A Versatile Toolset for Genetic Manipulation of the Wine Yeast Hanseniaspora uvarum

Hanseniaspora uvarum is an ascomycetous yeast that frequently dominates the population in the first two days of wine fermentations. It contributes to the production of many beneficial as well as detrimental aroma compounds. While the genome sequence of the diploid type strain DSM 2768 has been largely elucidated, transformation by electroporation was only recently achieved. We here provide an elaborate toolset for the genetic manipulation of this yeast. A chromosomal replication origin was isolated and used for the construction of episomal, self-replicating cloning vectors. Moreover, homozygous auxotrophic deletion markers (Huura3, Huhis3, Huleu2, Huade2) have been obtained in the diploid genome as future recipients and a proof of principle for the application of PCR-based one-step gene deletion strategies. Besides a hygromycin resistance cassette, a kanamycin resistance gene was established as a dominant marker for selection on G418. Recyclable deletion cassettes flanked by loxP-sites and the corresponding Cre-recombinase expression vectors were tailored. Moreover, we report on a chemical transformation procedure with the use of freeze-competent cells. Together, these techniques and constructs pave the way for efficient and targeted manipulations of H. uvarum.


Introduction
The yeast Saccharomyces cerevisiae is undoubtedly the first and most frequently employed biotechnological production organism used by mankind for millennia in the manufacture of bread and alcoholic beverages [1]. However, the quality of fermented products is invariably determined by its co-existence with other yeast and bacterial species and their metabolic activities [2][3][4]. Especially in the production of wine, beer, and cider, Hanseniaspora uvarum is often predominating in the early stages of spontaneous fermentations, worldwide, as it is readily found on the surface of grapes and apples, but also on various other fruits, such as plums and strawberries [5][6][7][8]. Even in coffee and chocolate bean fermentations, H. uvarum contributes significantly to the final product quality [9,10]. In addition, H. uvarum and other endophytic yeasts have been shown to promote growth of their host agricultural fruit plants by producing phytohormones [5]. Moreover, its potential as a biological control agent to diminish the loss of fruit harvests by competing with other yeasts and fungi is increasingly being recognized [6,[11][12][13][14]. Hanseniaspora uvarum can also be employed in another type of biological control to fight Drosophila suzukii, which already causes fruit harvest losses in Southern Europe, especially of grapes and cherries [15]. The strong attraction of the fruit flies to H. uvarum, presumably due to the efficient production of ester compounds like ethyl acetate, is thus employed to build traps for efficient pest management [16,17]. Interestingly, but not yet applied in oenology or medicine, H. uvarum also produces a killer toxin with activity against Candida species [18].

Characterization of the HHO1 Recipient Strain
The Hanseniaspora uvarum strain used in this work, further designated as HHO1, was previously obtained by selecting an isolate from the German stock collection, DSM 2768, for single-celled growth [31]. The sequence of its diploid genome has largely been determined [31]. A comparison of growth rates in rich medium showed that HHO1 generally grows faster than Saccharomyces cerevisiae at temperatures between 15-30 • C (µ = 0.25-0.53 h −1 for HHO1, as compared to µ = 0.15-0.45 h −1 for S. cerevisiae strain HD56-5A), but is more sensitive to warmer conditions, with very poor growth at 37 • C (µ = 0.07 h −1 as opposed to µ = 0.35 h −1 for HD56-5A), and loss of viability at 45 • C [38]. Regarding their morphology, cells are considerably smaller than those of the baker's yeast and display the typical, lemon-like shape (Figure 1a; hence the former designation as Kloeckera apiculata; [24]). The bipolar budding pattern with consecutive rings of bud scars at both cell poles is more readily observed in scanning electron micrographs (Figure 1b-d). [24]). The bipolar budding pattern with consecutive rings of bud scars at both cell poles is more readily observed in scanning electron micrographs (Figure 1b-d).

Cloning and Characterization of Chromosomal Replication Origins
Genetic manipulations of yeasts depend on a method to efficiently introduce homologous or heterologous genes. This is greatly facilitated by the availability of selectable, self-replicating episomal plasmids. To obtain such vectors for propagation in H. uvarum, the hygromycin resistance cassette from pJB2 [37] was first cloned into pUK15JJH as an XhoI/HindIII fragment. The resulting plasmid, pJJH3058, was employed to construct genomic libraries for the isolation of replication origins. For this purpose, it was digested in four independent restrictions with XhoI/BglII, SalI/BglII, EcoRI/BglII, and EcoRI/BamHI (Figure 2a). The upper and lower panels show cells in different stages to demonstrate the typical differences in the budding pattern of the two yeast species. The upper image for H. uvarum shows cells believed to be at the end of cytokinesis, whereas the lower image represents a cell after budding, presumably in the G1 phase of the cell cycle. Size bars are indicated and applicable to both images of the same yeast. (b-d) Images from scanning electron microscopy (SEM) of a logarithmically growing culture of H. uvarum HHO1, showing single-celled growth by bipolar budding, generating the typical lemon-shaped cells. Size bars are indicated. (b) Overview of cells in different budding states. (c) A cell during cytokinesis, which produced five buds at this pole in previous cytokineses (note the five ring-like bud scars on the surface, indicated by arrows). (d) A cell with the typical lemon-shape and bud scars (indicated by arrows) at both poles.

Cloning and Characterization of Chromosomal Replication Origins
Genetic manipulations of yeasts depend on a method to efficiently introduce homologous or heterologous genes. This is greatly facilitated by the availability of selectable, self-replicating episomal plasmids. To obtain such vectors for propagation in H. uvarum, the hygromycin resistance cassette from pJB2 [37] was first cloned into pUK15JJH as an XhoI/HindIII fragment. The resulting plasmid, pJJH3058, was employed to construct genomic libraries for the isolation of replication origins. For this purpose, it was digested in four independent restrictions with XhoI/BglII, SalI/BglII, EcoRI/BglII, and EcoRI/BamHI (Figure 2a). Genomic DNA of strain HHO1 was digested with the restriction enzymes indicated and cloned into the respective sites of pJJH3058. The genomic libraries obtained after transformation and amplification in E. coli were introduced into HHO1 and plasmids from colonies growing on selective medium with hygromycin B were again isolated and characterized by restriction mapping and sequencing. Protein-coding genes on plasmids are indicated by arrows, regulatory regions, and replication origins by bars. (b) Putative ACS-motif of the isolated ARS-sequences. The sequences of the inserts of potential ARS-carrying plasmids were first searched for ARS-sequences using the "ori finder 3" online-tool [39]. Because the identified sequences strongly varied in length (between 70 and 725 bps), the MEME suite 5.5.0 [40] was used to identify a common ACS-motif that is present in all sequences. Note that the ARS sequences of the cloning vectors pJJH3181 and pJJH3200, further designated as HuARS1, were derived as shorter versions from the original isolate pJJH3152, that of pJJH3182 (HuARS2) originated from pJJH3160. Genomic DNA obtained from the H. uvarum strain HHO1 was digested with the same combination of restriction enzymes, ligated with the respective vector fragments, and transformed into E. coli, selecting for kanamycin resistance. Approximately 5000 independent clones were obtained for each library, as judged from colony counts of appropriate dilutions on plates. Cells were washed from plates with the undiluted transformants with 2 mL of LB-kanamycin, each, collected in sterile tubes and used to inoculate fresh overnight cultures in selective medium. From those, plasmid DNA was prepared and introduced into strain HHO1 by electroporation (Figure 2a). Selection on rich medium with hygromycin yielded almost confluent growth on plates obtained from all four libraries, i.e., more than 1000 colonies, each. Three single colonies were picked from the borders of each plate and plasmids were isolated from the twelve overnight cultures grown in rich medium supplemented with 300 mg/L hygromycin B. Clones were readily obtained from (a) Strategy for cloning of HuARS elements. Genomic DNA of strain HHO1 was digested with the restriction enzymes indicated and cloned into the respective sites of pJJH3058. The genomic libraries obtained after transformation and amplification in E. coli were introduced into HHO1 and plasmids from colonies growing on selective medium with hygromycin B were again isolated and characterized by restriction mapping and sequencing. Protein-coding genes on plasmids are indicated by arrows, regulatory regions, and replication origins by bars. (b) Putative ACS-motif of the isolated ARSsequences. The sequences of the inserts of potential ARS-carrying plasmids were first searched for ARS-sequences using the "ori finder 3" online-tool [39]. Because the identified sequences strongly varied in length (between 70 and 725 bps), the MEME suite 5.5.0 [40] was used to identify a common ACS-motif that is present in all sequences. Note that the ARS sequences of the cloning vectors pJJH3181 and pJJH3200, further designated as HuARS1, were derived as shorter versions from the original isolate pJJH3152, that of pJJH3182 (HuARS2) originated from pJJH3160. Genomic DNA obtained from the H. uvarum strain HHO1 was digested with the same combination of restriction enzymes, ligated with the respective vector fragments, and transformed into E. coli, selecting for kanamycin resistance. Approximately 5000 independent clones were obtained for each library, as judged from colony counts of appropriate dilutions on plates. Cells were washed from plates with the undiluted transformants with 2 mL of LB-kanamycin, each, collected in sterile tubes and used to inoculate fresh overnight cultures in selective medium. From those, plasmid DNA was prepared and introduced into strain HHO1 by electroporation (Figure 2a). Selection on rich medium with hygromycin yielded almost confluent growth on plates obtained from all four libraries, i.e., more than 1000 colonies, each. Three single colonies were picked from the borders of each plate and plasmids were isolated from the twelve overnight cultures grown in rich medium supplemented with 300 mg/L hygromycin B. Clones were readily obtained from all preparations after transformation of E. coli and selection for kanamycin resistance and plasmids from single colonies were isolated and subjected to restriction analysis and sequencing. Three clones yielded the same plasmid, designated as pJJH3152, which carried the GIN4 homologue of S. cerevisiae and approximately 1.7 kb of its 5 -flanking sequence (Figure 2a). This and pJJH3160, one of the remaining plasmids carrying HuSTB5 with its 5 -flanking sequence, were chosen for re-introduction into H. uvarum HHO1, to confirm high transformation frequencies on hygromycin medium. Insertions from pJJH3152, pJJH3160 and six further isolates could be assigned to different contigs of the original genomic sequence [31]. We concluded that these plasmids carried autonomously replicating sequences (ARS), i.e., chromosomal replication origins, and applied the "ori finder 3" program [39], followed by an analysis through "MEME suite" [40] to determine their location in the inserted H. uvarum DNA. As shown in Figure 2b, a consensus sequence could be defined, which consists of 11 bp. To confirm the functionality of the postulated HuARS sequences, fragments of 367 bp and 414 bp were obtained by PCR from pJJH3152 and pJJH3160, respectively, and inserted as XhoI/BglII fragments into the original cloning vector pJJH3058. Both resulting plasmids, pJJH3181 (HuARS1) and pJJH3182 (HuARS2), autonomously replicated in H. uvarum, as judged by their high transformation frequencies on hygromycin plates (see Section 2.4).

Construction of Recyclable Deletion Cassettes and Cloning Vectors
As our strain of H. uvarum is diploid and two alleles of a given gene would thus generally need to be deleted to obtain a selectable phenotype, it was most unfortunate that only the hygromycin resistance cassette was available as a dominant selection marker [37]. Once used, the resultant strain would be resistant to the antibiotic without a possibility to remove the marker for deletion of the second allele. Moreover, an increase in the hygromycin B concentration to select for deletion of the second allele, as reported for the homozygous knockout of the HuATF1 gene [37], did not work in our hands with the deletion of genes for auxotrophic markers. To test its suitability, we therefore inserted the G418 resistance cassette from pUG6 [41], which carries the kanMX gene flanked by the promoter and terminator of TEF2 from Ashbya gossypii, into pJJH3181 to yield pJJH3197. Transformants first selected on hygromycin plates were also able to grow on plates with G418, confirming that this marker cassette, which is commonly used for deletion of genes in other yeasts, also works for H. uvarum. Importantly, it is flanked by loxP sites, thus enabling marker recycling upon expression of the gene encoding the bacteriophage P1 Cre-recombinase [42].
Consequently, we proceeded by constructing plasmids with the HuARS1 sequence and different selection markers for production of the Cre-recombinase in H. uvarum ( Figure 3a). First, pJJH3192 was obtained by simultaneous cloning of two PCR fragments carrying the HuTEF1 promoter and the coding sequence for the Cre-recombinase into pJJH3181, which can be selected for resistance to hygromycin and constitutively expresses the recombinase gene. Next, pJJH3203 was obtained with the same constitutive expression construct, but the kanMX cassette for selection on G418 instead of the hygromycin resistance. Finally, a set of Cre expression plasmids with auxotrophic markers were obtained, comprising pJJH3228 (HuADE2), pJJH3231 (HuURA3), pJJH3238 (HuHIS3), and pJJH3243 (HuLEU2). To employ these constructs with a recyclable deletion cassette for the hygromycin resistance marker, we amplified the TEF2 promoter from A. gossypii from pUG6 and the coding sequence of the hygromycin resistance gene from pJB2 for insertion between two loxP sites to yield pJJH3204 ( Figure 3b). This vector carries the same sequences flanking the deletion cassette as pUG6 and the other plasmids of this series [41], so that the same oligonucleotides can be employed for PCR-based one-step gene replacements with different selection markers.
Antibiotics like G418 and hygromycin are a fairly expensive way of selection in yeast. Based on the autonomously replicating plasmid pJJH3200, we therefore substituted the sequences conferring hygromycin resistance by different auxotrophic markers for H. uvarum to generate a set of cloning vectors. Hence, pJJH3249 (HuURA3), pJJH3252 (HuLEU2), and pJJH3253 (HuHIS3) were constructed, with several unique restriction sites, as depicted in Figure 3c. Their applicability is demonstrated in the following Section 2.4. The cloning vector pJJH3181 was derived from the original screen for HuARS elements and first used to assemble the strong and constitutive HuTEF1 promoter with the Cre coding sequence. The resulting plasmid pJJH3192 was then used to introduce the other selectable markers as indicated. While constructs with the auxotrophic markers HuURA3, HuHIS3, and HuLEU2 also retained the hygromycin resistance cassette, it was removed from the vector conferring G418 resistance (pJJH3203). Note that the HuADE2 marker was introduced in a different backbone of the S. cerevisiae vector YEp352 [43] and can be selected for ampicillin instead of kanamycin resistance in E. coli. (b) A hygromycin resistance cassette comprised of sequences entirely heterologous to the H. uvarum genome. The cassette was assembled from PCR-generated DNA fragments between the loxP sites of an E. coli cloning vector derived from pUG6 [41]. The forward and reverse sequences can be added to the 3′-end of longer oligonucleotides providing homologies to the target gene of interest. They are universally applicable to all templates of the pUG-series [41]. bla-ori indicates the ß-lactamase gene conferring Figure 3. Summary of DNA tools constructed for the genetic manipulation of H. uvarum. (a) Plasmids for expression of the gene encoding the bacteriophage P1 Cre recombinase. The cloning vector pJJH3181 was derived from the original screen for HuARS elements and first used to assemble the strong and constitutive HuTEF1 promoter with the Cre coding sequence. The resulting plasmid pJJH3192 was then used to introduce the other selectable markers as indicated. While constructs with the auxotrophic markers HuURA3, HuHIS3, and HuLEU2 also retained the hygromycin resistance cassette, it was removed from the vector conferring G418 resistance (pJJH3203). Note that the HuADE2 marker was introduced in a different backbone of the S. cerevisiae vector YEp352 [43] and can be selected for ampicillin instead of kanamycin resistance in E. coli. (b) A hygromycin resistance cassette comprised of sequences entirely heterologous to the H. uvarum genome. The cassette was assembled from PCR-generated DNA fragments between the loxP sites of an E. coli cloning vector derived from pUG6 [41]. The forward and reverse sequences can be added to the 3 -end of longer oligonucleotides providing homologies to the target gene of interest. They are universally applicable to all templates of the pUG-series [41]. bla-ori indicates the ß-lactamase gene conferring resistance to ampicillin and the ability to replicate in E. coli. (c) A set of H. uvarum/E. coli shuttle cloning vectors. pJJH3200 was derived from pJJH3181 by restoring the unique HindIII recognition sequence to the multiple cloning site. The hygromycin resistance cassette was then substituted by the different auxotrophic marker genes, as indicated at the left. The table at the right indicates which restriction sites remain unique, i.e., available for cloning, in each vector (+), or those where a second site has been introduced by the marker gene (-).

Construction of Strains with Auxotrophic Selection Markers and Determination of Transformation Efficiencies
In order to show that these constructs are functional in H. uvarum, and to obtain a set of recipient strains with auxotrophic selection markers, we proceeded by first deleting the HuADE2 gene. The coding sequence was substituted for the kanMX cassette by in vivo recombination with an episomal plasmid in S. cerevisiae, as described in Materials and Methods. The resulting vector, pJJH3207, was digested with EcoRI/BamHI to generate the deletion cassette with large flanking sequences and introduced by electroporation into strain HHO1, selecting for G418 resistance (Figure 4). Two clones were confirmed by PCR to carry the heterozygous deletion and HHO4 (Huade2::kanMX/HuADE2) was used as a recipient for a PCR-generated deletion cassette with the hygromycin resistance marker from pJJH3204. Less than 1% of the colonies obtained displayed poor growth and turned red after one week of incubation. Four of these clones were checked by PCR and shown to contain a homozygous deletion of the two HuADE2 alleles, one substituted by the kanMX and the other by the hph genes ( Figure 4). As both dominant markers were flanked by loxP sites, pJJH3228 (HuADE2-HuTEF1p-Cre) was then introduced into strain HHO7 (Huade2::kanMX/Huade2::hph) and selected for growth on medium lacking adenine. Fifty colonies were picked onto a master plate on the same medium, incubated over night at 30 • C and replica-plated onto rich medium with either G418 or hygromycin B. All were sensitive to both antibiotics, showing that the Cre recombinase was expressed and efficiently removed the marker cassettes. Subsequent PCR and sequencing confirmed that both HuADE2 loci carried a single loxP site instead of the open reading frame in strain HHO20 (Huade2::loxP/Huade2::loxP; Figure 4). By similar approaches, heterozygous and homozygous deletions were obtained for HuHIS3, HuURA3, and HuLEU2 (Table 1).
For all further deletions, except Huhis3::kanMX, the PCR generated marker cassettes were directly employed for transformation of H. uvarum, without prior addition of longer flanking homologies by in vivo recombination in S. cerevisiae. Further, in the case of HuURA3 and HuLEU2 the hygromycin resistance cassette generated from pJJH3204 was exclusively employed, with removal from the heterozygous deletions prior to deletion of the second allele with the same PCR product. It should also be noted that the efficiency of the desired homologous recombination was low, generally yielding correct replacements in less than 3% of all clones examined. In contrast, once substituted by the selection markers, homozygous deletions leaving only loxP sites in both alleles of the gene in question were readily obtained by introducing one of the plasmids for Cre recombinase constructed in Section 2.3. More than 95% of clones grown on a master plate selecting for maintenance of the Cre plasmid had lost the deletion markers upon replica-plating onto the respective antibiotics (G418 and/or hygromycin). Subsequently, non-selective growth in liquid rich medium overnight and streaking out for single colonies on plates yielded more than 80% of clones that also had lost the plasmid. Figure 5 shows the growth of the different homozygous deletion mutants on drop-out media, confirming the auxotrophic requirements. It should be noted that Huura3::loxP/Huura3::loxP mutants failed to grow on all synthetic media, even if supplemented with uracil, but did grow when uridine was added, instead. The auxotrophic mutants may thus serve as recipients for further manipulations, i.e., for introduction of heterologous genes on the vectors carrying the respective wild-type genes from H. uvarum (Figure 3c).  [43] was introduced for in vivo recombination into S. cerevisiae DHD5 together with a PCR-generated deletion cassette, to yield pJJH3207. The latter carries long flanking sequences of the HuADE2 locus (approximately 800 bp and 150 bp 5′ and 3′ to the coding sequence, respectively). The deletion fragment excised as an EcoRI/BamHI fragment was introduced into H. uvarum strain HHO1 by electroporation and colonies grown on selective G418 medium were picked onto a master plate. A clone carrying the respective deletion identified by PCR with flanking primers and confirmed by two further PCRs generating upstream and downstream fragments with primers specific for the deletion cassette, was designated as HHO4 and chosen for the next step. A PCR-generated hygromycin resistance cassette was then used directly to delete the second HuADE2 allele by homologous recombination, to generate strain HHO7. Introduction of the Cre-recombinase plasmid pJJH3228 selecting for complementation of the adenine auxotrophy finally yielded HHO20, with both alleles retaining a single loxP site, each.  [43] was introduced for in vivo recombination into S. cerevisiae DHD5 together with a PCR-generated deletion cassette, to yield pJJH3207. The latter carries long flanking sequences of the HuADE2 locus (approximately 800 bp and 150 bp 5 and 3 to the coding sequence, respectively). The deletion fragment excised as an EcoRI/BamHI fragment was introduced into H. uvarum strain HHO1 by electroporation and colonies grown on selective G418 medium were picked onto a master plate. A clone carrying the respective deletion identified by PCR with flanking primers and confirmed by two further PCRs generating upstream and downstream fragments with primers specific for the deletion cassette, was designated as HHO4 and chosen for the next step. A PCR-generated hygromycin resistance cassette was then used directly to delete the second HuADE2 allele by homologous recombination, to generate strain HHO7. Introduction of the Cre-recombinase plasmid pJJH3228 selecting for complementation of the adenine auxotrophy finally yielded HHO20, with both alleles retaining a single loxP site, each.

Figure 5.
Growth of different auxotrophic mutants on synthetic media. Cells were streaked out for single colonies on standard rich medium (YEPD) and allowed to grow over night at 30 °C. They were then replica-plated onto different drop-out media and incubated for the times indicated. Strain designations in the upper right corner correspond to those listed in Table 1. Uridine and uracil were added as indicated at final concentrations of 1.7 mM and 0.2 mM, respectively.

Establishment of a Chemical Transformation Procedure
During these studies, electroporation proved to be very efficient and reliable to introduce deletion cassettes but is fairly laborious and expensive for the introduction of episomal plasmids. We therefore investigated whether (i) competent cells can be stored for electroporation, and (ii) if they can be employed for an alternative transformation protocol. As shown in Table 2, electroporation of fresh competent cells with episomal plasmids yielded transformation frequencies in the order of 10 5 transformants per μg of DNA. This frequency was only reduced by half after storage of the electrocompetent cells for up to 17 days at −80 °C, which is more than sufficient for the introduction of episomal plasmids.   Table 1. Uridine and uracil were added as indicated at final concentrations of 1.7 mM and 0.2 mM, respectively.

Establishment of a Chemical Transformation Procedure
During these studies, electroporation proved to be very efficient and reliable to introduce deletion cassettes but is fairly laborious and expensive for the introduction of episomal plasmids. We therefore investigated whether (i) competent cells can be stored for electroporation, and (ii) if they can be employed for an alternative transformation protocol. As shown in Table 2, electroporation of fresh competent cells with episomal plasmids yielded transformation frequencies in the order of 10 5 transformants per µg of DNA. This frequency was only reduced by half after storage of the electrocompetent cells for up to 17 days at −80 • C, which is more than sufficient for the introduction of episomal plasmids.
Employing these stored electrocompetent cells in a variation of the freeze-transformation method by [44] resulted in similar transformation frequencies in the range of 10 5 transformants per µg of plasmid DNA (Table 2), which were also observed for episomal plasmids in the respective auxotrophic recipient strains (Table 2), permitting long-term storage and fast transformation, if the same recipient strain is to be used for various manipulations.

Discussion
This work was initiated with the aim of providing a useful set of strains, plasmids, and deletion cassettes for the genetic manipulation of Hanseniaspora uvarum. In this context, one of the key achievements was the isolation and tailoring of autonomously replicating sequences (ARS), i.e., chromosomal replication origins. These were readily obtained from genomic libraries by their ability to confer relatively high transformation frequencies, with approximately 10 5 transformants per µg of plasmid DNA. While that is still at least two orders of magnitude lower than those routinely obtained with similar plasmids in Saccharomyces cerevisiae [45], it clearly allows the fast and easy introduction of either homologous or heterologous genes for expression in H. uvarum. The ARS consensus sequence (ACS) defined herein (Figure 2b) differs considerably from that of S. cerevisiae [46], explaining why previous attempts to introduce episomal vectors from this yeast failed (this laboratory, unpublished results). Besides an ACS, functional ARS elements in S. cerevisiae also comprise at least a B1 and a B2 region with even less sequence conservation, together spanning approximately 100 base pairs [46]. As we located a functional HuARS1 sequence on a 367 bp fragment, we decided that this suffices for all practical purposes and did not do further molecular analyses.
Instead, as a proof-of-principle and to facilitate the multiple use of deletion markers, we cloned the bacteriophage P1 Cre-recombinase coding sequence on such an episomal plasmid, expressed from the HuTEF1 promoter. The Cre-recombinase has been widely employed in the construction of deletion mutants in the model yeasts S. cerevisiae and Kluyveromyces lactis to eliminate target genes flanked by direct repeats of short loxP sequences [42,47]. It also worked very efficiently in H. uvarum, as demonstrated by the fact that almost all clones growing on selective medium for the maintenance of the Cre expression plasmid had lost the respective deletion marker cassettes by the time they formed colonies and were transferred to a master plate. We attribute this behaviour to the fact that the HuTEF1 promoter allows for high constitutive expression of the recombinase gene, as it drives the expression of the gene encoding a translational elongation factor required in high quantities at its native locus. Moreover, plasmids carrying ARS sequences are known to reside in high copy numbers and are preferentially retained in the mother cells in S. cerevisiae, in fact constituting a cause of aging when carrying ribosomal DNA genes [48,49]. The mitotic instability of such ARS plasmids is presumably conserved in H. uvarum, as the Cre-recombinase constructs were readily lost within one overnight growth in non-selective medium, further facilitating the use of this system.
We here demonstrated that gene deletions can be obtained by introducing PCRgenerated, loxP-flanked dominant selection markers. Two features were improved on the previously reported use of hygromycin B resistance [37]: (i) Besides the addition of flanking loxP sites, the hph gene encoding the hygromycin phosphatase was placed under the control of the heterologous AgTEF2 promoter from Ashbya gossypii, instead of the HuSUI2 promoter, to minimize the unwanted integration of circularized PCR products at the native locus. (ii) The commonly used kanMX cassette residing between the AgTEF2 promoter and terminator sequences, and flanked by loxP sites [50], was shown to work as a second dominant selection marker in H. uvarum. As the pUG6 vector used as a template to generate this cassette is part of a series also carrying heterologous marker genes for complementation of his3, leu2, and ura3 mutants in S. cerevisiae [41], we expect that these can be exploited in future studies for gene deletions using the auxotrophic mutants generated here, with the advantage that all can be amplified by PCR with one pair of oligonucleotide primers.
As expected for the auxotrophic markers, the deletion of one allele from the diploid genome did not result in a notable phenotype, but homozygous deletions failed to grow on the respective drop-out media. The Huade2 mutants are especially of interest in this context, as they accumulate a red pigment, so that genetic screens for colony colour and sectoring can be used to assess the stability of plasmids and chromosomes equipped with the wildtype HuADE2 gene, similar to S. cerevisiae [51]. For instance, this could be employed to isolate centromeric sequences from the H. uvarum genome, similar to the isolation of HuARS sequences from libraries as described herein.
The homozygous Huura3 deletions also offer a range of future applications, for example, counter-selection of plasmids carrying the HuURA3 gene will be possible on medium with 5-fluoro-acetate (5-FOA), allowing for plasmid-shuffling [52]. In fact, we previously isolated Huura3 mutants from an aneuploid strain induced by nocodazole, based on its resistance to 5-FOA [31]. A puzzling result then was that it not only failed to grow on medium lacking uracil, but generally on synthetic medium. This feature is shared by the homozygous deletion obtained here and could be explained by a similar metabolism to A. gossypii. There, Agura3 mutants were also found to be sensitive to the presence of uracil in synthetic media, leading to inhibition of the uracil phosphoribosyltransferase reaction in the pyrimidine salvage pathway [53]. As observed in that work, growth could be restored in Huura3 mutants by substituting uracil for uridine.
The heterozygous deletion mutants listed in Table 1 may also be of value in future studies. One may want to stably integrate heterologous genes into the H. uvarum genome, e.g., to increase the sulphite resistance for improved use in must fermentations. Using an internal fragment of the coding sequence, e.g., of the HuURA3 gene, on a non-replicating plasmid, an integration into the remaining allele in the heterozygous mutant (Huura3::loxP/HuURA3) would result in an auxotrophy for uracil selectable on 5-FOA. In addition, the low efficiency of homologous recombination could be improved by diminishing the activity of the nonhomologous end joining repair, e.g., deleting the two copies of the HuKU80 gene, as proven valuable in K. lactis [54]. Alternatively, taking advantage of both the high expression levels and the mitotic instability of the HuARS vectors described here, ScRAD51, a component of the homologous repair system of S. cerevisiae, could be transiently over-expressed from the HuTEF1 promoter in co-transformations with deletion cassettes, as exercised in lager yeast [55].
As examplified by these few remarks, the toolset provided in this work allows a wealth of future applications for H. uvarum, which can be employed not only to improve its performance in the production of alcoholic beverages like wine and cider, but also in its use as a biocontrol agent in fighting fungal infections, insect pests, or enhance the growth of agricultural plants. Clearly, a deeper understanding of its basic metabolic capacities lies at the heart of such applications, which may also be broadened with the molecular genetic techniques now at hand. Finally, given the close phylogenetic relationship with other Hanseniaspora/Kloeckera varieties [36], the tools developed herein may probably also be of use for the genetic manipulation of other members of the genera.

Strains, Media, and Growth Conditions
All strains of Hanseniaspora uvarum applied and constructed in this work are based on HHO1, which is derived from strain DSM 2768, also designated as CBS 2584 or NRRL 115. It was obtained from the German stock culture collection and selected for singlecelled growth, as described previously [31]. Saccharomyces cerevisiae strains HD56-5A and its diploid derivative DHD5 have been described [56,57]. Yeast cell culture and genetic techniques followed standard procedures devised for S. cerevisiae [58]. Rich medium (YEPD) contained 1% yeast extract, 2% Bacto peptone (Difco Laboratories Inc., Detroit, MI, USA), and 2% glucose. Synthetic media were prepared as described in [58], with the omission of amino acids or bases as required for selection of plasmids or deletion markers and 2% glucose (SCD). If required, uracil was substituted for 1.7 mM uridine. For selection of the kanMX marker, 50 mg/L of G418 were added to the rich medium after sterilization, for selection of hygromycin resistance, 300 mg/L of hygromycin B were employed (both antibiotics were purchased from Carl Roth GmbH, Karlsruhe, Germany).
For manipulations in E. coli, strain DH5α was employed with standard media as described in [59].

Transformation Procedures
For electroporation, the method described by [37] was adopted and modified as follows: To prepare five aliquots of competent cells, a pre-culture of H. uvarum was inoculated in 3 mL of YEPD and incubated with shaking over night at 30 • C. The culture was diluted 1:30 in fresh YEPD and 5 µL were used to inoculate 50 mL of YEPD in a 300 mL Erlenmeyer flask. After shaking for 16 h at 180 rpm at 30 • C the culture reached an OD 600 of 3.5-4.0, determined with a Beckman spectrophotometer DU800 (Beckman Coulter, Krefeld, Germany). If higher densities were observed, cells were diluted again to an OD 600 of 1.0 and incubated for approximately 2 h until they reached the required final OD 600 . Cells were collected in a sterile Falcon tube by centrifugation for 5 min at 1700× g at room temperature and resuspended in 25 mL of sterile water. After another centrifugation step (5 min/1700× g) the pellet was resuspended in 25 mL of lithium acetate buffer (10 mM Tris, pH 8.0, 1 mM EDTA, 100 mM lithium acetate, 10 mM dithiothreitol), with the components freshly mixed from 10fold concentrated stock solutions. After incubation for 1 h at room temperature, cells were collected by centrifugation, washed with 25 mL of ice-cold sterile water, and resuspended in 20 mL of cold sterile 1 M sorbitol. Cells were again pelleted, resuspended in 500 µL of cold sterile 1 M sorbitol and aliquoted in 100 µL samples into sterile electroporation cuvettes (Gene Pulser Cuvette 0.2 cm electrode gap from BioRad/Feldkirchen/Germany). Alternatively, 100 µL aliquots of the sorbitol suspension were stored in sterile Eppendorf tubes at −80 • C and directly used for electroporation after thawing on ice. For electroporation, 10 µL of plasmid DNA, or 20 µL of PCR product or digested DNA purified with Roche High-Pure PCR purification kit (Roche/Mannheim/Germany), were added to the cuvettes at room temperature and mixed well (note that products from restriction digests or PCR reactions cannot be applied directly for electroporation without damaging the cuvettes). After an incubation of 5-10 min at room temperature, electroporation was performed with an Electro Cell Porator 600 (BTX Electroporation Systems, Fisher Scientific GmbH, Schwerte, Germany). A program designed for electroporation of E. coli cells was applied, with 1.55 kV for 5.18 msec at a resistance of 2.5 kV. Cuvettes were then kept on ice and 500 µL of ice-cold sterile 1M sorbitol were added. The cell suspension was transferred into 1.5 mL of fresh YEPD medium and either incubated with shaking at 30 • C for 3-4 h for regeneration when dominant markers (G418 or hygromycin B) were employed, or 100 µL were directly plated onto selective media in the case of auxotrophic markers (HuADE2,  HuURA3, HuHIS3, HuLEU2), from appropriate dilutions, if required.
For introduction of episomal plasmids, a chemical transformation procedure was developed. It followed the preparation of competent cells described above for electroporation, except that 200 µL aliquots of the sorbitol suspension prior to electroporation were stored at −80 • C until use. For DNA addition and further treatment, the freeze-transformation method [44] was modified. Thus, 5 µL of herring-sperm DNA (10 µg/µL from Roche, Mannheim, Germany) were added as carrier DNA to the frozen cells, together with 10-20 µL of plasmid solution, corresponding to 0.1-0.5 µg of DNA. Cells were thawed and heat-shocked with occasional shaking at 32 • C, and 500 µL of solution B (40% polyethylene glycol 1000, 100 mM bicine pH 8.35) were added. After an incubation for 1 h at 28 • C, cells were collected by centrifugation (2 min at 13,000× g), the supernatant discarded, and the pellet resuspended in 500 µL of YEPD. One hundred microliters from appropriate dilutions (usually 1:50 to 1:200) were then either plated directly onto selective media for auxotrophic markers or regenerated for 4 h with shaking at 30 • C in 2 mL of YEPD in case of selection for antibiotic resistance.

Construction of Plasmids and Deletion Mutants
The basic vectors employed were derived from pUK15JJH, a variant of pUK1921 [60], with an extended multiple cloning site, and pJB2, a plasmid for selection of hygromycin resistant deletions reported by [37], (note that the resistance gene is under the control of the HuSUI2 promoter instead of the HuTEF1 promoter, as originally reported). One-step gene replacements [61] were either obtained by transformation of marker cassettes with long flanking sequences homologous to the target loci and excised from plasmids with appropriate restriction enzymes, or directly by the introduction of PCR products with flanking homologies of 50 ± 7 bais pairs. For cloning in S. cerevisiae, standard vectors were used [43,62]. The kanMX deletion cassette was derived from pUG6 [41], which was also used as a source for the AgTEF2 promoter. Restriction enzymes were obtained from Thermo Fisher Scientific, Schwerte, Germany, and used according to the manufacturer's instructions. Plasmids were generated either by standard restriction-ligation reactions from previously cloned DNA or PCR products, or by in vivo recombination in S. cerevisiae. Details on the manufacture of specific constructs, maps, and sequences of all plasmids and genomic loci with the deletion constructs are available upon request. Oligonucleotides used as primers for cloning, deletions and PCR verifications can be found in Supplementary  Table S1. All PCR-generated fragments further used in cloning and expression studies were verified by Sanger sequencing (Seqlab, Göttingen, Germany).

DNA Preparations and PCR
Genomic DNA from H. uvarum was prepared by a rapid, phenol-free extraction procedure as follows: 500 µL of a late-logarithmic overnight culture in YEPD were collected by centrifugation in a 1.5 mL Eppendorf tube for 1 min at 10,000× g and resuspended in 500 µL of spheroblast buffer (0.9 M sorbitol, 0.1 M EDTA, pH 8.0). Next, 2 µL of a Zymolyase stock solution (10 mg/mL of Zymolyase 20T stored at −20 • C) and 10 µL of a 100 mM stock solution of DTT were added and the cells were incubated for 15 min at 37 • C. Spheroblasts were collected by centrifugation for 1 min at 10,000× g and the pellet was suspended in 300 µL TE/SDS (12.5 mM Tris pH 7.5, 5 mM EDTA, 0.25% SDS). The suspension was incubated for 10 min at 65 • C to inactivate DNases, transferred to ice and 80 µL of a 5 M potassium acetate solution were added. After incubation for 5 min, the tubes were centrifuged for 10 min at 10,000× g and the 350 µL of the supernatant was removed into fresh Eppendorf tubes. Then, 800 µL of pure ethanol were added for precipitation of the genomic DNA, mixed well, and subjected to centrifugation (10 min at 10,000× g). The pellet was dried and dissolved in 100 µL of an RNase solution (solution 1 from the High Pure Plasmid Purification kit from Roche, Mannheim, Germany, supplemented with 150 mM sodium acetate, pH 7.0, to avoid DNA nicking activity). Tubes were incubated for 10 min at 37 • C with occasional vigorous shaking. The DNA was again precipitated by addition of 5 µL from a 3 M stock solution of sodium acetate and 250 µL of pure ethanol, followed by 5 min of centrifugation at 10,000× g. The pellet was again dried and dissolved in 100 µL of the column elution buffer from the plasmid preparation kit. Finally, 1 µL was then applied to PCR reactions or 10 µL for digestions with restriction enzymes. Plasmids were isolated from H. uvarum by disruption with glass beads, purification through the Roche High-Pure PCR purification kit, followed by transformation into E. coli DH5α for amplification, as described for S. cerevisiae in a previous work [63].
The PCR reactions were routinely performed in a volume of 50 µL in 0.2 µL Eppendorf tubes in a thermocycler (Biometra GmbH, Göttingen, Germany). For analytical purposes, i.e., confirmation of deletion constructs, the Dream Taq preparation (Thermo Fisher Scientific, Schwerte, Germany) was used according to the instructions of the manufacturer. For cloning purposes, the High Fidelity Taq Polymerase from Roche (Mannheim, Germany) was employed. Custom-made oligonucleotides were used as primers for sequencing and the PCR were purchased from Biolegio BV, Nijmegen, The Netherlands. For sequences of H. uvarum, the annotated genome sequence was consulted ( [31]; deposited in GenBank at https://www.ncbi.nlm.nih.gov/GenBank/under accession number APLS01000000; accessed on 30 December 2022).

Imaging
Brightfield micrographs were obtained from cultures growing logarithmically in synthetic complete medium with a Zeiss Axioplan 2 microscope (Carl Zeiss, Jena, Germany) with differential interference contrast (DIC) as detailed in [64].
For scanning electron microscopy, the H. uvarum strain HHO1 was grown in 3 mL liquid rich medium (YEPD) overnight at 30 • C with shaking, and the late logarithmic culture was used to inoculate 3 mL of synthetic complete medium at an OD 600 of approximately 0.3 and grown for another four hours into the logarithmic phase. Cells were harvested and treated for imaging exactly as described for Kluyveromyces lactis in [65].