Tuning Fatty Acid Profile and Yield in Pichia pastoris

Abstract

Fatty acids have been supplied for diverse non-food, industrial applications from plant oils and animal fats for many decades. Due to the massively increasing world population demanding a nutritious diet and the thrive to provide feedstocks for industrial production lines in a sustainable way, i.e., independent from food supply chains, alternative fatty acid sources have massively gained in importance. Carbohydrate-rich side-streams of agricultural production, e.g., molasses, lignocellulosic waste, glycerol from biodiesel production, and even CO₂, are considered and employed as carbon sources for the fermentative accumulation of fatty acids in selected microbial hosts. While certain fatty acid species are readily accumulated in native microbial metabolic routes, other fatty acid species are scarce, and host strains need to be metabolically engineered for their high-level production. We report the metabolic engineering of Pichia pastoris to produce palmitoleic acid from glucose and discuss the beneficial and detrimental engineering steps in detail. Fatty acid secretion was achieved through the deletion of fatty acyl-CoA synthetases and overexpression of the truncated E. coli thioesterase ‘TesA. The best strains secreted >1 g/L free fatty acids into the culture medium. Additionally, the introduction of C16-specific ∆9-desaturases and fatty acid synthases, coupled with improved cultivation conditions, increased the palmitoleic acid content from 5.5% to 22%.

1. Introduction

Fatty acids (FAs) are fundamental constituents of the metabolic processes in all living organisms, serving as pivotal precursors for phospholipids essential in cell membrane synthesis. Additionally, FAs play a vital role in the formation of carbon- and energy-storage molecules, such as triacylglycerols (TAGs) and polyhydroxyalkanoates (PHAs) [1,2,3,4]. In addition to their crucial cellular significance, FAs and their derivatives (e.g., alka(e)nes, fatty alcohols, fatty acid esters, hydroxy fatty acids) exhibit significant industrial importance as versatile building blocks for a wide array of commercially relevant compounds, including detergents, soaps, lubricants, biofuels, cosmetics, and pharmaceutical products, among others [5,6]. Traditionally, the sourcing of FAs and their derivatives has heavily relied on animal fats and plant oils. However, the escalating reliance on edible oils, particularly palm oil, as feedstock for chemical and biofuel production has raised substantial environmental concerns.

Addressing these sustainability issues, research has embarked on investigating microbial free fatty acid (FFA) production, especially from non-edible carbon sources (e.g., lignocellulosic biomass, waste streams, glycerol), leading to promising eco-friendly production routes. Microorganisms offer numerous advantages over higher eukaryotic organisms for FFA production, including a simpler metabolism, being more amenable to genetic engineering, and demonstrating faster growth rates and safe production, making them well-suited for large-scale industrial applications. Furthermore, the enhanced accessibility of reaction/pathway and omics databases, genetic engineering tools, and high-throughput screening methods has facilitated the development of synthetic microbial cell factories proficient in the synthesis of diverse industrially relevant compounds with remarkable efficiency [7,8]. Nonetheless, the efficient microbial production of FFAs remains a major challenge due to the tight control and regulation of FA metabolism, as most native microorganisms do not support the biosynthesis of excess FAs beyond their metabolic demands through a set of regulatory mechanisms. Hence, in order to produce FFAs and derivatives thereof for commercial exploitation, non-oleaginous and oleaginous organisms have to be extensively engineered in their metabolic pathways.

Escherichia coli, Saccharomyces cerevisiae, and oleaginous yeasts like Yarrowia lipolytica and Rhodosporidium toruloides are commonly utilized microbial chassis in engineering approaches for optimized FA production [5,9,10]. E. coli has been intensively investigated in regard to its fatty acid biosynthesis and regulation [11,12]. Beyond the abundant genetic information, the benefits of its fast growth and easy genetic manipulation by numerous available genetic tools make E. coli a frequently used host, engineered for the elevated production of FFAs and derivatives [5,6,13,14,15,16,17,18,19]. However, when considering products intended for food and pharmaceutical applications, it is important to use microorganisms that are classified as “generally recognized as safe” (GRAS). This approach helps to address any potential safety concerns raised by the public. S. cerevisiae is one prime example of well-studied GRAS strains. It has been extensively employed in research studies focused on enhancing FA production through metabolic engineering strategies. Several combinatorial, multi-pathway attempts yielded considerable amounts of extracellular and intracellular lipids in S. cerevisiae, as recently reviewed in [20].

Oleaginous yeasts, such as Yarrowia lipolytica, Rhodosporidium toruloides, or Lipomyces starkeyi, are known for their natural ability to accumulate lipids up to 70% of their CDW under nutrient-limiting conditions, and most of the lipids are stored in the form of triacylglycerols (TAGs) [21]. These yeasts, which also have the GRAS status, harbor robust lipid synthesis pathways and have been further engineered for enhanced FA and oleochemical production [22,23,24] or for tailoring the chain-length of produced FAs and their derivatives [25,26].

The properties and nutritional value of natural fats, oils, and their derived oleochemicals depend primarily on the carbon chain-length and degree of saturation of the fatty acids they contain [27,28,29]. Recently, the biotechnological production of tailored chain-length fatty acids has gained importance, as not all fatty acid species are highly abundant in natural sources. Oleic acid (C18:1, OA) is abundantly found in various natural sources, especially plants and fruits (e.g., olive (55–83%), canola (62–64%), or avocado (59–62%)) [30]. Oleic acid’s presence contributes to the fluidity and stability of cell membranes, affecting cell signaling and physiological functions [31]. In contrast, the content of palmitoleic acid (C16:1, POA) in main vegetable oils is very low (<2%) [32], and alternative sourcing of POA has gained significant interest from industries, as several potential health benefits have been ascribed to POA, such as the attenuation of inflammation and the prevention of cardiovascular diseases or diabetes [33,34,35,36]. Furthermore, POA exhibits promise for utilization as a cleansing agent in skincare and medical products, owing to its specific bactericidal activity against Staphylococcus aureus [37]. Currently, its primary sources involve marine organisms and selected plants, particularly Macadamia integrifolia (15–22% POA in macadamia oil) and sea buckthorn (25–35% POA), as highlighted by Hu et al. [34] and Solà Marsiñach and Cuenca [38]. The cultivation and commercialization of plants rich in POA on a larger scale is hindered by low yield, small seed size, and limited geographic distribution; hence, there is a need to establish a sustainable and cost-effective production method as an alternative to the current POA extraction procedures.

The yeast Pichia pastoris (syn. Komagataella phaffii) has gained significant attention in biotechnological production processes due to its efficient protein expression system, especially when using its strong alcohol oxidase (AOX1) promoter. Products made by P. pastoris have been classified as GRAS. Thus, this methylotrophic yeast has become a preferred host organism for the production of recombinant proteins, enzymes, and other biomolecules [39,40,41,42]. In contrast to S. cerevisiae, P. pastoris is a Crabtree-negative yeast [43], which facilitates the application of this yeast for high-cell-density fermentation in industrial processes. Here, we aim to develop a P. pastoris strain for high-level fatty acid production from sustainable carbon sources, such as glucose or glycerol [39,44]. Our objectives encompass two linked approaches: firstly, the enhancement of overall FFA production, and secondly, the specific synthesis of POA at high titers. Thus, we have manipulated the metabolic pathway of the yeast P. pastoris towards a high-POA phenotype and have evaluated different strategies for FFA over-production in P. pastoris.

2. Materials and Methods

2.1. Chemicals, Media, and Cultivation Conditions

All the media components and laboratory reagents were sourced from Carl Roth GmbH & Co. KG (Karlsruhe, Germany), unless stated otherwise. Bacto^TM yeast extract was purchased from BD (Becton Dickinson, Heidelberg, Germany) while the yeast nitrogen base without amino acids and without ammonium sulfate was obtained from ForMedium^TM (Swaffham, UK). Antifoam 204 was obtained from Sigma-Aldrich, Vienna, Austria. Zeocin^® was purchased from InvivoGen (Toulouse, France). For plasmid assembly and propagation, E. coli TOP10 F′ obtained from Life Technologies (Vienna, Austria) was utilized. Phusion^TM High-Fidelity DNA polymerase, restriction enzymes, CloneJet PCR cloning Kit, Phire^TM Plant Direct PCR Master Mix, Gel Extraction Kit, and GeneJET Plasmid Miniprep Kit were purchased from Thermo Fisher Scientific^TM (Waltham, MA, USA). Additionally, primers and short synthetic DNA fragments up to 500 bp were synthesized by Integrated DNA Technologies (IDT, Leuven, Belgium) and codon-optimized genes were obtained from Twist Bioscience (South San Francisco, CA, USA). Gibson Assembly^® Master mix kits were purchased from New England Biolabs Inc. (Frankfurt am Main, Germany). Detailed lists of the synthetic genes and primers are provided as Supplementary Information (Supplementary Tables S1 and S2).

E. coli was cultivated in LB medium (Lennox) purchased from Carl Roth GmbH & Co. KG (Karlsruhe, Germany).

Pichia pastoris pre-cultures were cultivated in YPD medium containing 20 g/L peptone, 10 g/L yeast extract, and 20 g/L glucose. P. pastoris cultivations for fatty acid production were performed in 96-deep well plates (DWPs) containing 0.5 mL nitrogen-limited, buffered minimal dextrose medium (BMD11) containing 1.9 g/L ammonium chloride, 3.8 g/L yeast nitrogen base without amino acids, and without ammonium sulfate; 0.1 M potassium phosphate buffer, pH 6; 30 g/L glucose; 0.4 mg/L biotin; and 40 mg/L L-histidine, if required. The DWPs were covered with gas-permeable AeraSeal™ films, purchased from Sigma-Aldrich (Vienna, Austria). The DWPs were inoculated to an OD₆₀₀ of 0.2 from precultures grown for 48 h to a stationary phase in YPD and were shaken at 320 rpm and 28 °C for 72–96 h. For strain selection, the YPD or LB agar plates were supplemented with the respective antibiotics (Zeocin^®: 50 µg/mL for LB and 100 µg/mL for YPD or ampicillin: 100 µg/mL). The media for the plates were solidified by the addition of agar to 1.5%.

2.2. Strains and Synthetic Genes

E. coli TOP 10F′ (F′[lacIq, Tn10(TetR)] mcrA Δ(mrr-hsdRMS-mcrBC) φ80lacZΔM15 ΔlacX74 recA1 araD139 Δ(ara-leu)7697 galU galK rpsL(StrR) endA1 nupG), purchased from Thermo Fisher Scientific (Waltham, MA, USA), was used for all the cloning experiments and propagation of expression vectors. The P. pastoris (Komagataella phaffii) wild-type strain CBS7435 (NRRL Y-11430 [45,46,47]) was used as the host strain for all the further strain constructions (

Table 1. List of P. pastoris strains used/generated in this study. All strains are based on the P. pastoris CBS7435 wild-type strain. All numbered P. pastoris strains were generated in this study (except for the wild-type strain). Gene abbreviations: DAS1, dihydroxyacetone synthase 1; FDH1, formate dehydrogenase 1; FBP1, fructose-bisphosphatase; ARG4, argininosuccinate lyase; AOX1, alcohol oxidase 1. Abbreviations for integration sites: Chr1–Chr4, chromosomes 1 to 4; NS1-10 neutral site (integration site) on respective chromosome (i.e., Chr1_NS10: neutral site 10 on chromosome 1); TEFup, integration site upstream of transcription elongation factor 1 locus; FLDup, integration site upstream of S-(hydroxymethyl)glutathione dehydrogenase.

Strain (Pp#)	Genetic Background	Further Modifications
Wild-Type (WT)	CBS7435
12	WT	Δfad12
14	WT	Δole1-1
16	WT	Δole1-2
28	Pp#16	his4::PGAP-‘TesA-TTARG4
29	Pp#28	Δfaa2
32	Pp#29	Δfaa1
35	Pp#32	Δpox1
37	Pp#35	Δfat1
39	Pp#35	FLDup::PTEF1-MmSCD3-TTARG4
44	Pp#35	Δdga1
45	Pp#35	Δlro1
46	Pp#35	Δdga1 Δlro1
47	Pp#39	his4::PGAP-MtFAT-A-TTARG4
48	Pp#39	his4::PGAP-AtFAT-B-TTARG4
49	Pp#39	acc1::PTEF1-ACC1-TTACC1
50	Pp#39	acc1::Acc1S1151A-TTACC1
51*	Pp#35	Δdga1 Δlro1 Δare2
51	Pp#39	TEFup::PPGK1-CeFAT-TTARG4
53	Pp#39	his4::PGAP-Δ‘tesA-TTARG4
54	Pp#39	acc1::PTEF1-ACC1S1151A-TTACC1
67	Pp#39	TEFup::PHTX1-BaFAS1-TTARG4-BaPPT-TTAOX1
85	Pp#39	PpFAS1::PHTX1-ScFAS2-TTPpFAS1-ScFAS1-TTPpFAS1
89	Pp#39	TEFup::PHTX1-ScFAS2-TTARG4-ScFAS1-TTAOX1
91	Pp#39	TEFup::PHTX1-hFAS-TTARG4-EcACPS-TTAOX1
112	Pp#50	Chr1_NS6::PHHX1-ZWF1-TTARG4-SOL3 -TTTEF1
113	Pp#50	faa2::PGAP-RAD52-TTAOX1 Chr1_NS10_PGAP-CaGDH-TTAOX1
114	Pp#50	faa2::PGAP-RAD52-TTAOX1
115	Pp#85	PpFAS2::PHIS4-HIS4-TTPpFAS2
121	Pp#50	faa2::PGAP-RAD52 Chr2_NS3::PTPI-RtACL-TTDAS1-PGAP-MmME-TTHTB-PRP-PpCTP1-TTPGK1-PpMDH3-TTTEF1
127	Pp#35	FLDup::PTEF1-MmSCD3-T2A1-PpCypb5-T2A2-PpCyb5R-TTARG4
129	Pp#35	FLDup::PTEF1-MmSCD3-T2A1-MmCypb5-T2A2-MmCyb5R-TTARG4
135	Pp#39	Δpxa1
136	Pp#39	Δelo3
138	Pp#39	Δelo100
140	Pp#39	Δfad15
141	Pp#50	faa2::PGAP-RAD52 Chr4_NS7::PCAT-DGA1-TTTEF1-PHHX2-TGL3-TTHTA-TGL4-TTGAP
142	Pp#50	faa2::PGAP-RAD52-TTAOX1 Chr2_NS3::PENO1-RtACL-TTDAS1
145	Pp#50	faa2::PGAP-RAD52 Chr3_NS7::PHHX2-BbPK--TTFDH1-CkPTA-TTFBP1
148	Pp#85	Chr4_NS7::PGAP-ScOLE1-TTDAS1
150	Pp#39	Δfad12
153	Pp#50	faa2::PGAP-RAD52-TTAOX1 Chr2_NS3::PENO1-RtACL PPpIDH1-RtIDH1-TTPpIDH1PPpIDH2-RtIDH2-TTPpIDH2

).

For heterologous gene expression in P. pastoris, codon-optimized variants of the following genes were designed by applying the P. pastoris codon usage: ‘TesA (E. coli thioesterase I without N-terminal signal-peptide, AAC73596.1); MtFatA (acyl-ACP thioesterase from Macadamia tetraphylla, EU383030.1); AtFAT-B (Thioesterase from Arabidopsis thaliana, Z36911.1). All the thioesterases were expressed as truncated proteins without their native signal-peptides: MmSCD3 (stearoyl-CoA desaturase from Mus musculus, AF272037.1); CeFAT-5 (palmitoyl-CoA desaturase from Caenorhabditis elegans, AF260242.1); RtACL (ATP-citrate lyase from R. toruloides, CDR44680.1); MmME (malic enzyme from M. musculus NP_032641.2); CaGDH (NADP-dependent glyceraldehyde-3-phosphate dehydrogenase from Clostridium acetobutylicum WP_010966919.1); CkPA (phosphate acetyltransferase from Clostridium kluyveri WP_012101779.1); BbPKA (D-fructose 6-phosphate phosphoketolase from Bifidobacterium breve ERI86329.1); RtIDH1 (mitochondrial NAD+-specific isocitrate dehydrogenase subunit 1 from R. toruloides, AEK32870.1); RtIDH2 (mitochondrial NAD+-specific isocitrate dehydrogenase subunit 2 from R. toruloides, AEK32871.1); hFAS (fatty acid synthase from Homo sapiens AAH63242.1); EcAcpS (holo-acyl-carrier-protein synthase from E. coli, WP_219378987.1); BaPPT1 (holo-acyl-carrier-protein synthase from Corynebacterium (Brevibacterium) ammoniagenes, EFG81440.1); BaFAS (fatty-acid synthase from Corynebacterium ammoniagenes, CAA46024.1). The sequences are listed in Supplementary Table S1.

2.3. Strain Design and Genetic Manipulations

Genomic modifications of P. pastoris were generated by CRISPR/Cas9-induced integration of knockout or expression cassettes as previously described [48]. Modifications were introduced iteratively by co-transforming Cas9-gRNA plasmids (pPpT4_pHTX1_hsCa) and linearized donor cassettes, followed by screening for successful integration. The donor cassettes were flanked by approximately 1 kb up- and downstream DNA stretches of the target genes to facilitate genomic integration through double crossover events. The donor cassettes and Cas9 plasmids with novel gRNAs were constructed as follows: individual building blocks for expression cassettes, such as promoters, structural genes, transcription terminators, and homologous regions, were amplified from genomic DNA (P. pastoris CBS7435 wild-type or S. cerevisiae CEN.PK2–1C) or from synthetic DNA fragments, equipped with 30 bp overhangs between adjacent segments to facilitate in vitro assembly. Then, the PCR fragments were assembled with the linearized pJET1.2 vector of the CloneJet PCR cloning kit in one-pot Gibson assembly^® reactions to form plasmids for propagation in E. coli. For gene assembly reactions, the Gibson assembly^® Master Mix was applied according to the supplier’s protocol. Integration sites for the expression cassettes were selected based on the work of Cai et al. and Liu et al. [49,50] with some modifications. For seamless gene knockouts, donor cassettes were constructed from 1 kb up- and downstream DNA stretches of the target genes, which had also been amplified from the P. pastoris genome. These fragments were directly fused in an assembly with the linear pJET1.2 vector. Specific genes (FAD15, PXA1, LRO1, ARE2, ELO100) were knocked out through Cas9-induced frameshift mutations (indels) without supplying knockout cassettes [48]. Additional information on the assembly of certain constructs is given in Supplementary Method S3.

The pPpT4_pHTX1_hsCas vector, kindly provided by the group of Prof. Anton Glieder (Institute of Molecular Biotechnology, Graz University of Technology, Austria), was modified to target new genetic loci. The N20 sequence and ribozyme loop region were exchanged by newly assembling pPpT4_pHTX1_hsCas from two PCR fragments with compatible overhangs. PCR fragments 1 and 2 were amplified from the original pPpT4_pHTX1_hsCas with the primers Fw_T4pHTX1 and Gibson_Rv_insert_HH, Fw_Gibson_insert20N and Rv_T4pHTX1, respectively, with each PCR fragment representing one half of the vector. Primers Gibson_Rv_insert_HH and Fw_Gibson_insert20N introduced novel HH and N20 sequences in fragment 1 and 2 adjacent to one of the two overlapping regions, and subsequent Gibson assembly resulted in the reconstitution of the vector. A detailed vector map and primer sequences are provided in the supplementary file (Supplementary Figure S1, Table S2). Further details on vector function are given in the work of Weninger et al. [48]. Typically, 2–3 gRNAs per target site were designed using the online tool CRISPOR http://crispor.gi.ucsc.edu/ (accessed on 1 September 2023). The CRISPR target sites (N20 and PAM sequences) are listed in Supplementary Table S3. The donor cassette and pPpT4_pHTX1 assemblies were used to transform electro-competent E. coli Top10F′ cells and were selected on LB agar plates supplemented with 100 µg/mL ampicillin and 50 µg/mL Zeocin^®, respectively.

Electrocompetent P. pastoris cells were prepared according to the method described by Lin-Cereghino (2005) [51]. Prior to the transformation of P. pastoris cells, the donor cassette DNA fragments were excised from the backbone via either SmiI, SmaI, or EcoRV restriction sites previously introduced by PCR primers, followed by gel purification. The competent cells were co-transformed with 300–500 ng of purified donor cassette DNA and 100 ng of Cas9-gRNA DNA fragments. The transformants were plated on YPD agar supplemented with 100 µg/mL Zeocin^® to select for the presence of Cas9-gRNA plasmids. Genomic DNA was isolated from transformants using a rapid protocol [52], and PCRs and Sanger sequencing were performed to confirm successful cassette integration. The Cas9-gRNA plasmids were subsequently eliminated through repeated propagation on non-selective medium to enable further engineering cycles.

2.4. Bioreactor Cultivations

Fed-batch fermentations were carried out in Sartorius Biostat CT+ 5 L bioreactors with an initial volume of 3 L. Overnight cultures were grown at 130 rpm and 28 °C for 48 h in 100 mL baffled shake flasks with 10 mL YPD medium. These overnight cultures were then used to inoculate fermentation seed cultures with 300 mL YPD medium in 2 L baffled shake flasks to an OD₆₀₀ of 0.1. The seed cultures were cultivated at 130 rpm and 28 °C to an OD₆₀₀ of 30. Bioreactors with 2.7 L of basal salt medium (BSM) containing 0.17 g/L CaSO₄ × 2 H₂O, 2.86 g/L K₂SO₄, 0.64 g/L KOH, 2.32 g/L MgSO₄ × 7 H₂O, 4.25 g/L H₃PO₄, 0.22 g/L NaCl, 33 g/L glucose monohydrate, 10 g/L NH₄Cl, 4.35 mL/L Pichia trace metal solution (PTM1), 80 mg/L L-histidine, and 0.1 g/L antifoam 204 (Sigma-Aldrich, Vienna, Austria) were subsequently inoculated to an OD₆₀₀ of 3. The trace metal solution consisted of 5.0 mL/L H₂SO₄ (69%), 5.99 g/L CuSO₄ × 5 H₂O, 1.18 g/L KI, 3 g/L MnSO₄ × H₂O, 0.2 g NaMoO₄ × 2 H₂O, 0.02 g/L H₃BO₃, 0.92 g/L CoCl₂ × 6H₂O, 42.18 g/L ZnSO₄ × 7 H₂O, 65 g/L FeSO₄ × 7 H₂O, and 0.2 g/L biotin. The fermentation temperature was set to 28 °C throughout the entire process and the dissolved oxygen (dO₂) was maintained above 40% saturation with initial agitation and aeration rates of 500 rpm and 1 vvm, respectively; the agitation rate was steadily increased to 800 rpm as oxygen consumption increased. In later stages of the fermentation, the agitation was increased to 1200–1350 rpm to break the foam, which spiked the dO₂ to ~80%. The pH was measured with an autoclavable pH-electrode (EasyFerm Plus PHI K8 120, Hamilton, Bonaduz, Switzerland) and was maintained at pH 5.6 through the automatic addition of 6 M NaOH. Dissolved oxygen was monitored with a dO₂ sensor (InPro6850i/12/120, Mettler Toledo, Vienna, Austria). Foaming was controlled by the addition of 10% antifoam 204 dissolved in 20% ethanol in water (w/v) when needed. All the sensors were calibrated before fermentation, and the addition of solutions (base, antifoam, feeding solutions) and the aeration rate were controlled by a Biostat CT+ controlling unit.

2.5. Fed-Batch Feeding

After an initial batch phase of 14 h, the bioreactors were fed with solution 1 (500 g/L glucose, 12 mL/L PTM1, and 80 mg/L L-Histidine) and solution 2 (250 g/L NH₄Cl) at rates of 2–8 g/L/h and 1–4 g/L/h, respectively, maintaining glucose levels between 1 and 10 g/L. As a result, 183 g/L (203 g/L total) glucose and 26 g/L (36 g/L total) ammonium chloride were fed until the end of the fermentation (134 h feeding period, 148 h total batch and fed batch). The cell wet weight (CWW) was measured by centrifuging one mL of the cell suspension for 1 min in pre-weighed microcentrifuge tubes at 4500× g, followed by removal of the supernatant and determination of the wet cell weight. Aliquots of the supernatant and pellet fractions were subjected to fatty acid extraction and subsequently analyzed by GC-MS or GC-FID.

2.6. Fatty Acid Methyl Ester (FAME) Analysis—Sample Preparation

Free fatty acids were methylated through acid-catalyzed transesterification with methanol and hydrochloric acid, followed by quantification of FAMEs using either GC-MS or GC-FID. For sample preparation, 0.2 mL of vigorously mixed suspension from bioreactor or deep-well plate cultivations was centrifuged for 1 min at 4500× g. The resulting supernatant and pellet (dissolved in 0.2 mL pure H₂O) were transferred into 10 mL Pyrex glass tubes with Teflon-lined plastic caps. These samples were then frozen at −80 °C and lyophilized overnight using an Alpha 1–4 LDplus freeze dryer (Martin Christ, Osterode am Harz, Germany).

To produce the FAMEs, 0.5 mL of (5%) hydrochloric methanol containing 0.1 mg/mL pentadecanoic acid as an internal standard were added to each tube. After vortexing briefly, the tubes were incubated at 85 °C for 1.5 h. Next, 0.5 mL of a 0.8% KCl solution and 1 mL of hexane were added, and the FAMEs were extracted into the hexane layer by shaking at 1500 rpm on a Vibrax^® mixer (Staunfen, Germany). Following a brief centrifugation step, 200 µL of the upper hexane layer were transferred to GC vials with glass inlets for subsequent GC analysis.

2.7. GC Analysis

The FAMEs were quantified using either a Shimadzu GC-2010 Plus device with a Shimadzu GCMS-QP2010 SE detector (GC-MS) or with a flame ionization detector FID-2010 Plus (GC-FID). Both devices were equipped with a ZB-5MSi column (5%-phenyl 95%-dimethylpolysiloxane phase 30 m × 0.25 mm × 0.25 µm) and operated with helium and nitrogen as carrier gases, respectively. For the GC-MS analysis, the temperature program started with an initial hold at 100 °C for 1 min, followed by a ramp to 300 °C with a linear increase of 15 °C/min and a final hold of 1 min. The injection, ion source, and interface temperatures for GC-MS were set to 240 °C, 250 °C, and 300 °C, respectively. For the GC-FID analysis, the temperature program began with an initial 1 min hold at 150 °C, followed by a ramp to 340 °C with a linear increase of 15 °C/min and a final hold of 1 min. Injection and FID temperatures of 240 °C and 320 °C were used for the GC-FID measurements. The flow rates for GC-MS and GC-FID were set to 0.92 mL/min (linear velocity 35.7 cm/s) and 0.81 mL/min (linear velocity 19.6 cm/s), respectively. Both devices utilized an injection volume of 1 µL with split ratios of 30 (GC-MS) and 20 (GC-FID). The described GC methods facilitated rapid sampling but did not enable clear separation of C18:1, C18:2, and C18:3 fatty acids. Consequently, the values for these fatty acids (C18:1, C18:2, and C18:3) were combined into a single peak for most of the analyzed strains. Selected strains were analyzed with an extended GC method that allowed for clear separation of all the peaks (Method S1).

2.8. Glucose and Ethanol Concentration Analysis

The glucose and ethanol concentrations were monitored using HPLC measurements. Fermentation samples were centrifuged for 2 min at 16,000× g, filtered through 0.45 µm syringe filters, and diluted if necessary. An analysis was performed on a Merck-Hitachi LaChrome HPLC System (Merck, Darmstadt, Germany) with a Bio-Rad Aminex HPX-87H column (Bio-Rad, Hercules, CA, USA) (precolumn: Micro-Guard Cation H-Cartridge 125-0129) equipped with a Merck LaChrome L-7490 refractive index detector. The samples were eluted with 5 mM sulfuric acid at a flow rate of 0.6 mL/min and a constant temperature of 65 °C.

3. Results and Discussion

Our primary objective was to develop a P. pastoris strain capable of efficiently producing and secreting free fatty acids, especially palmitoleic acid (POA), using glucose as the carbon source. The efficient secretion of fatty acids is of particular significance, as it streamlines downstream processing, which often constitutes a substantial portion of the production costs in biotechnological applications [53]. To achieve fatty acid secretion, we implemented the following engineering steps: (i) deletion of acyl-CoA synthetases, (ii) overexpression of heterologous thioesterases, (iii) enhancement of precursor supply, and (iv) alleviation of competing pathways in accordance with the push–pull block theorem commonly employed in metabolic engineering [54]. Additionally, we aimed to shift the fatty acid profile towards POA to achieve higher product titers in the supernatant by expressing heterologous fatty acid desaturases and fatty acid synthases in combination with engineering of the endogenous fatty acid elongation system. Figure 1 summarizes the general fatty acid metabolism in P. pastoris and highlights our engineering targets.

3.1. Establishing Free Fatty Acid Secretion in P. pastoris

Fungal fatty acid synthesis de novo is catalyzed by a cytosolic type I FAS complex, typically composed of two polypeptides: FAS1 and FAS2. Acyl-chain elongation occurs iteratively, adding two carbon atoms at a time until the final chain length of 16 to 18 carbon atoms is reached [55]. Each cycle consumes one molecule of malonyl-CoA and two molecules of NADPH. Malonyl-CoA, the primary building block in FA synthesis, is produced through the reaction catalyzed by the cytosolic acetyl-CoA carboxylase (ACC1) from acetyl-CoA [55]. The key activities in these processes are modulated by feedback inhibition from long-chain fatty-acyl-CoAs [56], the major products of the FA synthase complex. The release of FFAs from the acyl-CoA pool, i.e., via heterologous cytosolic thioesterase expression, partially alleviates this feedback inhibition [57]. However acyl-CoA synthetases, encoded by FAA genes, readily counteract this reaction by reactivating released FFAs, thereby preventing overflow metabolism [58]. In FAA null mutants, acyl-CoA pools are continuously depleted through cellular processes, leading to a relief of feedback inhibition on FA synthesis, resulting in an overproduction phenotype that secretes FFAs [59]. To enable this effect in P. pastoris, we performed sequential deletions of the acyl-CoA synthetase genes FAA2 and FAA1, the two FAA variants present in P. pastoris, while simultaneously overexpressing a leaderless E. coli thioesterase ‘TesA [60] from the strong constitutive P_GAP promoter. It was previously shown that a similar engineering approach implemented in S. cerevisiae enabled free fatty acid overproduction, reaching FFA titers of up to 0.67 g/L in shake flask cultivation media [58,61]. In our study, the overexpression of ‘TesA in the ∆faa2 strain background (strain Pp#29) resulted in a slight increase in the intracellular FA content; however, no considerable amounts of FFAs were secreted into the medium (Figure 2). The FFA secretion phenotype was achieved by additionally deleting the FAA1 gene in strain Pp#29 (Δfaa2 ‘TesA), leading to a 2.3-fold increase in the total cellular FA content compared to the wild-type strain (Figure 2). The resulting strain Pp#32 secreted 0.69 g/L of FFAs in the DWP cultivations. These findings are consistent with a recent study performed with P. pastoris GS115 [62], where the ∆faa1 knockout yielded the most significant improvement, while the implementation of ∆faa2 caused a minor but still noticeable increase in the FA content. Notably, strain Pp#32 displayed reduced biomass yield compared to the wild-type strain (CWWs are listed in Supplementary Table S4), which may be attributed to the metabolic burden induced by FFA production. To further enhance FA production, we deleted the fatty acyl-CoA oxidase 1 gene (POX1), encoding the first step in β-oxidation [63], to prevent the degradation of newly synthesized fatty acids. This strategy was successfully implemented in an FFA-producing E. coli strain (deletion of the POX1 homolog FadE), which led to 4-fold increase in productivity [57]. We additionally targeted the fatty acid importer Fat1p, which is responsible for FA uptake and exerts minor acyl-CoA synthetase activity [59,64]. The deletion of FAT1 in an S. cerevisiae FAA quadruple mutant was shown to promote elevated internal FFA accumulation [59]. Surprisingly, both sequential deletions (Strains Pp#35 and Pp#37) did not result in a significant increase in productivity compared to the background strain Pp#32 (Figure 2). The negligible change in productivity facilitated by the POX1 deletion is consistent with findings reported in prior research [62]. Nevertheless, we retained the deletion of POX1 in subsequent strains, as it may hold potential for synergistic interactions with other modifications.

3.2. Engineering the Fatty Acid Desaturase System in P. pastoris

The ∆9-desaturase Ole1p in S. cerevisiae plays a crucial role in converting unsaturated FAs to their monounsaturated counterparts, including the synthesis of palmitoleic acid [65]. In contrast to S. cerevisiae, P. pastoris CBS7435 possesses two distinct ∆9-desaturase isoforms, Ole1-1p and Ole1-2p [47].

Initial experiments indicated that Ole1-1p exhibits major ∆9-desaturase activity and that ∆ole1-1 mutants (wild-type background) were unable to grow without supplementation of exogenous fatty acids. Subsequent attempts to exchange OLE1-1 for a C16-specific ∆9-desaturease from C. elegans (CeFAT) [66] were unsuccessful. In the case of S. cerevisiae, ∆ole1 mutants also require FA supplementation. However, the replacement of OLE1 with a rat ∆9-desaturase yielded viable cells [65], which differs from our observations in P. pastoris. Interestingly, unlike the ∆ole1-1 P. pastoris strains, ∆ole1-2 mutants can thrive without the need for FA supplementation and exhibit modestly increased C16:1 content. Consequently, we retained this modification in all the subsequent strains.

Having established FFA secretion in strain Pp#35 (Figure 2), we set out to assess whether the overexpression of additional ∆9-desaturases could increase the POA content or overall FFA productivity.

Previous studies have demonstrated that overexpression of OLE1 in engineered S. cerevisiae leads to increased fatty alcohol production [54]. Furthermore, increased ∆9-desaturase activity has been associated with lipid accumulation phenotypes in Y. lipolytica and mammalian tissue [67]. This effect is a result of reduced feedback inhibition of the FA biosynthesis machinery by unsaturated fatty-acyl-CoAs [67]. Given that P. pastoris exhibits low palmitoleic acid (C16:1) content and high levels of oleic acid (C18:1) compared to other yeast species (Table 2), we devised a strategy to overexpress a heterologous ∆9-desaturase, SCD3 from M. musculus, known for its high specificity for palmitoyl-CoA [68]. The strong constitutive promoter P_TEF1 (translation elongation factor 1) was used to foster MmSCD3 expression (strain Pp#39). The heterologous expression of MmSCD3 did not enhance the overall fatty acid secretion significantly but it did raise the POA content from 8.3% to 13.1% (Figure 2). Consequently, we selected Pp#39 as the basis strain for further engineering efforts, like the introduction of CeFAT, another palmitoyl-CoA-specific ∆9-desaturase. In contrast to the expression of CeFAT in the basic Δole1-2 background (Pp#16), which exhibited minor changes in the FA content or profile, the overexpression of CeFAT from the glycolytic PGK1 promoter in the Pp#39 strain increased the titer of secreted fatty acids from 0.77 g/L to 0.86 g/L but could not enhance the POA content (Figure 3). Further attempts to overexpress a third heterologous desaturase from S. cerevisiae in later experiments did not result in any improvements in the productivity or POA content (data not shown).

Yeast ∆9-desaturases contain a self-sufficient cytochrome b5 domain that provides electrons for the desaturase complex, transferring electrons from NADH directly to the desaturase via a cytochrome b5 reductase, which facilitates efficient coupling of the cofactor supply and double bond formation [65]. However, this domain is not present in higher eukaryotic counterparts, so we hypothesized that electron transfer might be limiting in our engineered strains. To this end, we introduced additional copies of cytochrome b5 and NADH-dependent cytochrome b5 reductase genes, either as endogenous variants (Pp#127) or codon-optimized genes from M. musculus (Pp#129), linked to the MmSCD3 coding strand via T2A sequences [69]. Unlike a previous study with S. cerevisiae overexpressing FA desaturases from Kluyveromyces lactis [70], the CYB5 overexpression proved to be unsuccessful in P. pastoris (Pp#39) overexpressing MmSCD3 (data not shown).

Aside from ∆9-desaturases, P. pastoris possesses ∆12- and ∆15-desaturase activities, as confirmed by a previous analysis of the wild-type fatty acid profile (Table 2), which revealed the presence of linoleic acid (C18:2) and linolenic acid (C18:3). The responsible desaturases, encoded by FAD12 and FAD15 (FAD-3), were previously identified in P. pastoris GS115 and characterized through heterologous expression in S. cerevisiae [71,72]. Similarly, we confirmed Fad12p as the sole ∆12-desaturase in the P. pastoris CBS7435 wild-type strain [71], as evidenced by the absence of C18:2 in a FAD12 null mutant (Supplementary Figure S2). To explore the impact of ∆12- and ∆15-desaturases on the FA profile and secretion, we individually deleted FAD12 (strain Pp#150) and FAD15 (strain Pp#140), in our platform strain Pp#39. Surprisingly, these deletions increased the FA secretion by 19% and 24%, respectively, and led to modest increases in the POA content (Figure 3). Notably, the biomass yield for strain Pp#150 was compromised (Supplementary Table S4). We hypothesize that the increased productivity induced by the deletions of FAD12 or FAD15 is caused by a compensation effect, where a lack of membrane fluidity arising from a deficiency or reduction in PUFA is bolstered by enhanced monounsaturated acid synthesis. This should lead to a reduction in saturated acyl-CoAs, which otherwise inhibit FA synthesis. Indeed, we observed a marginal reduction in saturated fatty acids in these strains (Figure 3).

3.3. Analysis of Growth Conditions and Nutrient Manipulation

In addition to genetic factors, the fatty acid composition and secretory capacity of various yeast species are influenced by nutrient availability. Nutrient deficiencies, such as nitrogen or phosphate limitations, coupled with high carbon abundance have been shown to trigger lipid accumulation in oleaginous yeast and certain conventional yeast species [73,74]. Furthermore, oxidative stress can induce changes in the unsaturated fatty acid content as a protective response against ROS damage [75]. Although several biotechnologically relevant yeast species have been investigated for POA synthesis under various cultivation conditions, P. pastoris was not among the organisms studied [74]. Consequently, we conducted experiments to assess the FA production capacity and fatty acid composition of strain Pp#39 under different carbon-to-nitrogen ratios and nitrogen sources, using glucose as the carbon source, in batch cultivation format (96-DWPs; media compositions are given in Supplementary Table S5). We found that among the different N-sources (ammonium chloride, ammonium sulfate, yeast extract, and peptone), peptone yielded the highest FA secretory capacity (up to 2-fold higher than most other N-sources) and that the peptone or yeast extract led to the highest POA content of up to 16–18%, whereas other N-sources yielded 12–14% POA (Supplementary Figure S3). Notably, the use of peptone or yeast extract in the large-scale production of comparably low-priced compounds is not feasible. Hence, we chose ammonium chloride as the nitrogen source for further experiments. When considering nitrogen limitation, CN ratios ranging from 5 to 50 led to an FA secretory capacity that was similar to the value obtained for BMD11, whereas CN ratios above 50 resulted in compromised FA secretion, probably due to the reduced biomass yields as a consequence of low nitrogen availability. Interestingly, the increased CN ratios with ammonium sulfate and peptone as a nitrogen source led to a relatively increased C16:1 content, yet the overall FA titers were low. The minor impact of the CN ratio on lipid accumulation in P. pastoris may be attributed to the fact that P. pastoris does not possess ATP-citrate lyase [76] or cytosolic malic enzymes, which catalyze key reactions required for lipid accumulation in oleaginous yeasts [77]. We decided to set the CN ratio for subsequent experiments to ~25 to balance the biomass yield and fatty acid production. Lastly, it is important to note that these findings simulate batch cultivations and that optimal parameters for fed batch and continuous formats in bioreactors may differ from these conditions, as constant CN rations can only be consistently maintained in such controlled feeding setups.

3.4. Down-Regulation of Competing Pathways

Further strain construction efforts involved the elimination of pathways in strain Pp#39 (or Pp#35 for the interference of neutral lipid storage) that consume fatty acyl-CoAs (see Figure 1). In previous studies, deleting the peroxisomal long-chain fatty acid importer PXA1 in an S. cerevisiae ∆pox1 ∆faa1 strain resulted in increased FFA production by completely abolishing β-oxidation [78]. Pursuing the same strategy, strain Pp#135 improved FA secretion by 24% but, astonishingly, displayed slightly reduced POA content (Figure 3). The exact reason for this reduction remains unclear, as FA transport across membranes is still—at least partially—obscure [79].

The reduction of FA flux to neutral lipids was achieved by deleting the major genes involved in neutral lipid synthesis, DGA1 and LRO1 (diacylglycerol acyl-transferases), as well as ARE2, the sterol acyl-transferase, which are responsible for catalyzing acyl transfer to diacylglycerides and sterols, respectively. Individual deletions of LRO1 and DGA1, respectively, led to negligible changes in FFA secretion (increase from 0.74 g/L to 0.75 g/L and 0.76 g/L secreted FFA, respectively), but were accompanied by a reduction in stored neutral lipids (Supplementary Figure S4). The combinatorial strain Pp#51a (∆dga1 ∆lro1∆are2) showed impaired growth and a significant reduction in productivity, suggesting a crucial role of neutral lipid storage for strain fitness. Thin layer chromatography conducted with lipid extracts from strains Pp#35, Pp#44, Pp#46, and Pp#51a confirmed the reduction or absence of triglycerides and sterol esters in the respective strains (Supplementary Figure S5). In later experiments, we pursued an alternative strategy that involved overexpression of DGA1 and triacylglycerol lipases 3 and 4 (TGL3, TGL4) to divert FA flux through lipid bodies (see Section 3.6).

3.5. Expression of Heterologous Thioesterases

As previously mentioned, thioesterases play a crucial role in FFA biosynthesis, governing, among other reactions, the profile of released FFAs. Substrate specificities vary among thioesterases, with certain variants accepting a broad range of long chain acyl-CoAs, while others only accept a few or are highly specific towards single activated FA species [80]. Plenty of length-specific enzymes have been identified in plants, where thioesterases terminate plastidal fatty acid synthesis, rendering the fatty acid composition of plant kernel fats [81,82].

To increase the POA content in platform strain Pp#39, we overexpressed two plant thioesterases, namely, MtFAT-A from M. tetraphylla and AtFAT-B from A. thaliana, which have been shown to exhibit high specificity for C16:1 and C16:0 [81,82]. The FAT genes were integrated into the HIS4 locus, replacing ‘TesA to verify the individual thioesterase activity. The N-terminal leader peptides for translocation to plastids in their original hosts were omitted in the P. pastoris expression constructs to allow for cytosolic expression. Surprisingly, the substitution of ‘TesA with MtFAT-A or AtFAT-B resulted in a reduction in the FA titers by approximately 42% and led to alterations in the fatty acid profile (Figure 3). The resulting strains mirrored the FA composition and productivity of strain Pp#53 with a ‘TesA knockout in the same background. This indicates that the plant thioesterases were probably not active in this context and the changes in FA content were solely due to the ‘TesA removal. The poor activity of the plant thioesterases may be attributed to insufficient protein processing or low substrate acceptance. The latter may be particularly true since plastidal thioesterases hydrolyze acyl-ACP thioesters in their original host, while fatty acyl chains in yeast are attached to CoA. However, previous studies have shown that thioesterases from different origins can accept both substrate types [83]. Despite the absence of heterologous thioesterase activity in strain Pp#53, it still secreted 0.45 g/L of FFAs, with marginally elevated POA content and reduced C16:0 levels, potentially due to endogenous TE or lipase activities (Figure 3). A sequence similarity search [84] (BLAST) for thioesterases in the genome of P. pastoris CBS7435 only identified a homolog of the S. cerevisiae peroxisomal thioesterase, which is likely involved in β-oxidation [85]. Consequently, the release of FFAs in this strain is presumably a result of lipase activity.

3.6. Enhancing the Availablity of Fatty Acid Precursor Molecules

To push the flux towards FA production, we aimed to increase the supply of the FA precursors acetyl-CoA and malonyl-CoA (Figure 4). It is commonly reported that the carboxylation of acetyl-CoA to malonyl-CoA, catalyzed by the acetyl-CoA carboxylase (Acc1p), poses a major bottleneck for FA synthesis in yeast [54,86]. Acc1p activity is regulated at the transcriptional and post-translational levels. Previous studies of S. cerevisiae demonstrated that the overexpression of ACC1, along with mutagenesis of SNF1 regulation sites, led to a 3-fold increase in Acc1p activity [86]. Snf1p phosphorylation sites, identified as Serine 659 and Serine 1157 in Acc1p of S. cerevisiae, were conserved in the P. pastoris variant as Serine 693 (LRTPSPGKL) and Serine 1151 (MDRAVSVSDL). To boost the Acc1p activity in Pp#39, we introduced the S1151A substitution within the genomic ACC1 gene and replaced its native promoter region with the constitutive P_TEF1 promoter. Individual modifications, like promoter exchange PpP_ACC1::P_TEF-ACC1 (strain Pp#49) or deletion of the phosphorylation site PpACC1::ACC1^S1151A (ACC1*; strain Pp#50), improved productivity by 11% and 6%, respectively, whereas the combination of both strategies increased the titers by 36% (Figure 5). Interestingly, strain Pp#50 (ACC1^S1151A) displayed slightly elevated POA content, whilst strain Pp#49 (PpP_ACC1::P_TEF-ACC1) produced fewer C16 fatty acids. Early studies on fungal fatty acid synthases revealed that increased Acc1p activity promotes C18 synthesis by the FAS complex, as higher malonyl-CoA concentrations facilitate efficient malonyl loading rather than chain termination [87]. This pattern aligns with the results observed for strain Pp#49 but does not explain the C16:1 increase in strain Pp#50.

After enhancing the Acc1p activity, we aimed to increase the cytosolic acetyl-CoA abundance. In most Saccharomycotina, including S. cerevisiae and P. pastoris, cytosolic acetyl-CoA is primarily derived from pyruvate, which undergoes conversions to acetaldehyde and acetate before forming acetyl-CoA [76]. In contrast, oleaginous yeasts adopt a different approach for cytosolic acetyl-CoA synthesis. They redirect the carbon flux through the ATP citrate lyase (ACL) pathway by exporting citrate from the mitochondria and converting it to acetyl-CoA and oxaloacetate. This metabolic rearrangement efficiently diverts the carbon flux from the TCA cycle to FA synthesis in oleaginous yeasts [88]. P. pastoris lacks endogenous genes for ACL and cytosolic malic enzyme (ME; present only in mitochondria). To overcome this limitation, we integrated a heterologous ACL shuttle construct in strain Pp#50 composed of R. toruloides ACL, M. musculus malic enzyme (ME), endogenous malate dehydrogenase (MDH3), and a mitochondrial citrate exporter (CTP1). While this strategy was previously successful in an FA-producing S. cerevisiae strain [61], in strain Pp#50, it resulted in reduced FFA titers (from 0.82 g/L to 0.53 g/L) and lower biomass yields (Figure 5), suggesting a metabolic imbalance likely caused by the withdrawal of citrate from the TCA cycle or an imbalance at the pyruvate node. However, interestingly, recent research has demonstrated that overexpression of a single ACL gene (without ME) could enhance FFA production from methanol by 23% in an engineered P. pastoris strain [62]. Building on this knowledge, we introduced the R. toruloides ACL (RtACL) into the Pp#50 background, which raised the secreted FFA titers from 0.82 g/L to 1 g/L (Figure 5).

In oleaginous yeasts, nitrogen depletion triggers the redirection of the citrate flux from the TCA cycle to the ACL shunt, caused by the assimilation of adenosine monophosphate and the subsequent inactivation of isocitrate dehydrogenase (Idhp) [77]. To mimic this effect in P. pastoris, we aimed to replace the endogenous IDH1 and IDH2 genes with the variants from R. toruloides. Notably, the gene replacements were not successful, and we identified that the heterologous IDH genes were randomly integrated in the genome, suggesting an essential role of the endogenous IDH variants. The resulting strain displayed reduced fatty acid titers (Figure 5, Pp#153), presumably due to the perturbation of the TCA cycle.

We then explored another strategy for cytosolic acetyl-CoA synthesis, involving the expression of a phosphoketolase (PK) and phosphotransacetylase (PTA) shunt. This pathway splits fructose-6-phosphate into erythrose-4-phosphate and acetyl-phosphate, followed by the synthesis of acetyl-CoA [89] and was successfully utilized to raise the FFA production from 0.6 g/L to 0.7 g/L in a modified P. pastoris strain [62]. The co-expression of B. breve PK and C. kluyveri PTA in P. pastoris strain Pp#50 increased the FA production by 15%, yielding 0.94 g/L FFAs in Pp#145 (Figure 5).

Next, we aimed to divert the FA flux through the lipid bodies by overexpressing endogenous diacylglycerol acyltransferase DGA1 and triacylglycerol lipases TGL3 and TGL4 to increase the neutral lipid synthesis, leading to continuous fatty acid release from stored TAGs. Indeed, channeling the fatty acids through the lipid bodies increased the secreted fatty acid titers from 0.82 g/L (Pp#50) to 1.16 g/L (Pp#141), representing a 42% increase.

Subsequently, we focused on increasing NADPH availability for fatty acid synthesis. In yeast, NADPH is generated in the pentose phosphate pathway (PPP) [90] during the oxidation of acetaldehyde to acetate via an NADP+-dependent aldehyde dehydrogenase [54] or, if present, in the reaction of a cytosolic malic enzyme [88]. Since incorporating the latter reaction in an ACL shuttle reduced productivity in our engineered strains, we attempted to increase the flux through the PPP. Glucose-6-phosphate dehydrogenase (Zwf1p) and 6-phosphogluconolactonase (Sol3p), two enzymes that were previously shown to limit PPP flux in P. pastoris [91], were overexpressed from the constitutive bidirectional histone promoter P_HHX1 [92]. These modifications raised the fatty acid titers from 0.82 g/L in Pp#50 to 0.89 g/L in Pp#112 (Figure 5). Since yeast inherently exhibits high flux through glycolysis, we also tested the expression of an NADP+-dependent glyceraldehyde-3-phosphate dehydrogenase (CaGDH), aiming to reduce glycolytic NADH synthesis and instead produce NADPH, which had been successfully conducted in Y. lipolytica [90]. However, pursuing this approach in strain Pp#50 did not result in noticeable changes in FFA productivity (see strain Pp#113, Figure 5).

3.7. Expression of Heterologous Fatty Acid Synthases

Our previous engineering efforts increased the POA content from 8.3% to 12.0% and the total C16 content in the culture supernatant from 17.8% to 38.6% in strain Pp#39. However, these levels were still comparably low, particularly considering that other yeast species exhibit POA contents of 30–70% (Table 2). Aside from the thioesterase and desaturase activities, the fatty acid synthase (FAS) itself has a considerable impact on the fatty acid composition. In particular, the ketoacyl synthase, malonyl/palmitoyl-transferase (or malonyl acetyl transferase) domains, and attached thioesterase domains in mammalian FAS govern the final acyl-chain length [93]; thus, we anticipated that the endogenous P. pastoris FAS contributed to the high C18 content in our strains.

It has previously been shown that the co-expression of an Actinomyces type I FAS from B. ammoniagenes with a wax ester synthase in S. cerevisiae increased fatty acid ethyl ester production by 6.3-fold [94]. In the same study, bacterial type I FAS expression could rescue growth defects in a S. cerevisiae fas1 mutant and, astonishingly, the resulting strain displayed 2.7-fold higher palmitoleic acid titers as compared to the S. cerevisiae FAS wild-type. In a separate study, the introduction of human FAS was also shown to remedy the growth defect of a S. cerevisiae fas2 mutant—knockout of either FAS1 or FAS2 abolishes fatty acid synthesis [95]. Mammalian FAS offer the advantage of a relatively open structure that allows efficient access of thioesterases to the growing acyl-chain. Additionally, the human FAS possesses an attached thioesterase domain with high specificity for C16 fatty acyl ACPs [95]. Since strain Pp#39 predominantly produced C18 fatty acids, we attempted to replace the endogenous FAS complex with the FAS genes from B. ammoniagenes (BaFAS) or H. sapiens (hFAS), respectively. The required phosphopantetheine transferase proteins BaPPT and EcAcpS were co-expressed. However, attempting a complete replacement of P. pastoris FAS1 in our background resulted in lethality. This suggests that the metabolic burden induced by producing FFAs, combined with a substantial alteration in the FA profile, impeded cell growth. Interestingly, co-expression of individual BaFAS or hFAS and respective PPT (strains Pp#67 and Pp#91) with the intact endogenous PpFAS system increased the FFA titers by 33% and 24% (Figure 6), respectively, although the POA contents in the secreted fraction were only elevated by 2.3% and 1.7%, respectively. These findings align with results from Eriksen et al. [94], where alterations in the FA profile by heterologous FAS expression were not as evident when the endogenous FAS was active. The results furthermore suggest that the endogenous FAS or fatty acid elongases efficiently elongate C16 fatty acids provided by the heterologous FAS variants, which was indicated by the appearance of vaccenic acid mass fragments in GC-MS.

Fungal FAS belong to the cytosolic type I FAS complex and are typically composed of two polypeptides, Fas1p and Fas2p. The structural organization of the P. pastoris FAS complex has been elucidated recently, and several distinctions to the S. cerevisiae FAS in both the structural and enzymatic domains were identified [96].

Consequently, we explored whether the expression of the heterologous fungal FAS from S. cerevisiae, which primarily produces C16 and C18 acyl-CoAs [87], could replace endogenous FAS activity and how this approach influences the FA profile of P. pastoris, as previous reports have indicated that S. cerevisiae exhibits higher C16 and POA contents than P. pastoris (see Table 2). The seamless replacement of PpFAS1 with an ScFAS1–ScFAS2 expression cassette driven by the bidirectional promoter P_HTX1 in strain Pp#39 proved to be successful, generating strain Pp#85, which showed elevated POA and total C16 contents of 23% and 55%, respectively (Figure 6). The subsequent replacement of PpFAS2 by a HIS4 expression cassette in strain Pp#85 did not lead to further changes in the FA titer or POA content (strain Pp#115). Notably, strain Pp#85 exhibited reduced growth and compromised productivity, releasing only 0.35 g/L of FFAs into the supernatant, representing a 56% reduction compared to the parental strain. Chromosomal integration of the ScFAS1–ScFAS2 expression cassette in the TEFup locus with the native PpFAS intact (Pp#39 background) yielded strain Pp#89, which exhibited an identical fatty acid composition as strain Pp#85 but showed modestly increased FFA titers of 0.44 g/L, which is still 43% lower compared to the parental strain Pp#39. Importantly, Pp#89 displayed compromised growth, signifying cellular stress induced by alterations in the FA profile, particularly due to the expression of ScFAS.

3.8. Engineering the Endogenous Fatty Acid Elongase System

Following the substitution of the native P. pastoris FAS, the POA content still remained below 25%, and we hypothesized that the endogenous elongase system was likely responsible for extending the POA to form vaccenic acid (C18:1-Z11), resulting in diminished POA content and elevated C18 titers. Differentiating between vaccenic acid and oleic acid proved to be challenging due to their high structural similarities, and the presence of vaccenic acid could only be tentatively confirmed through mixed mass spectra. To unambiguously validate our hypothesis, we undertook the deletion of the FA elongases responsible for the extension of POA.

In S. cerevisiae, FA elongation is mediated by elongases Elo1p, Elo2p, and Elo3p [55]. While Elo1p is responsible for elongation of medium-to-long-chain fatty acids like C16:1 [97,98], the other elongases (Elo2p and Elo3p) are involved in extending fatty acids from C16 and C18 up to C22 and C26 for sphingolipid biosynthesis [99]. Consequently, all three isoforms presented viable deletion targets. A homology search in P. pastoris CBS7435 strain revealed three putative elongases: Elo2p (CAH2449582.1; 66% identity to ScElo2p, 50% identity ScElo1p, 48% identity to ScElo3p), Elo3p (CAH2449986.1; 58% identity to ScElo3p, 53% identity to ScElo2p, 47% identity to ScElo1p), and Elo100p (CAH2449019.1; 13–14% identity and 20–23% similarity to all S. cerevisiae elongase isoforms), which were previously annotated by Sturmberger et al. [46]. Since we could not unambiguously allocate the different isoforms based on homology, we decided to delete all three putative elongase genes in parallel in strain Pp#39. The deletions of ELO3 and ELO100 resulted in a modest increase in the POA content but, interestingly, the resulting strains displayed increased FA secretion (improvement of 35% and 30%, respectively; Figure 6: strains Pp#138 and Pp#136). Unfortunately, our attempts to delete ELO2, the isoform with the highest homology to ScElo2p, were unsuccessful, and subsequent efforts to obtain the deletion through supplementation with very long-chain fatty acids (VLCFA) to compensate for a possible lack in VLCFA also proved fruitless. It is important to note that the engineered strains exhibited low acyl-CoA synthase activity due to the deletions of the FAA1 and FAA2 genes, which hinders the incorporation of FAs into the metabolism. However, some degree of VLFA activation to CoA thioesters should be facilitated by Fat1p, the very long-chain fatty acid transport protein [59,100], which was still present in strain Pp#39.

3.9. Evaluation of FA Production of Strain Pp#85 in Bioreactor Cultivations

After implementing various strategies to enhance the FA secretion and increase the POA content, we conducted an in-depth analysis of the strain Pp#85 exhibiting the highest relative POA content in a glucose-fed batch fermentation. Throughout the fermentation process, we adopted different CN ratios increasing from CN5 to CN28 (Figure 7d). This strategy was designed to promote cell growth in the earlier stages of fermentation and enhance FFA production towards the later stages. After 148 h of cultivation, we obtained 0.37 g/L of free POA (~19% of total fatty acids) in the supernatant (0.61 g/L in total—pellet and supernatant), while 210 g/L of glucose was consumed. The total FFA titer reached 1.8 g/L (Figure 7a,b), and the cell wet weight reached 70 g/L, representing an approximately 5-fold increase in productivity (4.9-fold for POA) and biomass compared to the DWP cultivations. Specific productivity was not improved in the process. As the fermentation progressed, we observed a reduction in the total C16 content from 50% at 24 h to 40% at 148 h, while the C18:1 content increased over time. This observation aligns with previous studies that demonstrated that extended cultivations of engineered S. cerevisiae strains promote C18:1 synthesis, potentially due to the upregulation of FA elongases [61]. HPLC measurements of the fermentation supernatant revealed the excretion of lactate and ethanol during the initial growth phase (Figure 7c). These results suggest a potential imbalance in NAD+ cofactor recycling, especially during the initial growth phase. The excessive NADH generated in glycolysis is consumed by lactate synthesis to maintain physiological NAD+ concentrations. This imbalance could compromise the fatty acid yield, as lactate synthesis consumes pyruvate, an essential precursor for cytosolic acetyl-CoA synthesis.

3.10. Evaluation of Metabolic Engineering Strategies for POA Production and Secretion

Metabolic engineering studies have been focusing on either FA secretion, producing a wider FA spectrum [62,101,102], or the intrinsic production of POA [103,104], yet there are no studies specifically aiming at increasing POA secretion in recombinant microbes. To fill this knowledge gap, we have undertaken a comprehensive metabolic engineering approach to harness the potential of P. pastoris as a platform for the secretory production of POA. The summarized results (Figure 8) provide a valuable overview of the high-impact targets that promote FA secretion and profile alterations in this yeast.

Relevant improvements in FA secretion were facilitated through increased expression of ∆9-desaturases, enhancing precursor-supplying reactions, and manipulating the “lipid body bypass” pathway, among others. Plenty of these findings align with commonly employed engineering targets in the literature [61,102]. Regarding the engineering of P. pastoris’ FA profile, the expression of ∆-9 desaturases, human or bacterial type I FA synthases, and the deletion of ∆12- and ∆15-desaturases, as well as FA elongases, contributed to the improved POA content.

Among the various strategies employed, three approaches stood out in enhancing the overall FA secretion or elevating the POA content: the deletions of fatty acyl-CoA synthetases, the overexpression of the leaderless E. coli thioesterase, and the expression of S. cerevisiae FA synthase. The former strategies increased the FA synthesis by 130% and 70%, respectively, while the latter elevated the POA content from 12% to 22% of the total FA. There is a consensus in the literature that deletions of fatty acyl-CoA synthetases and expression of thioesterases are among the most effective strategies to increase FA synthesis [61,62,102].

Our observations, however, indicate that especially these high-impact modifications, which led to increased FA secretion and profound alterations in the intrinsic FA profile, coincided with reductions in the growth rate and final cell wet weight (CWW). This is underscored by the fed-batch bioreactor cultivation of strain Pp#85, heavily modified in the FA profile, which yielded only 1.8 g/L FFA and a final biomass of 70 g/L CWW. This is not surprising, as the FA composition has considerable impact on membrane fluidity, homeostasis, and various other cellular processes. We recognize that beyond a certain threshold, extensive modifications to the FA profile can rapidly lead to diminished cellular fitness.

To achieve significant changes in the product profile while maintaining cell viability, the focus should therefore shift to enhancing the specificity of the terminal reactions, i.e., thioesterases or lipases [59,80], which release FAs from acyl-CoA pools or neutral lipids. In this context, exploring a wide array of different thioesterases and lipases may be beneficial, allowing for identification of ideal gene candidates that can selectively release specific FAs. In this way, the intrinsic FA composition may largely remain intact. Thus, the secretion of FAs into the growth medium not only simplifies downstream processing but also holds immense potential for tailoring FA profiles.

Table 2. FA composition of wild-type and engineered yeasts producing elevated levels of palmitoleic acid according to the literature. Total FA titers are usually given as g/L FFAs. Notably, in most studies, internally bound FAs (e.g., in triglycerides or phospholipids) are included in the free FA titer according to the extraction method applied; however, downstream processing of internally stored FAs requires additional unit operations. We also list examples where POA is exclusively produced intracellularly. SF, shake flask; FB, fed-batch reactor cultivation; N-lim., nitrogen-limited cultivation; MM, minimal medium; POA, palmitoleic acid; PA, palmitic acid; αLEA, alpha linolenic acid; LA, linoleic acid; OA, oleic acid (or vaccenic acid for P. pastoris strains); SA, stearic acid; other FA, other FAs (C17:1 and C17:0 for our strains); *, alpha linolenic acid content included in linoleic acid content (due to lack of separation); **, g POA/g dry weight; (s), FFAs in supernatant; (t), total FAs (this includes FFAs in supernatant and covalently linked FAs in the pellet fraction; (i), intracellular production.

Strains (Wild-Type)	Conditions	Fatty Acids (%)							POA (g/L)	FA (g/L)	Ref.
		C16:1 POA	C16:0 PA	C18:3 αLEA	C18:2 LA	C18:1 OA	C18:0 SA	other FA
P. pastoris CBS7435	YPD, SF, (72 h)	5.5	5.1	10.9	29.5	45.7	0.8	2.5	0.05	1 (i)	this study
Kluyveromyces polysporus DBM 2171	N-lim. MM, SF (n/d~48–96 h est.)	74.5	7.2	0	0.9	17.1	0.3	0	0.16 g/g DW **	n/d	[74]
S. cerevisiae DBM 2115	N-lim. MM, SF (n/d~48–96 h est.)	59.8	11.5	0	2	24	2.7	0	0.08 g/g DW **	n/d	[74]
Strains (engineered)	Conditions	C16:1 POA	C16:0 PA	C18:3 αLEA	C18:2 LA	C18:1 OA	C18:0 SA	other FA	POA (g/L)	FA (g/L)	Ref.
P. pastoris Pp#85	N-lim. MM, FB (150 h)	18.7	27.9	*	6.3	40.7	3.4	3.0	0.37(s) 0.61 (t)	1.8 (s) 3.3 (t)	this study
P. pastoris PC124H	N-lim. MM, FB (220 h)	9	42	n/d	27	17	4	n/d	2.1	23.4 (t)	[62]
S. cerevisiae MK	N-lim. MM, FB (144 h)	57.5	9.4	0	0	23	2	8.1	6.56	11.4 (i)	[103]
S. cerevisiae Y & Z055E (CEN.PK113-5D)	N-lim. MM, FB (~250 h)	36.5	32.9	0.0	0.0	18.7	6.3	5.4	9.1	25 (t)	[101]
Scheffersomyces segobiensis SS-12	MM, FB (~185 h)	24.9	9.6	0.0	4.0	58.1	0.0	3.4	7.3	29.6 (i)	[104]
E. coli SBF50	N-lim. MM, FB (~55 h)	30.3	14.0	0.0	0.0	16.1	0.0	39.5	10	33.6 (s)	[18]

Table 2. FA composition of wild-type and engineered yeasts producing elevated levels of palmitoleic acid according to the literature. Total FA titers are usually given as g/L FFAs. Notably, in most studies, internally bound FAs (e.g., in triglycerides or phospholipids) are included in the free FA titer according to the extraction method applied; however, downstream processing of internally stored FAs requires additional unit operations. We also list examples where POA is exclusively produced intracellularly. SF, shake flask; FB, fed-batch reactor cultivation; N-lim., nitrogen-limited cultivation; MM, minimal medium; POA, palmitoleic acid; PA, palmitic acid; αLEA, alpha linolenic acid; LA, linoleic acid; OA, oleic acid (or vaccenic acid for P. pastoris strains); SA, stearic acid; other FA, other FAs (C17:1 and C17:0 for our strains); *, alpha linolenic acid content included in linoleic acid content (due to lack of separation); **, g POA/g dry weight; (s), FFAs in supernatant; (t), total FAs (this includes FFAs in supernatant and covalently linked FAs in the pellet fraction; (i), intracellular production.

Strains (Wild-Type)	Conditions	Fatty Acids (%)							POA (g/L)	FA (g/L)	Ref.
		C16:1 POA	C16:0 PA	C18:3 αLEA	C18:2 LA	C18:1 OA	C18:0 SA	other FA
P. pastoris CBS7435	YPD, SF, (72 h)	5.5	5.1	10.9	29.5	45.7	0.8	2.5	0.05	1 (i)	this study
Kluyveromyces polysporus DBM 2171	N-lim. MM, SF (n/d~48–96 h est.)	74.5	7.2	0	0.9	17.1	0.3	0	0.16 g/g DW **	n/d	[74]
S. cerevisiae DBM 2115	N-lim. MM, SF (n/d~48–96 h est.)	59.8	11.5	0	2	24	2.7	0	0.08 g/g DW **	n/d	[74]
Strains (engineered)	Conditions	C16:1 POA	C16:0 PA	C18:3 αLEA	C18:2 LA	C18:1 OA	C18:0 SA	other FA	POA (g/L)	FA (g/L)	Ref.
P. pastoris Pp#85	N-lim. MM, FB (150 h)	18.7	27.9	*	6.3	40.7	3.4	3.0	0.37(s) 0.61 (t)	1.8 (s) 3.3 (t)	this study
P. pastoris PC124H	N-lim. MM, FB (220 h)	9	42	n/d	27	17	4	n/d	2.1	23.4 (t)	[62]
S. cerevisiae MK	N-lim. MM, FB (144 h)	57.5	9.4	0	0	23	2	8.1	6.56	11.4 (i)	[103]
S. cerevisiae Y & Z055E (CEN.PK113-5D)	N-lim. MM, FB (~250 h)	36.5	32.9	0.0	0.0	18.7	6.3	5.4	9.1	25 (t)	[101]
Scheffersomyces segobiensis SS-12	MM, FB (~185 h)	24.9	9.6	0.0	4.0	58.1	0.0	3.4	7.3	29.6 (i)	[104]
E. coli SBF50	N-lim. MM, FB (~55 h)	30.3	14.0	0.0	0.0	16.1	0.0	39.5	10	33.6 (s)	[18]

Even though the total FFA secretion and relative POA content in our engineered strain could not match the highest reported values in the literature, amounting to 10–30 g/L FFA with POA contents of 20–60% (Table 2), we were still able to raise the POA content from 5.5% in the wild-type strain to above 20%, which is a 4-fold improvement. Moreover, we established FA secretion in P. pastoris, commonly generating FFA titers of 1 g/L in shake flask cultivation media. While it appears that host systems like S. cerevisiae, Kluyveromyces polysporus, and Scheffersomyces segobiensis should be favored for POA production due to their inherently high POA content or lipid accumulating phenotype, P. pastoris possesses unique advantages over these host systems. These advantages include P. pastoris’ ability to attain high cell densities in fermentations, the efficient expression of recombinant proteins, and the absence of the Crabtree effect [39,43]. These characteristics enable the generation of considerable numbers of whole-cell biocatalysts and facilitate rapid flux through heterologous pathways, while minimizing carbon loss and supporting the efficient expression of terminal enzymes for various biotechnological applications.

4. Conclusions

While our study primarily focused on enhancing POA secretion, the principles and targets identified herein can be readily adapted for the production of other long-chain FAs, i.e., oleic acid or PUFAs or tailored FA profiles, such as those found in palm oil, palm kernel oil, or cocoa butter. This research opens new avenues for the microbial production of valuable FAs with the potential to enable the synthesis of novel FA-derived products while reducing the environmental footprint associated with traditional production methods. As the demand for sustainable alternatives continues to grow, the development of efficient FA secretion systems represents a significant step forward in the field of metabolic engineering.