Pichia/Komagataella Research Notes

Creating this thread to post potentially useful info about Pichia pastoris/Komagataella phaffii that people come across.

Quick background: Pichia pastoris (now named Komagataella phaffii) is a yeast capable of eating methanol as its sole carbon source (methylotrophy); that can grow to incredible densities in oxygenated bioreactors (~700 g/L wet cell mass), thanks to 'non-Crabtree' metabolism that does not produce ethanol as a waste product; and that can secrete large amounts of recombinant proteins. Pichia is a very promising platform organism for Friendzymes' frugal protein manufacturing, if we can obtain/engineer open, IP-free strains.

Notes from Ahmad et al's review paper on recombinant protein production in Pichia:

There are several methanol-inducible promoters which appear to be off-patent: AOX1, DAS, and FLD1 in particular.
There are a number of good constitutive (always-on) promoters, including GAP, TEF1, PGK1, and GCW14. There are also engineered variants of these that have even higher expression levels.
For selecting and inserting many different gene cassettes, it's useful to have multiple auxotrophic markers. These markers include: -- ARG1, ARG2, ARG3, and ARG4 (arginine auxotrophy) -- HIS1, HIS2, HIS4, HIS5, and HIS6 (histidine auxotrophy) -- ADE1 (adenine/adenosine auxotrophy) -- URA3 and URA5 (uracil auxotrophy) -- MET2 (methionine auxotrophy) -- TYR1 (tyrosine auxotrophy) -- LYS2 (lysine auxotrophy) -- PRO3 (proline auxotrophy) -- GUT1 (glycerol utilization auxotrophy)
Secretion tags like alpha mating factor get cleaved off in Pichia by the membrane protease Kex2. Overexpressing Kex2 can increase protein secretion.
Co-overexpressing the HAC1 transcription factor with the adenosine A2 receptor can increase processing of the alpha mating factor signal sequence, and therefore secretion.
Deleting sections of alpha mating factor signal peptide (residues 57-70) increases secretion for some enzymes.
Knocking out VPS8, VPS10, and VPS13 (vacuolar sorting proteins) can increase secretion.
Overexpressing protein disulfide isomerase (PDI) can help with expression of proteins that have a bunch of disulfide bonds.
To overcome bottlenecks in protein secretion/folding, it can help to overexpress folding helper proteins BiP/Kar2p, DnaJ, PDI, PPIs and Ero1p, or overexpress HAC1. However, sometimes this doesn't work or actually reduces protein secretion.
Secretion tags, and proteins they efficiently secrete: -- KILM1 and CM cellulase -- Pp Pir1 and human a1-antitrypsin -- SUC2 and alpha-amylase, alpha-1-antitrypsin, and interferon -- PHO1 and porcine pepsinogen
Pichia Genome databases/browsers: -- http://bioinformatics.psb.ugent.be/orcae/overview/Picpa -- http://www.pichiagenome.org/
Widely used strains: X-33 prototrophic strain, GS115 ∆his4 auxotroph, KM71 and KM71H ∆aox1 strains, SMD1168 and SMD1168H protease-deficient strains, PichiaPink(TM) ∆ade2 auxotroph. But these are all patented or restricted by materials ownership policy.
Strains derived from CBS7435 are not covered by patents. -- CBS7435 Mut(S) strain is marker-free but grows slowly on methanol. -- Auxotrophic CBS7435 strains (∆ade1, ∆his4, ∆aox1, ∆met2, ∆arg4, ∆lys2, ∆pro3, ∆tyr1) exist. -- Protease deficient CBS7435 strains (∆prc1, ∆sub2, ∆pep4+∆his4, ∆prb1, ∆his4+∆pep4+∆prb1) exist. -- Homologous recombination-prone strains (∆ku70, ∆ku70+∆his4, ∆ku70+∆gut1, ∆ku70+∆ade1) exist.
Controlling temperature/pH can help control protease activity, as can adding casein/peptone fragments to compete as substrates.
PEP4 and PRB1 knockout are good for suppressing protease activity.
Ahmad et al. claim protease-deficient strains grow robustly.
Glycoengineering: Introduce Trichoderma reesei alpha-1,2-mannosidase, knock alpha-1,6-mannosyltransferase encoded by OCH1, co-overexpress targeted glycosyltransferases and glycosidases (Hamilton et. al. 2003), introduce sialic acid pathway (Hamilton et al. 2006).
Nett et al 2011 covers the best targeting signal/enzyme combinations for N-linked glycoengineering.
O-glycosylation is a problem in Pichia--different from humans, and knocking out the genes inhibits cell viability. -- Hamilton et al 2013 has addressed some of this. -- Can also add a chemical protein-O-mannyltransferase (PMT) inhibitor to reduce O-glycosylation.
Driving selection marker expression with weak ARG4 promoter reduces false positive colonies during colony selection.
For intracellular protein accumulation, there's a pretty linear relationship between copy number and expression level (relationship is less linear at at >6 copies for secretion). -- But, high copy number strains are less genetically stable due to recombination of the identical sequences in the copies.
Pichia gets used for making membrane proteins a lot, because even with low per-cell yield the high cell density (150 g dry cell weight/L) gives decent overall protein yield.
Proteins produced industrially in Pichia: -- Phytase (animal feed additive, increases phosphorus availability) -- Trypsin (industrially relevant protease) -- Nitrate reductase (water testing) -- Phospholipase C (degumming vegetable oils) -- Collagen (dermal filler, medical research) -- Proteinase K (industrially relevant protease)
Biomedicines in Pichia: -- Plasma kallikrein inhibitor, (treats hereditary angioedema) -- Recombinant insulin!
Products under development: -- Elastase inhibitor -- Nanobody (R) antibody fragments -- Plectasin antimicrobial peptide
Pichia-expressed recombinant proteins used for research: -- Cytokines -- Growth factors -- Angiostatin
Researchers are interested in finding alternatives to methanol for industrial scale fermentations.
Marker-free integration strategies are important for generating deletion strains.

Notes from the Vogl et al review paper on opportunities for synthetic biology in Pichia biopharmaceutical production:

Biopharmaceuticals are a >$80 billion industry.
Antibodies are the biggest group of biopharmaceuticals, but insulin, human growth hormone (hGH), and erythropoietin (EPO) are other big ones.
Hypermannosylation is less pronounced in Pichia than in S. cerevisiae.
Producing sialylated proteins in Pichia required deleting 6 genes and inserting 9 synthetic genes.
Gram/L antibody production in glycoengineered strains has been achieved.
Glycoengineered Pichia can make biopharmaceuticals with better pharmacokinetics than equivalents from mammalian cells.
AOX1 promoter has achieved 15 g/L secreted protein.
For deletions/marker recycling, use counter-selective markers like mazF, self-excising cassettes employing Cre/Lox recombinase (flank the selection marker and recombinase with recombination sites and put them under tight methanol inducible control).
During glycoengineering, only 5 out of 460 clones showed the desired targeted gene replacement. --ku70 deletion strains help with this, boosting homologous recombination rates to 100%. --∆ku70 strains are genetically stable, but grow 10-30% slower and are UV-sensitive. --Maybe a good strategy is to transform into a ∆ku70 strain, but to re-introduce ku70 after expression is verified/optimized?

Notes on the Macauley-Patrick et al paper on heterologous protein expression in Pichia:

Methanol metabolism starts in little organelles called peroxisomes.
Mut^S strains that grow slowly on methanol can actually be advantageous for some recombinant protein production, though Mut+ works well for many applications.
GAP constitutive promoter can actually produce higher cell specific protein yield, and faster protein accumulation, than AOX1 under some circumstances (at least as of 2005).
Combining constitutive and methanol inducible promoters can double yield--get expression while cells are growing to max density, and then more expression once methanol feeding begins.
AOX2 promoter is methanol inducible, but with expression levels 10-20-fold lower than AOX1.
FLD1 methanol inducible promoter can also be induced by methylamine, enabling induction with methylamine while growing on glucose or glycerol as a carbon source. --Good for methanol-independent fermentation strategies, assuming access to methylamine isn't a problem.
Methylamine can be used as a nitrogen source for Pichia, and co-inducing with methanol and methylamine boosts yields from FLD1 promoter.
Pichia can tolerate a broad range of pH (3.0-7.0) without impacts on growth rate. -- This is potentially useful for reducing contamination in fermentation. -- Different pH values are optimal for expression/secretion of different recombinant proteins. An important parameter to tune.
Lowering bioreactor temperature from 30ºC to 23ºC or 20ºC can increase stability, expression of some proteins. -- Lowering temperature can also reduce oxygen consumption rate, which can be good for cell viability and protein yield in oxygen-limited dense cultures. Reduces protease activity due to decreased cell lysis.
Adding methanol to growth-rate-limiting concentrations can also reduce proteolysis. -- But important to note this paper is from 2005; maybe at this point protease knockout strains are the way to solve these problems.
Gene copy number is the rate limiting step in recombinant protein production in Pichia pastoris.
Pichia really behaves better in bioreactors than in shake flasks--better control of oxygen, methanol/carbon substrate, etc.
To minimize batch-to-batch variation, it's better to eliminate complex media ingredients like peptones/caseins.
Ammonium is the commonly used nitrogen source, but its concentration must be carefully calibrated. Too little and recombinant proteins might get degraded; too much (>0.4 M NH4+) and cell growth can be inhibited.
Example defined medium for Pichia fermentation: -- 40 g/L glycerol -- 0.93 g/L calcium sulfate -- 18.2 g/L potassium sulfate -- 14.9 g/L magnesium sulfate heptahydrate -- 4.13 g/L potassium hydroxide -- 7.0 g/L dipostassium monophosphate -- 22.7 mL/L 85% phosphoric acid -- 12 mL/L trace element solution --- Trace salt stock solution: ---- 6.0 g/L Copper sulfate pentahydrate ---- 0.08 g/L sodium iodide ---- 3.0 g/L manganese sulfate monohydrate ---- 0.2 g/L sodium molybdate dihydrate ---- 0.02 g/L boric acid ---- 0.5 g/L cobalt chloride ---- 20.0 g/L zinc chloride ---- 65.0 g/L iron(II) sulfate heptahydrate ---- 0.2 g/L biotin ---- 5 mL/L concentrated sulfuric acid.
Glycerol is preferred to glucose as a carbon source because glucose can generate a small amount of ethanol as a byproduct even in Pichia (???).
Methanol concentrations above 5 g/L are toxic to Pichia, but methanol levels that are too low won't induce gene transcription/expression.
Measuring methanol concentration while Pichia is growing in a bioreactor is doable, but nontrivial. Fourier Transform Infrared (FT-IR) spectroscopy is a commonly used method.
Again, decreasing fermentation temperature during induction seems to help with yields for a lot of proteins. Pichia is really good for secreting soluble proteins.

Braun-Galleani et al analysis of Pichia/Komagataella genetic diversity

The name Pichia pastoris was discontinued in 2009; it was being used for two different species, which were renamed Komagataella phaffii and Komagataella pastoris.
Pichia/K. phaffii is only distantly related to S. cerevisiae (baker's yeast).
Widely used strains GS115 and X-33 are derivatives of CBS7435.
Pichia/K. phaffii has four chromosomes, grows primarily as a haploid.
Mating is induced by nitrogen depletion.
Mating occurs between mating types MATa and MATα.
Strains can switch mating type by inverting 138 kb on chromosome 4.
Pichia/K. phaffii spores are very small (1-2 μm), tend to clump together (makes genetic analysis of mating crosses difficult).
Five isolates of Pichia/K. phaffii in the NRRL culture collection: -- NRRL Y-7556 (alias CBS2612, nickname Pp3) --- Essentially identical to CBS7435. --- CBS7435 and Pp5 are duplicate accessions of Pp3. -- NRRL Y-12729 (nickname Pp1) --- Essentially identical to Pp3 and CBS7435. -- NRRL Y-17741 (nickname Pp2) --- Differs from CBS7435 by 42,000-44,000 SNPs. -- NRRL YB-378 (nickname Pp4) --- Differs from CBS7435 by 42,000-44,000 SNPs. -- NRRL YB-4289 (nickname Pp5) --- Differs from CBS7435 by 16,000 SNPs.
Pichia/K. phaffii genetic diversity and SNP density is lower than for S. cerevisiae isolates.
GS115 is a mutagenized derivative of CBS7435, differs from CBS7435 at 69 sites.
Pichia/K. phaffii is less prone to chromosomal crossovers/recombinantion during mating and meiosis than S. cerevisiae. -- On average, 25 (11-37) crossing over events per meiosis, 3.5 times less than S. cerevisiae. -- On average, one crossover per 369 kilobase (kb) of genome.
In Pichia/K. phaffii, meiotic recombination is suppressed for ~150 kb on either side of the centromeres. -- As a result, there is very low sequence diversity across the strains for the regions around the centromeres of chromosome 1 (CEN1, 362 kb region) and chromosome 2 (CEN2, 240 kb) -- These low diversity regions are 20 times larger than the actual centromeres. -- By contrast, in S. cerevisiae recombination is suppressed for only 10 kb around the centromeres.
There are many protein coding genes in the recombination-suppressed regions around the centromeres.

Idea: strain engineering by mating different Pichia strains Idea: integration at recombination-suppressed regions as safe-harbor loci for keeping multiple integrated gene cassettes together during mating

Love et al analysis of Pichia/K. phaffii genome and transcriptome

Parent (un-mutagenized) strain of GS115 is also in NRRL, under the label NRRL Y-11430.
Sister species K. pastoris has strain ID NRRL Y-1603. -- K. pastoris is also used industrially for recombinant protein production.
Love et al sequenced GS115, NRRL Y-11430, NRRL Y-1603. -- With long reads (PacBio) and short reads (Illumina) -- cDNA sequenced from all strains while growing on different carbon sources
K. pastoris genome is 9.6 Mbp long, K. phaffii genome is 9.4 Mbp.
Ribosomal DNA (rDNA) and telomeric DNA are the only sequence regions that couldn't be assigned to specific chromosomes, because they're so repetitive.
All the sequenced Pichia genomes have 41.3%-41.5% GC content.
K. pastoris has 5241 annotated genes, K. phaffii has 5167.
Wild K. phaffii (NRRL Y-11430) has a linear plasmid with 7 genes on it.
One gene in NRRL Y-11430 is inactivated in GS115 by a frameshift mutation.
There are ~398 genes that are species-specific for either K. phaffii or K. pastoris.
There are ~4600 ortholog genes shared between K. phaffii and K. pastoris, with ~90% nucleotide identity across those genes. -- alpha mating factor is <85% identical, K. phaffii gene has repeated sequence motifs that K. pastoris lacks.
Commonly used promoters P_AOX1 and P_GAPDH have 90% and 88% identity across the two species.
There are significant structural rearrangements in the chromosomes (synteny) between the two species.
Both species have very similar codon usage, and 122-123 tRNA genes.
Telomeres consist of 100-340 bp repeats of TGGATGC. -- Similar to related yeasts Yarrowia lipolytic and H. polymorpha, different from S. cerevisiae ([TG]2-3[TG]1-6)
rDNA cluster with 18S, 5.8S and 16S rRNA genes located at close to the telomere on chromosome 1 in GS115. -- Found on chromosomes 1, 3 and 4 in K. phaffii and K. pastoris
5S rRNA genes are dispersed around the genome, 21-23 copies.
A linear plasmid was detected in both Y-11430 and GS115, at very low copy numbers/expression levels (even lower in GS115).
Only 21 genes in either species had isoforms derived from alternative mRNA splicing.
Highly expressed genes are distributed evenly across all chromosomes during growth on all carbon sources.
Most highly expressed genes are highly expressed during growth on all carbon sources.
Only 10 out of 24 most highly expressed genes have been described as useful promoter sequences in the past! -- These are primarily likely housekeeping genes, involved in central metabolism, transport and stress response
List of most highly expressed genes in all strains (and hour of incubation when they're most highly expressed while growing on glycerol): -- GCW14 (6) -- HSP12 (24, 48) -- OLE1 (48) -- TMA10 (6) -- ADH2 (48) -- RGI2 (6, 24, 48) -- ALD4 (48) -- WT_04506 (48) -- PCK1 (48) -- POR1 (48) -- ACO1 (48) -- TEF1 (6) -- TDH3 (6) -- G7 (6) -- SSA4 (6) -- ICL1 (48) -- CIT1 (24) -- CTA1 (48) -- QCR7 (6) -- ATP1 (6) -- ATP2 (6) -- WT_01845 (48) -- FDH1 (24, 48) -- ATO2 (24, 48)
Protein folding machinery increased expression over time, except in K. phaffii growing on glycerol.
'Secretory pathway genes are inversely correlated with increasing expression phenotypes, particularly in glycerol and methanol. These results together suggest that optimizing Komagatella strains as expression hosts for sustained protein expression and secretory function during cultivation will require genome engineering and concomitant optimization of fermentation.'
Secretory function (or at least, expression of genes for secretory function) declines over time in batch fermentation.
Of the strains and conditions analyzed, K. phaffii grown on methanol had the lowest overall expression of native secreted proteins, which could be good for downstream purification.
GS115 has 35 single nucleotide polymorphisms (SNPs) predicted to influence gene expression
GS115 is more sensitive to UV than WT K. phaffii NRRL Y-11430.
GS115 grows more quickly than WT K. phaffii NRRL Y-11430 on glucose.
GS115 genes with non-synonymous mutations: -- Chr1: TGL1, CWH43, YMR087W, MSS4, PIR_HUMAN, HIS4, YDR248C, SMI1, VPS35 -- Chr2: DOP1, YH93_SCHIPO, MRPS28, DBP8, YMR253C, LOS1, YDR131C, SMC3, MTG1, UTP10 -- Chr3: NAB3, RAD5, LCB5, PEX22_PICPA, RPO31, AVT2, EXO2, SFB3, CRH1, COS111
Specific mutations, and their predicted impact: -- C557R in HIS4 --> histidine auxotrophy -- S752F in RAD5 --> radiation (UV) damage sensitivity -- Mutation in YDR248C --> lower YDR248C expression in GS115, higher growth rate
Telomere sequences combined with centromeres could enable construction of artificial chromosomes in K. phaffii and K. pastoris.
NCBI accession numbers for the sequencing data and assemblies from this study: -- PRJNA304627 (K. pastoris) -- PRJNA304977 (K. phaffii wildtype) -- PRJNA304986 (K. phaffii GS115)
NCBI accession number for RNA-seq data from this study: -- PRJNA304627
RNA-seq data was highly correlated (R = 0.969-0.995) for the same growth conditions at low cell density and at high cell density

Cregg et al 1985 paper that first reported genetic engineering of Pichia

Alcohol oxidase (AO) protein is required for consumption of methanol as sole carbon source.
AO is undetectable during growth on glucose; but when grown on methanol, AO can be 35% of total cellular protein!
Original transformation system used the histidine auxotrophy of GS115, and complementation with the S. cerevisiae HIS4 gene, as a selection marker.
To get the DNA into the yeast, the protocol enzymatically degrades the cell walls of the yeast, leaving 'spheroplasts.' -- Essentially, grow the cells up, pellet them, and wash them with (1) water, (2) sorbitol-EDTA-DTT solution, and (3) twice with sorbitol solution, then re-suspend them in SCE solution (1 M sorbitol/0.1 M sodium citrate (pH5.8)/10 mM EDTA) and mix in lyticase enzyme (also called Zymolyase) and let the enzyme chew up the cell walls for 30 minutes. -- Then wash the cells 2 more times in 1 M sorbitol, and once in CaS solution (10 mM CaCl2, 1 M sorbitol) and resuspended in CaS. Then plasmid DNA is added, along with random E. coli genomic DNA (I believe to overload Pichia's native nucleases). DNA and yeast are incubated for 20 min at room temperature. -- Then 1 mL of 20% polyethylene glycol (PEG) 3350 in 10 mM Tris buffer (pH 7.4) with 10 mM CaCl2 is added in, and samples are incubated for 15 more minutes. Then cells are centrifuged, suspended in 150 uL SOS medium (1 M sorbitol, 0.3X YPD culture, 10 mM CaCl2), and incubated for 30 minutes. -- Cells were diluted to 1 mL total volume with 1 M sorbitol solution, and dispensed onto 10 mL 'regeneration agar' plates: 3% agar, 1 M sorbitol, SD medium, 0.4 ug/mL biotin. Plates were incubated at 30ºC for 3-5 days.
At least in the early days, Pichia plasmids that integrate into the genome have lower transformation efficiencies than self-replicating plasmids.

Notes from Ravinder Kumar's 2019 paper in the journal Yeast, describing a simplified Pichia transformation protocol

Several methods for transforming Pichia have been developed: -- Electroporation (Becker & Guarante, 1991) -- Alkali cation method (Ito, Fukuda, Murata, & Kimura, 1983) (this one actually looks like the earliest method!) -- PEG treatment (Dohmen, Strasser, Höner, & Hollenberg, 1991) -- Spheroplast generation (Cregg, Barringer, Hessler, & Madden, 1985)
However, all of the methods listed above require the generation of competent cells, and growth of cells in 50-100 mL of media.
This protocol grows out Pichia cells in patches on agar plates (2 cm x 1.5 cm patch for two transformations), instead of in shaker flasks. -- YPD agar plates -- 18-24 hours of outgrowth -- 2 cm x 1.5 cm patch of cells for two transformations
In a 1.5 mL microcentrifuge tube, the following is mixed: -- 1 mL YPD media -- 40 µL of freshly prepared or well-frozen (-20ºC, no freeze-thaw cycles) 1 M DTT solution -- 40 µL of 1 M HEPES-NaOH (pH 8)
Cells from the patch are scraped and transferred into the 1.5 mL tube with the mixture prepared above, and resuspended. -- Protocol resuspends by vortexing; not clear if resuspending by pipetting is an acceptable alternative.
Tube is gently shaken at 30ºC for 15 min.
While tube is shaking, prepare the following: -- At least 1 mL of 1 M sorbitol, on ice. -- Electroporation cuvette, on ice (labeled, if doing multiple transformations).
Cells are pelleted and washed. -- Pellet at 3000g for 3 minutes. -- Wash with by re-suspending pellet in sterile water and pelleting again. Do this twice. -- Final volume of resuspended cells after pelleting and washing should be ~50 µL. -- If you have a -80ºC freezer, cells at this stage can be washed in 1 M sorbitol and stored at -80ºC for later transformations.
After washing, incubate tube on ice for 3-5 minutes.
Add DNA to transform, and mix. -- 250 ng DNA for plasmid -- 300 ng DNA for PCR product
Transfer cells/DNA mix to ice-cold electroporation cuvette, while the cuvette is still on ice.
Wipe down ice-cold electroporation cuvette, to remove moisture from the electrical contacts.
Load cuvette into electroporator, and shock the cells. -- Voltage: 1500 VH -- Resistance: 200 Ω -- Capacitance: 25 µF -- Original protocol uses BTX, ECM20, version 1.04
Add 1 mL of ice-cold 1 M sorbitol immediately after electroporation, and mixed with cells.
If using auxotrophy as a selection marker, proceed immediately to plating cells. -- If using antibiotics as a selection marker, incubate cuvette at 30ºC for 2-3 hours, then proceed to plating. -- To plate, first pellet the cells and resuspend them in no more than 100 µL sterile water. -- In this protocol, cells are spread on the selection plates using glass beads.
Age of the patched cells matters a lot. -- 2-day-old cell patches give fewer transformed colonies, while 5-day-old cell patches yield zero transformed colonies. -- Thicker cell patches are worse, because they're older.
Adding the DTT and HEPES (final concentrations of 40 mM) significantly increase transformation efficiency. -- Lower DTT/HEPES concentrations lowered transformation efficiency, while higher DTT/HEPES concentrations did not improve transformation efficiency.
Recovery medium after electroporation can be chilled 1 M sorbitol, or chilled 2% YPD.
For high efficiency transformation, plasmids are linearized by digesting in the middle of the selection marker -- The Pichia selection marker? Or the E. coli selection marker? Not clear.
PCR products had lower transformation efficiencies than plasmids.
Surprisingly, cells frozen at -80ºC after pelleting and washing had ~2/3 more transformation efficiency compared to freshly prepared cells. -- 425 colonies vs 625 colonies.
This protocol does, however, have a lower transformation efficiency than previously published, more complicated electroporation protocols.
This protocol requires an electroporator. However, it does NOT require a refrigerated centrifuge.
Incubating cells after mixing with DNA but before electroporation had little effect on transformation efficiency.

friendzymes / reading

Pichia/Komagataella Research Notes #1