Engineering of molybdenum-cofactor-dependent nitrate assimilation in Yarrowia lipolytica.

Engineering a new metabolic function in a microbial host can be limited by the availability of the relevant cofactor. For instance, in Yarrowia lipolytica, the expression of a functional nitrate reductase is precluded by the absence of molybdenum cofactor (Moco) biosynthesis. In this study, we demonstrated that the Ogataea parapolymorpha Moco biosynthesis pathway combined with the expression of a high affinity molybdate transporter could lead to the synthesis of Moco in Y. lipolytica. The functionality of Moco was demonstrated by expression of an active Moco-dependent nitrate assimilation pathway from the same yeast donor, O. parapolymorpha. In addition to 11 heterologous genes, fast growth on nitrate required adaptive laboratory evolution which, resulted in up to 100-fold increase in nitrate reductase activity and in up to 4-fold increase in growth rate, reaching 0.13h-1. Genome sequencing of evolved isolates revealed the presence of a limited number of non-synonymous mutations or small insertions/deletions in annotated coding sequences. This study that builds up on a previous work establishing Moco synthesis in S. cerevisiae demonstrated that the Moco pathway could be successfully transferred in very distant yeasts and, potentially, to any other genera, which would enable the expression of new enzyme families and expand the nutrient range used by industrial yeasts.

INTRODUCTION

As Saccharomyces cerevisiae, the Dipodascaceae yeast Yarrowia lipolytica has gained interest as a model yeast for dimorphism studies and as an industrial work horse (Barth and Gaillardin 1997). This strictly aerobic oleaginous Saccharomycetales yeast has been traditionally exploited for its ability to efficiently degrade a wide variety of abundant and cheap hydrophobic substrates such as n-alkanes, fatty acids and oils that was coupled to its remarkable high enzyme secretion capacity and production of organic acids such as citric and α-ketoglutaric acids (Tsugawa et al. 1969).

The fast development of dedicated molecular tools including the addition of CRISPR technology, enabled to propel Y. lipolytica as a potential contender of S. cerevisiae for the biosynthesis of commodity and speciality chemicals. Y. lipolytica has become a reliable platform for metabolic engineering as illustrated by the synthesis of flavor compounds (Marella et al. 2020), insect sex pheromones (Holkenbrink et al. 2020), non-caloric sweeteners (Rumelhard et al. 2016), itaconic acid (Blazeck et al. 2015) and terpenoids (Arnesen et al. 2020) as a few examples.

Engineering of Y. lipolytica with increasingly more complex pathways will require the expression of an even broader range of enzymes as its attractiveness as metabolic engineering platform grows. Many enzyme activities require the presence of one or more essential cofactors (Broderick 2001; Champe, Harvey and Ferrier 2005). Therefore, the successful expansion of the enzyme repertoire in a microbial host may require the parallel broadening of its cofactor set. Whenever a cofactor requirement cannot be met by media supplementation because either (1) the cofactor is not commercially available or too expensive, (2) is unstable or (3) cannot be imported by the organism, metabolic engineering is required to enable its de novo biosynthesis or its transport. This approach was successful in the model yeast Saccharomyces cerevisiae as exemplified with the implementation of high affinity Ni2+transport, an inorganic cofactor of Ni-dependent urease (Milne et al. 2015), or with the engineering of tetrahydrobiopterin pathway, that was instrumental in the implementation of de novo biosynthesis of opioids (Galanie et al. 2015; Li and Smolke 2016) and melatonin (Germann et al. 2016). More recently, the molybdenum cofactor biosynthesis pathway from the methylotrophic yeast Ogataea parapolymorpha was introduced in S. cerevisiae allowing expression of a functional Moco-dependent nitrate reductase that could support growth on media containing nitrate as sole nitrogen source (Perli et al. 2021). Although these pioneering studies demonstrated how metabolic landscape could be expanded beyond the natural ability of a microorganism, these approaches have not yet been transposed in other industrially relevant yeast species. Enabling Moco biosynthesis and nitrate assimilation in a evolutionarily distant yeast such as Y. lipolytica would confirm that the selected gene-set may be used in wide range of yeast species but would also contribute to further expand the metabolic capabilities of this microbial cell factory. Previous research has shown how Y. lipolitica accumulates lipids in nitrogen starving conditions (Ratledge 2002). Introduction of a heterologous nitrogen assimilation pathway such as the nitrate assimilation pathways could be applied to mimic a nitrogen-starving condition without actually having to limit the amount of nitrogen in the media.

As S. cerevisiae and more than 75% of the Saccharomycotina yeast species, Y. lipolytica is not able to synthesize molybdopterin cofactor (Shen et al. 2018) which precludes the possibility to harness enzyme families of biotechnological relevance.

Moco-dependent enzymes encompass more than 30 different catalytic activities divided in three main families, based on the Moco variant they require (Cvetkovic et al. 2010; Hille, Hall and Basu 2014). Moco can be directly used as cofactor or modified by either sulfuration of the molybdate ion to form sulfurated Moco or, in prokaryotes, by the covalent attachment of either GDP or CDP to molydopterin (MTP) to form Bis(Molybdopterin Guanine Dinucleotide) Molybdenum Cofactor or MTP-cytosine dinucleotide cofactor, respectively. In a large majority, Moco-dependent enzymes catalyse oxido/reduction reactions, often involving oxygen and are implicated in nutrient e.g. carbon, nitrogen and sulfur cycles or in detoxification of growth-inhibiting compounds thanks to the redox versatility of the Mo atom (Hille 2002).

Moco is composed of a molybdate (MoO42–) oxyanion coordinated by two sulfur atoms on MTP, a tricyclic pterin scaffold. Moco cannot be supplemented in the media since the molecule is too unstable due to its oxygen sensitivity (Fischer et al. 2006). For this reason, Moco is, in the majority of cases, de novo synthesized intracellularly. The Moco biosynthetic pathway is very well conserved and it has been extensively studied in both prokaryotes and eukaryotes (Rajagopalan and Johnson 1992; Iobbi-Nivol and Leimkuhler 2013; Mendel 2013). The first step, which takes place in the mitochondria of eukaryotic cells, is catalysed by the heterodimer Cnx1/Cnx2, that cyclizes GTP onto pyranopterin monophosphate (cPMP). The molecule is then exported to the cytosol through a yet uncharacterized transporter, and it is converted to MPT by action of the MPT synthase complex (Cnx52/Cnx62) which donates two sulfur atoms present on a conserved Cysteine residue in the Cnx5 protein. Cnx5 is then reloaded with sulfur via a sulfur mobilization route that includes the adenyltransferase Cnx4 and a cysteine desulfurase that is typically involved in iron–sulfur cluster biosynthesis and tRNA thiolation, Nfs1 (Leimkuhler, Buhning and Beilschmidt 2017). In the final step, molybdate is inserted in a two-steps reaction catalyzed by the multi-domain protein Cnx3 (Iobbi-Nivol and Leimkuhler 2013). Previous studies showed that the yeast S. cerevisiae lacks a high-affinity molybdate transport system and that it is able to import the oxyanion only with low affinity, unless a high affinity transporter such as CrMot1 from Chlamydomonas reinhardtii is expressed (Tejada-Jimenez et al. 2007; Perli et al. 2021).

The goals of this study were to investigate whether the heterologous Moco biosynthetic pathway from the nitrate-assimilating yeast O. parapolymorpha could be functionally engineered in Y. lipolytica, together with a Moco-dependent nitrate assimilation pathway. To this end, a total of 11 genes encompassing Moco biosynthesis, molybdate transport and nitrate reduction functions were introduced in the oleaginous yeast Y. lipolytica using CRISPR/Cas9 gene-editing technology. The engineered strain was then subjected to adaptive laboratory evolution (ALE) in synthetic media with nitrate as the sole nitrogen source to evolve a fast-growing population. After that, single cell lines were isolated, phenotyped on nitrate containing medium and characterized by whole-genome resequencing.

MATERIAL AND METHODS

Strains, media and maintenance

All strains used and constructed in this study are shown in Table 1. All Y. lipolytica strains were derived from the strain ST6512 (W29, MATa ku70∆::Spycas9-EcDsdAMX4; Marella et al. 2020). Yeast strains were grown on either YP (10 g/L Bacto yeast extract and 20 g/L Bacto peptone) or SM medium (Verduyn et al. 1992) with either 5 g/L KNO3, or 2.3 g/L urea (SMNO3 and SMurea, respectively) as sole nitrogen source. In all SM media variants, 6.6 g/L K2SO4 was added as a source of sulfate (Solis-Escalante et al. 2013). YP or SM media were autoclaved at 121°C for 20 min. After sterilization, SM was supplemented with 1 mL/L of filter-sterilized vitamin solution as previously described (Verduyn et al. 1992). A concentrated glucose solution was autoclaved at 110°C for 20 min and then added to the YP and SM medium at a final concentration of 20 g/L, yielding SMD and YPD, respectively. 500-mL shake flasks containing 100 mL medium and 100-mL shake flasks containing 20 mL medium were incubated at 30°C and 200 rpm in an Innova Incubator (Brunswick Scientific, Edison, NJ). Solid media were prepared by adding 1.5% (w/v) Bacto agar and, where indicated, 250 mg/L nourseothricin or 250 mg/L hygromycin B. Escherichia coli strains were grown in LB (10 g/L Bacto tryptone, 5 g/L Bacto yeast extract and 5 g/L NaCl) supplemented with 100 mg/L ampicillin. Yarrowia lipolytica and E. coli cultures were stored at −80°C after the addition of 30% v/v glycerol.

Table 1.

Strains used in this study.

Name	Relevant genotype	Parental strain	Reference
ST6512	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2	W29, Y-63746 and ATTC-20460	Marella et al. (2020)
IMX2264	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2	ST6522	This study
IMX2265	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10	IMX2264	This study
IMX2266	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10 pGPD-OpCNX3_tLIP2 E_1Δ::pTEFin-OpCNX3-tPEX10 pGPD-CrMoT1_tLIP2	IMX2265	This study
IMX2267	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10 pGPD-OpCNX3_tLIP2 E_1Δ::pTEFin-OpCNX3-tPEX10 pGPD-CrMoT1_tLIP2 C_3Δ::pTEFin-OpYNR1-tPEX10 pGPD-OpYNT1_tLIP2 pTEFin-OpYNI1-tPEX10	IMX2566	This study
IMX2565	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10 pGPD-OpCNX3_tLIP2 E_1Δ::pTEFin-OpCNX3-tPEX10 pGPD-CrMoT1_tLIP2 C_3Δ::pTEFin-OpYNR1-tPEX10 pGPD-OpYNT1_tLIP2 pTEFin-OpYNI1-tPEX10 E_3Δ::pTEFin-OpCNX6-tPEX10 pGPD-OpCNX5-tLIP2	IMX2267	This study
IMS1174	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10 pGPD-OpCNX3_tLIP2 E_1Δ::pTEFin-OpCNX3-tPEX10 pGPD-CrMoT1_tLIP2 C_3Δ::pTEFin-OpYNR1-tPEX10 pGPD-OpYNT1_tLIP2 pTEFin-OpYNI1-tPEX10 E_3Δ::pTEFin-OpCNX6-tPEX10 pGPD-OpCNX5-tLIP2 (Evolved on SMDNO3 for 21 transfers. Line 2–Colony 1)	IMX2565	This study
IMS1175	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10 pGPD-OpCNX3_tLIP2 E_1Δ::pTEFin-OpCNX3-tPEX10 pGPD-CrMoT1_tLIP2 C_3Δ::pTEFin-OpYNR1-tPEX10 pGPD-OpYNT1_tLIP2 pTEFin-OpYNI1-tPEX10 E_3Δ::pTEFin-OpCNX6-tPEX10 pGPD-OpCNX5-tLIP2 (Evolved on SMDNO3 for 21 transfers. Line 2–Colony 2)	IMX2565	This study
IMS1176	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10 pGPD-OpCNX3_tLIP2 E_1Δ::pTEFin-OpCNX3-tPEX10 pGPD-CrMoT1_tLIP2 C_3Δ::pTEFin-OpYNR1-tPEX10 pGPD-OpYNT1_tLIP2 pTEFin-OpYNI1-tPEX10 E_3Δ::pTEFin-OpCNX6-tPEX10 pGPD-OpCNX5-tLIP2 (Evolved on SMDNO3 for 21 transfers. Line 2–Colony 3)	IMX2565	This study
IMS1177	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10 pGPD-OpCNX3_tLIP2 E_1Δ::pTEFin-OpCNX3-tPEX10 pGPD-CrMoT1_tLIP2 C_3Δ::pTEFin-OpYNR1-tPEX10 pGPD-OpYNT1_tLIP2 pTEFin-OpYNI1-tPEX10 E_3Δ::pTEFin-OpCNX6-tPEX10 pGPD-OpCNX5-tLIP2 (Evolved on SMDNO3 for 21 transfers. Line 3–Colony 1)	IMX2565	This study
IMS1178	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10 pGPD-OpCNX3_tLIP2 E_1Δ::pTEFin-OpCNX3-tPEX10 pGPD-CrMoT1_tLIP2 C_3Δ::pTEFin-OpYNR1-tPEX10 pGPD-OpYNT1_tLIP2 pTEFin-OpYNI1-tPEX10 E_3Δ::pTEFin-OpCNX6-tPEX10 pGPD-OpCNX5-tLIP2 (Evolved on SMDNO3 for 21 transfers. Line 3–Colony 2)	IMX2565	This study
IMS1179	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10 pGPD-OpCNX3_tLIP2 E_1Δ::pTEFin-OpCNX3-tPEX10 pGPD-CrMoT1_tLIP2 C_3Δ::pTEFin-OpYNR1-tPEX10 pGPD-OpYNT1_tLIP2 pTEFin-OpYNI1-tPEX10 E_3Δ::pTEFin-OpCNX6-tPEX10 pGPD-OpCNX5-tLIP2 (Evolved on SMDNO3 for 21 transfers. Line 3–Colony 3)	IMX2565	This study
IMS1180	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10 pGPD-OpCNX3_tLIP2 E_1Δ::pTEFin-OpCNX3-tPEX10 pGPD-CrMoT1_tLIP2 C_3Δ::pTEFin-OpYNR1-tPEX10 pGPD-OpYNT1_tLIP2 pTEFin-OpYNI1-tPEX10 E_3Δ::pTEFin-OpCNX6-tPEX10 pGPD-OpCNX5-tLIP2 (Evolved on SMDNO3 for 21 transfers. Line 1–Colony 1)	IMX2565	This study
IMS1181	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10 pGPD-OpCNX3_tLIP2 E_1Δ::pTEFin-OpCNX3-tPEX10 pGPD-CrMoT1_tLIP2 C_3Δ::pTEFin-OpYNR1-tPEX10 pGPD-OpYNT1_tLIP2 pTEFin-OpYNI1-tPEX10 E_3Δ::pTEFin-OpCNX6-tPEX10 pGPD-OpCNX5-tLIP2(Evolved on SMDNO3 for 21 transfers. Line 1–Colony 2)	IMX2565	This study
IMS1182	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10 pGPD-OpCNX3_tLIP2 E_1Δ::pTEFin-OpCNX3-tPEX10 pGPD-CrMoT1_tLIP2 C_3Δ::pTEFin-OpYNR1-tPEX10 pGPD-OpYNT1_tLIP2 pTEFin-OpYNI1-tPEX10 E_3Δ::pTEFin-OpCNX6-tPEX10 pGPD-OpCNX5-tLIP2 (Evolved on SMDNO3 for 21 transfers. Line 1–Colony 3)	IMX2565	This study
IMS1183	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10 pGPD-OpCNX3_tLIP2 E_1Δ::pTEFin-OpCNX3-tPEX10 pGPD-CrMoT1_tLIP2 C_3Δ::pTEFin-OpYNR1-tPEX10 pGPD-OpYNT1_tLIP2 pTEFin-OpYNI1-tPEX10 E_3Δ::pTEFin-OpCNX6-tPEX10 pGPD-OpCNX5-tLIP2 (Evolved on SMDNO3 for 50 transfers. Evolved population line 2)	IMX2565	This study
IMS1184	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10 pGPD-OpCNX3_tLIP2 E_1Δ::pTEFin-OpCNX3-tPEX10 pGPD-CrMoT1_tLIP2 C_3Δ::pTEFin-OpYNR1-tPEX10 pGPD-OpYNT1_tLIP2 pTEFin-OpYNI1-tPEX10 E_3Δ::pTEFin-OpCNX6-tPEX10 pGPD-OpCNX5-tLIP (Evolved on SMDNO3 for 50 transfers. Evolved population line 3)	IMX2565	This study
IMS1185	MATa ku70Δ::pTEF1-Spcas9-tTEF12::pGPD-EcdsdAMX4-tLIP2 E_4Δ::pTEFin-OpCNX1-tPEX10 pGPD-OpCNX2-tLIP2 C_2Δ::pTEFin-OpNFS1-tPEX10 pGPD-OpCNX3_tLIP2 E_1Δ::pTEFin-OpCNX3-tPEX10 pGPD-CrMoT1_tLIP2 C_3Δ::pTEFin-OpYNR1-tPEX10 pGPD-OpYNT1_tLIP2 pTEFin-OpYNI1-tPEX10 E_3Δ::pTEFin-OpCNX6-tPEX10 pGPD-OpCNX5-tLIP2 (Evolved on SMDNO3 for 50 transfers. Evolved population line 1)	IMX2565	This study

Molecular biology techniques

Primers used in this study are shown in Table 2. DNA was amplified using either Phusion Hot Start II High Fidelity Polymerase (Thermo Fisher Scientific, Waltham, MA) or Phusion U (Thermo Fisher Scientific) and desalted or PAGE-purified oligonucleotide primers (Sigma-Aldrich, St-Louis, MO) according to manufacturers’ instructions. Diagnostic PCR reactions were performed with DreamTaq polymerase (Thermo Fisher Scientific). PCR products were separated by gel electrophoresis on a 1% (w/v) agarose gel (Thermo Scientific) in TAE buffer (40 mM Tris, 20 mM acetic acid and 1 mM EDTA; Thermo Scientific) and purified with a Zymoclean Gel DNA Recovery Kit (Zymo Research, Irvine, CA). Plasmids were isolated from E. coli using a NucleoSpin Plasmid kit (Macherey-Nagel, Düren, Germany), and verified by either restriction digestion or diagnostic PCR. Escherichia coli DH5-α (New England BioLabs, Ipswich, MA) was used for cloning procedures (Inoue, Nojima and Okayama 1990). Yeast genomic DNA used for diagnostic PCR reactions was isolated by using the SDS/LiAc protocol (Looke, Kristjuhan and Kristjuhan 2011). Yarrowialipolytica transformation was performed with the LiAc method as previously described (Chen, Beckerich and Gaillardin 1997). A total of four to eight colonies were re-streaked on selective medium to select for single clones and diagnostic PCRs were performed to verify the correct genotypes.

Table 2.

Primers used in this study.

Primer number	Primer sequence	Product(s)
22956	AGTACTGCAAAAAGUGCTGGTCGG	PrTEFin-PrGPD_USER_Biobrick
24013	ATCAGTAGCUAGAGACCGGGTTGGCGGCG	PrTEFin-PrGPD_USER_Biobrick
15529	AGCTACTGAUGACGCAGTAGGATGTCCTGCACGG	PrTEFin-PrGPD_USER_Biobrick
15528	ATGACAGAUTGTTGATGTGTGTTTAATTCAAGAATG	PrTEFin-PrGPD_USER_Biobrick
27208	ACACGCGAUAGAGACCGGGTTGGCGG	PrTEFin_USER_Biobrick
22956	AGTACTGCAAAAAGUGCTGGTCGG	PrTEFin_USER_Biobrick
24479	ACTTTTTGCAGTACUAACCGCAGCCCGTGCGACACCTG	OpCNX1_USER_Biobrick
24480	CGTGCGAUTTAGCCGCCGATCAGG	OpCNX1_USER_Biobrick
24481	ATCTGTCAUGCCACAATGGTGGCCATCCACG	OpCNX2_USER_Biobrick
24482	CACGCGAUTTACTTGAAGATGGTAGACAGGTCG	OpCNX2_USER_Biobrick
24483	ACTTTTTGCAGTACUAACCGCAGTACCGATTCCGAATTGGAGC	OpNFS1_USER_Biobrick
24484	CGTGCGAUTTAGTGTCCGGCCCACTC	OpNFS1_USER_Biobrick
24485	ATCTGTCAUGCCACAATGTCTCTGTCTCTGAACGAGTAC	OpCNX4_USER_Biobrick
24486	CACGCGAUTTAGTAGATGGGGAAGTTAGGGTC	OpCNX4_USER_Biobrick
24487	ACTTTTTGCAGTACUAACCGCAGTCTATCTTCGTGGACATCACC	OpCNX6_USER_Biobrick
24488	CGTGCGAUTTAGGTTCGAGACAGCACG	OpCNX6_USER_Biobrick
24489	ATCTGTCAUGCCACAATGGTGGCCGTGGC	OpCNX5_USER_Biobrick
24490	CACGCGAUTTAGCCAGAGGACACAGGAG	OpCNX5_USER_Biobrick
24491	ACTTTTTGCAGTACUAACCGCAGGCCCTGCAGAACGCC	CrMoT1_USER_Biobrick
24492	CGTGCGAUTTAGGCTCGGCCGC	CrMoT1_USER_Biobrick
24493	ATCTGTCAUGCCACAATGACCGTGGGCATCC	OpCNX3_USER_Biobrick
24494	CACGCGAUTTACACGTAGATCTGGTCGATC	OpCNX3_USER_Biobrick
24495	ACTTTTTGCAGTACUAACCGCAGGACTCTGTGGTGACCGAGGT	OpYNR1_USER_Biobrick
24496	CGTGCGAUTTAGAAGTAGACCACGTACTGCTTGTC	OpYNR1_USER_Biobrick
24497	ATCTGTCAUGCCACAATGCGACTGTCTACCCTGTG	OpYNT1_USER_Biobrick
24498	CACGCGAUTTAGATCTCGGCCTTTCGG	OpYNT1_USER_Biobrick
24499	ACTTTTTGCAGTACUAACCGCAGACCTGCTCTGTGCCTC	OpYNI1_USER_Biobrick
24500	CGTGCGAUTTACCAGTCGAAAGAGATGGC	OpYNI1_USER_Biobrick
24517	TAGATAAATTTACACTCCCTCAGATGCATTCTTGGGCGGT	pCfB9006-7_backbone_Gibson_fragment
24518	TCATGGGCCTTCCTTTCACTCAGATGCATTCTTGGGCGGT	pCfB9006-7_2-genes_insert_Gibson_fragment
24520	ACCGCCCAAGAATGCATCTGAGGGAGTGTAAATTTATCTATACAGAGGTAA	pI774_GFPmut3b_spacer_Gibson_fragment
24521	ACCGCCCAAGAATGCATCTGAGTGAAAGGAAGGCCCATGA	pI774_GFPmut3b_spacer_Gibson_fragment
24522	TTCATTCATGTTAGTTGCGTTCTGCGTCTGCTGTTTGTGTC	pCfB9006_backbone_Gibson_fragment
24523	ACACAAACAGCAGACGCAGAACGCAACTAACATGAATGAATACGATATACA	pCfB9006_2-genes_insert_Gibson_fragment
24524	TTCATTCATGTTAGTTGCGTGCCATAGCACTATTGTAGAGTGGCC	pCfB9007_backbone_Gibson_fragment
24525	CTCTACAATAGTGCTATGGCACGCAACTAACATGAATGAATACGATATACA	pCfB9007_2-genes_insert_Gibson_fragment
17887	TCACTTCCCCATCCACACTTTTAGGTTCGAGACAGCACGT	pUDI264_insert_Gibson_fragment
17888	AGGTTGATTCCGAACAGAAGTTAGCCAGAGGACACAGGAG	pUDI264_insert_Gibson_fragment
17889	CTCCTGTGTCCTCTGGCTAACTTCTGTTCGGAATCAACCTC	pUDI264_backbone_Gibson_fragment
17890	ACGTGCTGTCTCGAACCTAAAAGTGTGGATGGGGAAGTGA	pUDI264_backbone_Gibson_fragment

Plasmid construction

Plasmids used in this study are shown in Table 3. Gene sequences coding for proteins involved in Moco biosynthesis (OpCNX1, OpCNX2, OpCNX3, OpCNX4, OpCNX5, OpCNX6 and OpNFS1) nitrate assimilation (OpYNT1, OpYNR1 andOpYNI1) were retrieved from O. parapolymorpha DL-1 genome sequence (Ravin et al. 2013; Perli et al. 2021) BioProject PRJNA60503). A gene coding for a previously characterized high-affinity molybdenum transporter, CrMoT1, from C. reinhardtii was also included in the gene-set (Tejada-Jimenez et al. 2007). Each gene was codon-optimized for expression in Y. lipolytica using the GeneOptimizer tool (Thermo Fisher Scientific) and ordered as synthetic DNA from GeneArt (Thermo Fisher Scientific) resulting in plasmids pUD1057 (OpCNX6), pUD1058 (OpCNX4), pUD1059 (OpCNX5), pUD1060 (OpNFS1), pUD1061 (OpCNX1), pUD1062 (OpCNX2), pUD1063 (CrMoT1) pUD1064 (OpCNX3), pUD1065 (OpYNR1), pUD1066 (OpYNI1) and pUD1067 (OpYNT1). Single-gene Biobricks compatible with USER cloning (New England BioLabs) were amplified from pUD1057, pUD1058, pUD1059, pUD1060, pUD1061, pUD1062, pUD1063, pUD1064, pUD1065, pUD1066 and pUD1067 using primer pairs 24487/24488, 24485/24486, 24489/24490, 24483/24484, 24479/24480, 24481/24482, 24491/24492, 24493/24494, 24495/24496, 24499/24500 and 24497/24498, respectively. Single promoter Biobricks pTEFin and pGPD were amplified from ST6512 genomic DNA using primer pairs 22956/24013 and 15529/15528, respectively. Then, a Biobrick carrying the back-to-back promoter pair pTEFin-pGPD was cloned by USER cloning (Bitinaite et al. 2007) fusion of the two single-promoter Biobricks and primer pair 22956/15528. A single promoter pTEFin Biobrick was amplified using ST6512 genomic DNA and primer pair 27208/22956.

Table 3.

Plasmids used in this study.

Name	Characteristics	Reference
pCfB6371	bla ColE1 NotIC_3–3′homology tPEX20-tLIP2 C_3–5′homologyNotI	Holkenbrink et al. (2018)
pCfB6677	bla ColE1 NotIE_1–3′homology tPEX20-tLIP2 E_1–5′homologyNotI	Holkenbrink et al. (2018)
pCfB6679	bla ColE1 NotIE_4–3′homology tPEX20-tLIP2 E_4–5′homologyNotI	Holkenbrink et al. (2018)
pCfB6681	bla ColE1 NotIE_3–3′homology tPEX20-tLIP2 E_3–5′homologyNotI	Holkenbrink et al. (2018)
pCfB6682	bla ColE1 NotIC_2–3′homology tPEX20-tLIP2 C_2–5′homologyNotI	Holkenbrink et al. (2018)
pCfB6684	bla ColE1 NotID_1–3’homology tPEX20-tLIP2 D_1–5′homologyNotI	Holkenbrink et al. (2018)
pCfB6627	bla ColE1 NAT gRNA_C_2	Holkenbrink et al. (2018)
pCfB6630	bla ColE1 NAT gRNA_C_3	Holkenbrink et al. (2018)
pCfB6631	bla ColE1 NAT gRNA_D_1	Holkenbrink et al. (2018)
pCfB6633	bla ColE1 NAT gRNA_E_1	Holkenbrink et al. (2018)
pCfB6637	bla ColE1 NAT gRNA_E_3	Holkenbrink et al. (2018)
pCfB6638	bla ColE1 NAT gRNA_E_4	Holkenbrink et al. (2018)
pI774	bla ColE1 Gfpmut3	Unpublished
pUD1057	bla ColE1 OpCNX6*	GeneArt
pUD1058	bla ColE1 OpCNX4*	GeneArt
pUD1059	bla ColE1 OpCNX5*	GeneArt
pUD1060	bla ColE1 OpNFS1*	GeneArt
pUD1061	bla ColE1 OpCNX1*	GeneArt
pUD1062	bla ColE1 OpCNX2*	GeneArt
pUD1063	bla ColE1 CrMoT1*	GeneArt
pUD1064	bla ColE1 OpCNX3*	GeneArt
pUD1065	bla ColE1 OpYNR1*	GeneArt
pUD1066	bla ColE1 OpYNI1*	GeneArt
pUD1067	bla ColE1 OpYNT1*	GeneArt
pCfB8966	bla ColE1 NotIE_4–3′homology tPEX20-OpCNX1*-pTEF1in-pGPD-OpCNX2*-tLIP2 E_4–5′homologyNotI	This study
pCfB8967	bla ColE1 NotIC_2–3’homology tPEX20-OpNFS1*-pTEF1in-pGPD-OpCNX4*-tLIP2 C_2–5′homologyNotI	This study
pCfB8968	bla ColE1 NotIE_1–3′homology tPEX20-OpCNX6*-pTEF1in-pGPD-OpCNX5*-tLIP2 E_1–5′homologyNotI	This study
pCfB8969	bla ColE1 NotIE_3–3′homology tPEX20-CrMoT1*-pTEF1in-pGPD-OpCNX3*-tLIP2 E_3–5′homologyNotI	This study
pCfB8970	bla ColE1 NotIC_3–3′homology tPEX20-OpYNR1*-pTEF1in-pGPD-OpYNT1*-tLIP2 C_3–5′homologyNotI	This study
pCfB8971	bla ColE1 NotID_1–3′homology tPEX20-OpYNI1*-pTEF1in- tLIP2 D_1–5′homologyNotI	This study
pCfB9006	bla ColE1 NotIE_1–3′homology tPEX20-OpCNX6*-pTEF1in-pGPD-OpCNX5*-tLIP2 Gfpmut3 tPEX20-CrMoT1*-pTEF1in-pGPD-OpCNX3*-tLIP2 E_1–5′homologyNotI	This study
pCfB9007	bla ColE1 NotIC_3–3′homology tPEX20-OpYNR1*-pTEF1in-pGPD-OpYNT1*-tLIP2 Gfpmut3 tPEX20-OpYNI1*-pTEF1in- tLIP2 C_3–5′homologyNotI	This study
pUDI264	bla ColE1 NotIE_3–3′homology tPEX20-OpCNX6*-pTEF1in-pGPD-OpCNX5*-tLIP2 E_3–5′homologyNotI	This study

Codon optimized for expression in Y. lipolytica.

Backbone Biobricks for the integration at the E_4, C_2, E_1, E_3, C_3 and D_1 integration sites were prepared by digestion and nicking of plasmids pCfB6679, pCfB6682, pCfB6677, pCfB6681, pCfB6371 and pCfB6684, respectively, using endonuclease FastDigest AsiSI (Life Technologies, Carlsbad, CA) and Nb.BsmI (New England BioLabs) followed by gel purification as previously described (Holkenbrink et al. 2018). A backbone E_4 Biobrick was combined with Biobricks pTEFin-pGPD, OpCNX1 and OpCNX2 in a USER cloning reactions as previously described (Holkenbrink et al. 2018) to yield plasmid CfB8966. Similarly, plasmids pCfB8967, pCfB8968, pCfB8969, and pCfB8970 were cloned by combining a pTEFin-pGPD promoter Biobrick with a C_2, E_1, E_3 or C_3 backbone, respectively, and gene Biobricks OpNFS1/OPCNX4, OpCNX6/OpCNX5, CrMOT1/OpCNX3 and OpYNR1/OpYNT1, respectively. Plasmid pCfB8971 was cloned by combining a D_1 backbone Biobrick with a pTEFin promoter and OpYNI1 gene Biobricks. Gibson assembly (Gibson et al. 2009) was used to construct the integration plasmid pCfB9006 that carries overexpression cassettes for OpCNX3, CrMoT1, OpCNX5 and OpCNX6. First, the plasmid pCfB8968 carrying the OpCNX5-OpCNX6 cassette was linearized with primers 24522/24517 and the OpCNX3-CrMoT1 cassette was amplified using primers 24518/24523 and pCfB8969 as template. Secondly, the GFPmut3 spacer cassette (Andersen et al. 1998) was amplified from plasmid pI774 with primer pairs 24520/24521. Each fragment was gel-purified and combined in equimolar amounts in a Gibson reaction following manufacturer's instructions to form pCfB9006. In a similar way, the integration plasmid pCfB9007, carrying nitrate assimilation pathway was cloned by combining a OpYNT1-OpYNR1 cassette with an OpYNI1 cassette and a GFPmut3 spacer. Plasmid pCfB8970 was linearized with primers 24524/24517 while the OpYNI1 cassette was amplified with primers 24525/24518 and pCfB8971 as template. Plasmid pUDI264 for the integration of OpCNX5-OpCNX6 at the E_3 integration site was cloned by Gibson assembly by combining a backbone fragment amplified with primers 17889/17890 and pCfB6681 as template and a fragment carrying the OpCNX5-OpCNX6 cassette amplified with primers 17887/17888 and pCfB8968 as template. Correct plasmid assembly was verified by Sanger sequencing.

Strain construction

CRISPR/Cas9-mediated marker-free gene integration in Y. lipolytica was performed following the EASYcloneYali method as previously described (Holkenbrink et al. 2018). In brief, 1 µg of NotI (Thermo Fisher Scientific) digested integrative plasmid carrying two or more genes flanked by long-homology arms to the integration locus was co-transformed together with 500 ng of a plasmid for the expression of a gRNA targeting the locus of interest. After selection and correct genotyping of transformants via diagnostic PCR, one colony of correct clone was inoculated in a 50-mL Greiner tube containing 20 ml YPD and incubated overnight at 30°C, 200 rpm to allow the gRNA plasmid loss. The next day, cells were streaked to single colonies on a YPD plate and then, after incubation at 30°C overnight, single colonies were patched on both selective and non-selective media. One plasmid-free colony was then picked and grown overnight in a YPD flask prior stocking at −80°C.

First, strain ST6512 was transformed with NotI-digested pCfB8966 and the E_4 targeting gRNA expression plasmid pCfB6638 to yield IMX2264. Then, IMX2264 was transformed with NotI-digested pCfB8967 and the C_2 targeting gRNA expression plasmid pCfB6682 to yield IMX2265. IMX2266 was obtained by transforming IMX2265 with NotI-digested pCfB9006 and the E_1 targeting gRNA expression plasmid pCfB6633. IMX2264 was transformed with NotI-digested pCfB9007 and the C_3 targeting gRNA expression plasmid pCfB6630 to yield IMX2267. Whole genome re-sequencing of IMX2267 revealed that OpCNX5 and OpCNX5 were not integrated but instead CrMoT1 and OpCNX3 were present in two copies (PRJNA704845). To correct this absence, IMX2267 was subsequently transformed with NotI-digested pUDI264 together with the E_3 targeting gRNA expression plasmid pCfB6637, to yield the final strain IMX2565. The presence of all integrated genes in IMX2565 was confirmed by whole genome re-sequencing (PRJNA704845).

ALE

To evolve IMX2565 for fast growth in nitrate-containing media, the strain was inoculated in triplicate in 100-mL flasks containing 20 ml SMDNO3. Flasks were incubated at 30°C, 200 rpm until OD660 reached a value above 5. Then, 0.2 mL of each culture was transferred in a new shake flask containing the same medium and incubated again. This process was repeated for 50 times, corresponding to approximately 335 generations, after which the evolved population were stocked and named IMS1183 (line 1), IMS1184 (line 2) and IMS1185 (line 3). Glycerol stocks for each evolution line were also prepared at intermediate steps after 3, 6, 9, 12, 21, 27, 38 and 50 transfers. Before transferring the cultures from batch 21, single colonies were isolated by restreaking each culture on SMDNO3 agar plates. After incubation of the plates at 30°C in a static incubator for 48 h, three single colonies for each evolution line were picked and inoculated in 100-mL shake flasks containing 20 mL SMDNO3. Flasks were incubated for 48 h at 30°C, 200 rpm and the grown biomass was then stocked at −80°C by the addition of 30% v/v glycerol and named IMS1174 (line 2, colony 1), IMS1175 (line 2, colony2), IMS1176 (line 2, colony 3), IMS1177 (line 3, colony 1), IMS1178 (line 3, colony 2), IMS1179 (line 3, colony 3), IMS1180 (line 1, colony 1), IMS1181 (line 1, colony 2) and IMS1182 (line 1, colony 3).

High-throughput strain cultivation and growth rate estimation

The growth rate of strains IMS1174, IMS1175, IMS1176, IMS1177, IMS1178, IMS1179, IMS1180, IMS1181 and IMS1182, together with the three evolved populations IMX1183, IMS1184 and IMS1185 at batch number 3, 6, 9, 15, 21, 38 and 50, were estimated by cultivation in 96 deep-well plates in a Growth Profiler 960 instrument (Enzyscreen, Heemstede, The Netherlands). A glycerol stock for each strain was inoculated in a 100-mL shake flask containing 20 mL SMDNO3 and incubated overnight at 30°C, 200 rpm. The next day, each culture was centrifuged at 3000 g for 5 min, supernatant was discarded and cell pellets were resuspended in SMDNO3 to and OD660 of 5. Then, two 96 deep-well plates were filled with 250 µL of SMDNO3 medium and each well was inoculated with 5 µL of one of the tested strains. Each strain was inoculated in triplicate with the exception of IMS1174, IMS1176, IMS1178, IMS1179, IMS1181 and IMS1182 that were inoculated in duplicate. Plates were incubated at 30°C, 250 rpm until the cultures reached stationary phase. To convert the measured ‘green’ cell density values into OD660 equivalent, a calibration curve was prepared by correlating the ‘green’ value of a IMS2565 cultures in SMDUrea at eight different OD660 values that were measured externally with a 7200 Jenway Spectrophotometer (Jenway, Stone, UK). Moreover, values measured for each position in the plate were normalized by a factor that was calculated by measuring the green value of a IMS2565 culture in SMDUrea of OD660 = 5 and by dividing that value by the average value measured across the whole plate (position normalization). After normalizing each green value time point by its own position normalization factor, OD660 equivalent values were calculated by fitting values with the calibration curve. Growth rate of each culture was calculated by fitting the exponential growth function with points of OD660 equivalent values between 0.5 and 2.

Aerobic cultivation in shake flasks

For the determination of the growth rate of IMS1174, IMS1175, IMS1176, IMS1177, IMS1178, IMS1179, IMS1180, IMS1181, IMS1182, IMX1183, IMS1184 and IMS1185, frozen stock cultures were used to inoculate 20 mL SMDNO3 starter cultures. These were subsequently used to inoculate 100 mL SMDNO3 flask cultures to initial OD660 values between 0.1 and 0.2. Growth of these cultures was monitored with a 7200 Jenway Spectrophotometer (Jenway). Specific growth rates were calculated from at least five time points in the exponential growth phase of each culture. At each time point, 2 mL of liquid culture was centrifuged for 5 min at 14 000 g, and supernatant was collected for HPLC and nitrate, nitrite, and ammonia analysis.

Whole-genome sequencing

Genomic DNA of strains ST6512, IMX2267, IMX2565, IMS1175, IMS1177 and IMS1180 was isolated with a Blood and Cell Culture DNA Kit with 100/G Genomics-tips (QIAGEN, Hilden, Germany) following manufacturer's instructions. Illumina-based paired-end sequencing with 150-bp reads was performed on 550-bp TruSeq DNA PCR-free insert libraries with a minimum resulting coverage of 50 x (Macrogen-Europe, Amsterdam, The Netherlands). Data mapping was performed using bwa 0.7.15-r1142-dirty against the Y. lipolytica CLIB122 genome (Kerscher et al. 2001; Dujon et al. 2004) to which five extra contigs containing the relevant integration cassettes had been previously added. Data processing and chromosome copy number variation determinations were done as previously described (Nijkamp et al. 2012; Perli et al. 2020) except for VCF file intersection and annotation that was performed with VCFtools_0.1.13 (vcf-isec command) and SnpEff, respectively (Danecek et al. 2011; Cingolani et al. 2012).

In vitro nitrate reductase activity measurements from cell extract

Nitrate reductase activity was measured from cell extract of strain ST6512, IMX2565, IMS1175, IMS1177, IMS1180 and IMX2565 evolution lines 1, 2 and 3. Cell extract preparation and activity measurements were performed as previously described (Perli et al. 2021). In brief, frozen stock cultures of each tested strain and evolved populations were used to inoculate 20 mL starter cultures, which were then used to inoculate 100 mL shake flask cultures on the same medium, to an initial OD660 of 0.2. Shake flasks were incubated until the OD660 exceeded 40. All strains were grown on SMDurea with the exception of the evolution lines 1, 2 and 3 that were grown on SMDNO3. Cultures were then centrifuged at 3000 g for 5 min, supernatant was discarded and cell pellets were resuspended in 2 mL lysis buffer (100 mM KPO4 pH 7 supplemented with complete ULTRA EDTA-free protease inhibitor cocktail, Roche, Basel, Switzerland). Cell resuspensions were aliquoted in 1.5 mL bead-beating tubes along with 0.75 g of 400–600 µm acid-washed glass beads (Sigma Aldrich) per tube. Cells were disrupted by six 1-min cycles at 5 m/s speed in a Fast-Prep 24 cell homogenizer (MP Biomedicals, Santa Ana, CA), with 5-min cooling on ice between cycles. Glass beads were separated from the cell extract by centrifuging the tubes for 10 min at 4°C and 15 000 g on a tabletop centrifuge. Supernatant was recovered in a fresh tube and clarified by centrifuging for 1 h at 4°C, 20 000 rpm. Clarified cell extracts were recovered and diluted 10 times in lysis buffer and kept on ice prior to analysis. Nitrate reductase activity was measured by monitoring substrate dependent NADPH consumption at 340 nm using a spectrophotometer (Jasco, Easton, MA). Reactions were performed in 1 mL final volume of 100 mM KPO4 buffer pH 7 supplemented with 200 μM NADPH, 20 μM FAD, 1 mM KNO3 as substrate and either 50 or 25 μL of clarified cell extract. Protein contents of cell extracts were quantified with a Quick Start Bradford Assay (Bio-Rad Laboratories, Hercules, CA) following manufacturer's instructions. Specific activities of nitrate reductase in cell extracts were expressed in μmol NADP+ min-1mgprotein-1.

In silico mitochondrial targeting prediction

The likelihood of mitochondrial targeting for the wild-type and mutated OpCnx1 and OpCnx2 sequences were calculated using five different web-based tools: TargetP 2.0 (Almagro Armenteros et al. 2019), DeepLoc 1.0 (Almagro Armenteros et al. 2017), MitoFates (Fukasawa et al. 2015), Predotar (Small et al. 2004) and PredSL (Petsalaki et al. 2006). When possible, non-plant or fungal database was selected as option.

Analytical methods

Metabolite concentrations in culture supernatants were analysed by high-performance liquid chromatography (HPLC) on an Agilent 1260 HPLC (Agilent Technologies, Santa Clara, CA) fitted with a Bio-Rad HPX 87 H column (Bio-Rad). The flow rate was set at 0.6 mL/min, 0.5 g/L H2SO4 was used as eluent and the column temperature was set at 65°C. An Agilent refractive-index detector and an Agilent 1260 VWD detector were used for metabolite quantification (Verhoeven et al. 2017). Nitrate, nitrite and ammonium concentrations in culture supernatants were measured with a Hach DR3900 spectrophotometer and Hach kits LCK 339, LCK 341 and LCK 304 (Hach Lange, Düsseldorf, Germany), according to the manufacturer's instructions.

Statistical analysis

Statistical significance of differences between measurements from replicate samples were calculated by using a two-tailed t-test assuming unequal variances (Welch's correction).

Data availability

All measurement data and calculations used for each figure in the manuscript are available at the 4TU.Centre for research data repository (https://researchdata.4tu.nl/) under URL:https://doi.org/10.4121/14230238.v1. DNA sequencing data of Y. lipolytica strains ST6512, IMX2267, IMX2565, IMS1175, IMS11777 and IMS1180 were deposited at NCBI (https://www.ncbi.nlm.nih.gov/) under BioProject accession number PRJNA704845.

RESULTS

Design and engineering of Moco biosynthesis and nitrate assimilation in Y. lipolytica

The absence of molybdenum-dependent enzymes in Yarrowia metabolism strongly suggested that the engineering of Moco biosynthesis may not only require functional expression of Moco biosynthesis genes, but also of a high-affinity molybdate transporter (Fig. 1). Based on a previous work in S. cerevisiae (Perli et al. 2021), it is not fewer than 11 genes from the yeast O. parapolymorpha and the algae C. reinhardtii that would be required to introduce Moco biosynthesis and nitrate assimilation in Y. lipolytica. The gene-set comprises seven genes coding for Moco biosynthesis proteins (OpCNX1, OpCNX2, OpCNX3, OpCNX4, OpCNX5, OpCNX6 and OpNFS1), three genes coding for the nitrate assimilation pathway (OpYNT1, OpYNR1 and OpYNI1) from O. parapolymorpha, and one gene coding for the high-affinity molybdate transporter (CrMoT1) from C. reinhardtii (Fig. 1). The codon-optimized genes were integrated in the chromosome of the Yarrowia strain ST6512 (W29, ku70Δ::pTEF1-Cas9-tTEF12::pGPD-DsdA-tLIP2) by using CRISPR/Cas9 gene-editing and the EASYcloneYali promoter parts and integrative plasmids (Holkenbrink et al. 2018). Genes were sequentially integrated in five different integration sites (E_4, C_2, E_1, C_3 and E_3) that were previously tested for heterologous gene expression (Holkenbrink et al. 2018). At each transformation, two genes were integrated, with the exception of the second last transformation in which all three genes encoding for nitrate assimilation pathway were integrated in one step. After five consecutive transformation rounds, the successful construction of the final strain IMX2565 was confirmed by Illumina short-read sequencing (PRJNA704845; Fig. 2).

Figure 1.

Schematic representation of the Moco biosynthesis pathway coupled to the nitrate assimilation pathway. GTP is converted to cyclic pyranopterin phosphate (cPMP) in the yeast mitochondria by OpCnx1 and OpCnx2. In the cytosol, cPMP is converted to molybdopterin (MPT) by OpCnx5 and OpCnx6. The sulfur moiety on OpCnx5 is restored by OpCnx4 that transfers the sulfur atom obtained by action of the cysteine desulfurase OpNfs1. Molybdate (MoO42–) is imported through the high affinity transporter CrMot1 and is inserted in MPT by OpCnx3 to form Moco. Nitrate is imported via OpYnt1 and reduced by the Moco-dependent nitrate reductase OpYnr1 to nitrite. OpYni1 converts nitrite to ammonia that finally enters the native nitrogen assimilation pathway. OpCnx4, OpCnx5 and OpCnx6 are shown in light blue, teal and magenta, respectively. A question mark indicates a yet unknown cPMP transporter and a dashed line indicates multiple enzymatic steps.

Figure 2.

Schematic representation of strain construction. Strain ST6512, which constitutively expresses a Spcas9, was sequentially transformed five times with a targeting gRNA plasmid and a NotI-linearized plasmid carrying the integration cassettes as repair fragment, yielding IMX2565. Intermediate strains and the final strain IMX2565 are shown in light-blue and purple, respectively.

ALE on nitrate containing media

Similarly to what was observed in S. cerevisiae, adaptation was required to observe growth of the Y. lipolytica strain IMX2565 on nitrate medium (Perli et al. 2021). We inoculated strain IMX2565 in SMDNO3 and incubated the flasks at 30°C in triplicate. Reproducibly, after 2 weeks the Yarrowia strain showed full growth (OD660 above 20) in all flasks. To improve the strain growth rate, IMX2565 was subjected to ALE by sequential transfers in flasks containing SMDNO3 for 335 generations (50 consecutive batches). Evolving cultures were periodically stocked and used to monitor the growth rate of the evolving populations (Fig. 3A). At the onset of the ALE experiment, the initial batches exhibited growth rates ranged from 0.04 to 0.05 h-1. Throughout the first 21st transfers the specific growth rate of the yeast populations increased and levelled off to 0.11–0.12 h-1. At that time-point, corresponding to a about 140 generations, three single colonies were isolated from each evolving culture and named IMS1180 (line 1, colony 1), IMS1181 (line 1, colony 2), IMS1182 (line 1, colony 3), IMS1174 (line 2, colony 1), IMS1175 (line 2, colony2), IMS1176 (line 2, colony 3), IMS1177 (line 3, colony 1), IMS1178 (line 3, colony 2) and IMS1179 (line 3, colony 3). The specific growth rates of the single colony isolates on SMDNO3 ranged from 0.03 to 0.09 h-1 (Fig. 3B). Clones IMS1180, IMS1175 and IMS1177 were empirically selected for further characterization based on the similar growth performance to the relative evolved populations. The evolution experiment was prolonged for 29 additional sequential batches, summing up to a total of 50 batches (335 generations) to further probe evolvability of the phenotype. However, the growth rate of evolving populations stabilized to a value of about 0.13 h-1 and did not further increase (Fig. 3A).

Figure 3.

Maximum specific growth rate (µ) values of IMX2565 throughout the adaptive laboratory evolution experiment (A), growth rates of single colony isolates (B) on SMDNO3 and Venn diagram highlighting genes affected by non-synonymous mutations and/or INDELs in independently evolved isolates IMS1175, IMS1177 and IMS1180 (C). An arrow in panel A indicates the time-point when single colonies were isolated from the three independent evolving populations while asterisks in panel B indicate single colony isolates selected for whole-genome re-sequencing and further characterization. Evolution line 1, 2 and 3 are shown in black, magenta and teal, respectively. Error bars represent the standard error of the mean for replicate cultures (n = 4 with the exception of IMS1181, IMS1182, IMS1174, IMS1176, IMS1178 and IMS1179 where n = 2). The Venn diagrams shows genes that acquired one or more non-synonymous mutations or INDELs in multiple independent evolution experiments as well as genes that were affected in a single replicate. Apparent mutations also found in the genome of the parent strain ST6512 and/or IMX2565 were subtracted and not shown.

Whole-genome sequencing of evolved strains and mutations identification

To identify mutations responsible for the increase in growth rate, the genomes of clones derived from each evolution line (IMS1175, IMS1177 and IMS1180) and well as those of parental strains (ST6512 and IMX2565) were re-sequenced by Illumina short-read technology. After aligning the reads to the Y. lipolytica W29 reference genome (PRJNA601425; Kerscher et al. 2001; Dujon et al. 2004), mapped data were analysed for the presence of either copy number variations (CNVs), single nucleotide variations (SNVs) and/or insertions/deletions (INDELs) that occurred in the annotated coding sequences. As opposed to what happened in S. cerevisiae (Perli et al. 2021), no gene or chromosome CNVs were observed between the parental strain IMX2565 and evolved isolates IMS1175, IMS1177 and IMS1180. SNV and INDELs analysis was systematically performed and data from the three sequenced colony isolates were then compared. To minimize the number of miscalls caused by mapping artefacts, SNV and INDELs that were also detected in the two sequenced parental strains ST6512 and IMX2565, mapped to the same reference sequence, were systematically removed. After curation of SNVs and INDELs miscalls, a total of five, two and four mutations were found in evolved isolates IMS1175, IMS1177 and IMS1180, respectively (Fig. 3C and Table 4). While no genes were found mutated in all three independently evolved isolates, three genes (Spcas9, OpCNX1 and YALI0_E24167g) were found differently mutated in two sequenced isolates. The mutations identified in OpCNX1 that were found in IMS1175 and IMS1177, were identical and located in the 5′ end of the coding sequence (G22A). Since the evolution lines were started independently and the mutation was not present in the engineered strain IMX2565, the recurrence of this mutation in OpCNX1 might be critical in the acquisition of faster growth of Y. lipolytica on nitrate as N-source.

Table 4.

SNVs and INDELs found in single colony isolates IMS1775, IMS1177 and IMS1180 obtained from the serial transfer evolution experiment of strain IMX2565 in SMDNO3.

Mutated gene	Mutation type	Base change	Amino acid change	Gene annotation
IMS1175
YALI0_A04697g	INDEL	T970TG	Gln324Frameshift	Similar to uniprot|O42626 Neurospora crassa Serine/threonine-protein kinase nrc-2 (Nonrepressible conidiation protein 2)
YALI0_D25784g	SNV	T457C	Ser151Pro	weakly similar to uniprot|O74782 Schizosaccharomyces pombe Hypothetical protein
YALI0_E00638g	SNV	G868T	Gly290Cys	Similar to uniprot|Q9TEM3 Emericella nidulans MCSA Methylcitrate synthase precursor
OpCNX1	SNV	G22A	Glu8Lys	GTP 3',8-cyclase
Spcas9	SNV	G2959A	Ala987Thr	CRISPR-associated endonuclease Cas9
IMS1177
YALI0_E24167g	SNV	G743A	Trp248Stop	Weakly similar to uniprot|Q2VQ77 Saccharomyces cerevisiae YPL092w SSU1 Plasma membrane sulfite pump and required for efficient sulfite efflux
OpCNX1	SNV	G22A	Glu8Lys	GTP 3',8-cyclase
IMS1180
YALI0_E24167g	SNV	A440G	His147Arg	Weakly similar to uniprot|Q2VQ77 Saccharomyces cerevisiae YPL092w SSU1 Plasma membrane sulfite pump and required for efficient sulfite efflux
YALI0_F05346g	SNV	G337A	Ala113Thr	Weakly similar to uniprot|Q00858 Fusarium solani cutinase gene palindrome-binding protein
OpCNX2	SNV	G4A	Val2Met	Cyclic pyranopterin monophosphate synthase
Spcas9	SNV	C3020A	Ala1007Asp	CRISPR-associated endonuclease Cas9

The second gene (YALI0_E24167g) mutated in two different isolates IMS1180 and IMS1177 encoded a putative sulfite transporter that shared similarity to the S. cerevisiae sulfite efflux pump Ssu1 (Avram and Bakalinsky 1997). The nature of the mutations found in YALI0_E24167g would suggest a loss of function. In IMS1177 the mutation G743A resulted in the introduction of a stop codon at position 248. Consequently, the mutated protein would be truncated to 45% of the original sequence and thus likely not functional.

The last gene found mutated in two different isolates was Spcas9. Although different, these two mutations occurred in a section of the gene encoding amino acids located in the same functional domain. The Spcas9 mutation in IMS1175 (G2959A) and in IMS1180 (C3020A) led to amino acids change in the RuvC-III domain of SpCas9 (Jinek et al. 2014).

Whereas IMS1175 and IMS1177 both harbored mutations in OpCNX1, interestingly IMS1180 had a mutation in OpCNX2. As observed for OpCNX1, the OpCNX2 mutation was located in the 5′ end the gene (G4A).

On top of that, four more mutations were identified in only one isolates, it included YALI0_D25784g encoding a hypothetical protein, YALI0_A04697g a gene encoding a putative serine/threonine protein kinase, YALI0_E00638g a gene encoding a putative methyl citrate synthase in IMS1175 and YALI0_F05346g a gene encoding a protein exhibiting similarity with a cutinase from Fusarium solani cutinase.

ALE for fast growth on nitrate was associated with an increased Moco-dependent nitrate reductase activity

To investigate the effects of the ALE experiment, NADPH- and Moco-dependent nitrate reductase activity was assayed from cell-extracts of single colony isolates IMS1180, IMS1175 and IMS1177, evolved populations 1–2–3 after 50 transfers in SMDNO3, the parental un-evolved engineered strain IMX2565 and the Cas9-expressing parental strain ST6512 (Fig. 4). While as expected, cell extract from ST6512 which does not carry any of the heterologous genes for Moco biosynthesis and nitrate assimilation, showed no significant nitrate reductase activity, the engineered strain IMX2565 showed an activity of 0.004 ± 0.001 µM NADP+/min/mg total protein. Although significant, this activity was up to 55-fold lower than the one observed for the evolved single colony isolates IMS1180, IMS1175 and IMS1177 that reached up to 0.22 ± 0.01 µM NADP+/min/mg total protein. Finally, cell extracts from the evolved populations showed activity values that were comparable for the 335 generations old, prolonged evolution line 1 and within 2-fold for the 335 generations old, prolonged evolution 2 and 3 with those measured for the single colony isolates, reaching up to 0.41 ± 0.03 µM NADP+/min/mg total protein.

Figure 4.

NADPH-dependent nitrate reductase activity measured from cell extracts. ST6512, IMX2565, isolate IMS1180 and evolution line 1 at T50, isolate IMS1175 and evolution line 2 at T50 and isolate IMS1177 and evolution line 3 at T50 are shown in light blue, purple, black, magenta and teal, respectively. Cell extract was prepared from 100 mL of stationary phase cell cultures in SMDurea for ST6512, IMX2565, IMS1180, IMS1175 and 1177 and in SMDNO3 for evolution lines T50 1, 2 and 3. P-values for two-tailed Welch's t-test are shown above the tested pairs. Error bars represent the standard error of the mean of technical replicates (n = 3).

To see whether the increase in nitrate reductase activity was associated with improved growth performance on nitrate, the strains ST6512, IMS11780, IMS1175, IMS1177, as well as the prolonged evolved populations at T50 1–2–3 were cultivated in chemically defined medium with nitrate as N-source (SMDNO3-). The strains were grown in aerobic shake flasks and OD 660, glucose, nitrate, nitrite and ammonium levels were monitored over time (Fig. 5). As expected, the parental strain ST6512 did not show any growth and/or glucose/nitrate consumption (Fig. 5B). All the other strains and evolved populations showed a growth rate ranging between 0.07 ± 0.01 and 0.16 ± 0.01 h -1. Although slightly different in absolute values when compared to the growth rates measured in 96-wells format (Fig. 3B), the ones measured in 100 mL shake flasks followed the same trend with IMS1180 being the slowest and IMS1175 being the fastest growing isolates, respectively. Growth rate of the evolved populations, that were kept evolving for about 195 generations after single colonies were isolated, was significantly higher when compared to the respective single clones, with up to 2-folds increase (Fig. 5A). Notably, both evolved isolates and further evolved populations excreted moderate amounts of the intermediates nitrite and ammonia during the exponential phase of growth, suggesting that nitrate assimilation was not growth-limiting (Fig. 5C–H). Similarly to what happened to engineered nitrate-assimilating S. cerevisiae strains (Perli et al. 2021), ammonia and nitrite were also excreted toward the end of the batch fermentation, resulting from continuous conversion or possible cell lysis.

Figure 5.

Growth rates (A) and growth curves of ST6512 (B), IMS1180 (C), evolution line 1 T50 (D), IMS1175 (E), evolution line 2 T50 (F), IMS1177 (G) and evolution line 3 T50 (H) on SMDNO3. Symbols indicate biomass (●), glucose (◇), nitrate (○), nitrite (□) and ammonium (△). Statistical analysis was based on a two-tailed Welch's t-test and P-values are reported for tested pairs. Error bars represent the standard error of the mean for replicate cultures (n = 3).

DISCUSSION

Moco-dependent enzymes catalyse many redox reactions involved in the carbon, nitrogen and sulfur cycles (Leimkuhler and Iobbi-Nivol 2016) and might be harnessed to expand substrates range for microbial growth. This study firmly established that although Moco pathway is predominantly absent in Saccharomycotina yeasts, this function can be implemented by metabolic engineering. This study is, after introduction in S. cerevisiae, the second successful engineering of the pathway in yeasts. The large phylogenetic distance between the two tested hosts would also suggest that this pathway engineering could be extrapolated to an even larger set of budding yeasts (Shen et al. 2018). While in both biological systems functionality through growth coupling with nitrate assimilation required an evolution step, the resulting mutations identified were different. In the precedent example in S. cerevisiae, adaptation to fast utilization of nitrate involved segmental aneuploidy of the whole heterologous cluster, accompanied with a strong increase of nitrate reductase activity (Perli et al. 2021). However, since the increase in copy number affected both genes of the Moco biosynthesis and the nitrate assimilation pathways it was not really possible to discriminate which metabolic branch was the most limiting. In Y. lipolytica instead, the results would indicate that Moco synthesis is most limiting. This was supported by the absence of gene dosage but instead by recurrent mutations in the mitochondrial enzymes OpCnx1 andOpCnx2 that catalyse the formation of cPMP in the very first step of Moco biosynthesis (Fig. 1). Evolved isolates IMS1175 and IMS1177 carried a mutation in the OpCnx1 whereas the third isolate IMS1180 carried a mutation in OpCnx2. But more specifically, the non-synonymous mutations led to amino acid change in N-terminus of both proteins changing the sequence of the mitochondrial targeting signal. Although theoretical, the presence of the mutation systematically improved the in silico prediction of mitochondrial targeting likelihood using five different algorithms (Small et al. 2004; Petsalaki et al. 2006; Fukasawa et al. 2015; Almagro Armenteros et al. 2017, 2019; Table 5). These mutations could contribute to improve OpCnx1 and 2 translocations into the organelle and optimize the supply of cPMP for Moco biosynthesis. The higher Moco availability would then be at the origin of the massive increase by 1 to 2 order of magnitude of nitrate reductase activity (Fig. 4). These results seem to indicate that although the O. parapolymorpha biosynthetic genes are sufficient to implement the synthesis of the new cofactor in other yeast species, the adaptation required to tune its supply optimally might be species dependent.

Table 5.

Likelihood of mitochondrial targeting in WT and mutated OpCnx1-OpCnx2 protein sequences across different prediction tools.

	OpCnx1		OpCnx2
Prediction tool	WT	Mutated	WT	Mutated
TargetP 2.0	0.000208	0.000649	0.000525	0.001879
DeepLoc 1.0	0.0875	0.1946	0.1252	0.1261
MitoFates	0.080	0.280	0.000	0.000
Predotar	0.01	0.14	0.00	0.00
PredSL	0.006826	0.097417	0.004088	0.004091

Among the three genes that were mutated in at least two isolates we found YALI0_E24167g (mutated in IMS1177 and IMS1180), a protein containing a predicted SLAC1 domain characteristic of voltage-dependent anion transporters such as nitrate and sulfite exporters (Vahisalu et al. 2008). The protein is similar to the S. cerevisiae sulfite transporter Ssu1 which has been previously shown to be able to also export nitrate and nitrite (Cabrera et al. 2014). While the IMS1180 mutation in YALI0_E24167g led to an His to Arg change, the mutation found in strain IMS1177 introduced a stop codon (Trp248Stop), creating a truncation of the translation product by 180 amino acids, representing 45% of the protein length, that included not fewer than four transmembrane domains that undoubtedly caused a loss of function. Assuming that YALI0_E24167g shared function with its S. cerevisiae ortholog, the loss of function might represent a mechanism to avoid nitrogen loss through excessive nitrite export; it is worth noticing that this transport system would not be the only one as moderate extracellular nitrite concentration could still be measured. Maintenance of a higher intracellular nitrite could also partially explain the increase in nitrate assimilation and growth rate after evolution in nitrate-containing medium.

Interestingly, the third gene that was affected by non-synonymous mutations in at least two isolates (IMS1175 and IMS1180), was Spcas9. Although resulting in amino acid changes, it is not obvious to predict the impact of the mutations on the endonuclease activity. However, both mutations (Ala987Thr and Ala1007Asp) were found in the RuvC-III nuclease domain of Cas9 protein that extends from 925 to 1101. Several other mutations in RuvC-III at position 982, 983 and 986 have been described (Nishimasu et al. 2014) and all yielded either a decrease or an alteration of the endonuclease activity resulting in mutants able to only cleave one strand instead of two (Fonfara et al. 2014; Jinek et al. 2014). This was confirmed experimentally, as our attempts to further engineer the evolved isolates systematically failed. We cannot exclude that recurrent mutagenesis of Cas9 might take place to counteract potential endonuclease toxicity.

The Moco platform strain constructed and characterized in this study represents a stepping stone towards the exploitation of a new class of enzymes that might contribute to expand the metabolic capabilities of yeast microbial cell factories by enabling new metabolic engineering strategies.

FUNDING

This work was supported by the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska–Curie action PAcMEN (Grant agreement number 722287). IB acknowledges the financial support from the Novo Nordisk Foundation (Grant agreement numbers NNF20OC0060809 and NNF20CC0035580), and from the European Research Council under the European Union's Horizon 2020 research and innovation program (YEAST-TRANS project, grant agreement number 757384).

Click here for additional data file.