A supernumerary synthetic chromosome in Komagataella phaffii as a repository for extraneous genetic material

Abstract

Background

Komagataella phaffii (Pichia pastoris) is a methylotrophic commercially important non-conventional species of yeast that grows in a fermentor to exceptionally high densities on simple media and secretes recombinant proteins efficiently. Genetic engineering strategies are being explored in this organism to facilitate cost-effective biomanufacturing. Small, stable artificial chromosomes in K. phaffii could offer unique advantages by accommodating multiple integrations of extraneous genes and their promoters without accumulating perturbations of native chromosomes or exhausting the availability of selection markers.

Results

Here, we describe a linear “nano”chromosome (of 15–25 kb) that, according to whole-genome sequencing, persists in K. phaffii over many generations with a copy number per cell of one, provided non-homologous end joining is compromised (by KU70-knockout). The nanochromosome includes a copy of the centromere from K. phaffii chromosome 3, a K. phaffii-derived autonomously replicating sequence on either side of the centromere, and a pair of K. phaffii-like telomeres. It contains, within its q arm, a landing zone in which genes of interest alternate with long (approx. 1-kb) non-coding DNA chosen to facilitate homologous recombination and serve as spacers. The landing zone can be extended along the nanochromosome, in an inch-worming mode of sequential gene integrations, accompanied by recycling of just two antibiotic-resistance markers. The nanochromosome was used to express PDI, a gene encoding protein disulfide isomerase. Co-expression with PDI allowed the production, from a genomically integrated gene, of secreted murine complement factor H, a plasma protein containing 40 disulfide bonds. As further proof-of-principle, we co-expressed, from a nanochromosome, both PDI and a gene for GFP-tagged human complement factor H under the control of P_AOX1 and demonstrated that the secreted protein was active as a regulator of the complement system.

Conclusions

We have added K. phaffii to the list of organisms that can produce human proteins from genes carried on a stable, linear, artificial chromosome. We envisage using nanochromosomes as repositories for numerous extraneous genes, allowing intensive engineering of K. phaffii without compromising its genome or weakening the resulting strain.

Background

Numerous platforms for producing heterologous proteins are deployed across the biotherapeutics-manufacturing sector. Often purpose-designed, proprietary and under continuous development, protocols chiefly rely on mammalian cells and can be expensive and complex. An economical, high-yielding, general-purpose microbial platform could lower barriers to entry and widen availability of vaccines and protein therapeutics globally [1]. Robust evidence of efficacy and versatility would, however, be needed for it to supersede the biomanufacturing strategies to which much of the industry is now committed [2]. Prerequisites include strong promoters of gene expression, high-density low-cost cell culture and, crucially, a toolbox for precise genetic manipulations of the protein-producing cells. We wondered if the “biotech” yeast Komagataella phaffii (Pichia pastoris) [3], supplemented with a synthetic chromosome dedicated to genetic engineering, could meet all of these criteria.

Engineered yeast cells possess multiple attributes considered essential for industrial protein production [4]. Model yeast species such as Saccharomyces cerevisiae and non-conventional yeast such as Hansenula polymorpha, Yarrowia lipolytica and Kluyveromyces lactis [5] have been used commercially but K. phaffii [6, 7] is the most prominent example. Advantages include its potential to achieve exceptionally high cell density in simple media, efficient protein-secretory machinery, ability to perform some post-translational modifications, strong natural promoters and track record of pharmaceutical protein production [8]. Note that the P. pastoris genus was reclassified as Komagataella, which includes the subspecies K. phaffii and K. pastoris. Widely used strains are GS115 and K. phaffii CBS7435, as employed herein [9].

A hindrance to biomanufacture in K. phaffii and other yeasts is the need to integrate new genes into chromosomes. Hence sites within the native genome must serve as targets for the insertion of extraneous double-stranded DNA. This drawback is compounded in K. phaffii by a shortage of established tools for genetic editing compared to S. cerevisiae [10], and the nature of its DNA-repair machinery. In K. phaffii, unlike S. cerevisiae, the process of non-homologous end joining (NHEJ) for double-strand break (DSB) repair dominates over homologous recombination (HR) [11]. Compared to HR, NHEJ results in lower success rates for genetic deletions and insertions, and more off-target effects that can impact viability. On the positive side, lower efficiency of HR makes unintended recombination less likely [12].

One study [13] assessed the success of an HR-driven GFP-gene insertion designed to knock out AOX1 (creating a Mut^S strain) in K. phaffii CBS7435. Fewer than half of 800 clones, were, in fact, Mut^S strains with a gene-copy number of 1. About 5% were multi-copy clones and some had a gene-copy number > 10. Crucially, whole-genome sequencing (WGS) revealed non-canonical events including off-target gene disruption, co-integration of E. coli plasmid DNA and relocation of the AOX1 target locus to another chromosome [14]. Such outcomes are acceptable for one-off gene insertions because screening numerous colonies is feasible while an elevated gene-copy number might boost yields. But, because clonal variability per manipulation is multiplied by the number of manipulations, attempts to insert multiple genes in K. phaffii are laborious and risky. Attempts have been reported to use 1000 + base-pair long HR-promoting regions (LHRs), antimicrobial-resistance (AMR) markers, and genotyping to screen multiple clones following each manipulation [11]. However, reliance on LHRs restricts genomic sites suitable for integration, the set of available selection markers for K. phaffii is limited, and genotyping may fail to spot accumulating off-target effects. Since damage-risk escalates with each inserted gene, resulting strains become progressively depleted of available selection markers and are less likely to grow quickly to high densities or yield as much product. These considerations hamper extensive engineering of the K. phaffii genome.

Numerous strategies for facilitating genetic engineering in K. phaffii have been reported [7]. Many involve deleting or overexpressing genes aimed at suppressing NHEJ and enhancing HR. Shifting the balance of DSB repair towards HR also improved the efficacy of CRISPR/Cas9-type protocols in K. phaffii [11]. Regardless of mechanisms exploited to improve the accuracy of editing and repair, genomic integration relies on identification of a suitable neutral chromosomal locus for insertion of each gene. This must be a region that is accessible to transcriptional apparatus and compatible with efficient gene expression, and which can be disrupted without adversely impacting cellular physiology or metabolism [15].

In S. cerevisiae [16, 17] and Y. lipolytica [18] it was reported that multiple foreign genes could be incorporated into small, stable artificial chromosomes. The native 9.4-Mb K. phaffii genome consists of four chromosomes of 1.8, 2.3, 2.4 and 2.9 Mb with modular centromeres reminiscent of those in higher organisms [19], and telomeres consisting of 100–350 bp of homogenous repeated sequences [7]. We hypothesised that an additional or complementary route to facilitating genetic engineering in K. phaffii, while minimising damage to the native genome, would be to introduce a fifth, or supernumerary, chromosome, containing an appropriate origin of replication, centromere and telomeres, to serve as a dedicated receptacle for new genes.

Herein, we assembled a framework plasmid containing a centromere and an autonomously replicating sequence (ARS) (as well as a bacterial origin of replication, suitable for propagation in Escherichia coli cell culture). To this, we added a landing zone for new genes, and a pair of proto-telomeres (flanking an I–SceI cleavage site), to create a precursor plasmid. We planned to transform K. phaffii cells with this precursor plasmid that could be I-SceI-linearized in vivo, but subsequently reverted to in vitro, pre-transformation, I-SceI linearization to create our tiny (hence “nano”) prototype chromosome. On a wild-type background strain, WGS revealed nanochromosome instability. This was largely resolved by adding a second ARS and switching to a NHEJ-compromised strain of K. phaffii. We eventually arrived at a stable chassis upon which a series of more elaborate nanochromosomes were constructed in vivo by swapping or adding genes within the integration array. Finally, we showed that genes overexpressed from nanochromosomes supported production of hard-to-manufacture human proteins.

Results

Outline of the design and construction phase

Our goal was a stripped-down K. phaffii “nano”chromosome consisting only of a dedicated gene-landing zone plus those components essential for survival within the nucleus, for efficient and accurate duplication, and for reliable mitotic segregation. As reported below, we first ligated a K. phaffii ARS and a centromere part, along with an AMR gene, into pUC19 to create a framework plasmid (Fig. 1a and Additional file 1: Fig. S1). We added an initial version of the gene-landing zone and a proto-telomeres part, creating a nanochromosome-precursor plasmid. This was linearised by cleavage between the future telomeres, creating nanochromosome 1 (nChr 1) (Fig. 1b) that we used to transform K. phaffii cells. Subsequently we had to redesign our precursor plasmids (Fig. 1a) to improve nanochromosome stability, and switch to an NHEJ-deficient host strain of K. phaffii. In parallel, we established an assembly line for arrays of DNA segments (herein called integration and insertion arrays) designed to facilitate in vivo engineering of nanochromosomes through double-crossover HR.

Fig. 1

Nanochromosome construction and engineering (also see Additional file 1: Fig. S1). (a) In vitro and in E. coli, we assembled a centromeric, K. phaffii ARS-containing, framework. We added an initial gene-integration array (landing zone) and proto-telomeres, creating nanochromosome (nChr)-precursor version (v)1. We cleaved this between its proto-telomeres, and used the linear product - nChr 1 - to transform K. phaffii cells. To improve nChr stability, we extended nChr-precursor v1 and added a second ARS (for precursor v2), then replaced the initial integration array with new ones, creating precursors v2A and v2B. These yielded nChr 2A and nChr 2B, which were stable only in a ΔKU70 strain. Finally we explored the feasibility of engineering nChr 2A and nChr 2B (to Chr 2A.1 etc.) in vivo. Plasmids are not drawn to scale. (b) Schematic representations (drawn approximately to scale) of the synthetic, linear nanochromosomes as constructed in the current study. Lists of oligos, plasmids and strains may be found in Additional file 2: Tables 1, 2 and 3 respectively. Promoters and terminators are not shown. CEN = centromere, ncDNA = non-coding DNA, LHR = long HR-compatible region, mCH = mCherry gene, GFP:FH = gene coding for a GFP-FH fusion

A centromeric, ARS-containing framework plasmid

We opted to use DNA corresponding in sequence to one of the four native centromeres of K. phaffii [20] to serve as the nanochromosome centromere. We subsequently decided to proceed with an initial version of the nanochromosome in which we had succeeded in incorporating a centromere sequence identical to that of native chromosome 3 (henceforth, CEN3), and we did not test the other three possibilities. When creating the EcoRI-cleavage site, an additional nucleotide, G, was introduced in the CEN3 core region (Fig. 2a and Additional file: Fig. S2). This subsequently aided distinction between the native centromere and (nanochromosomal) CEN3 when analysing WGS results. Two segments were generated by PCR, from K. phaffii genomic (g)DNA, prior to combination by sequential cloning to create CEN3 in plasmid pUC19, yielding plasmid eDA24 (Fig. 2a). To supply an origin of replication, we selected DNA corresponding to PARS-A76 on K. phaffii Chr 1 because PARS-A76 was classified as efficient in comparison to other putative ARSs identified in the same study [21]. The sequence of PARS-A76 was inserted into pUC19 creating eDA26. We amplified zeocin- (Zeo^R, ble) and hygromycin-resistance (Hyg^R, hph) genes from commercial plasmids and inserted them into eDA26, forming episomal plasmids eDA40 (Zeo^R) and eDA37 (Hyg^R) (Additional file 1: Fig. S1). Gibson assembly of the following yielded the centromeric ARS-containing 10.5-kb framework plasmid, eDA53 (Fig. 2b): (i) a 1.4-kb Zeo^R part; (ii) 0.51-kb ARS-76; and (iii) the linearized, 8.85-kb, centromeric, plasmid eDA24 that thereby contributes CEN3, some pUC19-derived scaffolding DNA, a bacterial origin of replication and an ampicillin resistance (Amp^R, bla) gene.

Construction of the telomeres part (tel)

We used a “telomerator”-inspired approach [22] entailing insertion into the framework plasmid of a DNA sequence, which we called the telomeres part (Tel), comprised of native-like proto-telomeres flanking a restriction site. In our case, two sets of 16 inverted telomere-repeats (5’-TGGATGC-3’) bracketed an I-SceI-cleavage site (Fig. 2c and Additional file 1: Fig. S4). We chose 16 repeats as an intuitive compromise between the minimum number of repeats likely to be needed to protect the centromere, and the maximum length of DNA [7, 23] that could be assembled in the following convenient approach: two pairs of similar-length synthetic oligos were each annealed to form its 5’- and 3’-halves with complementary overhangs, and then the two halves were tandemly ligated into eDA40, forming eDA131 (Fig. 2c). I-SceI can linearise the resultant plasmid DNA molecule, yielding a molecule with telomeres at either end.

Fig. 2

Preparation of key parts.(a) Preparation of CEN3, using K. phaffii from gDNA. Left-hand gel: PCR amplicons of segment (Seg) 1 (oligos 197/198), and Seg 2 (oligos 199/200). This approach introduces one extra, functionally silent, base pair (G:C) in the core region (see Additional file 1: Fig. S2). DNA bands of expected sizes (boxed) were extracted and digested with appropriate restriction enzymes before sequential cloning into pUC19 to yield eDA24. Right-hand gel: restriction digestion of eDA24 confirming presence of 6.3-kb CEN3 (IR = inverted repeats). (b) Assembly of framework plasmid in a pUC19-derived scaffold (ori = bacterial origin of replication) (eDA53 validation shown in Additional file 1: Fig. S3). (c) The oligos used to construct the telomeres part (Tel, Additional file 1: Fig. S4). Each proto-telomere contains 16 copies of TGGATGC and the two are linked by the 18-bp I-SceI-recognition site. Lower schematic: cloning of Tel into eDA40 (eDA131 validation shown in Additional file 1: Fig. S5)

An assembly line for integration and insertion arrays

We built insertion arrays (donors) and integration arrays (acceptors) to allow precise nanochromosome engineering in K. phaffii cells. Each array includes Hyg^R or Zeo^R and features genes-of-interest (GoIs) flanked by LHRs. GoIs are expression cassettes containing a gene plus its promoter and terminator regions. LHR^A, LHR^B etc. are unique sequences of ~ 1000 bp from a library (LHRs^A−Z), with LHR^Z reserved for the terminus of each array. Thus, it should be possible to (for example) swap Hyg^R within the nanochromosome-resident integration array LHR^A-GoI¹-LHR^E-Hyg^R-LHR^Z for GoI²-LHR^D-Zeo^R by transforming cells with insertion array LHR^E-GoI²-LHR^D-Zeo^R-LHR^Z (Additional file 1: Fig. S6a). Subsequently, Zeo^R would be available for exchange with GoI³-LHR^E-Hyg^R, and so forth. Theoretically, repeated cycles of this “inch-worming” process would generate a string of evenly spaced genes (and their regulatory sequences). Recycling of a single pair of selection markers (Hyg^R and Zeo^R) is inherent to this strategy.

The following parts were deployed in ligase-assisted directional assembly of arrays within digested pUC19 (Additional file 1: Fig. S6b): LHRs^A−Z; synthetic codon-optimised GoIs; and Hyg^R or Zeo^R fused to LHR^Z. The assembled arrays were: (i) PCR-amplified using primers carrying 5’-blunt-end and 3’-SalI/XhoI restriction-enzyme sites to create integration arrays; or (ii) digested with AsiSI or FspI for insertion arrays.

Introduction of the initial gene-landing zone into the precursor plasmid

We introduced into our CEN3- and ARS-containing framework plasmid (eDA53) an initial integration array, LHR^A-Zeo^R-(P_AOX1)I-SceI(T_AOX1)-LHR^Z. The gene I-SceI encodes a restriction enzyme that recognises the cleavage site [24] located between proto-telomeres within Tel. Thus we anticipated that a future plasmid containing both I-SceI and Tel would self-linearise in vivo after I-SceI induction. This one-off integration array (unlike those used subsequently) was not preassembled, but was introduced stepwise by: (i) inserting LHR^A into eDA53 (yielding eDA71); (ii) constructing (P_AOX1)I-SceI(T_AOX1) in pPICZα B and amplifying by PCR; (iii) amplifying LHR^Z by PCR; and (iv) ligating the purified, BsmBI-digested products from step (ii) and (iii) with BsmBI-linearized, gel-purified, eDA71, yielding eDA83 (Fig. 3a).

Fig. 3

Production and testing of precursor plasmid v1 and nChr 1. (a) Incorporation of an initial, one-off, gene-landing pad into the precursor plasmid. Work on eDA83 was halted due to leaky I-SceI expression in E. coli. (b) Following deletion of I-SceI in eDA83 by insertion of Kan^R (creating eDA110), Tel was inserted to yield eDA137, the precursor plasmid (v1) of nChr 1. The plasmid was linearized in vitro and used to transform (“L&T” in figure) CBS7435 K. phaffii cells creating strain yDA122. The agarose gel shows I-SceI digestion of Tel-containing eDA137 (versus control, i.e. no-Tel, eDA110). (c)  Despite nChr 1 appearing stable in chromosome-loss assays performed on yDA122, WGS (and subsequently, colony PCR, see gel) revealed major chromosomal rearrangements as indicated in the schematic. Note, in yDA122, the loss of PCR-product 3 (of nChr 1), but retention of PCR-product 2 of the suspected translocation product nChr 1* (i.e. nChr 1(p):Chr 3(q)) 

A prototype nanochromosome

We aimed to make a nanochromosome-precursor plasmid by excising Tel from eDA131 and inserting it into eDA83’s Amp^R site. Disappointingly, the ligation product was unstable in E. coli due to leaky (P_AOX1)I-SceI expression. We abandoned in vivo plasmid linearization in favour of in vitro linearization by I-SceI, while opting to salvage our initial integration array. To this end, we replaced, in eDA83, P_AOX1 and part of the I-SceI ORF, with a bacterial Kan^R expression cassette, yielding eDA110 containing the array: LHR^A-Zeo^R-Kan^R(ΔI-SceI)-LHR^Z. We inserted Tel into eDA110, deleting Amp^R and yielding I-SceI-cleavable eDA137 (Fig. 3b). Unlike eDA83, eDA137 (with I-SceI effectively deleted) could be propagated in E. coli. We called this nanochromosome-precursor plasmid version 1 (precursor v1).

We transformed wild-type K. phaffii cells with I-SceI-linearized precursor v1 creating our prototype nanochromosome (nChr 1). We re-plated colonies that grew on zeocin and, based on colony-PCR, selected yDA122 for liquid culture in the presence of zeocin and used it to make a glycerol stock. We took cells from the stock and cultured them, to assay zeocin resistance after about ten generations on zeocin-free media, providing an indication of nChr 1 mitotic stability. Nine out of ten cells remained zeocin-resistant (Additional file 2: Table 5). For a more direct test of nChr 1 integrity, we submitted yDA122 cells, recovered from glycerol stock and cultured overnight, for WGS. Analysis by de novo assembly and visualization [25] revealed that nChr 1 had undergone a substantial, rearrangement (Fig. 3c and Additional file 1: Fig. S7a) entailing fusion of the p arm of nChr 1 with a copy of the Chr 3 ~ 34-kb p arm. While this conflicted with the PCR-based genotyping performed shortly after transformation, it agreed with PCR-based genotyping of yDA122 after multiple generations with no antimicrobial selection (Fig. 3c). Instability of nChr 1 was likewise evident from WGS of two other colonies, similarly prepared.

Extension of backbone failed to improve integrity

We wondered whether the telomere-to-centromere proximity within the p arm of nChr 1 might impair maintenance, or promote chromosomal aberrations [26]. We also conjectured that a second ARS might expedite duplication and/or segregation [23]. To address these possibilities, we reverted to eDA110 and inserted K. phaffii PARS-B413 [21] in tandem with some (1.3-kb) ncDNA, into the future p arm of the precursor plasmid to yield eDA146 (Fig. 4a). The AT-rich PARS-B413 was selected because it was classified as less efficient than the GC-rich PARS-A76 [20] already present in the q arm. Thus the resultant arrangement was designed to resemble (native) Chr 3 in which the centromere is sandwiched between (stronger) GC-rich PARS-C2216 and (weaker) AT-rich PARS-C2204 [27]. Into eDA146 we inserted Tel to create I-SceI-cleavable eDA155 (precursor v2) (Fig. 4a,b). PCR mapping of linearized eDA155 (i.e. nChr 2, Fig. 4c) validated its integrity.

Fig. 4

Observed aberrations of an extended neochromosome (nChr 2) in wild-type cells (a) The shorter (p) arm of nChr 1 was extended with non-coding (nc)DNA and a second ARS, added by cloning components into precursor v1 to create v2. (b) Agarose gel showing I-SceI digestion products for eDA146 (control) and eDA155. (c) Validation of nChr 2 by PCR before K. phaffii transformation (for oligonucleotides sets see Additional file 2: Table 1). (d) De novo assembly analysis of the WGS for the resultant strains (yDA174 and yDA175) revealed a preponderance of fused chromosomes indicating instability (see Fig. 5 and Additional file 1: Figs. S7 and S8). Strain yDA174 carries a triple-fusion between a nChr 2 q arm (i.e. lacking its p arm), another nChr 2 q arm; and a copy of nChr 2 lacking the p arm telomere; the junction formed between q arms involves the inverted-repeat regions (IRs, 99% sequence identity) of the centromeres but it was not possible to ascertain the nature of this fusion event. In strain yDA175 an additional copy of nChr 2, lacking its p arm telomere, appears fused to the triple-fusion structure seen in yDA174.

Thus, we linearized precursor v2/eDA155 in vitro and used the product, nChr 2, to transform K. phaffii cells. From colonies growing on zeocin, we picked yDA174 and yDA175. Chromosome-loss assays (Additional file 2: Table 5) were promising and the integrity of nChr 2 in these strains was supported initially by superficial analysis of PCR-based genotyping. However, an unanticipated PCR amplicon was noted (Additional file 1: Fig. S8) while subsequent WGS (Fig. 5), de novo assembly (Additional file 3) and coverage analysis revealed fused end-to-end, di- or tricentric DNA molecules, consistent with this anomaly, in which telomere repeats occur exclusively at termini (Fig. 4d).

Fig. 5

Whole-genome sequencing coverage data for examples of wild-type and ΔKU70 K. phaffii cells containing nanochromosomes (Also see Additional file 1: Fig. S7) The y-axes show “normalised coverage” values obtained by dividing the coverage of each base pair by the mean coverage of all base pairs in the respective sample. This allows comparisons, and identification of duplicated segments or deletions. The x-axes show the position (genomic coordinate) of each base pair in its respective chromosome (values in kilobases for native chromosomes). Each nanochromosome is shown as a schematic (colour-coding as in Fig. 1b), drawn to the same scale as the x-axis in each case. * Indicates a centromere. (.) Indicates anomalies attributable to native genome rearrangements (e.g. duplicated loci) or artifacts arising from highly repetitive regions such as found in telomeres. (X) Indicates a ~ 1-kb duplicated region of Chr 2 corresponding to an ORF encoding an unknown protein (inexplicably observed exclusively in ΔKU70 strains transformed with insertion arrays). Red boxes, drawn on the nanochromosome schematics, indicate CEN3 core regions characterised by the signature presence of an extra G:C base pair. Upper panel: For yDA175, on wild-type background, a plot of normalized coverage indicates two-to-four copies of nanochromosomal sequence per cell. This elevated normalised coverage values for nChr 2 suggest multiple copies per cell of its DNA content. The high copy-number for the Chr 3 centromere correlates with a high copy-number of nChr 2 and supports the existence of chimeric multi-centric nanochromosomes (a tri-centric model is suggested in Additional file 1: Fig. S8). This observation is compatible with yDA175 and yDA177 de novo assembly results obtained from long-read WGS (Additional file 3). Middle and lower panels: Both these strains on a ΔKU70 background are consistent with a single copy per cell of its nanochromosome, specifically: yDA253 with a single copy of nChr 2A; and yDA275 with a single copy of nChr 2A.2 (i.e. after the inch-worming proof-of-principle experiment). [(BioProject PRJNA971544)]

Switch to an NHEJ-deficient host strain

While HR events could occur between telomeres, these should lead to elongation or shortening rather than fusion, provided that telomere length exceeds a threshold value [28]. Given that HR repair is, in any case, inefficient in K. phaffii, we suspected that the observed end-to-end fusion of nanochromosomes in wild-type cells was a result of NHEJ. Multicentric nanochromosomes are regarded as unstable and, if not resolved, a source of further genomic abnormalities. Although, our multicentric entities appeared to persist over generations, they were deemed unsuitable as expression platforms. We therefore explored the use of a host strain in which NHEJ is impaired by KU70-knockout [11, 29] (Additional file 1: Fig. S9a) but which remained viable in cultures over multiple days at 30 °C (Additional file 1: Fig. S9b).

Attempts to maintain nChr 1 in ΔKU70 cells proved no more successful than they had been in wild-type, whereas nChr 2 appeared to be stable in the NHEJ-compromised cells (Additional file 1: Fig. S10). We therefore focused on nChr2. To better compare nanochromosome engineering in wild-type versus ΔKU70 strains, we chose to create more useful, versatile versions. We therefore excised the integration array (LHR^A-Zeo^R-Kan^R(ΔI-SceI)-LHR^Z) of precursor v2 and replaced it with either: (i) a triple-LHR integration array expressing a methanol-inducible PDI gene, and Hyg^R, i.e. LHR^A-PDI^H-LHR^E-Hyg^R-LHR^Z (for precursor v2A); or (ii) a double-LHR array carrying constitutively expressed (P_TEF1)GFP(T_CYC1) paired with Zeo^R, i.e. LHR^E-GFP-Zeo^R-LHR^Z (for precursor v2B) (Fig. 6). The rationale for introducing the PDI^H gene, encoding His-tagged protein disulfide isomerase (PDI), into the first array is explained below. We linearized precursor v2B (creating nChr 2B), and precursor v2A (creating nChr 2A), and transformed both wild-type and ΔKU70 K. phaffii cells with the products. Colony PCRs (on freshly transformed cells growing on agar plates) and chromosome-loss assays (Additional file 2: Table 5) suggested initial stability of nChrs 2A and 2B in both strains. Subsequently, however, WGS and PCR genotyping revealed that nChrs 2A and 2B were stable, longer term, in ΔKU70 strains (yDA253 and yDA218) (Fig. 5 and Additional file 1: Figs. S7 and S11) but were unstable on the wild-type background. Moreover, after protracted cell culture, the wild-type background strains that initially passed PCR-based genotyping, failed (Additional file 1: Fig. S11) despite performing well in chromosome-loss assays, suggesting integration of the marker into the genome.

Fig. 6

New versions of precursor plasmids created by swapping out the original integration array. PCR products (see agarose gel) from eDA197 or eDA227, digested with AfeI and XhoI were ligated into appropriately digested eDA155. Precursors v2A and v2B were linearized with I-SceI then used to transform wild-type or ΔKU70 cells to create a total of four strains (see also Additional file 1: Fig. S1)

The ability to engineer nanochromosomes depended on knocking out KU70

To exemplify nanochromosome engineering, we sought to replace GFP and Zeo^R within the integration array (LHR^E-GFP-Zeo^R-LHR^Z) of nChr 2B, by mCH (codes for mCherry) and Hyg^R delivered to the cell within an insertion array. We thus transformed yDA177 (despite subsequently emerging evidence for low nChr stability on the wild-type background) and yDA218 (ΔKU70 background) with the array LHR^E-mCH-Hyg^R-LHR^Z, aiming to create nChr 2B.1 (Additional file 1: Fig. S12a), and we selected colonies (34 from the wild-type, and 16 from ΔKU70 strains, in two independent experiments) on hygromycin-agar plates. We screened these for a combination of hygromycin resistance/zeocin sensitivity, and for red fluorescence/lack of green fluorescence, and then by PCR genotyping (data not shown). A total of 15 of 16 colonies on the ΔKU70 background (including yDA226) were positively verified, but only 1 of 34 colonies on the wild-type background, passed all tests. This is consistent with the subsequently analysed PCR genotyping results and WGS analysis (data not shown) for yDA177 showing that the nChr 2B had already been compromised by aberrant fusions (Additional file 1: Figs. S11 and S12a). A qPCR analysis of yDA226 (ΔKU70) indicated one copy of nChr 2B.1 per cell (Additional file 2: Table 6).

In related proof-of-principle experiments we swapped Hyg^R in LHR^A-PDI^H-LHR^E-Hyg^R-LHR^Z of nChr 2A - in yDA232 (wild-type) and yDA253 (ΔKU70) - for the tandem pair GFP-Zeo^R within insertion array LHR^E-GFP-Zeo^R-LHR^Z, aiming to produce nChr 2A.1 with an extended landing zone (Additional file 1: Fig. S12b). We obtained no positively verified colonies on the wild-type background but validated (zeocin-resistant/hygromycin-sensitive, GFP-producing) eleven colonies (Additional file 1: Fig. S12b) for ΔKU70 strains (including yDA263). Subsequent WGS of yDA263 confirmed integrity of nChr 2A.1 (Additional file 1: Fig. S7e) while qPCR was consistent with a single copy per cell (Additional file 2: Table 6).

Nanochromosomes support methanol-induced protein production

We investigated the potential of a nanochromosome-resident chaperonin-coding gene to facilitate protein production from a methanol-inducible heterologous gene that was integrated (using classical techniques) into the native K. phaffii genome (Fig. 7a). We had previously found, in K. phaffii cell cultures, that overexpression of PDI as well as a gene (mFH) for murine complement factor H (mFH) (both of which were genome-integrated and under control of P_AOX1) was required for detectable levels of secretion of mFH, which has 40 disulfide bonds [30]. Hence, we tested if (P_AOX1)PDI^H(T_AOX1) on nChr 2 A could enhance mFH production in a similar manner, which would require the nanochromosome to persist throughout both growth and induction phases of the culture. We created as a positive control, nanochromosome-null strain yDA264, in which both PDI^H and mFH are integrated into the native genome of the ΔKU70 parental strain. For the experimental strain we took yDA250, in which only mFH had been integrated into the native genome, and transformed it with nChr 2A generating strain yDA260. Integrity and stability of nChr 2A in yDA260 and yDA250 was verified by PCR-based genotyping (Additional file 1: Fig. S13) and chromosome-loss assay (Additional file 2: Table 5) and WGS (Additional file 1: Fig. S7c,d). We used yDA250 (expressing mFH at a low yield, without PDI support) as a negative control.

Small-scale mFH-production tests were performed in baffled flasks, with methanol induction at day 3 (Fig. 7b). As expected, Coomassie staining of SDS-PAGE gels revealed no mFH-candidate bands for yDA250, which lacks PDI^H (Fig. 7b). Conversely, a band for mFH was detected in the culture medium of mFH⁺ strains providing its production is supported by PDI^H and regardless of whether PDI^H is carried within the genome (yDA264) or on nChr 2A (yDA260) (Fig. 7b). The presence of His-tagged PDI in cultures of these strains was confirmed by Western blotting (Fig. 7c). Note that qPCR assay showed single copies of mFH and PDI^H genes in both yDA260 and yDA264 strains (Additional file 2: Table 6). The recombinant mFH, secreted into the yDA264/yDA260 culture medium but not purified, is functionally active in an assay in which it acts as a cofactor for complement factor I (FI)-catalysed cleavage of the α’-chain of the complement protein, C3b, generating cleavage products detectable on an SDS-PAGE gel (Fig. 7e). No cleavage products were obtained in the absence of FI or when testing pre-induction cultures. Importantly, WGS validated the stability of nChr 2A (Additional file 1: Fig. S7c,d), in cells even after the induction phase, as well as the presence of a single copy of the mFH-expression cassette integrated into the P_AOX1 site on (native) Chr 4, in agreement with qPCR.

Demonstration of “inch-worming” for gene integration and protein production

We excised the insertion array LHR^E-GFP:FH-LHR^D-Zeo^R-LHR^Z (from eDA250) that includes the 4.4-kb GoI (pAOX1)GFP:FH. This GoI encodes a fusion protein in which green fluorescent protein (GFP) is fused to the Nterminus of human complement factor H (FH). Like mFH, this is a challenging protein to produce [30] recombinantly because it contains 40 disulfides. We used this insertion array to transform strains yDA232 (wild-type) and yDA253 (ΔKU70), both containing nChr 2A carrying the integration array LHR^A-PDI^H-LHR^E-Hyg^R-LHR^Z. Double-crossover HR should preserve PDI^H but promote insertion into nChr 2A of GFP:FH as well as Zeo^R (replacing Hyg^R), yielding nChr 2A.2 (Additional file 1: Fig. S14). It should also introduce a new landing site between LHR^E and LHR^Z. We selected for transformed cells with zeocin, and then screened for single colonies possessing dual zeocin resistance/hygromycin sensitivity. We then screened colonies using PCR-genotyping (Additional file 1: Fig. S14) and chromosome-loss assays (Additional file 2: Table 5). No colonies were obtained on the wild-type background but in ΔKU70 strains, ten colonies passed screening (Additional file 1: Figs. S14 and S15). Two of these, yDA275 and yDA277, were submitted for WGS, and data analysis showed a stable, single copy of nChr 2A.2 (Fig. 5; Additional file 1: Fig. S7g, Additional file 3), also confirmed by qPCR (Additional file 2: Table 6). We cultured yDA275 and yDA277 in small-scale trials of methanol-induced GFP:FH production. Yields of protein (resolved, after treatment with EndoH_f, on SDS-PAGE, and verified by Western blot) (Fig. 7d) were comparable with those obtained previously from strains of K. phaffii where PDI^H and FH were integrated into the native genome (data not shown). Moreover, the recombinant GFP:FH secreted into the yDA275/7 culture medium is functionally active in FI-cofactor assays (Fig. 7e). Crucially, the newly created landing site in Chr 2A.2, between LHR^E and LHR^Z, would be available for the next of what could be multiple rounds of sequential gene insertions and hence an “inch-worming” mode of nanochromosome extension.

Fig. 7

Gene expression from the nanochromosome supports production of a traditionally integrated extraneous gene, or can be used directly to produce recombinant protein.(a) Expression of PDI^H on nChr 2A could assist formation of the 40 disulfide bonds of mFH encoded by DNA integrated into Chr 4. (b) SDS-PAGE (Coomassie staining) of crude cell-culture media (30 µL, reducing conditions, treated with Endo H_f) before (d0), and three days after (d3), methanol induction. As expected, co-expression of PDI^H located on either nChr 2A or (as a control) a native chromosome, is required for detectable quantities of mFH to be secreted. (c) A Western blot was used to demonstrate PDI^H production in cells containing nChr 2A three days post-induction. (d) A 165-kDa fusion GFP:FH fusion protein is detectable three days after induction in the cell-culture media from two strains containing nChr 2A.2 carrying both PDI^H and GFP:FH. In each lane 40 µL of EndoH_f–treated crude culture medium (after cell removal) was loaded under reducing conditions. Bands were detected with Coomassie blue and by Western blot using an anti-GFP primary antibody that also revealed the presence of a GFP-containing proteolytic fragment. (e) Both mFH and GFP:FH produced herein are cofactors for (30 nM) complement FI-catalysed cleavage of the 110-kDa α’-chain of complement protein C3b into 63-kD and 39-kDa fragments. In each case 12 µL of crude culture medium (at d(ay) 3 or, as a control, at d(ay) 0) was assayed. A potential contribution of non-specific protease activity is excluded by performing control reactions lacking FI.

Discussion

We have demonstrated stable protein production from DNA on a supernumerary chromosome in a ΔKU70 strain of the “biotechnology” yeast K. phaffii. This miniscule synthetic linear chromosome persists even after growing cells to high density on glucose followed by transfer to methanol-rich media for induction of heterologous gene expression. At 15–25 kb, it is smaller than even the smallest S. cerevisiae chromosome (250 kb), while so-called “microchromosomes” of birds and reptiles contain > 1000 kb [31]. Hence, we adopted the term “nano” chromosomes for our constructs. Telomere-capped gene-sized DNA molecules in the macronucleus of some single-celled ciliates were also called nanochromosomes, but lack centromeres and do not segregate [32]. In S. cerevisiae, which has highly efficient HR-based DSB-repair mechanisms, “yeast artificial chromosomes” were developed as vehicles for Mb-sized segments of foreign DNA and used early in the human genome project and for generating transgenic animals [33].

Recent work, demonstrated the feasibility of introducing centromeric and telomere-protected “neo”chromosomes into S. cerevisiae and then exploiting this organism’s highly efficient HR machinery to assemble multiple-gene orthogonal expression platforms [17, 34] A study in Y. lipolytica achieved a similar goal despite this organism being less efficient at HR than S. cerevisiae and more efficient at NHEJ. In the case of Y. lipolytica, HR proceeded more efficiently in the synthetic neochromosome than in native chromosomes [18].

To our knowledge, ours is the first report of a stable synthetic chromosome in K. phaffii, and the first WGS-verified example that survives without selective pressure or the possession of essential genes. We did not establish how many copies of the nanochromosome exist per cell immediately following transformation or how many reach and enter the nucleus, but our WGS and qPCR suggest an average of one copy per cell in cultures after a few dozen generations. During mitosis, its small size will restrict the number of cohesin molecules that can encircle a nanochromosome, which might cause premature separation of sister chromatids and spindle detachment [35]. Hence it will be important to check copy numbers over time, and retain the option of transferring an essential gene onto the nanochromosome. The presence of an ARS in both arms of the nanochromosome (flanking CEN3) enhanced stability but we did not explore relationships between persistence and the number, spacing, sequence or positioning of ARSs in a rigorous or systematic way. We have not explored nanochromosome behaviour during meiosis but this could be relevant to future prospects for using mating as a strategy for transferring it between strains. Of note is that K. phaffii has been suggested as a model organism for research on eukaryotic molecular cell biology. For example, its large, modular centromeres are reminiscent of those of higher organisms, unlike the 125-bp centromeres of S. cerevisiae [7]. Hence these and other studies of the nanochromosome (e.g. of chromatin structure and remodelling, and structure-function relationships of telomeres, and subtelomeric and pericentromeric regions) might have relevance for chromosome biology more widely.

On a wild-type background, analysis of WGS suggested nanochromosomes undergo translocation or fusion, creating aberrant new chromosomes. In the case of nChr 1 we observed translocation with the acrocentric K. phaffii Chr 3, reminiscent of centric translocation (Fig. 3). We found that nChr 1 was also unstable in ΔKU70 strains. Compared to nChr 1, nChr 2 has a longer p arm containing the second ARS (Fig. 1b). Interpretation of WGS results revealed that - on a wild-type background - nChr 2 formed chromosomes with more than one centromere (Fig. 4). In the breakage-fusion-bridge (BFB) cycle [36], chromosomes that lose telomeres may, following duplication and during anaphase, form end-to-end bridged dimeric chromatids [37]. Indeed, NHEJ-mediated formation of dicentric chromosomes has been reported in yeast [36, 38, 39]. Multimeric chromosomes are normally resolved, over multiple BFB cycles, because two centromeres on a mutual chromatid have a 50% chance of being pulled towards opposing spindle poles [37, 40], inducing a DSB and hence opportunities for repair via various mechanisms. In our fused nanochromosomes, the centromeres (~ 10 kb apart) may be too close together for attachment to different kinetochores, so are not resolved in this way [37]. No fused versions of nChr 2 were identified in the ΔKU70 strain although we did not confirm that this was a direct consequence of the knockout. Nor did we yet try restoring KU70 once the nanochromosome has become established in the nucleus.

Thus, a current limitation of our platform is the need to delete KU70, which encodes a DNA-binding protein within Ku70/Ku80 that is important for NHEJ. This might compromise its use in biomanufacture as it was reported that ΔKU70 strains of K. phaffii may lack robustness and be prone to large genome deletions or depressed colony numbers when attempting genetic manipulations [41, 42]. Moreover, deletion- or insertion-promoting alternative DSB-repair mechanisms, such as microhomology-mediated end joining or single-strand annealing, might prevail when neither NHEJ (due to knockout) nor HR (due to inherent low efficiency) are adequate [43]. We encountered no such issues with our strains (Additional file 1: Fig. S9B) but we did not test stability in a commercial setting and hence we cannot exclude the possibility that alternatives to depleting Ku70 might be needed in a biomanufacturing context. Fortunately, there are numerous possibilities. Recognition of DSBs by Ku70/Ku80 both recruits other NHEJ-critical proteins and blocks 5’-resection. Removing this block, by Ku70 depletion, allows resection to commence. Rapid binding of the resultant single-stranded DNA by recombinase Rad51, and its facilitator Rad52, launches HR and shuts down NHEJ [44]. Interestingly, in K. phaffii, genes for Ku70/Ku80 are more strongly expressed than the gene for Rad52 [42], while the inverse pertains in S. cerevisiae [45]. Consistently, overexpression of RAD51 and RAD52, or deletion of MPH1 (that unwinds D-loops made by Rad51) enhances HR in K. phaffii and improves success rates of CRISPR/Cas9-assisted gene integration [42]. A further gain in HR-mediated repair, post-Cas9 cleavage, was obtained by fusing Cas9 with endogenous Mre11 that (as part of Mre11-Rad50-Xsr2) trims back one of the strands at a DSB to create 3’-tails. Deletion of NHEJ-related protein, DNA ligase IV (Dnl4p), also improved HR efficiency in K. phaffii [46] and the ΔDnl4p strain was amenable to in vivo assembly of DNA fragments with concurrent integration at target sites [47]. Thus multiple alternatives, or complements, to depriving nanochromosome-possessing cells of Ku70 could be explored.

Despite the limitations of this study, our nanochromosome design has significant potential as a future robust and versatile protein-production platform. The small size of precursor plasmids required for nanochromosome construction ensures easy assembly. Insertion arrays are modular and readily constructed (Additional file 1: Fig. S6). In both cases, the process could be automated. The landing-zone design allows for effectively indefinite expansion using just two selection markers accompanied by the positioning of defined non-coding, DNA spacers between integrated genes, could facilitate metabolic engineering in K. phaffii. Further landing zones could, presumably, be added. There is also potential to employ, for convenience, shorter LHRs/spacers, while the prospect of in vivo sequential assembly, via transformation with overlapping insertion arrays, could be tested. The use of IRES-like sequences or bi-directional promoters to co-express multiple genes could be explored as could editing of the nanochromosome with CRISPR-Cas9.

The chemically defined nature of nanochromosomes could enhance the precision with which artificial genetic circuits and networks are assembled, facilitate progress in biocatalyst design, and improve quantitative comparisons of genetic variants. But K. phaffii strains containing supernumerary chromosomes will probably be most useful in biotherapeutics and vaccines applications that require expression of multiple foreign genes and where there is a mandate to cost-effectively biomanufacture large quantities of homogenous high-quality materials. Examples might include proteins with complex and extensive post-translational modifications (PTMs) such as disulfides and N-glycans, multi-protein or multi-subunit complexes [28] and virus-like particles [40]. For example, a library of selectively inducible genes for chaperonins and PTM-pathway enzymes, such as glycosyltransferases [48], could, in theory, be installed on an adjunct or “supporter” nanochromosome, although we have not yet explored how many genes a nanochromosome could carry. Subsequently, a codon-optimised gene coding for a protein of choice could be integrated into the native genome, using classical techniques. Alternatively, a bespoke array of heterologous genes could be positioned in the nanochromosome’s landing zone without engineering the native genome at all. We demonstrated minimalist versions of both strategies in the current study. Secreted yields of mFH, from a codon-optimised gene within the native genome, became detectable following PDI expression from a supporting nanochromosome, while human (GFP-tagged) FH and (His-tagged) PDI were co-produced from adjacent genes on a bespoke nanochromosome. We have not so far conducted systematic, side-by-side, quantitative comparisons of the protein production yields achievable from our strains versus more traditionally engineered ones. To maximise yields of recombinant proteins, the following could be varied and optimised, potentially in an automated fashion, and with more control and precision compared to equivalent investigations in native chromosomes: choice of nanochromosomal centromere and ARSs, promoter selection, synonymous codon usage, gene-copy numbers, spacing between genes, and the positioning of landing zone(s) relative to centromere, ARSs, and telomeres.

Conclusions

In conclusion, we have added the biotech-friendly yeast K. phaffii to a small but growing list of organisms in which a synthetic linear chromosome serves as a stable source of genetic information that is orthogonal to the native genome. While highlighting the need for further development and optimisation, we have demonstrated a potential application of nanochromosome-containing strains of K. phaffii for the manufacture of challenging-to-make proteins.

Methods

Materials

The following were sourced from New England Biolabs (NEB): type II and type IIS restriction enzymes, Q5 high-fidelity DNA polymerase, T4 and T7 DNA ligases, T4 DNA polynucleotide kinase and Antarctic phosphatase. Synthetic genes were ordered from GeneArt and plasmids purchased from Thermo Fisher Scientific. Oligonucleotides (oligos) (Additional file 2: Table 1) were ordered from Integrated DNA Technologies. E. coli TOP10 cells purchased from Thermo Fisher Scientific were used for sub-cloning. Wild type K. phaffii cells were sourced from the ATCC strains library. Non-coding DNA segments of nanochromosomes, used for HR and as fillers or spacers, were derived from a proprietary library of 8-kb synthetic DNA chunks deposited in a GeneArt plasmid (ordered from Thermo Fisher Scientific) that was custom designed for construction of yeast synthetic chromosomes (unpublished).

Gibson assembly

Gibson-assembly reactions were performed according to manufacturer’s (NEB) protocols. Sanger sequencing allowed verification of all DNA constructs and gel-extracted PCR products (Azenta Life Sciences). Plasmidsaurus (SNPsaurus) performed full-length sequencing of plasmids. All DNA constructs sequences are available in Additional file 3.

Transformations

Rubidium chloride-preparation of chemically competent TOP10 E. coli cells (Thermo Fisher Scientific), and their heat-shock transformations with DNA, was conducted according to standard protocols (http://currentprotocols.onlinelibrary.wiley.com). Transformed E. coli cells were selected on LB agar medium with an appropriate antibiotic. Transformations by electroporation of K. phaffii cells with precursor plasmids were performed according to a modification of a published protocol [49]. Briefly, 1–10 µg of DNA in 10 µL were added to 90 µL of pretreated competent K. phaffii cells prior to electroporation, and the volume adjusted to 1.0 mL with yeast extract peptone dextrose (YPD) medium containing 1.0 M sorbitol (YPDS). Following recovery (four hours, 30 °C), cells were spread on YPDS agar plates with a suitable antibiotic and incubated (30 °C) for between two and (exceptionally) five days. Single colonies were then transferred to fresh YPD agar plates in preparation for PCR-based genotype screening. Finally, verified strains were grown on YPD medium ahead of the next experimental step or for deep-freeze storage in glycerol.

Cell cultures

Cultures of E. coli were performed (30 °C) on LB medium supplemented with antibiotic at 50 µg/mL in the cases of zeocin and ampicillin or 100 µg/mL for hygromycin. Plasmid- and DNA-gel extractions were done with kits (Qiagen) as per manufacturer’s protocols. Yeast cells were cultivated on yeast extract, peptone with a carbon source that depended on the desired outcome (glucose for propagation; glycerol or methanol for induction). This was supplemented with 0.3 mg/mL zeocin or hygromycin.

Yeast genomic DNA extraction

Extraction of genomic DNA (gDNA) from K. phaffii cells was carried out according to a protocol supplied with the Epicentre MasterPure Yeast DNA-Purification Kit from Bioresearch Technologies, but modified by inclusion of an additional RNase A treatment and a purification step using Zymo-Spin III columns (Zymo-Research). Purity of gDNA was assessed using either a Denovix DS-11 spectrophotometer or the Qubit 3 fluorometer (Thermo Fisher Scientific) and dsDNA Broad Range Assay Kit (Thermo Fisher Scientific).

PCR-based genotyping of K. phaffii strains

Yeast-colony PCR was performed using green GoTaq Master Mix (Promega) as per the manufacturer’s protocol (https://doi.org/10.17504/protocols.io.bp2l69p95lqe/v1) and employing oligos (Additional file 2: Table 1) as forward and reverse primers to target appropriate regions of nanochromosomes or their precursors. The sizes of the resulting amplicons were confirmed on agarose gels.

qPCR for gene-copy number estimation

Estimates of gene-copy number were obtained using qPCR reactions, and according to the Luna Universal qPCR Master Mix protocol (NEB). Reaction volumes of 10 µL, in 96-well plates, were analysed using the StepOnePlus real-time PCR System (Thermo Fisher Scientific). Values for C_T were inferred from two biological repeats with three technical duplicates, and calculated using StepOnePlus software. Gene-copy number values were inferred by comparisons with two “housekeeping” genes, ACT1 and PDH-PDA, assumed to be single-copy.

Chromosome-loss assays

A protocol similar to one reported previously [50] enabled quantification of nanochromosome loss over multiple mitotic events. Briefly, K. phaffii cells with the nanochromosome were initially inoculated into rich media containing the appropriate antimicrobial agent and incubated (30 °C) overnight in a shaking incubator. Cells were then transferred to a medium [1% w/v yeast extract (Sigma-Aldrich), 2% w/v peptone (Sigma-Aldrich), 0.7% w/v yeast nitrogen base without amino acids (Formedium Ltd) with 2% v/v glycerol and 1% v/v methanol], to emulate inducible recombinant protein production, with no antibiotic, and incubated (30 °C) for at least ten cell divisions. Then triplicate aliquots (500–600 cells) were spread on agar plates with or without antibiotic. Retention of AMR was equated to retention of the nanochromosome (on which the AMR gene had been delivered into the cell). Hence the numbers of colonies growing on the two plates were counted and compared, with the help of ImageJ64 software [51]. Note that this assay does not distinguish between scenarios in which (i) the AMR gene remains on the nanochromosome versus (ii) its integration into the native genome.

Whole-genome sequencing

Illumina NextSeq (Novogene) performed WGS to generate “short” 150-nucleotide paired-end reads. Samples of gDNA were prepared as above. WGS-library preparation, and bioinformatic analysis, were performed by Novogene. The Integrative Genome Viewer 2.12.2 software (Broad Institute) was used for visualization, in Genome Browser, of mapped WGS data. Alternatively, BigWigs-formatted files were created from bedGraph files. Coverage plots were generated with custom R scripts based on data extracted from BAM files. Values were normalized to the mean in order to visualize relative coverage. To smoothen graphs a moving-window average function was deployed with a window of 2.5 kb and a step size of 200 bp. For longer reads, WGS was performed using a MinION Mk1B device (Oxford Nanopore Technologies). Briefly, gDNA was obtained using the NucleoBond HMW DNA kit (Macherey-Nagel) as per manufacturer’s guidelines, using lyticase (Sigma) for cell lysis and incubated (one hour, 37 °C) in Y1 buffer (1.0 M sorbitol, 100 mM EDTA, pH 8.0, 14 mM 𝛽-mercaptoethanol). DNA quality and concentration were assessed using a Qubit 3 Fluorimeter. Library preparation for Nanopore sequencing was performed using the Ligation Sequencing Kit SQK-LSK109 with Native Barcoding Kits EXP-NDB104 and EXP-NDB114 for multiplexing (Oxford Nanopore Technologies). Kits were used according to the manufacturer’s guidelines, except that DNA shearing was not applied and hence input DNA was increased fivefold to match the molarity expected in the protocol. Sequencing on the MinION Mk1B used MinION Flow Cells [FLO-MIN111 (R10)] or Flongle Cells [FLO-FLG001 (R9.4.1)]. Base calling was performed using Guppy [version:6.0.1 + 652ffd179] and reads mapped against the reference using minimap2 [version: 2.17-r941] [52]. Canu [version: 2.2] was used for de novo assembly [53]. All raw reads are deposited at BioProject PRJNA971544 and an overview of de novo assembly sequenced strains may be found in Additional file 2: Table 3 and Additional file 3.

Small-scale protein production

Cell culture for small-scale tests of recombinant protein production was performed as described [30]. To assess production of GFP:FH, samples were taken from the culture medium, after spinning out cells, and Endo H_f-treated to trim back yeast N-glycans. Then SDS-PAGE loading buffer was added and gel electrophoresis performed under reducing, denaturing conditions using 4–12% w/v Bis-Tris NuPAGE gels (Thermo Fisher Scientific) with Coomassie Blue staining. Bands of appropriate size were verified by Western blot (electrophoretic wet transfer performed on nitrocellulose membranes, in Tris-CAPS buffer using, the BioRad Mini-PROTEAN system) with anti-GFP primary antibody (Abcam, ab38689) and IRDye 800CW donkey anti-mouse IgG as a secondary (Li-COR Biosciences). Samples of secreted mFH were similarly prepared and analysed, but gel bands were identified by reference to a sample of human plasma-derived FH (Complement Technologies). To identify recombinant His-tagged PDI, cell pellets were lysed in SDS-PAGE-loading prior to gel electrophoresis. Bands were verified by Western blot, using anti-His tag monoclonal Ab, H1029 (Scientific Labs) as a primary, and IRDye 800CW as a secondary antibody. Signal was detected on the Odyssey CLx Imaging System (Li-COR Biosciences).

Assay of FH cofactor activity

Recombinant mFH was assayed for its activity as a cofactor for cleavage of the human C3b α′-chain by human FI. The 21-µL reaction volume contained 1.7 µM C3b, 14 nM FI (Complement Technology) in PBS buffer, pH 7.0, and various volumes of culture medium. The reaction mixture was incubated (one hour, 37 °C), before addition of SDS-PAGE loading buffer. After boiling (five minutes), electrophoresis was performed on NuPAGE (4–12%) gels and stained with Coomassie Brilliant Blue G250. Densitometry values were inferred using Image Lab software (Bio-Rad). Plasma-derived human FH, at 10–80 nM, served as a positive control and for calibration.

The construction of parts

Note that all DNA parts used for assembly of nanochromosome precursors are free of restriction sites for SalI and/or XhoI (needed to produce compatible cohesive ends), as well as AsiSI, BsaI and/or BsmBI (used in Golden Gate-like assembly). Oligonucleotides sequences were deposited in Additional file 2: Table 1. Plasmids are listed in Additional file 2: Table 2. Selected plasmids were sequenced; DNA sequences are available in Additional file 3. Additional file 1: Fig. S1 includes a schematic summarizing the workflow used to generate our nanochromosomes.

To create the centromere part (CEN3), a DNA sequence corresponding to the centromere of K. phaffii Chr 3 was generated initially as two segments, in separate PCRs, using K. phaffii genomic DNA as a template. The 3.2-kb segment 1 was amplified (oligos 197/198) and ligated between pUC19 KpnI and EcoRI sites, generating plasmid eDA10. The 3.1-kb segment 2 was amplified (oligos 199/200) and ligated into eDA10 between EcoRI and NdeI sites. This generated plasmid eDA24 carrying the 6.3-kb CEN3.

Based on previous reports [54], two K. phaffii ARSs [23, 27] were eventually selected to serve as origins of replication, one per nanochromosome arm. A 233-bp DNA segment corresponding to PARS-A76 was PCR-amplified from K. phaffii gDNA (oligos 217/218) and inserted into the BamHI site of pUC19 creating eDA26. Separately, a 530-bp molecule equating to PARS-B413 was likewise PCR amplified (oligos 238/239) and inserted into pUC19 yielding eDA41.

Expression cassettes for AMR selection-markers were amplified from commercial vectors. Zeo^R, between P_TEF1 and T_CYC1, was PCR-amplified from pGS-GnT-I (a GlycoSwitch plasmid) (oligos 308/309), digested with SalI and BamHI and inserted into SalI/BamHI-linearized eDA26, creating eDA40. Hyg^R, between P_TEF1 and T_TEF1, was amplified from pGS-GnT-II (GlycoSwitch) (oligos 306/307), digested with SalI and BamHI and inserted into SalI/BamHI-linearised eDA26 yielding eDA37.

The telomeres part (Tel) comprises an 18-bp I-SceI-cleavage site flanked by arrays of 16 inverted telomere repeats [27]. To construct it, two pairs of complementary oligos were annealed separately: (i) Oligo 246 (127-nt) with oligo 247 (120-nt); (ii) oligo 248 (121-nt) with oligo 249 (128-nt). In each case, oligos (100 µM) were incubated (30 min, 37 °C) with five units of T4 polynucleotide kinase in T4 polynucleotide kinase buffer supplemented with 2.5 mM ATP. Then, NaCl was added to 50 mM and the mixture heated (300 s, 95 °C), then cooled to room temperature. The oligos were designed so that annealing would yield two ds-DNA molecules with mutually complementary overhangs that, upon ligation, form the Tel with overhangs at its distal ends compatible with PvuI. Thus, products of processes (i) and (ii), together with PvuI-linearized eDA40 (deleting its Amp^R gene) were ligated (T4 ligase) to yield eDA131. After transformation of E. coli with eDA131, zeocin (50 µg/mL)-resistant, ampicillin (100 µg/mL)-sensitive cells were verified (colony PCR, not shown), and eDA131 confirmed (by Sanger sequencing) as the Tel-donor plasmid from which a blunt-ended Tel could be excised by blunt-ended restriction enzymes FspI/PvuII for insertion into nanochromosome precursor plasmids.

Codon-optimised for K. phaffii genes encoding fluorescent proteins were delivered in plasmids pM-GFP and pM-mCH (Thermo Fisher Scientific). Constitutive expression cassettes for GFP and mCherry (mCH) were PCR-amplified (oligos 302a/303a, targeting P_TEF1 and T_CYC1), and overhangs released after BsmBI digestion. The products were ligated into plasmid eDA40, replacing the Zeo^R gene between P_TEF1 and T_CYC1. The resultant plasmids (eDA189 for GFP and eDA191 for mCH) were used to transform E. coli. Cells were selected on LB plates with ampicillin.

To prepare the PDI^H (coding for His-tagged protein disulfide isomerase) component, a previously synthesised K. phaffii-codon optimized K. phaffii PDI gene was re-amplified (oligos 447/448) from the in-house plasmid pPIC3.5 K-PDI [30]. The primer 448 introduces six His-codons followed by the TAA stop-codon. After excision with NotI and MfeI, the 1.64-kb DNA (PDI^H) fragment was cloned into NotI and MfeI-digested pPICZ A (between P_AOX1 and T_AOX1) forming plasmid eDA226.

A gene cassette (P_AOX1)GFP:FH(T_AOX1) for the GFP:FH fusion protein was prepared as follow: PCR-amplified GFP (from pM-GFP using oligos 276/277), digested with NsiI was inserted into PstI (NsiI- and PstI-recognition sites are compatible)-linearized pPICZα B-cFH plasmid prepared previously [55] forming eDA89. The DNA sequence for a flexible seven-amino acids linker (GGGGSNA) between fusion partners was delivered by oligo 277.

Assembly of arrays

Integration (receptor) and insertion (donor) arrays contain LHRs alternating with gene-expression cassettes. Examples include LHRⁿ-GoI-AMR-LHR^Z and LHRⁿ-GoI-LHR^m-AMR-LHR^Z, wherein (as described above): LHRs are unique sequences from a library, LHRs^A−Z; GoI contains a gene plus promoter and terminator regions; and AMR is Hyg^R or Zeo^R (with its own promoter and terminator).

Arrays were built from the following pre-prepared components: (i) one or two different LHRs (other than LHR^Z); (ii) the GoIs; and (iii) (common to all arrays) either Hyg^R-LHR^Z or Zeo^R-LHR^Z tandem pairs. These components were prepared as follows. (i) LHRs were amplified from source plasmids e.g. 0.87-kb LHR^E from eDA9, and amplicons digested with BsmBI or BsaI to release overhangs. (ii) Synthetic GoIs were amplified from their delivery plasmid, then digested with restriction enzymes before insertion into pUC19. (iiia) For the Zeo^R-LHR^Z tandem pair, LHR^Z was amplified by PCR from eDA8 and inserted into pUC19, to generate eDA101. The Zeo^R was PCR-amplified from eDA40 for adjacent insertion into eDA101, to create eDA105. The resultant Zeo^R-LHR^Z construct was re-amplified and digested with BsmBI. (iiib) To make the Hyg^R-LHR^Z pair, LHR^Z was generated by PCR and inserted into pUC19, yielding eDA99. The Hyg^R cassette was PCR-amplified from eDA37 for insertion into eDA99 to create eDA103. An unwanted AsiSI-recognition site in Hyg^R was modified by site-directed mutagenesis to generate the AsiSI-null plasmid eDA115. Then Hyg^R-LHR^Z was re-amplified and digested for subsequent assembly steps. See protocols [https://doi.org/10.17504/protocols.io.4r3l271o3g1y/v1] and [https://doi.org/10.17504/protocols.io.bp2l69p95lqe/v1] for details.

The integration array carrying constitutively expressed GFP and Zeo^R, i.e. LHR^E-(P_TEF1)GFP(T_CYC1)-Zeo^R-LHR^Z, was assembled by ligation of LHR^E, a GFP-expression cassette from eDA189, and Zeo^R-LHR^Z pair from eDA105. Assembly was instigated by ligating LHR^E with GFP. The gel-extracted (LHR^E-GFP) product and the Zeo^R-LHR^Z pair were then ligated tandemly into pUC19 forming eDA197.

Insertion array LHR^E-(P_TEF)mCH(T_CYC1)-Hyg^R-LHR^Z - carrying mCH paired with Hyg^R - was assembled, by ligation, from LHR^E, the mCH-expression cassettes from eDA191, and the Hyg^R-LHR^Z pair from eDA103. LHR^E was first ligated with the mCH-expression cassettes forming LHR^E-mCH, which was gel extracted. Then LHR^E-mCH plus Hyg^R-LHR^Z were ligated into pUC19 forming eDA199.

A longer array, LHR^A-PDI^H-LHR^E-Hyg^R-LHR^Z was assembled from: LHR^A; PDI^H(encoding His-tagged K. phaffii PDI under P_AOX1); LHR^E; and Hyg^R-LHR^Z. First LHR^A-(P_AOX1)PDI^H(T_AOX1) and LHR^E-Hyg^R-LHR^Z were prepared by pairwise ligations and gel purification. Then the two ligated parts were inserted, sequentially, into pUC19 forming eDA227.

Another triple-LHR array LHR^E-GFP:FH-LHR^D-Zeo^R-LHR^Z was formed from: LHR^E; an expression cassette (P_AOX1)GFP:FH(T_AOX1) encoding GFP:FH (see above) amplified from eDA89; LHR^D amplified from eDA9 and suitably digested; and Zeo^R-LHR^Z prepared as above. In two separate reactions, LHR^E was ligated to GFP:FH and LHR^D was ligated with Zeo^R-LHR^Z. Subsequently, gel-purified products, i.e. LHR^E-GFP:FH and LHR^D-Zeo^R-LHR^Z, were further ligated into pUC19 forming eDA250.

Assembly of parts into framework and precursor plasmids

A circular, telomere-null, framework plasmid (eDA53, 10.5-kb) was constructed by Gibson assembly from: (i) a 1.46-kb part corresponding to Zeo^R that was PCR amplified (oligos 253/254) using eDA40 as a template; (ii) a 0.57-kb PARS-A76 part that was generated by PCR (oligos 255/256) from eDA26; and (iii) the 8.63-kb centromeric plasmid eDA24 linearized by SmaI/PvuII.

We initially engineered into eDA53 an integration array consisting of LHR^A-I-SceI-Zeo^R-LHR^Z where I-SceI (encoding the meganuclease, I-SceI) was under control of P_AOX1. This initial integration array was not pre-assembled in the same way as the arrays described above, but was introduced in the following steps that converted eDA53 into 14.5-kb eDA83.

• (i) PvuII-linearized eDA53 was ligated with PCR-generated 0.8-kb LHR^A (from eDA8 template, oligos 264/273) carrying PvuII restriction site yielding 11.3-kb eDA71, which was BsmBI-linearized creating -CCCT and -CGCG overhangs on antisense and sense strands, respectively.

• (ii) An I-SceI-expression cassette was made by amplifying the I-SceI gene from plasmid eDA22 (oligos 223/224, carrying EcoRI and SacII sites), and inserting it into EcoRI/SacII-linearized pPICZα B forming eDA27. Then, using eDA27 as a template, a 2.2-kb (P_AOX1)I-SceI(T_AOX1) cassette was PCR-amplified (oligos 267/268) and BsmB1-digested, creating –GGGA and –CCAA overhangs on sense and antisense strands, respectively.

• (iii) A 0.95-kb LHR^Z part was PCR-generated from eDA9 (oligos 269/270) and BsmBI -digested, creating –GGTT and –GCGC overhangs on sense and antisense strands, respectively.

• (iv) Purified BsmBI-digested PCR products from steps (ii) and (iii) were ligated, using T7 DNA ligase, with gel-purified, linearized, eDA71 from step (i). After E. coli transformation, the product (eDA83) was verified by restriction-enzyme digestion and DNA sequencing.

The next step would have been to introduce a pair of proto-telomeres into eDA83. With this in mind, a blunt-ended 0.46-kb DNA segment containing the telomeres part (Tel) was gel-purified from PvuII/FspI-digested eDA131. Tel was then ligated with FspI-linearised eDA83 (thereby deleting a bla gene in eDA83). However, the ligation product was consistently lost when attempting propagation in E. coli.

Creation of nanochromosomes

The P_AOX1 (in addition to ~ 50-bp of the I-SceI open reading frame) in eDA83 was replaced with the bacterial kanamycin-resistance (Kan^R) cassette to create 14.5-kb eDA110, containing the modified integration array we called LHR^A-Kan^R(ΔI-SceI)-Zeo^R-LHR^Z. To achieve this, the 13.5-kb SphI-digested and gel-extracted eDA83 was ligated with a 980-bp PCR-amplified Kan^R (using oligos 321/322 and vector pUC57-Kan (Addgene) as a template). Following extraction from kanamycin-resistant E. coli cells, plasmid eDA110 was verified by Sanger sequencing. A blunt-ended 0.46-kb DNA segment containing the Tel was gel-purified from PvuII/FspI-digested eDA131, as before, but this time it was ligated with FspI-linearised eDA110 to yield 14.9-kb eDA137. Plasmid eDA137, verified by Sanger sequencing, was re-amplified in E. coli cells, linearized in vitro by I-SceI and used to transform K. phaffii (wild type). Transformants were selected on YPDS plates with 0.3 mg/mL zeocin.

The plasmid eDA110 was extended by tandem insertion of (i) a second ARS - the 529-bp K. phaffii ARS-B413 - and (ii) an extended (1.3-kb) non-coding spacer segment of DNA (ncDNA), as follows (See Fig. 4a). PARS-B413 was generated by PCR (oligos 238/239) from gDNA, and BamHI-digested then ligated into the multi-cloning site of pUC19 forming eDA41. Meanwhile, ncDNA was amplified (oligos 379/380) from eDA9 and then NdeI/EcoRI-digested for insertion immediately downstream of PARS-B413 in eDA41, generating intermediate plasmid eDA144. Subsequently, the combined 1.9-kb ARS-B413-ncDNA part was amplified from eDA144 (oligos 379/381), digested by NdeI, and inserted into the NdeI site of eDA110, yielding eDA146 (16.8 kb). The Tel was inserted into FspI-linearized eDA146, generating eDA155. Plasmid eDA155, verified by Sanger sequencing, was propagated in E. coli and then isolated and linearised with I-SceI in vitro. The linear product was used to transform K. phaffii strain CBS7435 using zeocin for selection.

The modified integration array (LHR^A-Kan^R(ΔI-SceI)-Zeo^R-LHR^Z) of eDA155 was excised with PvuI and SalI, releasing 5’-blunt and 3’-SalI/XhoI compatible ends. Subsequently, the PCR-amplified integration arrays (i) 4.4-kb LHR^E-GFP-Zeo^R-LHR^Z (from template eDA197 with oligos 397/409), or (ii) 7.1-kb LHR^A-(P_AOX1)PDI^H(T_AOX1)-LHR^E-Hyg^R-LHR^Z (from template eDA227 with oligos 384/409), were digested with AfeI and XhoI/SalI, then ligated with the PvuI/SalI-treated eDA155. The products ((i) eDA201 or (ii) eDA229) were used to transform E. coli cells. Following sequence verification these nanochromosome-precursor plasmids were linearized in vitro by I-SceI and the products used to transform K. phaffii cells, with selection on YPDS plates with 0.3 mg/mL hygromycin or zeocin.

Proofs of principle

• (i) The insertion array LHR^E-(P_TEF)mCH(T_CYC1)-Hyg^R-LHR^Z was amplified by PCR (oligos 409/397) from eDA199, and used to transform K. phaffii cells (creating yDA177 on the CBS7435 background, or yDA218 on the CBS7435::ΔKU70 background) already carrying nChr 2B (i.e. linearised eDA201). The aim here was to replace GFP-Zeo^R (in LHR^E-GFP-Zeo^R-LHR^Z) with mCH-Hyg^R.

• (ii) The insertion array LHR^E-(P_TEF)GFP(T_CYC1)-Zeo^R-LHR^Z was amplified by PCR (oligos 397/409) from eDA197 and used to transform K. phaffii cells (strains CBS7435 (to create yDA232) and CBS7435::ΔKU70 (to create yDA253) carrying nChr 2 A (i.e. linearised eDA229). In this case the aim was to replace Hyg^R (in LHR^A-(P_AOX1)PDI^H(T_AOX1)-LHR^E-Hyg^R-LHR^Z) with GFP-Zeo^R.

Transformed cells (strains yDA218 and yDA253 for (i) and (ii), respectively) were selected on YPDS plates containing 0.3 mg/mL zeocin or 0.3 mg/mL hygromycin. Thus, efficiency of gene replacement was estimated from the number of colonies that were fluorescent protein-producing, and resistant or sensitive to the expected antimicrobial.

To test our “inch-worming” strategy for gene insertion and landing zone extension, LHR^E-GFP:FH-LHR^D-Zeo^R-LHR^Z was excised from eDA250 using AsiSI. Following concentration by isopropanol precipitation, the array was used to transform K. phaffii cells containing nChr 2 A so as to replace Hyg^R (in LHR^A-(P_AOX1)PDI^H(T_AOX1)-LHR^E-Hyg^R-LHR^Z) with GFP:FH-LHR^D-Zeo^R. Transformed cells were selected on YPDS plates with zeocin. Gene integration/replacement efficiency was estimated from the number of colonies that were green-fluorescent, zeocin-resistant, and hygromycin-sensitive.