerevisiae, these functionalities are clustered with a similar expression pattern [36]. Martinez et al. chose ScMup1 as reference transporter, which shows a similar expression pattern as SbMup1.1 in this analysis: it is highly expressed in the exponential phase only to go down in the stationary phase. Later in the growth curve, reaching the stationary phase and therefore SL production, transporters involved in nitrogen metabolism are more highly expressed. A similar observation was made in Pseudomonas aphidis, during the production of mannosylerythritol lipids, just as sophorolipids a type of glycolipid [37]. This can be explained by the high carbon to nitrogen ratio at that timeframe; transcriptional changes occur with the release of nitrogen catabolite repression [38]. Nitrogen depletion switches on the transcription of S. bombicola‘s nitrogen transporters (SbGap1.2, SbGap1.3, SbDur3, SbUga4.2, SbFur4, SbHnm1,SbMep2.1) similarly to what is seen in Y. lipolytica and S. cerevisiae [39]. The biggest effect of this high C/N ratio however, is observed on the SL biosynthetic genes, including the transporter SbSLMdr.1. This is supported by previous results and S. bombicola’s natural niche, in the midst of flowers, bees, and therefore high sugar content [40]. SLs are produced extracellular as excess carbon storage, unavailable for consumption by competitive microorganisms. Because of this, SL production is only observed in the stationary phase. Indeed, all genes present in the SL biosynthetic gene cluster are induced in the stationary phase [41]. The expression of the entire SL biosynthetic cluster is regulated simultaneously, at least partially by a subtelomeric effect [11]. How exactly the C/N ratio is steering the cluster regulation is still to be determined in detail.
After analysis of the differential expression between the metabolizing sample and the stationary phase sample in Fig. 4b, it is clear that the cells are in survival mode. Because all nutrients enter the cell via transport proteins, the cell heavily invests in the presence of transporters with a large variety in capacity and affinity to increase the chances of gating the scarce nutrients that are left in the media into the cell. The best example of this phenomenon was found in S. cerevisiae and its hexose transport family encoded by HXT genes [42]. This family of glucose transporters contains 20 members that are differentially expressed depending on the glucose concentration in the medium. In excess of glucose, low affinity transporters are prevalent, while they are downregulated to be replaced by high affinity transporters when glucose becomes limited. How this regulation occurs in cooperation with glucose sensors ScSnf3 and ScRgt2, on transcriptional level, but also on post-translational level, was reviewed by Bisson et al. [43]. Following this reasoning, it can be speculated that SbHgt1.1, SbItr1, SbCpaT and SbMFS.9 are high affinity transporters for S. bombicola’s preferred carbon source, glucose. SbHgt1.3 on the other hand, that is very highly expressed in the exponential phase, will more likely be a high capacity and low affinity glucose transporter. Besides glucose, S. bombicola is in search for other forms of carbon and/or energy, which is reflected by the upregulation of amino acid and peptide transporters SbPtr2, SbMtr.1, SbOpt2.1 and SbMup1.1.
The same nitrogen transporters, i.e. SbGap1.3, SbDur31.1, SbMep3, under influence of the nitrogen regulation are again repressed when the C/N ratio drops to growth-limiting conditions in the metabolizing phase. Yet, some orthologues of the same amino acid transporters are actually upregulated (SbGap1.2, SbGap3.3). Most likely, the cell turns to a second strategy besides the higher glucose affinity; it upregulates transporters that can scavenge amino acids as alternative carbon source [42]. More information on how amino acid transporters are degraded by endocytosis in response to nitrogen excess or starvation in S. cerevisiae and Aspergillus spp. can be found in the recent review of Barata-Antunes et al. [44]. The SLs that are stored extracellularly are now actively consumed for the cell’s survival, so the SL biosynthetic genes and its central transporter are largely silenced in response to this lower C/N ratio. Other members of the ABC transporter family are also significantly downregulated. This can be explained by the functional mechanism of ABC transporters in eukaryotes that are mainly – but not all, as was believed until recently [45] - exporters. MFS transporters belonging to the Drug:H + Antiporter-1 Family and therefore mainly responsible for export, such as SbQdr2, are downregulated. The same holds true for SbFfz1; this transporter previously described by Goncalves et al. as a specific, low affinity and high capacity fructose transporter that allows a fructophilic metabolism [46]. SbCtp1, the mitochondrial citrate carrier that was linked to SL production is also downregulated in SL metabolizing conditions [12].
Highlighting the ABC transporter family
Yeast ABC transporters are shown to translocate structurally and functionally unrelated molecules. These include fungicides, antibiotics, flavonoids, detergents etc. [17]. For this reason, this group of transporters is investigated first in the search for the multidrug resistance phenotype culprits. Even though this yeast is not pathogenic, overcoming the multi-resistance phenotype is still beneficial from a biotechnological point of view. In recent years, some progress has been made toward engineering this rather unconventional yeast, yet genetic modification of it is still in its infancy compared to the myriad of engineering tools that are available in model yeasts nowadays [9]. Only two dominant drug selective markers can be used and this in rather high concentrations: 500 µg/mL hygromycin or 600 µg/mL nourseothricin which can be counteracted by the introduction of the hygR gene or nat gene respectively [47]. Understanding and restricting this multidrug resistance can help the development of new selective tools that can be used for genetic modification. Members of the MDR family within the ABC Superfamily are also known to transport lipids, lipopolysaccharides and/or lipoproteins [48]. It is therefore not a surprise that both the previously identified SbSLMdr.1 and the newly discovered SbSLMdr.2 can be categorized in this family.
Four transporters were chosen to study more in detail: SbCdr4, SbYcf1.2, SbYor1.1 and SbYor1.2 From Table S5, it can be seen that the selected transporters are not essential for viability, but they do have an influence on the yeast’s fitness. The closest homolog to SbCdr4, CDR4 from C. albicans is one of the seven full-sized members of the Pdr protein subfamily that is largely responsible for the notorious multidrug resistance of that yeast [49]. However, CDR4 does not have a role in antifungal resistance, yet it does function as a phospholipid translocator [50]. The second highest homology is shared with Yarrowia lipolytica ABC1, the long chain (C14/C16) alkane importer. Three other highly similar Y. lipolytica ABC transporters - ABC2-ABC4 - were also suggested to be involved in alkane transport, both in- and outward [19]. SbYor1.2 in its turn shows the closest homology to YOR1-like transporters from S. cerevisiae that are known to export a broad range of substrates, including hydrophobic molecules [17]. Indeed, Cui et al. (1996) showed that S. cerevisiae ΔYOR1 is especially sensitive to carboxyl group containing compounds such as propionic acid or reveromycin A [28].
S. bombicola wild type is still able to grow in the presence of 200 µg/mL decanol, and only moderately affected by 100 µg/mL undecanol, whereas the model yeast S. cerevisiae already stops growing at 50 µg/mL and 25 µg/mL respectively [51]. The liquid toxicity assay suggests that this four times higher tolerance to medium chain fatty alcohols is attributed to the activity of ABC transporters. This knowledge can be of use in an industrial biotechnology set-up: medium chain fatty alcohols are broadly used in various industries, yet are currently extracted from vegetable oils and animal fat. The transition to a more ecological, microbial production of fatty alcohols would be beneficial. To that end, genetically engineered E. coli was already developed in 2009 [52]. Better results were obtained more recently using the oleaginous yeast Y. lipolytica, that reached maximum undecanol production titers of 550 mg/L [53]. In these works, the aspect of export was not taken into account, leading to growth impairing accumulation of the fatty alcohols inside the cell [39]. Furthermore, costly extraction procedures prohibit upscaling to industrial levels. The chief reason for this neglect is the lack of functional knowledge of transporters. Hu et al. (2018) even turned to domain shuffling of human fatty alcohol transporters, to improve fatty alcohol secretion in S. cerevisiae [32]. The S. bombicola transporters described in this manuscript are the first experimentally confirmed transporters that participate in the efflux of medium chain fatty alcohols and could therefore be used when aiming at higher production and efflux of these compounds.
SLMdr.2: the second SL transporter
Secondary metabolites are non-essential for growth and reproduction, but are commonly produced to function as communication signals. Transport systems are often found within gene clusters encoding their biosynthetic genes, being responsible for the secretion of the secondary metabolite where they can interact with other (micro)organisms [54]. Furthermore, several ABC-transporters associated with secondary metabolism are highly conserved among fungi [23]. All fungal glycolipid biosurfactants produced by yeasts that are characterized so far are also organized in such clustered architecture with a transporter in its center, as was recently reviewed by our group [55]. Taken together, it is not surprising that SLMdr.1 is under heavy transcriptional regulations that coincide with the rest of the cluster genes. From the RNAseq results depicted in Fig. 5, it can be clearly seen that the SLMdr.1 transporter gene expression is highly upregulated in the stationary growth phase, governed by a high C/N ratio (1757.52 cpm exponential vs 10943.07 cpm stationary). In metabolizing conditions, when the C/N ratio drops and SLs are being consumed rather than produced, the SLMdr.1 centrally located in the SL biosynthetic gene cluster, is again largely downregulated, yet its expression remains rather high (781.68 cpm). Nevertheless, secondary metabolite clusters are among the most mobile genetic elements, driven by processes such as gene loss, duplication and horizontal gene transfer [56]. This might explain the existence of a paralogous gene, SbSLMdr.2, that has the same function but is subject to different regulatory rules. SbSLMdr.2 is more steadily expressed in all three studied stages of growth, albeit slightly - but not significantly- higher in the metabolizing stage (852.00 cpm exponential, 941.01 cpm stationary and 1479.31 cpm metabolizing). These results clearly indicate a different regulatory system between these two highly similar transporters. Transcription of SLMdr.1 most likely is regulated as part of the SL biosynthetic gene cluster, a mechanism that is up to now only partially resolved [11]. A high C/N ratio is needed for induction of the cluster enzymes, and at least partially, the subtelemoric positioning of the cluster has an effect. SLMdr.2, located on another chromosome not surrounded by any other SL biosynthetic enzymes is no subject to this cluster regulation.
Analysis of SL formation with either S. bombicola ΔSLMdr.1 or S. bombicola ΔSLMdr.2 demonstrated a drop in product titer to approximately one third of what is achieved with wild type S. bombicola. The double knockout strain yielded even less (17%) compared to both single SL transporter knockouts, which means there is redundancy in their functionality. This can also be derived from the fact that the drop in SL export is equally large for both knockouts. Based on the UPLC chromatograms, no shift in composition of the SL mixture could be observed compared to the SL mixture produced by wild-type S. bombicola, which indicates that there is no marked substrate preference within the SL spectrum for either of the transporters. Even in the double deletion mutant, SL molecules could still be detected in the extracellular environment. Alternative secretion routes are probably still in play. Most likely they are less specific and consequently less efficient. A different outcome can be observed when no rapeseed oil is added to boost SL production. As expected, the overall yield is much lower (5–10%). Moreover, the SL production titers of both wild-type S. bombicola and S. bombicola ΔSbSLMdr.1 is approximately the same. Only in S. bombicola ΔSbSLMdr.2 and S. bombicola ΔSbSLMdr.1ΔSbSLMdr.2, a reduction in SL titer could be observed. This suggests that when no rapeseed oil is added to the fermentation broth, the SL export is predominantly governed by SbSLMdr.2, as opposed to the SbSLMdr.1, the transporter present in the biosynthetic gene cluster. Altogether, these results confirm the hypothesized role of SLMdr.2 as a SL exporter.