Novel genetic tools for Hansenula polymorpha

Authors

  • Ruchi Saraya,

    1. Molecular Cell Biology, Groningen Biomolecular Sciences and Biotechnology Institute, Kluyver Centre for Genomics of Industrial Fermentation, University of Groningen, Groningen, The Netherlands
    Search for more papers by this author
  • Arjen M. Krikken,

    1. Molecular Cell Biology, Groningen Biomolecular Sciences and Biotechnology Institute, Kluyver Centre for Genomics of Industrial Fermentation, University of Groningen, Groningen, The Netherlands
    Search for more papers by this author
  • Jan A.K.W. Kiel,

    1. Molecular Cell Biology, Groningen Biomolecular Sciences and Biotechnology Institute, Kluyver Centre for Genomics of Industrial Fermentation, University of Groningen, Groningen, The Netherlands
    Search for more papers by this author
  • Richard J.S. Baerends,

    1. Molecular Cell Biology, Groningen Biomolecular Sciences and Biotechnology Institute, Kluyver Centre for Genomics of Industrial Fermentation, University of Groningen, Groningen, The Netherlands
    Search for more papers by this author
  • Marten Veenhuis,

    1. Molecular Cell Biology, Groningen Biomolecular Sciences and Biotechnology Institute, Kluyver Centre for Genomics of Industrial Fermentation, University of Groningen, Groningen, The Netherlands
    Search for more papers by this author
  • Ida J. van der Klei

    Corresponding author
    • Molecular Cell Biology, Groningen Biomolecular Sciences and Biotechnology Institute, Kluyver Centre for Genomics of Industrial Fermentation, University of Groningen, Groningen, The Netherlands
    Search for more papers by this author

Correspondence: Ida J. van der Klei, Molecular Cell Biology, Groningen Biomolecular Sciences and Biotechnology Institute, Kluyver Centre for Genomics of Industrial Fermentation, University of Groningen, PO Box 11103, 9700 CC Groningen, the Netherlands. Tel.: +31 50 363 2179; fax: +31 50 363 2348; e-mail: i.j.van.der.klei@rug.nl

Abstract

Hansenula polymorpha is an important yeast in industrial biotechnology. In addition, it is extensively used in fundamental research devoted to unravel the principles of peroxisome biology and nitrate assimilation. Here we present an overview of key components of the genetic toolbox for H. polymorpha. In addition, we present new selection markers that we recently implemented in H. polymorpha. We describe novel strategies for the efficient creation of targeted gene deletions and integrations in H. polymorpha. For this, we generated a yku80 mutant, deficient in non-homologous end joining, resulting in strongly enhanced efficiency of gene targeting relative to the parental strain. Finally, we show the implementation of Gateway technology and a single-step PCR strategy to create deletions in H. polymorpha.

Introduction

Hansenula polymorpha (also designated as Ogataea angusta or Pichia angusta) is a methylotrophic yeast that can grow on methanol as a sole carbon and energy source. Growth on this unusual substrate involves the oxidation of methanol into formaldehyde and hydrogen peroxide by the peroxisomal enzyme alcohol oxidase (AO). Peroxisomes in addition harbour catalase, which decomposes hydrogen peroxide, as well as dihydroxyacetone synthase (DHAS), an enzyme of the xylulose-5-phosphate cycle, essential for formaldehyde assimilation. The other enzymes of the xylulose-5-phosphate pathway as well as the enzymes involved in formaldehyde dissimilation are cytosolic (van der Klei et al., 2006).

As is evident from the above, peroxisomes play an important role in methanol metabolism. Peroxisomes are cell organelles that consist of a proteinaceous matrix surrounded by a single membrane. During growth of H. polymorpha cells on glucose, the peroxisomal enzymes involved in methanol metabolism are fully repressed and only a few small peroxisomes are present per cell. However, upon a shift of the cells to methanol medium, peroxisomes are massively induced (van der Klei et al., 2006). Conversely, upon a shift of methanol-grown cells to glucose, the peroxisomes are selectively degraded by autophagy (Sakai et al., 2006). These properties make H. polymorpha a very attractive model organism to study the molecular mechanisms involved in peroxisome biogenesis and degradation (van der Klei & Veenhuis, 2006).

In addition, H. polymorpha can assimilate nitrate as sole nitrogen source. In fact, nitrate assimilation gene regulation in yeast has been studied for the first time using H. polymorpha (for a review see Siverio, 2002). Furthermore, the molecular mechanisms involved in the regulation of nitrate transport in H. polymorpha have been studied (see Martín et al., 2011; and the references therein). In addition to fundamental molecular cell biology research, H. polymorpha is also extensively used in industrial biotechnology, predominantly for the production of various pharmaceuticals. One reason is that several of the genes encoding key enzymes of methanol metabolism (e.g. AO, DHAS and formaldehyde dehydrogenase) are controlled by very strong, inducible promoters, which makes this organism very attractive for heterologous protein production. Another important property for industrial fermentations is the thermotolerant nature of H. polymorpha, which reduces the need for expensive cooling. In fact, some strains have been described that are able to grow at temperatures above 50 °C. Hansenula polymorpha is well suited to the production of secreted heterologous proteins, as it does not hyper-glycosylate proteins (as observed in Saccharomyces cerevisiae). Moreover, this yeast secretes only minor amounts of endogenous proteins, which facilitates downstream processing of secreted products. For more detailed information on the use of H. polymorpha in biotechnology the reader is referred to excellent reviews on this topic (Gellissen, 2002; Stockmann et al., 2009).

For both fundamental research and biotechnological applications, genetic engineering is crucial. Over the last 20 years several genetic tools have been developed for H. polymorpha. This includes, among others, efficient transformation procedures (electrotransformation and the lithium-acetate method), the construction of (over)expression cassettes as well as vectors for targeted integration enabling gene deletion (Ito et al., 1983; Sudbery et al., 1988; Faber et al., 1992, 1994; Raschke et al., 1996; Gellissen, 2002; Haan et al., 2002; Gellissen et al., 2005; Kang & Gellissen, 2005; Böer et al., 2007), including the use of the Cre-loxP system for marker rescue (Krappmann et al., 2000; Qian et al., 2009).

Moreover, the genomes of two widely used H. polymorpha strains (CBS4732 and NCYC495) have been sequenced (Ramezani-Rad et al., 2003; http://genome.jgi-psf.org/Hanpo2). This sequence information has strongly facilitated molecular research and has enabled systems biology approaches such as transcriptomics analysis (van Zutphen et al., 2010). The genome sequence of H. polymorpha DL-1 is also known, but this strain is only approximately 95% identical to CBS4732 and NCYC495, implying that this may represent a different species (see GenBank AEOI00000000.1).

Here we present an overview of the current H. polymorpha genetic toolbox, which includes several novel markers, expression plasmids as well as a yku80 strain, which shows strongly enhanced efficiency for targeted integration. Also, we present the implementation of the Gateway technology as well as a single-step PCR strategy to create gene deletions.

Expression vectors

Several expression vectors are currently available for H. polymorpha (Table 1). Plasmids pHIPX4 and pHIPM4 are based on the lower-copy vector pOK12 (Vieira & Messing, 1991) and contain the kanamycin-resistance gene for selection in Escherichia coli. All other plasmids are based on the high-copy pBLUESCRIPT (Stratagene) backbone and carry the ampicillin-resistance marker for selection in E. coli (Table 1).

Table 1. Hansenula polymorpha expression vectors based on the strong, methanol-inducible H. polymorpha AOX promoter and AMO terminator regions
Vector nameMarker in H. polymorphaMarker typeMarker Escherichia coliE. coli originReference
  1. Hp, Hansenula polymorpha; Kp, Klebsiella pneumoniae; Sc, Saccharomyces cerevisiae; Sh, Streptoalloteichus hindustanus; Sn, Streptomyces noursei; Sv, Streptomyces viridochromogenes; Tn, Escherichia coli transposon Tn903.

  2. a

    In these plasmids the marker used in H. polymorpha is also preceded by the synthetic Em7 promoter, which should allow gene expression in E. coli as well. So far, for pHIPZ4 and pHIPN4 the zeocin and nourseothricin resistance markers have been demonstrated to be functional also in E. coli.

  3. b

    Bialaphos is a peptide that can be used only in combination with synthetic mineral medium and not with rich yeast extract peptone dextrose (YPD) plates.

pHIPX4Sc-LEU2Leucine-auxotrophicKanamycinpOK12Gietl et al. (1994)
pHIPM4Hp-MET6Methionine-auxotrophicKanamycinpOK12Gidijala et al. (2007)
pHIPA4Hp-ADE11Adenine-auxotrophicAmpicillincolE1Haan et al. (2002)
pHIPZ4Sh-bleZeocinRAmpicillinacolE1Salomons et al. (2000)
pHIPN4Sn-nat1NourseothricinRAmpicillinacolE1Saraya et al. (2011)
pHIPH4Kp-hphHygromycin BRAmpicillinacolE1Saraya et al. (2011)
pHIPK4Tn-KanMXG-418/geneticinRAmpicillinacolE1Unpublished
pHIPB4bSv-patBialaphosRAmpicillinacolE1Unpublished

Vectors containing the strong, regulatable promoter of the alcohol oxidase gene (PAOX) and the terminator of the amine oxidase gene (AMO) are most commonly used (Fig. 1 and Table 1). This promoter is repressed by glucose, derepressed during growth on glycerol and strongly induced when cells are grown in media containing methanol as sole carbon and energy source. Replacement of PAOX by alternative promoter fragments has resulted in a series of vectors, which enable alternative expression profiles. For instance, pHIPZ15 carries an alternative, strong and methanol-inducible promoter from the H. polymorpha DHAS gene (Kiel et al., 2005). Like PAOX, this promoter is repressed by glucose. Other examples are the vectors pHIPX5, pHIPZ5 (Kiel et al., 1995; Faber et al., 2001) and pHIPN5, which all contain the H. polymorpha AMO promoter (PAMO). This promoter is somewhat weaker than PAOX or PDHAS. Promoter PAMO is induced by culturing H. polymorpha cells on media containing primary amines as sole nitrogen sources (methylamine, ethylamine) and fully repressed by ammonium sulphate. pHIPX7/pHIPZ7 both contain the constitutive H. polymorpha TEF1 promoter (Baerends et al., 1997; Kurbatova et al., 2009).

Figure 1.

Schematic representation of integrative expression vectors for use in Hansenula polymorpha. For strong, methanol-inducible expression of (heterologous) genes these vectors contain the H. polymorpha AOX promoter and AMO terminator regions. The available auxotrophic and dominant markers for H. polymorpha are indicated in Table 1.

Because no stable extrachromosomally replicating plasmids are available for H. polymorpha, the expression vectors are routinely integrated into the host genome. For vectors containing PAOX, PAMO or PTEF1 this is accomplished by linearizing the expression plasmid in the promoter region to ensure directed integration at the AOX, AMO or TEF1 locus, respectively.

For H. polymorpha vectors with a large range of selectable markers are available. These include both auxotrophic as well as dominant resistance markers. One of the first expression vectors that has been constructed in our laboratory and which is still commonly used is pHIPX4 (Gietl et al., 1994). This vector contains the heterologous S. cerevisiae LEU2 gene under control of its endogenous promoter as a selection marker in H. polymorpha. A single copy of this gene, however, is not sufficient to fully functionally complement the leucine auxotrophy of the H. polymorpha leu1.1 strain. Consequently, fully complemented transformants generally contain multiple copies of the expression plasmid (up to about seven; Baerends et al., 1997), whereas transformants with a single copy integration form small colonies. This property greatly facilitates the selection of single or multiple copy integrants. Unexpectedly, increased copy numbers do not always lead to enhanced expression. In isolated cases, protein production from a single-copy integrant has been shown to be superior to that of transformants that carry more copies of the expression cassette (Gellissen, 2002).

A major advantage of H. polymorpha is that despite the presence of (large) direct repeats in the genome resulting from the targeted integration of multiple expression cassettes, these strains are stable and can be maintained on nonselective media without problem. Apparently, loss of the integrated plasmid by replication slippage during genome replication is not very efficient in H. polymorpha.

When using expression plasmids with a homologous auxotrophy marker (e.g. H. polymorpha MET6 or ADE11), generally a single copy of the homologous gene suffices to fully complement the amino acid auxotrophy. Where transformants with multiple copy integration of the expression cassette are required with these vectors, such transformants have to be selected by using laborious methods such as colony PCR or Southern blotting (Haan et al., 2002).

As indicated in Table 1, currently five dominant markers are available for H. polymorpha. The genes encoding these marker enzymes are constitutively expressed under control of the S. cerevisiae TEF1 promoter. When applying standard antibiotic concentrations, generally single copy integrants are obtained. However, an increase in antibiotic concentration may result in selection of cells with enhanced copies of the expression cassette. This has been demonstrated for the ble marker of pHIPZ4 (Salomons et al., 2000). Thus, it is possible to select for transformants with different copy numbers of these expression cassettes to address specific needs.

The availability of multiple expression vectors (Table 1) has allowed the introduction of multiple different genes in one host strain. For instance, we recently introduced the entire penicillin (PEN) biosynthesis pathway (five genes) from the filamentous fungus Penicillium chrysogenum into H. polymorpha using (derivatives of) the vectors indicated in Fig. 1 and Table 1 (Gidijala et al., 2009). We demonstrated that all PEN biosynthetic enzymes were satisfactory produced and allowed production of the antibiotic penicillin G, which was efficiently secreted into the culture medium. As different promoters and markers are available, the levels of the various proteins can be adapted by selecting the proper promoters and copy numbers.

The use of the Gateway™ recombination system in H. polymorpha

During the last decade new technologies for genetic engineering have been developed that rely less on standard cloning procedures. One of these is the Gateway™ system designed by Invitrogen, which involves recombination cloning and specific genetic modules (promoters, terminators, tags, markers, etc.). We recently designed and implemented novel modular Gateway-compatible vectors for genetic engineering of H. polymorpha. These include modules and vectors for gene expression (promoters, terminator), for gene deletion and for creating hybrid genes encoding fusion proteins with green fluorescent protein (GFP) (for an overview see Fig. 2 and Table 2).

Figure 2.

Recombination cloning using Gateway technology in Hansenula polymorpha. (a) Schematic representation of the multi-site Gateway cloning system. Three modules in Entry vectors are recombined at the indicated attachment (att) sites with a Destination vector resulting in a new plasmid. Depending on the modules chosen, the system can be utilized for (heterologous) gene expression (when required with a tag like a GFP moiety) (b), in which case the Destination vector needs an additional marker for selection on integration into the H. polymorpha genome. Additionally, recombination cloning allows us to create cassettes for gene deletions in the H. polymorpha genome (c). The currently available standard modules for use in H. polymorpha are listed in Table 2.

Table 2. Modules in Gateway vectors for use in gene expression and gene deletion in Hansenula polymorpha
NameModuleReference
  1. Hp, Hansenula polymorpha; Kp, Klebsiella pneumoniae; Sc, Saccharomyces cerevisiae; Sg, Streptomyces griseochromogenes; Sh, Streptoalloteichus hindustanus; Sn, Streptomyces noursei; Sv, Streptomyces viridochromogenes; Tn, Escherichia coli transposon Tn903.

(a) (Heterologous) gene expression in H. polymorpha
Promoter (+tag)
pDONR P4-p1R+PAOXH. polymorpha AOX promoterSaraya et al. (2011)
pDONR P4-p1R+PAMOH. polymorpha AMO promoterNagotu et al. (2008b)
pDONR P4-p1R+PTEF1H. polymorpha TEF1 promoterUnpublished
pDONR P4-p1R+PPEX14H. polymorpha PEX14 promoterUnpublished
pDONR P4-p1R+PPEX11H. polymorpha PEX11 promoterCepinska et al. (2011)
pDONR P4-p1R+PAOX-eGFPH. polymorpha AOX promoter and the eGFP gene lacking a stopcodonUnpublished
pDONR P4-p1R+PAMO-eGFPH. polymorpha AMO promoter and the eGFP gene lacking a stopcodonNagotu et al. (2008a)
pDONR P4-p1R+PTEF1-eGFPH. polymorpha TEF1 promoter and the eGFP gene lacking a stopcodonUnpublished
pDONR P4-p1R+PPEX14-eGFPH. polymorpha PEX14 promoter and the eGFP gene lacking a stopcodonUnpublished
Terminator (+tag)
pDONR P2R-P3+TAMOH. polymorpha AMO terminatorNagotu et al. (2008a)
pDONR P2R-P3+eGFP-TAMOThe eGFP gene and the H. polymorpha AMO terminatorNagotu et al. (2008b)
pDONR P2R-P3+mGFP-TAMOThe mGFP gene and the H. polymorpha AMO terminatorCepinska et al. (2011)
pDONR P2R-P3+mCherry-TAMOThe mCherry gene and the H. polymorpha AMO terminatorSaraya et al. (2011)
Destination vectors
pDEST R4-R3/natGateway Destination vector with the Streptomyces noursei nat1 gene in the vectorSaraya et al. (2011)
pDEST R4-R3/zeo (=pRSA07)Gateway Destination vector with the Streptoalloteichus hindustanus ble gene in the vectorSaraya et al. (2011)
(b) Gene deletion in H. polymorpha
Markers
pENTR221-URA3Hp-URA3 gene; uracil auxotrophyNagotu et al. (2008b)
pENTR221-CaLEU2Ca-LEU2 gene; leucine auxotrophyNagotu et al. (2008b)
pENTR221-bsdSg-bsd gene; blasticidinRThis study
pENTR221-hphKp-hph gene; hygromycin BRSaraya et al. (2011)
pENTR221-kanMXTn-KanMX gene; G-418/geneticinRUnpublished
pENTR221-nat1Sn-nat1 gene; nourseothricinRSaraya et al. (2011)
pENTR221-bleSh-ble gene; zeocinRUnpublished
pENTR221-patSv-pat; bialaphosRUnpublished

For (heterologous) gene expression in H. polymorpha five different promoter modules are now available, either for high-level inducible expression (PAOX, PAMO), relatively high constitutive expression (PTEF1) or relatively low constitutive expression (PPEX14 or PPEX11). GFP fusion proteins are extensively used for protein localization by fluorescence microscopy in molecular cell biology research.

So far, only two derivatives of the standard pDEST R4-R3 destination vector (Invitrogen) carrying dominant markers (nat1 and ble) have been constructed (Table 2). However, as other dominant markers utilized in S. cerevisiae also function in H. polymorpha, destination vectors constructed for the former in other laboratories are expected to be applicable as well.

In the past, gene deletion in wild-type H. polymorpha was cumbersome and time-consuming because checking for correct deletion required screening many colonies (because of random integration in the H. polymorpha genome). In our laboratory, routinely gene deletion would be accompanied by selection on uracil-prototrophic transformants of a ura3 host strain. This was advantageous, as uracil-auxotrophic H. polymorpha strains have the tendency to grow with a much enhanced doubling time on glucose medium, probably because of limited uracil uptake from the culture medium. Recently, we constructed eight Gateway modules with different auxotrophic and dominant markers that allowed us to quickly construct deletion cassettes for modification of the H. polymorpha genome (Table 2). Indeed, strains harbouring multiple deletions have been constructed without problem (Saraya et al., 2011).

The frequency of targeted gene deletion is improved in H. polymorpha yku80

A major bottleneck for rapid and efficient construction of H. polymorpha deletion strains is the high frequency of random integration in the host genome. Consequently, identifying the correct deletion strain often requires PCR screening of large numbers of transformants. In various yeast and fungal species the level of random integration was highly reduced upon deletion of KU70 or KU80 genes (Snoek et al., 2009). Yku80 is the yeast homologue of human KU80, which together with KU70 forms the Ku heterodimer (KU70/KU80) that is required for double strand break (DSB) repair by non-homologous end joining (NHEJ; for a review see Pastwa & Blasiak, 2003). In eukaryotes, DSBs can also be repaired by homologous recombination (HR). Integration of a (foreign) DNA fragment into the yeast genome involves DSB repair. By HR, targeted integration is obtained, whereas NHEJ results in random integration of the fragment. Hence, deletion of a gene required for NHEJ, such as KU70 or KU80, should strongly diminish random integration.

The orthologue of YKU80 in H. polymorpha was identified using YKU80 of S. cerevisiae as query in a BlastP search against all protein sequences of H. polymorpha. The best hit was found with Hp39g181 (24% identities, 45% positives, E-value: 5e-020). A reciprocal blast at NCBI using the Hp39g181 protein as input, in turn yielded hits with KU80/Yku80 proteins from many species. The Hp39g181 gene was replaced by the URA3 gene, allowing selection for uracil prototrophs.

Growth experiments revealed that the resulting H. polymorpha yku80 strain has identical growth characteristics (doubling times, growth yield) as the wild-type control on mineral media containing different carbon (glucose, ethanol, glycerol and methanol) or nitrogen sources (ammonium sulphate and methylamine). Also, we did not obtain any indications that the genomic stability of the strain has decreased relative to the parental strain.

The efficiency of obtaining gene deletions in H. polymorpha wild-type and yku80 strains was determined by transformation with AOX deletion cassettes, containing the up- and downstream flanking regions of the AOX gene on either side of one of the four dominant markers. As summarized in Table 3, the frequency of gene deletion was highly enhanced compared with the wild-type control. Using zeocin as selection marker, a success percentage of over 50% was achieved, while using any of the other markers the percentages were above 90%, where values of 0.5–40% were obtained for the wild-type control. Hence, as in other species, also in H. polymorpha deletion of the YKU80 gene is sufficient to strongly decrease the level of NHEJ dramatically, resulting in a significant improvement in isolating gene deletions.

Table 3. Efficiency of gene deletion in Hansenula polymorpha wild-type and yku80 deletion strain using dominant antibiotic markers
MarkerDeletion efficiency (%)
Wild-typeHp39g181 (yku80)
BlasticidinR293
Hygromycin BR4093
NourseothricinR0.598
ZeocinR356

Gene deletion in H. polymorpha yku80 by PCR

We recently tested a single-step PCR strategy for creating gene deletions in H. polymorpha, as is widely used in S. cerevisiae (Manivasakam et al., 1995). PCR fragments with flanking regions of different lengths (for making a deletion of the AOX gene) were transformed to H. polymorpha wild-type and yku80 strains. Transformants were tested for their ability to grow on methanol, indicative of AOX deletion (Tables 4 and 5). Indeed, in all transformants that were deficient in methanol utilization the deletion cassette was correctly integrated as confirmed by colony PCR. The data show that the single-step PCR strategy functions properly to create deletions in H. polymorpha. Flanking regions of 31/32 bp are sufficient although the efficiency is relatively low (12% in the wild-type). In the yku80 strain, however, efficiency is much higher using these fragments (60%). The efficiency increases also with the size of the flanking regions to 31% (wild-type) and 88% (yku80) using flanking regions of 245/247 bp.

Table 4. Deletion of the Hansenula polymorpha alcohol oxidase (AOX) gene using DNA fragments containing the hph gene flanked by homologous flanking regions of different sizes in H. polymorpha yku80
Fragment nameFlanking regions size (bp)AOX deletion efficiency (%)
AOX-231/3260
AOX-341/4167
AOX-452/5478
AOX-5114/11183
AOX-6148/14281
AOX-7181/18277
AOX-8245/24788
Table 5. Deletion of the Hansenula polymorpha alcohol oxidase (AOX) gene using DNA fragments containing the hph gene flanked by homologous flanking regions of different sizes in wild-type H. polymorpha
FragmentFlanking regions size (bp)AOX deletion efficiency (%)
AOX-231/3212
AOX-341/4111
AOX-452/5413
AOX-5114/11117
AOX-6148/14320
AOX-7181/18226
AOX-8245/24731

Concluding remarks

Here we discuss novel resistance markers suitable for use as selection markers for genetic modification in H. polymorpha. In addition to the auxotrophic markers previously available (leu, ura, ade, met), these additional markers allow multiple gene manipulations of a single strain.

We also showed that using a yku80 mutant greatly increased the ease of isolating gene deletions and is further improved by using a single-step PCR strategy. With these achievements we added important new tools to the H. polymorpha genetic toolbox that allow integration of complete complex metabolic pathways in the organism (as before for the complete PEN biosynthesis pathway; Gidijala et al., 2009). Also, multiple deletions are now possible, allowing us to remove (parts of) specific pathways, thus rendering the organism highly suitable for synthetic biology approaches.

Acknowledgements

We thank Christiaan P. Postema, Kim Susanna and Jean-Paul Twagirumuhuza for skilful assistance in various parts of the research. The construction of some of the plasmids described in this manuscript was outsourced to Sylphium Life Sciences, Groningen, the Netherlands. Ruchi Saraya is supported by the Netherlands Organisation for Scientific Research/Chemical Sciences (NWO/CW). This project was carried out within the research programme of the Kluyver Centre for Genomics of Industrial Fermentation, which is part of the Netherlands Genomics Initiative/Netherlands Organization for Scientific Research.

Ancillary