Although DNA repair mechanisms in S. cerevisiae have been deeply understood, there are still many challenges in the exploration of the alternative yeast P. pastoris (Ahmad et al., 2014). Compared to S. cerevisiae, knock-out cassettes or knock-in expression fragment present a more strenuous and low-efficient targeted integration with homologous arms in P. pastoris (Fischer and Glieder, 2019). One reason is that P. pastoris uses non-homologous end-joining (NHEJ) as a main DNA repair mechanism when DNA double strand broke (Naatsaari et al., 2012). DNA repair by homologous recombination (HR) offers several advantages over NHEJ, including more efficient screening for positive clones and improved genomic stability (Dalvie et al., 2022). However, P. pastoris presents a low HR rate, and a majority of the introduced DNA cassettes are integrated at random sites in the genome by virtue of NHEJ (Wang et al., 2016). Hence, P. pastoris gives more favor to the NHEJ pathway than recombination events (HR) for repairing double-strand breaks (DSB) (Fischer and Glieder, 2019). Currently, this issue was addressed by a double knockout of the homogeneous genes dnl4 (coding for DNA ligase IV) of Ku 70, overall increased efficiency to a comparable level of other eukaryotic cells (Ito et al., 2018). Relative to S. cerevisiae, in which short overhangs are capable of targeting strongly specific integrations, in P. pastoris, long overhangs present a larger efficiency and foreign DNAs may undergo ectopic integration (Vogl et al., 2018a). In order to overcome this issue, HR machinery with multiplex genome integration was introduced from S. cerevisiae to P. pastoris strains. Core genes related to HR were overexpressed under strong constitutive promoters (i.e., PGAP and PTEF1) and inserted into the P. pastoris chromosome. Thus, HR efficiency was enhanced to realize effective one-, two-, and three-loci genome integration as high as 100, ∼98, and ∼81% using ∼40 bp homology arms (Gao et al., 2022b).

The appearance of novel genome editing tools eased the gene disruption and specifically improved the HR efficiency. One good example is CRISPR/Cas9. CRISPR/Cas9 is a powerful genetic editing tool, and has been developed into a mature laboratory technique capable of disrupting certain genes with high efficiency in prokaryotic and eukaryotic systems (Sander and Joung, 2014). It utilizes a nuclease, introducing a DNA double-strand break at the regions complementary to the gRNA sequence under the guidance of a short RNA (guide RNA: gRNA or single guide RNA: sgRNA). Recruiting the cellular repair mechanism helps to seal these breaks, which allows to introduce difference genome modifications, i.e., NHEJ in the case of no template and homology directed repair (HDR) with appropriate repair template (Jiang and Doudna, 2017). CRISPR/Cas9 can be reprogrammed, and different loci can be targeted by changing 20 bp of gRNA (Fischer and Glieder, 2019). In spite of its convenience, CRISPR/Cas9 was first adapted to P. pastoris in 2016 (Weninger et al., 2016). Expression of Cas9 usually relies on the use of regular genomic vectors, while nuclear localization as well as 5′–3′ trimming is needed for the single guide RNAs. The identification of this RNA polymerase-III (PolIII) mediated transcription in P. pastoris can be easily achieved by small RNA sequences, highly reducing the difficulty of multiplex cloning and expression (Dalvie et al., 2020). Later research further explored the potential of this technique (Weninger et al., 2018; Yang et al., 2020). The deletion of the Ku70 involved in NHEJ would remarkably improve the HR repair, achieving nearly 100% efficiency for site-directed of gene insertion (Weninger et al., 2018; Liu et al., 2019), but with the help of CRISPR/Cas9 system, HR efficiency could also be 100% for Mxr1 (methanol expression regulator 1) using the same sgRNA without deleting Ku70 due to the help of stable Cas9 expression through integrative expression (Yang et al., 2020). Furthermore, the HR efficiency was improved by even over 54 times in NHEJ deficient strain by the CRISPR-Cas9 system, which assisted in the incorporation of donor DNA to express heterologous gene (Cai et al., 2021). However, hard to knock out genes such as OCH1, where CRISPR/Cas9 showed only 50% mutation efficiency (Weninger et al., 2016). Normal approaches solve this problem such as increasing HR efficiency by addition of hydroxyurea during transformation to stop cell division at the S/G2 phase where the status of HR is rather more active than NHEJ (Lee et al., 2018). Compared to the Ku70 mutation strain, this method does not reduce transformation efficiency and it is more suitable for indel mutation (Ahmad et al., 2019; Gassler et al., 2019). Recently, another study revealed that Ku70 mutation negatively impacted the chromosome terminal stability that caused loss of colonies, but over-expression of RAD52 can drastically improve the efficiency of HR (Cai et al., 2021).

Besides the gene replacement and the targeted insertion, CRISPR/Cas9 can serve for creating indel mutations at targeted genomic loci by virtue of the NHEJ DNA repair mechanism under the condition of no homology target (donor DNA). HDR mediated by CRISPR/Cas9 allows DNA fragments to be effectively integrated into the genome, which is marker-free in selection, and is likely to replace, destroy or label a certain genomic locus (Singh et al., 2017). Recently, one synthetic biology toolkit based on CRISRP was established to ensure that multigene biosynthetic pathways could be integrated as well as assembled into the chromosome of P. pastoris with the selection of integration sites for the efficient and stable genome integration. Also, it characterized a panel of constitutive, methanol-inducible promoters, enabling the integration of one locus, two loci, and three loci (efficiency: ∼100, ∼93, and ∼75%, respectively) (Gao et al., 2022a). The most recent applications for CRISPR/Cas9 were shown below in Table 1.

In addition to functional studies regarding gene deletion, there are many kinds of knock-out strains that have been applied in the field of industrial biotechnology (Ito et al., 2018; Karbalaei et al., 2020; Aw et al., 2021; Guo et al., 2021). However, targeted gene knock-out of P. pastoris is still time-consuming (Ahmad et al., 2019). Typical genomic integration as well as gene replacement by HR expression cassettes for selecting markers was used extensively caused by a large number of recent patents in the field of genome editing based on CRISPR/Cas9 (Fischer and Glieder, 2019; Liao et al., 2021). Considering the shortage of selective markers, multiple genetic modifications of P. pastoris are limited (Liao et al., 2021). In the past, approaches are usually combined with zeocin-resistance marker recycling vectors that employ FRT/FLP recombinase or Cre/loxp recombination system (Li C. et al., 2017), which removes the marker sequences between two recognition sites (Kobalter et al., 2022). Recently, acetamidase (amdS) was found as a kind of effective selection marker for the integrative transformation in P. pastoris when requiring many times of gene deletion or insertion. It was allowed to easily recycle markers by counter-selection with fluoroacetamide (Piva et al., 2018). However, after the employment of the CRISPR/Cas9 system, it is also easy to achieve such recycling by CRISPR/Cas9 geneticin plasmids that contain gRNAs for targeting the Zeocin resistance gene. P. pastoris strains generated by this approach can be retransformed by Zeocin resistance markers that contain plasmids (Yang et al., 2020). Marker-free system was also possible by the use of multi loci gene integration approach mediated by CRISPR/Cas9 with efficient gRNA targets in P. pastoris. The use of high efficient sites with a 100 bp range of upstream promoter and downstream terminator enables the integration of multiple gene cassettes into genome simultaneously (Liu et al., 2019).

Plasmids can be usefully observed in bacteria, but only in a few yeast species. Generally, an episomal plasmid was not considered as an optimal choice to express the heterologous genes because of segregational instability in non-selective conditions, but it is an excellent choice regarding the expression of Cas9 (Pena et al., 2018). However, episomal plasmids for heterologous gene expression and the CRISPR/Cas9 system for genome editing have not been well developed in P. pastoris (Gu et al., 2019a). Based on previous attempts to apply the CRISPR/Cas9 system in P. pastoris, Yang et al. (2020) developed a CRISPR/Cas9 system that contained episomal sgRNA plasmid, which reached 100% genome editing efficiency in 14.7% of sgRNA. High multicopy gene editing and stable multigene editing efficiency were obtained which did not exhibit a sharp decrease due to the multi-sgRNA expression cassette. Besides HR efficiency, the vector carrying sgRNA expression cassette could be eliminated easily, benefiting the editing of other genes (Yang et al., 2020). It is necessary to identify two genetic elements, namely, the centromere and the autonomously replicating sequence (ARS), to engineer a steady episomal plasmid in a microorganism without carrying a plasmid (Piva et al., 2020; Schultz et al., 2021; Gatjen et al., 2022). Several recently identified ARS enable maintenance of episomal plasmids and efficient expression of recombinant proteins (Camattari et al., 2016; Schwarzhans et al., 2017; Nakamura et al., 2018; Gatjen et al., 2022). For instance, the episomal plasmids can be maintained in P. pastoris by a 452 bp DNA sequence (pan ARS) recovered from Kluyveromyces lactis (Camattari et al., 2016). The presence of ARS activity was found in a short fragment (111 bp) of chromosomal centromere 2 (PpARS2) (Nakamura et al., 2018). In addition, it was discovered that a piece of mitochondrial DNA (mitoARS) was able to help stabilize maintenance in the nucleosome (Schwarzhans et al., 2017). Gu et al. (2019a) thoroughly compared these ARSs, especially for creating and improving the genome editing system based on CRISPR/Cas9. Compared to a previously described method that used endogenous ARS (PARS1), CRISPR/Cas9 genome editing using panARS demonstrated an increased editing efficiency by over ten times (Weninger et al., 2016; Gu et al., 2019a). Similar findings were also discovered when establishing a Pichia surface display (PSD) system. Highest display levels were achieved with panARS-based expression vectors (Gatjen et al., 2022).

The engineering of microorganisms is now led by the reduction of manufacturing costs, mainly reached by higher titers. To efficiently enhance the protein production by P. pastoris, a fusion protein that includes the alpha mating factor secretion signal, researchers constructed human serum albumin (HSA) for promoting the protein secretion in P. pastoris (Sigar et al., 2017; Cao et al., 2018). The most recent recombinant proteins produced by P. pastoris were referred to below in Table 2.

The promoter of the P. pastoris alcohol oxidase 1 gene (P_AOX1) remained the most commonly used natural promoter for the RP expression vector construction in P. pastoris (Vogl and Glieder, 2013). It can be strictly inhibited by glucose or glycerol and triggered by methanol, which enables cells to grow in methanol as solely carbon source. In other words, it decouples cell growth and protein production phase. Relying on these mentioned advantages, it is of big value for the high-level protein expression and can replace the constitutive P_GAP (promoter region of the glyceraldehyde-3-phosphate dehydrogenase) in certain cases (Gellissen, 2000).

The clarification of regulation mechanism promoted the development of P_AOX1 engineering. Methanol expression regulator (Mxr1) essentially regulates the utilization pathway of methanol and is capable of activating a lot of genes in response to methanol (Wang et al., 2016). The overexpression of Mxr1 has the function of promoting AOX1 expression through inhibiting glycerol transporter 1 (GT1) expression (Zhan et al., 2017). Partial rewiring of P_AOX1 transcriptional circuits can still maintain a basic output by overproducing a deregulated Mxr1 form in strains containing multiple P_AOX1-based expression cassettes (Camara et al., 2019). Targeted editing of Mxr1 can be successfully achieved by CRISPR/Cas9 technology with plasmids containing sgRNA and the methanol expression regulator 1 homology arms (Hou et al., 2020). In other words, Mxr1 protein frame-shift mutations may reduce AOX1 protein levels as well as weaken the enzyme activity (Hou et al., 2020). In one study, the P_AOX1 engineering by a mutation in the core promoter where wild type adjacent triplets were changed to cytosine or adenine triplets, a completely synthetic construction, demonstrating the strong tolerance of the core promoters to small mutations, which supported regulatory models regarding degenerate motifs, or redundant design in the future (Portela et al., 2018). Generally, modifications in P_AOX1 core promoter or around upstream regulatory sequences (URS) by synthetic promoter demonstrate regulatory effects, and the engineering effort is mostly focused on 5′ untranslated regions (5′ UTR) (Vogl and Glieder, 2013). In a more recent study, the genetic modifications of P_AOX1 were realized by using synthetic Adr1, Cat8, and Aca2 cis-acting DNA elements to replace specific cis-regulatory DNA elements for Mxr1, Cat8, and Aca1 binding, respectively. The combined promoter design with 3 Cat8, 3 palindromic Adr1, as well as Aca2 synthetic binding motifs can retrofit the strength of methanol to 1.97-fold that of P_AOX1 (Ergün et al., 2020).

Besides P_AOX1, some other methanol-inducible promoters have been characterized in P. pastoris. The expression level of the promoter of the dihydroxyacetone synthase 2 gene (P_DAS2) was higher than that of P_AOX1 (Vogl and Glieder, 2013), while the expression of P_PEX8 and P_AOX2 was much lower (Vogl, 2022). In order to identify alternative methanol inducible promoters, the transcriptional response of P. pastoris on microarray was observed under various carbon source conditions including glucose repression, derepression, and methanol-induction (Vogl et al., 2016). Fifteen different strengths of methanol-responsive promoters that participate the methanol utilization (MUT) pathway were identified. Clearly, the promoter of the CAT1 gene participating in the defenses against ROS, presented a strong methanol induction and the highest derepression level. P_CAT1 induction can be achieved by using oleic acid at a similar level to methanol. Therefore, P_CAT1 could be considered as an appropriate alternative derepressed, methanol-free promoter if its regulation mechanism was to be further elucidated (Vogl et al., 2016). Orthologous promoters from related yeast species were tested and even surpassed endogenous promoters in P. pastoris. Under methanol induction, the promoter of the HpMOX gene from Hansenula polymorpha (Hp) presented a similar expression levels as P_AOX1 and HpFMD promoter surpassed P_AOX1 by threefold (Vogl et al., 2020). These results indicated the potential of utilizing high-efficacy orthologous promoters from other eukaryotic hosts. Similar findings were revealed in the terminators, the terminator sequences from S. cerevisiae were confirmed to maintain function when transferred to P. pastoris (Vogl et al., 2020).

While methanol induction offers strong promotion and regulation of recombinant protein expression, it also contributes its own challenges. Due to the safety concern and strict process control for inducing methanol in large-scale fermentation, there have been some efforts in terms of the replacement of methanol as a single carbon source in P. pastoris (Wei et al., 2016). After the deletion of three transcription repressors and overexpression of one transcription factor MIT1, one methanol-free P_AOX1 start-up strain was successfully constructed (Wang J. J. et al., 2017). Accordingly, a glucose-glycerol-shift induction model was built for replacing conventional glycerol/methanol induction in the wild-type strain. It is safe, economic, and environmentally friendly, but exhibits less protein expression ability. Only 77% GFP expression level was detected in glycerol relative to the wide type in methanol (Wang J. J. et al., 2017). Later, another study confirmed that any P_AOX1-based strain can be converted to a methanol-free system with the simple overexpression of Mit1 or Mxr1 by derepression of P_CAT1, and no addition of an alternative inducer was required, however, low feasibility and stability remain to be the main concern of this method (Vogl et al., 2018b). As one transcription regulating element in the MUT gene of P. pastoris, P_DAS1, a strong methanol-inducible promoter driving dihydroxyacetone synthase expression, is also considered to be applied to RP production (Vogl et al., 2021). Constitutive expression of one single transcription factor KpTrm1 was enough to activate P_DAS1 without the addition of methanol, the simplicity of P_DAS1 regulation made it promising for the development of methanol-free protein expression system (Takagi et al., 2019). The protein yield remains to be a drawback in terms of these innovative methanol free system.

Despite the limitation of the selection of protein expression promoters in P. pastoris to P_AOX1 or P_GAP (Yang and Zhang, 2018), researchers have made efforts to find new synthetic promoters that are likely to replace conventional promoters. Engineered promoter variants (EPVs) exhibit an extremely stronger performance than natural promoters and allow to conduct “green-and-clean” production on a non-toxic carbon source in the first-choice utilization pathway of carbon source of yeast. One recent work to improve the RP expression with ethanol utilization pathway was done by Ergn et al. (2019). Transcription binding sites of alcohol dehydrogenase 2 promoter (P_ADH2) were engineered for designing the novel engineered promoter variants (NEPVs) (Ergn et al., 2019). The most popular method for the generation of EPVs is hybrid-promoter architectures. Using de novo synthetic sequence to replace native cis-acting DNA assists in achieving the architecture of the hybrid promoter (Ergün et al., 2020). The hybrid-architectured promoter design refers to collecting monodirectional double-promoter expression system (DPESs) with hybrid architecture composed of engineered promoter variants P_ADH2-Cat₈-L₂ and Pm_AOX1 and the natural promoter P_GAP for enhancing and upregulating deregulated gene expressions in P. pastoris in media free of methanol (Demir and Calik, 2020). Biofunctional DPESs exhibited higher transcription and expression upregulation power relative to twin DPESs (two-copy expression systems). The most advanced technology of P. pastoris promoter engineering provides the comprehensive insights into the cis-acting DNA motifs and functions to ensure a rationally designed non-conventional promoter libraries that possess improved strength and different regulation mechanisms (Ergun and Calik, 2021). Apparently, synthetic promoter variants are becoming a trend for improvement of protein expression in the P. pastoris system (Machens et al., 2017). Lastly, since suitable promoters can mediate dynamic regulation of biosynthetic pathways and maintain cellular robustness, the appearance of the inexpensive synthetic promoters still remains to be explored for future applications (Vogl et al., 2016).

Metabolic engineering regarding P. pastoris initially featured protein glycosylation humanization (Irani et al., 2016; Pena et al., 2018). There are two main types of glycosylation, including N- and O-glycosylation. Researches show that N-glycosylation had a great effect on protein stability (Zou et al., 2013), activity (Zou et al., 2012), and specificity (Yang et al., 2015). However, the removal of N-glycosylation didn’t make great changes on the secondary structure of proteins but tertiary structure showed differences (Wang et al., 2018). Another recent study found that the degree of O-glycosylation was remarkably higher when the protein was induced by methanol compared with glucose (Radoman et al., 2021), but details of machinery required to be investigated later.

In view of the proven history of P. pastoris as an industrial cell factory for protein production, there has been an increasing interest in exploring its potential to produce primary and secondary metabolites. The RP production is a costly metabolic process, and during the process, many cellular machineries deviate from the targets of evolution regarding cell development and maintenance, precursors substances are expelled from the central carbon metabolism, and redox and energy cofactors are consumed, leading to low energy efficiency in the metabolism (Kafri et al., 2016). Thousands of biochemical reactions occur coordinately, hence cells are capable of obtaining energy as well as building blocks from the environment for maintaining their lives (Ergun et al., 2021). The metabolic constraints of recombinant protein production are shown in Figure 1.

The accuracy of metabolic flux analysis depends on the biomass composition, which is obviously different in cells grown on various carbon sources. The differences of biomass composition were found to remarkably affect the lipid flux distribution, the biosynthesis of secondary alcohol and the oxidation-reduction processes (Cankorur-Cetinkaya et al., 2017; Tomas-Gamisans et al., 2018). Considering the abundant proteinogenic amino acids, NMR can serve for evaluating how RP production affects the metabolomic profiles (Tredwell et al., 2017). On that account, correlation between these unfolded protein response (UPR) markers and specific metabolic signals was unveiled by transcript analysis of gene transcripts associated with UPR (HAC1, KAR2, PDI1) (Tredwell et al., 2017). This approach allows for the high-throughput screening of large numbers of clones in the future.

In spite of the achievements in recombinant DNA technology, the protein expression by engineered P. pastoris, faces many limitations, indicating that expression system exhibits unpredictable physiology (Prabhu et al., 2017). Above protein production restrictions can be solved via the metabolic engineering strategies through increasing the precursor supply as well as enhancing the cellular redox and energy efficiency. Protein folding and processing in the endoplasmic reticulum (ER) mainly restrict the protein production and secretion from P. pastoris. Recombinant gene overexpression is capable of destroying the inherent secretory machinery regarding P. pastoris, and accumulation of proteins folded in incorrect way occur inside the ER. For restoring the normal protein folding, an UPR pathway was triggered naturally by cells, causing expression upregulation on genes coding for chaperones as well as other folding-assisting proteins (Raschmanova et al., 2021). From another application perspective of metabolomics, metabolic biomarkers could indicate the cellular stress resulted from the UPR under when the expression was high (Tredwell et al., 2017). The effect of vacuolar protein sorting (VPS) system on protein secretion was investigated by generating VPS mutant strains. The impairment of VRS is considered as an effective means of enhancing recombinant protein secretion (Marsalek et al., 2017). The N-terminal portion of the pre-pro-α-factor is a common secretion signal, but it is capable of partially leading to aggregation, hence limiting the export from the ER (Barrero et al., 2018). A hybrid secretion signal that possessed the S. cerevisiae Ost1 signal sequence was engineered in P. Pastoris to facilitate co-translational translocation into the ER, and then the α-factor pro region (αPR). When Ost1 signal sequence was paired with the αPR, its secretion efficiency was much higher relative to the α-factor signal sequence (Barrero et al., 2018).

The manipulation of genes gives more possibilities for metabolic engineering through different carbon utilization pathways. As described before, by overexpressing only one transcription factor such as MIT1 in P_AOX1-based expression strains, methanol-dependent strains could be transferred into a glucose/glycerol-regulated system (Vogl et al., 2018b). Furthermore, after adding 8 heterologous genes as well as deleting 3 native genes, it is possible to observe the transformation of the traditional P. pastoris methanol-assimilation pathway into a CO₂-fixation pathway similar to the Calvin–Benson–Bassham cycle (Gassler et al., 2020), which provides us more insights in terms of innovating novel green and carbon-neutral process. Most recently, in order to regulate the target gene expression level of metabolic pathway, fast and reliable tools are applied to precisely manipulate the expression of target genes. One practical example was dCas9, which can be fused via RNA scaffolds to trans-activator domains and thus regulate the gene expression when targeted to the promoter region of a gene (Baumschabl et al., 2020).