GoldenPiCS: a Golden Gate-derived modular cloning system for applied synthetic biology in the yeast Pichia pastoris

Abstract

Background: State-of-the-art strain engineering techniques for the host Pichia pastoris (syn. Komagataella spp.) include overexpression of homologous and heterologous genes, and deletion of host genes. For metabolic and cell engineering purposes the simultaneous overexpression of more than one gene would often be required. Very recently, Golden Gate based libraries were adapted to optimize single expression cassettes for recombinant proteins in P. pastoris. However, an efficient toolbox allowing the overexpression of multiple genes at once was not available for P. pastoris.

Methods: With the GoldenPiCS system, we provide a flexible modular system for advanced strain engineering in P. pastoris based on Golden Gate cloning. For this purpose, we established a wide variety of standardized genetic parts (20 promoters of different strength, 10 transcription terminators, 4 genome integration loci, 4 resistance marker cassettes).

Results: All genetic parts were characterized based on their expression strength measured by eGFP as reporter in up to four production-relevant conditions. The promoters, which are either constitutive or regulatable, cover a broad range of expression strengths in their active conditions (2-192% of the glyceraldehyde-3-phosphate dehydrogenase promoter P _GAP ), while all transcription terminators and genome integration loci led to equally high expression strength. These modular genetic parts can be readily combined in versatile order, as exemplified for the simultaneous expression of Cas9 and one or more guide-RNA expression units. Importantly, for constructing multigene constructs (vectors with more than two expression units) it is not only essential to balance the expression of the individual genes, but also to avoid repetitive homologous sequences which were otherwise shown to trigger "loop-out" of vector DNA from the P. pastoris genome.

Conclusions: GoldenPiCS, a modular Golden Gate-derived P. pastoris cloning system, is very flexible and efficient and can be used for strain engineering of P. pastoris to accomplish pathway expression, protein production or other applications where the integration of various DNA products is required. It allows for the assembly of up to eight expression units on one plasmid with the ability to use different characterized promoters and terminators for each expression unit. GoldenPiCS vectors are available at Addgene.

Keywords: Cell engineering; Golden Gate cloning; GoldenMOCS; GoldenPiCS; Pichia pastoris; Synthetic biology.

Fig. 1

Assembly strategy and hierarchical backbone levels of the cloning systems GoldenMOCS and GoldenPiCS. In the microorganism-independent general platform GoldenMOCS, DNA products (synthetic DNA, PCR products or oligonucleotides) are integrated into BB1 by a BsaI Golden Gate Assembly and fusion sites Fs1, Fs2, Fs3 and Fs4. Fusion sites are indicated as colored boxes with corresponding fusion site number or letter. Basic genetic elements contained in backbone 1 (BB1) can be assembled in recipient BB2 by performing a BpiI GGA reaction. The transcription units in BB2 are further used for BsaI assembly into multigene BB3 constructs. Single transcription units can be obtained by direct BpiI assembly into recipient BB3 with fusion sites Fs1-Fs4. Fusion sites determine module and transcription unit positions in assembled constructs. Thereby, fusion sites Fs1 to Fs4 are used to construct single expression cassettes in BB2 and are required between promoter (Fs1-Fs2), CDS (Fs2-Fs3) and terminator (Fs3-Fs4). Fusion sites FsA to FsI are designed to construct BB3 plasmids and separate the different expression cassettes from each other. The FSs are almost randomly chosen sequences and only FS2 has a special function, because it includes the start codon ATG. GoldenPiCS additionally includes module-containing BB1s specific for P. pastoris: 20 promoters, 1 reporter gene (eGFP) and 10 transcription terminators, and recipient BB3 vectors containing different integration loci for stable genome integration in P. pastoris and suitable resistance cassettes (Additional file 2)

Fig. 2

Comparison of conventional and Golden Gate based strain engineering strategies for P. pastoris. Overexpression of multiple genes (GOIs) in P. pastoris using conventional cloning plasmids requires several rounds of competent making (2 days), transformation (4 days including second streak-out) and clone screening (5 days) and takes at least 31 days for three genes with three selection markers. Alternatively, multiple vectors can be co-transformed at once, but resulting transformation efficiencies are usually very low and clonal variation increases. Appropriate flanking sites for the selection marker (loxP or FRT sites) enable marker recycling by recombinases (Cre or Flp, respectively), but require one more round of competent making and transformation which takes at least eight additional days. Golden Gate plasmids carry several transcription units at once and thereby enable transformation and screening in only nine days

Fig. 3

Relative eGFP levels obtained with various elements of the GoldenPiCS toolbox. Expression strength of different promoters in comparison to P _GAP tested on different carbon sources (a, b), expression levels for P _GAP-controlled expression in combination with different transcriptional terminators (c), and comparison of P _GAP variants with alternative ‘-1’ nucleotides (d). At least 10 P. pastoris clones were screened to test promoter and terminator function in up to four different conditions: glycerol and glucose excess as present in batch cultivation (“G”, “D”), limiting glucose (“X”) and methanol feed (“M”), both representing fed batch. P _GAP to P _SHB17 were tested in ‘G’, ‘D’, ‘X’ and ‘M’ (A). P _TEF2 to P _PFK300 were validated in ‘D’, while P _GUT1, P _THI11 and MUT-related promoters were tested in putative repressed and induced conditions (‘D’/‘G’, ‘D’+/−100 μM thiamine and ‘D’/M, respectively) (B). P _GAP variants with alternative ‘-1’ bases were analyzed in glucose excess (‘D’). Relative eGFP levels are related to P _GAP- controlled expression (A, B), terminator ScCYC1tt (B), or presented as relative value (C)

Fig. 4

Kozak consensus sequences of P. pastoris and S. cerevisiae. The sequences were retrieved by the regulatory sequence analysis tool (RSAT) and illustrated using weblogo

Fig. 5

Genome integration efficiency of Golden Gate vectors without repetitive homologous sequences (a) and expression of eGFP obtained after integration into different genomic loci (b). Genome integration efficiency (fraction of positive clones) are shown for P. pastoris transformed with different Golden Gate vectors containing P _GAP-eGFP_ScCYC1tt: control plasmid (single eGFP), eGFP in position 2 with a repetitive transcription unit (TU1) in position 1 and 3 (‘loop-out’ control) and 4 quadruple combinations with eGFP in position 1, 2, 3 and 5 next to four other transcription units TU1–4) (a). Genome integration was verified by analyzing eGFP expression of each 16 (controls) and 22 clones (quadruple combinations), respectively. TU1–4 were cloned with the promoters and transcription terminators P_POR1/RPS3tt, P_PDC1/IDP1tt, P_ADH2/RPL2Att and P_MDH3/TDH3tt, respectively. Coding sequences of TU1–4 were derived from various non-essential intracellular-protein coding genes of P. pastoris with a length of 500–2000 bp. The influence of the genomic integration locus was analyzed using vectors containing only one transcription unit (P _GAP_eGFP_ScCYC1tt) (b). In both cases clones were screened under glucose surplus

Fig. 6

Overview of CRISPR/Cas9-BB3 plasmids assembled using GoldenPiCS. Transcription units for sgRNA (fusion of crRNA and tracrRNA [42]) and hcas9 (with SV40 nuclear localization sequence at its C-terminus) with various promoters and transcription terminators were assembled in BB3cK_AC (CEN/ARS locus, KanMX selection marker, fusion sites FsA-FsC). The plasmids were successfully used to generate InDel mutations in different genomic loci and eGFP in P. pastoris. The effects of the tested constructs are summarized in the table below the vector scheme. Modules which worked most efficiently are indicated in bold in the vector scheme