The first transposable element, discovered by McClintock in 1947, belongs to the Ac/Ds transposons, a part of the hAt transposon superfamily, which were mostly studied in plants (Du et al., 2011). The dissociator (Ds) element can excise from its locus in response to the activator (Ac) element, which encodes an active transposase flanked by terminally inverted repeats (TIRs) similar to those in the Ds element. Upon excision a small scar is left at the excision site and an 8-bp duplication is created at the target site upon transposition (Yan et al., 1999). The transposon prefers insertion in unique low copy sequences and tends to avoid highly repetitive DNA. Therefore, it preferentially inserts into or near coding sequences and tend to insert into closely linked regions. This feature makes Ac/Ds transposons highly suitable for insertion mutagenesis (Vollbrecht et al., 2010). However, Ac/Ds transposition frequencies are variable and in several species too low for large scale mutagenesis. Therefore, a hyperactive Ac transposase, discovered by Lazarow et al. (2012) is now widely used. The AcTPase4x carries a combination of four amino acid substitutions which result in a 100-fold increase in the excision frequency in yeast. The four mutations did slightly affect insertion preference for the insertion site: Typically, AcTPase4x prefers a slightly lower GC-content in its environment compared to wild type AcTPase (Lazarow et al., 2012). Noteworthy, both AcTPase and AcTPase4x form inactive protein aggregates when expressed a high levels. This is probably to protect the host from harmful transposition frequencies (Heinlein et al., 1994; Lazarow et al., 2012). So far, Ac/Ds was used successfully in the model yeast S. cerevisiae and haploid cells of the human fungal pathogen Candida albicans (Michel et al., 2017; Segal et al., 2018).

The hAT superfamily of transposons contains several other transposons that are used in fungi, such as Hermes, from the house fly Musca domestica. This mobile element exists of one single component, i.e., a transposase enzyme flanked by TIRs. Unlike its distantly related cousin, the Ac element from maize, it does integrate into a specific “nTnnnnAn” target sequence and also leads to a target site duplication of 8-bp. Up to now, Hermes has been used in the model yeasts S. cerevisiae and Szichosaccharomyces pombe and the human fungal pathogen Candida glabrata (Subramanian et al., 2009; Gangadharan et al., 2010; Edskes et al., 2018). The native Hermes TPase has been used in fungi as well as a hyperactive version (Subramanian et al., 2009; Gangadharan et al., 2010; Edskes et al., 2018). Next to Hermes, the TcBuster transposon from the flour beetle Tribolium castaneum was recently discovered in the hAT transposon family (Woodard et al., 2012). It differs from Hermes mainly in its DNA binding and insertion domain and in their insertion target site, which is “nnnTAnnn” for TcBuster (Arensburger et al., 2011). TcBuster was only used in Komagataella phaffii, previously classified as Pichia pastoris (Zhu et al., 2018a).

The Tc1/mariner transposons are the best-known superfamily of class II transposable elements. As in the hAT family, the transposon encodes a transposase enzyme, which is flanked by TIRs. Upon insertion into its target sequence, TA, this dinucleotide is duplicated (Dornan et al., 2015). In 1997 Ivics et al. (1997) genetically revived Sleeping Beauty an inactive Tc1/Mariner family transposon from fish. The authors used molecular phylogenetic data to guide the elimination of inactivating mutations within the gene encoding the transposase (Ivics et al., 1997). This transposon was used in Komagataella phaffi (Zhu et al., 2018a).

Third, the piggyBac transposon, originally from the cabbage looper moth Trichoplusia ni, forms its own family. It is currently the most widely used transposon system for genetic manipulations. PiggyBac-like elements are present in other moth species, frogs and bats, making it the first active transposon in mammals. This wide host range makes it very popular for several applications, such as determination of gene essentiality and drug target identification (Yusa, 2015; Gao et al., 2018). It is a 2475-bp long autonomous mobile element, containing TIRs and subterminal inverted repeats on either end. It contains a single open reading frame, which encodes the piggyBac transposase (Elick et al., 1996). PiggyBac shows very precise excision, leaving no excision scars. Furthermore, it is a unique eukaryotic transposon, as no DNA synthesis is required and excision occurs through hairpin formation, as described for specific bacterial transposons (Bischerour and Chalmers, 2007; Mitra et al., 2008). PiggyBac always inserts in a TTAA target-site, as this palindromic sequence is required for its insertion (Yusa, 2015). In addition, it shows a strong trend to hop locally, thus inserting close to the donor site, preferably in the same chromosome and in accessible chromatin structures (Li et al., 2013a). An optimized engineered non-autonomous piggyBac transposon system was first constructed and used in the silk worm Bombyx mori (Tamura et al., 2000). Further optimization was performed over the years to adapt the system to different species. Different markers, with a size of up to 10 kb, can be inserted between the TIRs of the transposon and, as such, can be mobilized without losing transposition efficiency (Yusa, 2015). A hyperactive transposase, hyPBase, was developed to increase transposition efficiency. Error prone PCR was used in order to generate a pool of mutant PBase DNA. Upon a subsequent screening in a S. cerevisiae transposition assay, several mutants with excision frequencies up to seven times higher compared to the wild type were identified. The hyPBase maintains the scarless transposon excision typical for piggyBac (Yusa et al., 2011). The piggyBac transposon has been used in the model organism S. cerevisiae, the industrial yeast K. phaffii and the human fungal pathogen C. albicans (Gao et al., 2018; Mutumwinka et al., 2018; Zhu et al., 2018b).

In model yeasts such as S. cerevisiae and S. pombe, a lot of information is available on essential genes and even conditionally essential genes, which are only essential under certain growth conditions. This information was acquired using deletion collections, such as the yeast deletion collection in S. cerevisiae (http://www-sequence.stanford.edu/group/yeast_deletion_project/deletions3.html) (Mira et al., 2009). Making such collections is, however, very labor-intensive and molecular tools need to be available in the investigated species. Deletion collections can be used to identify conditionally essential genes by either screening every mutant separately in a high-throughput screen, or using competition assays, where pools of barcoded mutants are grown together and their relative presence is quantified. Transposon mutagenesis can be used to identify essential genes or conditionally essential genes if enough unique insertions are acquired and if a transposon system is used without specific target sites (Mira et al., 2009; Novo et al., 2013).

In the model yeast S. cerevisiae, Michel et al. (2017) succeeded in full functional mapping of the baker's yeast genome using saturated transposon mutagenesis. Seven libraries were constructed in different environmental conditions using a transposon inserted in the ADE2 gene, either encoded on a plasmid or endogenously. More than 280,000 independent insertions were identified, with no sequence preferences and no large genomic regions without insertions. The Ac/Ds transposon did favor inter-nucleosomal DNA and pericentromeric regions in case of excision from a centromeric plasmid or targeted close to the excision site in case of excision from the endogenous ADE2 locus. Of note was that some known non-essential genes did not have any or only very few insertions in this approach. This observation was explained by dubiously annotated ORFs that overlapped with essential genes, by conditional essentiality or by the data analysis procedure in which repeated sequences were filtered from the reads. On the other hand, essential genes contained some transposons, specifically in permissive domains. The mutant libraries were also compared with each other to highlight that the differences in insertion locations were linked to medium usage or to the background strain/mutant (Michel et al., 2017).

S. pombe is another very popular model yeast, in which deletion collections exist (Kim et al., 2010). One of the first transposon experiments in S. pombe used Hermes to generate 360,513 insertions in vivo and in a naked gDNA control in order to perform an integration profiling study. The aim of the study was to identify genes essential for cell division. Only 33% of the transposon insertions occurred in the ORFs, and low integration densities corresponded to essential genes. There was, however, an exception to that: several mitochondrial genes, encoded in the nucleus, had high insertion densities even though they were believed to be essential. This is probably due to the existence of large protein pools, which segregate together with the mitochondria and were sufficient to sustain several generations without the presence of the encoding genes. Furthermore, some genes that were marked as non-essential inferred from deletion collection screenings were found to be essential by transposon insertion. These genes were later found mistakenly non-deleted in the deletion collection (Guo et al., 2013). Both of these experiments show that transposon insertion mutagenesis libraries contain even more information compared to classical deletion collections.

Several yeast species are used in industry in order to produce specific molecules at high quantities. K. phaffii, previously classified as Pichia pastoris, is the most popular yeast system for expression of recombinant proteins (Schwarzhans et al., 2017). Its genome sequence is known, yet most genes had not been functionally annotated until recently. Two different transposons have been used in K. phaffii to determine gene essentiality: TcBuster and Sleeping beauty. Both of them were used to construct several insertion libraries. Insertion-containing cells were enriched in selective liquid medium before sequencing. Sleeping beauty inserted in almost 800,00 unique sites, while TcBuster resulted in more than 130,000 independent insertions. Almost 40% of these insertions occurred in intergenic regions, which make up <20% of the genome. When insertions occurred in ORFs, they often positioned near the end of the gene. Insertion sites preferred by Sleeping beauty differed significantly from the ones preferred by TcBuster, Sleeping beauty inserted slightly more at the 3′-untranslated sequences of genes inserting almost randomly, while TcBuster preffered insertion in promoter regions. The different insertion sites led to a total of more than 200,000 unique insertion sites. Most genes with low insertion densities in these libraries were orthologs of essential genes in S. cerevisiae or S. pombe, validating the functionality of the libraries (Zhu et al., 2018a).

Yarrowia lipolytica is an oleagineous yeast, which is often used for the production of hydrocarbon-based chemicals, as it can accumulate up to 90% of its cellular weight in lipids (Blazeck et al., 2014). For Y. lipolytica, only 25% of its genome features are presently functionally annotated, for many others their annotation is strictly based upon homology with S. cerevisiae or S. pombe (Magnan et al., 2016). The Hermes transposon was used to generate 534,000 independent insertions. This led to the classification of roughly 22% of the genes as essential, most of which were homologous to S. cerevisiae and S. pombe essential genes. More respiratory genes are essential in Y. lipolytica compared to the two model species. Most hits were intergenic, with a frequency of one hit every 24.37-bp. For intragenic insertions the frequency was about one hit every 97.5-bp. Insertion mutants were grown in competition over several generations, which showed that more stringent/competitive environments led to a lower variability in the mutants. For example, fewer independent insertion mutants were present after growth on glycerol compared to glucose. Furthermore, the insertion library was screened for accumulation of lipids and several mutants were identified. All the mutants exhibited insertions in intergenic regions, mostly between two divergent genes. This means that the insertion possibly affected not just one, but two adjacent genes (Patterson et al., 2018).

Yeasts can also negatively impact our daily lives, as is the case for pathogenic organisms. C. albicans is one of the most common human fungal pathogens. This fungus is present as a commensal in a significant part of the population and can cause severe invasive infections in immunocompromised individuals. Next to that, superficial infections, mainly on the oral or vaginal mucosa are quite common, even in immunocompetent people (Kim and Sudbery, 2011). Only very recently transposon mutagenesis was used in order to investigate gene essentiality and drug resistance in this fungus (Gao et al., 2018; Segal et al., 2018). The construction of transposon insertion mutants in this fungus was long thought to be compromised by its diploidy and lack of meiosis. In 2013, however, it was discovered that haploids could be obtained through concerted chromosome loss. These haploids were unstable and grew significantly slower than related diploids (Hickman et al., 2013). A derived haploid strain was described, that was selected for its faster growth and, as such, could be used for the construction of a transposon mutant library. The Ds element from maize, containing a selective marker for nourseothricin, was inserted into the 5'-region of ADE2, thus inactivating this gene. A hyperactive transposase was expressed from an inducible promoter, which effectively induced excision and insertion of the transposon (Mielich et al., 2018). This system was used to construct a library from which more than 500,000 independent insertion sites were mapped, with one third of them occurring in annotated features in the C. albicans genome. Around 95% of all annotated features contained at least one transposon insertion. This library was then used to determine gene essentiality under standard laboratory conditions, since essential genes represent potential drug targets. More than 1,500 genes were identified as essential, most of which were essential in S. cerevisiae and/or S. pombe. However, some non-essential orthologs of S. cerevisiae or S. pombe seemed to be essential in C. albicans. This is probably due to functional redundancy due to duplications and paralogs in the model yeasts. Some C. albicans essential genes did not have orthologs in either of the model yeasts and are poorly characterized. They could serve as specific drug targets, especially since some of them have no human homologs and are conserved in other fungal pathogens such as Aspergillus fumigatus and Cryptococcus neoformans (Segal et al., 2018). C. albicans haploids were also used for piggyBac transposition. An insertion frequency of about 5% was reached, leading to insertions in all chromosomes in nearly 5,000 genes. Genes without transposon insertions were probably essential (Gao et al., 2018).

Infections with a second Candida species, C. glabrata are increasing, especially in the western world. This haploid yeast is considered less virulent compared to C. albicans, yet rapidly acquires resistance to the most used antifungal fluconazole (Dyer and Paoletti, 2005). A Hermes transposon was used in a clinical isolate of C. glabrata, generating more than 500,000 independent insertions. These were used to identify essential genes. About 84% of essential genes in C. glabrata had orthologs that are essential in S. cerevisiae. Interestingly, genes from the spliceosome complex, which are essential in S. cerevisiae, were non-essential in C. glabrata. This could be explained by the lower presence of introns in C. glabrata compared to baker's yeast. On the other hand, genes involved in polyamine biosynthesis were essential in C. glabrata but not S. cerevisiae under the used growth conditions (Gale et al., 2020). In a separate study, Levitan et al. (2020) compared three different transposons in three different yeast species for their ability to predict gene essentiality. Data from previously published AcDs, Hermes and piggyBac transposon studies in S. cerevisiae, S. pombe and C. albicans were used in a machine learning approach (Li et al., 2011; Guo et al., 2013; Michel et al., 2017; Gao et al., 2018; Segal et al., 2018). Interestingly, the authors found that a library with many independent insertions would significantly improve performance, while increasing reads per insertion did not have a similarly strong effect. Gene essentiality predictions using the piggyBac-data in both S. pombe and C. albicans were the least efficient, since the specific target site, TTAA, is not present in all ORFs throughout the fungal genome. This also led to a lower transferability of the piggyBac data compared to Hermes or MiniDs data. This means the machine learning algorithm can be trained with a MiniDs dataset in order to analyze Hermes data, but not piggyBac data. Furthermore, the AcDs transposon exhibited cryptic promoter-enhancer activity, which can result in increased expression of downstream ORFs when inserted in promoter sequences. This leads to S. cerevisiae essential genes tolerating MiniDs insertion, yet not Hermes insertion. Interestingly the authors found that essential genes were species- and condition-specific (Levitan et al., 2020).

It is often interesting to identify genes which are necessary for growth in specific conditions, as this can create a more in-depth understanding of pathways needed to metabolize specific nutrients for example. Since K. phaffii is known for its ability to grow on methanol (De Schutter et al., 2009), the libraries made by TcBuster and Sleeping beauty mentioned before, were also used to screen for loss of growth on methanol. More than 100 genes with a possible function in methanol metabolism were thus identified, some of which were previously known to be involved in this process (Zhu et al., 2018a). Another method to identify genes necessary for growth on methanol in K. phaffii, is the use of 2-deoxyglucose (2-DG), non-metabolizable a glucose analog. Upon growth on 2-DG and methanol, only cells that can use methanol as a carbon source can grow. A piggyBac library was screened for genes involved in resistance to 2-deoxyglucose (2-DG) The authors identified five genes involved in 2-DG resistance, including SNF3, GRR1, and MIG1. Some of these genes had been linked to methanol metabolism previously, which confirmed the validity of the system (Hoepfner et al., 2014; VanHoute and Maxwell, 2014).

Specific mutations can be an advantage or a disadvantage for a yeast, or even both at the same time. A deletion that results in slower growth, yet better tolerance to a certain stressor, can help in processes in industry. Ideally, it would be possible to find additional modifications that stabilize the tolerance effect, meanwhile rescuing the growth defect. Transposon mutagenesis can be used to identify such genetic interactions (Li et al., 2011; Yusa et al., 2011; Michel et al., 2017; Edskes et al., 2018). Furthermore, large screening for genetic interactions can help establish networks of functional connections between genes and pathways. Transposon insertion mutagenesis can be used both for extensive screening, when the whole library is sequenced as well as for identifying specific mutations, by only sequencing the mutants that show the desired phenotype. The possibility to only sequence specific mutants makes the process cheaper and easier, as only few sequencing reactions are needed.

In the saturated transposon mutagenesis work of Michel et al. (2017) in S. cerevisiae, two proof-of-principle genetic interaction experiments were included. In one of them a transposon insertion library was made in strains containing a combination of a mutation in the ER-mitochondria encounter structure gene, MMM1, and its suppressor mutation in endosomal protein Vps13. Previously identified genes, that are necessary for proper functioning of this suppressor mutation, indeed became essential in this insertion library, even if they were not essential in the insertion library created in the wild type strain. In a second experiment, two libraries were created including one in a DPL1-deletion strain and the other in a dpl1/psd2 double deletion strain. These genes were previously demonstrated to function in phosphatidylethanolamine biosynthesis. In the double deletion strain, phosphatidylethanolamine can only be generated in the mitochondria by means of Psd1, so that PSD1 as well as genes for lipid shuttling to and from the mitochondria will be essential. This was confirmed using transposon insertion mutagenesis (Michel et al., 2017).

Hermes transposition was also used to identify genes that could rescue a mutant phenotype, more specifically the growth defect induced by the URE-3 prion. A comparison of mutagenesis libraries between a strain that was under constant selection for the presence of the prion and a control strain, resulted in genes with insertions in the control, but not in the prion-containing strain. It was shown that a mutation of GLN1, encoding glutamine synthetase, or a previously uncharacterized gene, LUG1, could rescue the growth defect (Edskes et al., 2018).

Since piggyBac is currently the most commonly used transposon system, it was also used in S. cerevisiae. The system was used to identify mutations rescuing the cell wall defects of a mutant in the OCH1 gene, which is used in glyco-engineered yeast cells for expression of human-compatible glycoproteins. The deletion eliminates yeast specific N-glycan structure production. One mutant was found with the ability to rescue the growth defect of the parental strain. It was proven that this phenotype was linked to transposon insertion, since removal of the element resulted in the parental defect. Further demonstration for the validity of the approach came from the nature of the investigated gene: the insertion occurred in the BEM4 promoter, which resulted in the upregulation of the downstream gene. Bem4 belongs to the same pathway as Rho1, a protein already known to be involved in rescue of the och1 phenotype (Mutumwinka et al., 2018).

The piggyBac transposon system was used in S. pombe to identify mutants that suppress the temperature sensitive growth defect of a cdc25-22 strain. All mutants that rescued the phenotype had insertions in the WEE1 gene, which had previously been identified using chemical mutagenesis (Li et al., 2011).

Several yeast species have negative effects either on our own bodies or on the production of our food. Antifungal drugs are used commonly in the clinic as well as the agricultural sector. Only few classes of antifungals are available and resistance to all of them is increasingly present. Therefore, it is of high importance to develop new drugs as well as to identify resistance mechanisms, as this can help to develop synergistic therapies.

In their extensive S. cerevisiae functional mapping study, Michel et al. (2017) also included a screening for targets of the drug rapamycin, which blocks cell proliferation. They showed interruption of several genes, among which FPR1, the rapamycin receptor, and other genes known to be involved in rapamycin signaling (Michel et al., 2017).

In S. pombe, the pre-constructed library using piggyBac was applied to screen for mutants resistant to thiabendazole, a drug affecting microtubules (https://pubchem.ncbi.nlm.nih.gov/compound/Thiabendazole). Seventy-three resistant mutants were isolated, revealing a novel gene involved in thiabendazole resistance, DAM1, a gene required for proper chromosome segregation. The insertion mutant in DAM1 caused a gain-of-function (thiabendazole resistance). It is therefore always important to verify whether or not specific transposon insertions cause a gain- or a loss-of-function. Interestingly, Li et al. (2011) found in this experiment several resistant mutants that were not caused by transposon insertion, but simply by spontaneous mutation. It is therefore important to verify whether removal of the transposon reverts the original phenotype (Li et al., 2011).

In the clinic, fluconazole is one of the most commonly used drugs to treat fungal infections. Unfortunately, resistance to this drug is becoming increasingly widespread. Gao et al. (2018) used piggyBac transposition in C. albicans haploids in order to screen for genes involved in fluconazole resistance. This led to the identification of several genes that improve fitness during growth in the presence of the drug. Several of them were already known to be involved in this process including members of the ergosterol biosynthesis pathway and its regulation including ERG3, ERG6, and ERG251. Additionally, identified genes were linked to ergosterol and sphingolipid biosynthesis. Upon further characterization of genes involved in sphingolipid biosynthesis (FEN1 and FEN12), a novel in vitro resistance mechanism to fluconazole was discovered (Gao et al., 2018).

Furthermore, the transposon insertion library created with Hermes in C. glabrata was also used to identify genes that regulate susceptibility to fluconazole. Several genes that were previously linked to fluconazole resistance, were identified in this approach, validating the use of insertion profiling for this purpose. On the other hand, several genes that were not linked to fluconazole resistance up to now were identified, among them α-ketoglutarate metabolic genes (KGD1, KGD2, IDH1, and IDH2). Their role in fluconazole susceptibility was confirmed using deletion mutants (Gale et al., 2020).

During their saturated transposon mutagenesis in S. cerevisiae, Michel et al. (2017) noticed that it was not only possible to identify essential genes, but also essential protein domains. As multiple insertions can be present in the same gene, attention should be paid to the position of transposon insertions within the coding sequence. An example is GAL10, which encodes a bifunctional enzyme with epimerase and mutarotase activities. In the experiment, cells were grown on a mixture of α- and β-galactose to avoid the need for conversion of one form into the other. The mutarotase domain exhibited several insertions in the transposon library and facilitated the identification of domains essential for its activity (Michel et al., 2017). During the aforementioned rapamycin screen insertions were found in PIB2. This was unexpected, since pib2 deletions are known to be sensitive to rapamycin. A more in-depth look showed that insertions were only present at the N-terminal end of the encoded protein and resulted in a gain-of function effect on PIB2, as such explaining rapamycin resistance (Michel et al., 2017).