CRISPR-Cas9 knockout screen informs efficient reduction of the Komagataella phaffii secretome

Strains were grown in 3 mL cultures in 24-well deep well plates (25 °C, 600 rpm) or 100 mL cultures in 500 mL shake flasks (25 °C, 300 rpm). Cells were cultivated in complex media (potassium phosphate buffer pH 6.5, 1.34% nitrogen base w/o amino acids, 1% yeast extract, 2% peptone). Cells were inoculated at 0.1 OD600, outgrown for 24 h with 4% glycerol feed, pelleted, and resuspended in fresh media with 3% methanol for HSA production, or 1% methanol, 40 g/L sorbitol, and 10 mM glutathione for trastuzumab production. Supernatant samples were collected after 24 h of production and analyzed.

For quantitative measurement of secreted protein titer, strains were cultivated in biological triplicate from frozen stocks. For growth assays, strains that produce trastuzumab were seeded at an optical density of 0.01 OD600 in 200 µL cultures (25 °C, 300 rpm) in either outgrowth or production media in biological triplicate from frozen stocks. OD600 was measured every hour for 48 h and plotted on a log axis. Growth rate was calculated manually as the slope of the curve during exponential or stationary growth.

All strains were derived from wild-type Komagataella phaffii (NRRL Y-11430). Strains for recombinant protein production were derived from a modified base strain [AltHost Research Consortium Strain S-63 (RCR2_D196E, RVB1_K8E)] described previously [17]. Genes containing recombinant protein products HSA and trastuzumab were synthesized (Integrated DNA Technologies) and cloned into a custom vector with the methanol-inducible promoter P_AOX1. To enable protein secretion, HSA was expressed with the signal peptide from the S. cerevisiae α-mating factor, and trastuzumab was expressed with the signal peptide from the S. cerevisiae α-mating factor for the light chain and the signal peptide from human HSA for the heavy chain. All vector sequences are listed in the Supplemental Materials.

K. phaffii strains were transformed as described previously [18]. After transformation of the HSA or trastuzumab expression vectors, 4–8 clones were selected and grown in 3 mL cultures. Supernatant samples were analyzed by SDS-PAGE, and the clone that exhibited the highest productivity was selected for quantitative growth and titer measurements.

Knockout of individual genes was performed with a custom knockout cassette as described previously [19]. Disruption of genes was confirmed by PCR and Sanger sequencing. Design, construction, and screening of the pooled knockout library is described in the Additional file 1.

SDS-PAGE was carried out as described previously [20]. HSA supernatant titers were measured by reverse phase liquid chromatography. Trastuzumab supernatant titers were measured by Protein A biolayer interferometry. Specific productivity was calculated as titer normalized to cell density by OD600, relative to the original base strain.

Wild-type K. phaffii was cultivated in 200 mL shake flask cultures in complex media. Cells were inoculated at 0.1 OD600, outgrown for 48 h with 4% glycerol feed, pelleted, and resuspended in fresh media with 1.5% methanol feed to simulate recombinant gene expression. Supernatant samples were collected after each phase of the cultivation.

Supernatant was reduced (10 mM dithiothreitol, 56 °C for 45 min) and alkylated (50 mM iodoacetamide, room temperature in the dark for 1 h). Proteins were subsequently digested with trypsin (sequencing grade, Promega, Madison, WI), at an enzyme/substrate ratio of 1:50, at room temperature overnight in 100 mM ammonium acetate pH 8.9. Trypsin activity was quenched by adding formic acid to a final concentration of 5%. Peptides were desalted using C18 SpinTips (Protea, Morgantown, WV), lyophilized, and stored at − 80 °C.

Peptides were labeled with TMT 6plex (Thermo) per manufacturer’s instructions. Lyophilized samples were dissolved in 70 μL ethanol and 30 μl of 500 mM triethylammonium bicarbonate (pH 8.5), and the TMT reagent was dissolved in 30 μl of anhydrous acetonitrile. The solution containing peptides and TMT reagent was vortexed and incubated at room temperature for 1 h. Samples labeled with the ten different isotopic TMT reagents were combined and concentrated to completion in a vacuum centrifuge. The samples were labeled using the TMT 10plex channels as follows: 126–4/27/16 48 h induction; 127N–4/29/16 48 h harvest; 127C–5/6/16 96 h harvest; 128N–5/4/16 48 h induction; 129N–5/13/16 96 h harvest; 129C–5/20/16 96 h harvest; 130N–5/18/16 48 h induction; 130C–5/27/16 96 h harvest; 131–5/25/16 48 h induction.

Peptides were loaded on a precolumn and separated by reverse phase HPLC (Thermo Easy nLC1000) over a 140 min gradient before nanoelectrospray using a QExactive Plus mass spectrometer (Thermo). The mass spectrometer was operated in a data-dependent mode. The parameters for the full scan MS were: resolution of 70,000 across 350–2000 m/z, AGC 3e⁶, and maximum IT 50 ms. The full MS scan was followed by MS/MS for the top 10 precursor ions in each cycle with a NCE of 34 and dynamic exclusion of 30 s. Raw mass spectral data files were searched using Proteome Discoverer (Thermo) and Mascot version 2.4.1 (Matrix Science). Mascot search parameters were: 10 ppm mass tolerance for precursor ions; 15 mmu for fragment ion mass tolerance; 2 missed cleavages of trypsin; fixed modification were carbamidomethylation of cysteine and TMT 10-plex modification of lysines and peptide N-termini; variable modification was methionine oxidation. TMT quantification was obtained using Proteome Discoverer and isotopically corrected per manufacturer’s instructions. Only peptides with a Mascot score greater than or equal to 25 and an isolation interference less than or equal to 30 were included in the quantitative data analysis. Relative abundance of each protein was defined by the log₁₀ of the total area under the curve for all peptide counts detected for each protein, summed over four independent replicate cultivations.

We determined which genes in the K. phaffii genome contained a signal peptide using SignalP 5.0 and filtering for Sec/SPI > 0.5 [21]. Comparison of the K. phaffii secretome to Valli et al. [22] was performed by manual comparison in SnapGene (snapgene.com) of protein coding sequences from both the Love et al. genome [23] and the genome from Pichiagenome.org [24]. Descriptions of protein functions were obtained using BLAST.

Cells were cultivated at 3 mL plate scale and harvested after 18 h of production in methanol medium. RNA was extracted and purified according to the Qiagen RNeasy 96 kit. RNA quality was analyzed on an Agilent BioAnalyzer to ensure RNA Quality Number > 6.5. RNA was reverse transcribed with Superscript III (ThermoFisher) and amplified with KAPA HiFi HotStart ReadyMix (Roche). RNA libraries were prepared using the Nextera XT DNA Library Preparation Kit with the Illumina DNA/RNA UD Indexes Set A. sequenced on an Illumina Nextseq to generate paired reads of 50 (read 1) and 50 bp (read 2). Sequenced mRNA transcripts were demultiplexed using sample barcodes, aligned to the WT.fa Komagataella phaffii genome (strain Y11430↗) and exogenous transgenes, and quantified using Salmon version 1.6.0 [25]. Gene level summaries were prepared using tximport version 1.24.0 [26] running under R version 4.2.1 [27]. Gene set enrichment analysis (GSEA) was performed with GSEA 4.1.0 using Wald statistics calculated by DESeq2 [28] and gene sets from yeast GO Slim [29].

To improve the production of recombinant proteins, we next sought to disrupt the most abundant secreted proteins based on a rank-ordering of our initial analysis and using a previously reported, host-informed strategy for CRISPR-Cas9 genome editing in K. phaffii [18, 19]. We used this tool to serially disrupt several of the most abundant secreted proteins. We encountered engineering challenges in this approach, however, including in-frame deletions in disrupted genes, and were unable to disrupt two of the first five targeted genes (Fig. 1B). (GQ67_01286 is a homolog of the gene ayr1 in S. cerevisiae—ayr1 and tdh3 are both non-essential in S. cerevisiae (Saccharomyces Genome Database).) We posited that identification of essential genes—particularly unannotated essential genes—would streamline further engineering of K. phaffii. We therefore sought to identify which genes among the identified secretome of K. phaffii are essential in laboratory conditions (under standard conditions for fermentation).

We next prioritized a short list of non-essential genes from the secretome as engineering targets. We filtered the secretome for 61 genes with an essentiality score less than 0.5 and with gene products experimentally detected in the culture supernatant (Fig. 2C). Then, we removed seven genes from the potential target pool known to contribute to methanol metabolism since disruption of these genes may affect cellular function during methanol-induced protein secretion (though appear non-essential when cultured on solid glucose medium). Finally, we constructed and separately transformed CRISPR-Cas9 vectors with multiplexed sgRNAs targeting up to four genes per vector (Fig. S5). These 14 vectors targeted all 54 gene targets at least once. On the first attempt, we observed disruption of 20 of the 54 gene targets. We hypothesized that certain combinations of multiplexed knockouts may cause unforeseen synthetic lethality. It may be feasible to disrupt more of these genes with further optimization. We chose to proceed, however, with engineering strains based on the 20 gene disruptions observed.

We next attempted to combine many disruptions to create strains with a reduced secretome. We performed gene disruptions sequentially in two lineages (S∆3a (∆tos1, ∆crh1, ∆exg1) and S∆3b (∆tfs1, ∆lsc2, ∆gq67_05326)), and then combined these sets to construct a new strain S∆6 (∆tos1, ∆crh1, ∆exg1, ∆tfs1, ∆lsc2, ∆gq67_05326). We also extended the S∆3a lineage to construct S∆9 with six additional disruptions (∆gq67_02848, ∆tef4, ∆cts1, ∆gq67_01368, ∆gq67_03274, ∆msc1). We added ∆lsc2 and ∆gq67_05326 to S∆9 to create S∆11, but we were unable to disrupt tfs1 in S∆11. This observation suggested that tfs1 may confer synthetic lethality with another disrupted gene in S∆11.

Throughout this engineering process, we noticed high disruption efficiencies (typically 80–100% of colonies were disrupted). During construction of S∆11, for example, we combined all 11 knockouts without screening more than 16 colonies at each step (Fig. 2D). We attributed this engineering efficiency to the additional knowledge of gene essentiality used to guide the selection of the targeted genes. We next sought to assess the utility of these engineered strains for production of recombinant proteins.

To assess the differences between the S∆6 and wildtype (control) strains producing HSA, we analyzed the transcriptional states of the cells by RNA-seq during recombinant protein expression. Genes related to ribosomal processing and translation were upregulated in the base strain, while genes related to cell division and genome replication were upregulated in S∆6 (Fig. 3C, Table S2). We hypothesize that the engineering changes in S∆6 may alleviate the translational burden experienced by the base strain, either by reduction of the overall translational load from knockout of abundant secreted proteins, or by specific functions of the disrupted genes. We also evaluated the expression of three chaperones that are commonly used as markers for endoplasmic reticulum stress (pdi1, ero1, kar2). We observed higher expression of pdi1 and kar2 in S∆6 compared to wild type during expression of HSA (Fig. S6). We hypothesize that translational capacity in S∆6 has been subsumed by the recombinant HSA, which may lead to secretion-related stress.

Given the improved production of HSA, we hypothesized that S∆6 may also have benefits for producing other large proteins (such as mAbs). We evaluated the secreted titer of trastuzumab, a mAb used to treat HER2 + breast cancer. S∆6 exhibited a ~ 30% increase in specific productivity compared to the wildtype strain in 3 mL cultures (p = 0.002, unpaired Welch’s t-test) (Fig. 3D). To evaluate the performance of engineered strains at higher cell densities, we cultivated the base strain and all four engineered strains producing trastuzumab in 100 mL cultures in shake flasks. At this scale, all strains reached an optical density of 40–50 OD600 after one day of production. In these growth conditions, all the engineered strains secreted two-fold more trastuzumab than the base strain, particularly S∆6 and S∆3a (Fig. 2E).

We also assessed the gene expression of S∆6 and the base strain during expression of trastuzumab. Like the strains that produced HSA, we observed that genes related to ribosomal and RNA synthesis were upregulated in the base strain, while genes related to cell and genome replication were upregulated in S∆6 (Fig. 2F, Table S2). An overall reduction of the translational load may also improve production of trastuzumab, similar to the results for producing HSA. We again observed higher expression of pdi1 and kar2 in S∆6 compared to wild type when expressing trastuzumab (Fig. S6). Interestingly, strains expressing trastuzumab had overall higher expression of all three chaperones pdi1, ero1, and kar2. We hypothesize that secretion of trastuzumab is also limited by other secretory process such as protein folding.

Finally, we measured the rate of growth by seeding engineered antibody-producing strains at low density in 200 µL cultures. Interestingly, in the glycerol-containing media used to accumulate biomass, S∆6 exhibited a higher growth rate than the base strain while S∆11 exhibited a lower growth rate (Fig. 2G). Similarly, in the methanol-containing media used to induce expression of the recombinant protein, we observed higher growth rates for S∆3a, S∆3b, and S∆6, and a lower growth rate for S∆11 (Fig. 2H). These results, together with the observed improvement in recombinant protein titers, demonstrate that strains of K. phaffii with a reduced secretome can improve the secreted productivity of multiple proteins relevant for biopharmaceutical and vaccine products without a decrease in growth rate compared to the base strain.

Additional file 1. Supplemental methods on construction of the knockout library. Additional file 2. Supplemental figures. Additional file 3: Table S1. Lists and data for secretome gene selection and essentiality. Additional file 4: Table S2. Raw data from Gene Set Enrichment Analysis of engineered strains. Additional file 5: Table S3. Lists and data for sgRNA library generation, amplicon sequencing, and analysis. Additional file 6. R-code for sgRNA KS-test. Additional file 7. Plasmid sequences used in this study.