Introduction

As production system, yeasts are valuable hosts due to important advantages, including the ability to accomplish proper posttranslational modifications, fast growth, simple genetic manipulation, scalable fermentation, high biomass concentrations and safe pathogen-free production [1,2,3]. Yeast expression systems could be divided into two groups: methylotroph and non-methylotroph [4] (Fig. 1). This article will review the essential features, current tools and new developments in the most common methylotrophic and non-methylotrophic yeast expression hosts.

Fig. 1
figure 1

Classification of common methylotroph and non-methylotroph yeasts

Non-methylotrophic Yeasts

Saccharomyces cerevisiae

The first and best well-known yeast expression system, S. cerevisiae, was primarily utilized as expression host in the 1980s and profited from its traditional use in brewing, wine making and baking [5, 6]. In 1996, the genome of S. cerevisiae (S288C strain) became the first completely sequenced eukaryotic genome [7]. Furthermore, considering an outstanding achievement in eukaryotic biology, sequencing of S. cerevisiae genome provided valuable information on numerous aspects of genome organization and evolution [8]. S. cerevisiae is a preferred host over bacteria, other yeasts, and filamentous fungi in numerous physiological features related to industrial ethanol production including tolerance to wide pH range, high concentrations of ethanol and sugar, and resistance to elevated osmotic pressure [9,10,11]. This expression system is one of the most commonly used eukaryotic organisms that has been utilized as a model to study gene expression regulation [12], signal transduction [13], aging [14], apoptosis [15], metabolism [16, 17], cell cycle control [18], programmed cell death [19] neurodegenerative diseases [20], autophagy [21], secreatory pathway [6] and numerous other important biological procedures [22]. Heterologous glycoprotein production in S. cerevisiae frequently results in a hyper-mannosylated glycan structure that can lead to lower activity and higher immunogenicity. Thus, reduction of hyper-mannosylated glycans is one of the major considerations in the production of heterologous glycoproteins in S. cerevisiae. It has been shown that disruption of Mnn2p and Mnn11p genes that are related to glycosylation modification pathways, improved the production of recombinant cellulases [23]. Removal of the α-1, 6-mannosyltransferase Och1p enhanced the production of active form of human tissue-type plasminogen activator. These studies indicated that N-glycosylation modification also leads to increased protein secretion [23]. Recombinant proteins can be expressed intracellularly or directed to the secretory apparatus using a secretory signal peptide. The frequently used signal sequence that is functional in all yeast expression systems, is prepro-sequence of mating factor α1 (MFα1) [24].

Saccharomyces cerevisiae has also some important advantages from the safety point of view and this property encourages the use of this system in various industrial processes. It is generally regarded as safe (GRAS) because S. cerevisiae is nonpathogenic and historically have used in various nutritional industries and production of biopharmaceuticals. On the other hand, current knowledge on genetics, physiology, and fermentation of yeasts facilitate the use of this organism in the production of useful products [5, 6]. Hepatitis B surface antigen, hirudin, insulin, glucagon, urate oxidase, macrophage colony-stimulating factor and platelet-derived growth factor are examples of products on the market from S. cerevisiae [3, 25]. However, several reasons have still limited the number of commercial products from S. cerevisiae, including the hyperglycosylation of proteins, low protein yield and plasmid instability [26]. These limitations have resulted in the development of alternative expression systems including methylotrophic yeasts Pichia pastoris and Hansenula polymorpha and non-methylotrophic yeast Yarrowia lipolytica, Kluyveromyces lactis and Arxula adeninivorans [6].

Saccharomyces cerevisiae Strains

Strains of S. cerevisiae that are used for protein expression include both wild-type and mutant strains. Wild-type strains include S. cerevisiae, S. boulardii, and S. uvaruium. S. boulardii is utilized as a probiotic for treatment of bacterial diarrhea. There are some S. cerevisiae strains [27, 28] that are used in industrial processes and research works including:

S288c

This strain has been isolated through genetic crosses in 1950 [29]. Most of the genomic information on S. cerevisiae has been acquired from S288c. This strain has some limitations [30] such as low rate of sporulation compared to W303 [7], inability to grow on maltose [31] and lack of filamentous growth on nitrogen-deficient media [32]. Other strains commonly used as lab strains include A364A, BY4716, CEN.PK, FL100, ∑1278b, SK1, BJ5464, W303 [30], BY4741, BY4742 and BY4743 [33].

A634A

This strain is originated from the cross of S288c and an unknown strain. It has a close relationship to S288c and commonly used in cell cycle studies [30].

BY4716

Since this strain is similar to S288c, it is frequently used as a reference or control strain [30].

CEN.PK

This strain is used as the secondary reference strain alongside with S288c for genome sequencing. It grows well on different carbon sources under anaerobic conditions. It is commonly used in studies related to the cell growth rates and product formation [34].

∑1278b

This strain comprises genes that are unique for nitrogen metabolism and it is identified by this feature [35].

SK1

This strain is used in meiotic studies due to high sporulation efficiency [36].

BJ5464

This strain predominantly used for recombinant protein expression. It lacks vacuolar degradation due to deletion of PEP4 and PRB1 [37].

BY4742

This strain is utilized for the evaluation of protein function and endogenous cellular processes [37].

W303

This strain is originated from the cross of S288c and an unknown strain, with a close genetic relationship to S288c [30]. This strain applied in biochemical and genetic studies [7] (Table 1).

Table 1 Common non-methylotroph yeast strains: S. cerevisiae, Y. lipolytica

One of the main drawbacks of S. cerevisiae-derived recombinant proteins is hypermannosylation of glycoproteins that leads to immunogenic reactions after long term administrations. In addition, hypermannosylation leads to the reduction of serum half-life and therapeutic efficiency of glycoproteins. Glycode, a biotechnology company, has developed technologies to increase production of heterologous glycoprotein by selective modification of glycosylation in S. cerevisiae. In the GlycodExpress™ technology (patent WO/2008/095797), S. cerevisiae has been modified through sequential removal of mannosyltransferases and glycosyltransferases [40]. In a study, a dual strategy has been used for enhancement of the N-glycan homogeneity in the glycoengineered S. cerevisiae by removing the mannosyltransferase MNN1 and increasing the transportation of UDP-GlcNAc into the Golgi apparatus [41]. In recent years, CRIPR/Cas9 has been utilized for successful engineering of S. cerevisiae genome and provided an efficient method for site-specific mutagenesis in this yeast strain [42].

Yarrowia lipolytica

Yarrowia lipolytica, as a hemiascomycetous yeast, is highly considered for industrial purposes due to its capability to grow on n-paraffin and to produce high levels of organic acids (e.g. citric acid) [43, 44]. Due to its ability to produce large quantities of heterologous proteins and metabolites, Y. lipolytica has been utilized for numerous industrial applications, including the production of recombinant proteins, erythritol, citric acid and lipids [45, 46]. This yeast can use only limited range of C6 sugars such as glucose, fructose, and mannose [47] but it can utilize acetate, alcohols and hydrophobic substrates including oils, alkanes, and fatty acids. In the early stages of secretion (in the cytoplasm and ER) proteins can enter into one of two different secretion pathways, co-translational and post-translational translocation pathways. As it is common in the complex eukaryotes, the first pathway is prominent in Y. lipolytica. This may provide an advantage for Y. lipolytica [43]. In S. cerevisiae, co-translational and post-translational pathways are equally important. The choice of the route depends on the signal sequence which is added to the protein. The co-translational pathway is dependent upon SRP that transport secretory proteins to the ER membrane whereas they are still being produced by ribosomes. In addition, post-translational pathway is dependent upon Hsp70 that fully synthesized precursor polypeptides are deliver via cytosolic molecular chaperones, belongs to a large family of proteins called Hsp70 chaperones [48].

Yarrowia lipolytica Strains

The major Y. lipolytica strains that have been applied for recombinant protein production are listed in Table 1. The most frequently used strains include E129, Po1d, Po1f, Po1g and Po1h [38]. These strains provide significant capabilities in recombinant protein production. Several Y. lipolytica host strains have recently been engineered, including marine strains that are engineered for production of citric acid, single cell oil (SCO) and single cell protein (SCP) [46, 49, 50]. The wild-type WSH-Z06 strain optimized for KGA production and Po1d derivatives improved for lipid metabolism and better growth on sucrose [46].

Methylotrophic Yeasts

Pichia pastoris

Pichia pastoris is an excellent expression host for the production of heterologous proteins including industrial enzymes and biopharmaceuticals [51]. So far, this methylotrophic expression system has been successfully utilized for the generation of numerous recombinant proteins including human erythropoietin, phospholipase C, phytase, human superoxide dismutase, trypsin, human serum albumin, collagen and human monoclonal antibody 3H6 Fab fragment [52]. With the extended application of P. pastoris, novel genetic tools was developed and the accessibility to new strains increased the need for an improved cultivation and production procedures with this yeast [51]. Compared to any other yeast species, P. pastoris is more efficient in the secretory production of recombinant proteins [53]. The industrial interest to this host is also attributed to powerful methanol-regulated alcohol oxidase promoter (PAOX1), highly efficient secretion mechanism, posttranslational modification capabilities [54], and high cell density growing on defined medium. However, P. pastoris cultivation is restricted by some limitations, including the high concentration of proteases, difficulties in the systematic study due to the product-specific effects and risks related to storage of large amounts of methanol [55]. Nevertheless, various recombinant proteins have been expressed using P. pastoris. In a previous study (2016), we have demonstrated the production of a nanobody (VHH) against Clostridium botulinum neurotoxin type E (BoNT/E) in P. pastoris. A product yield of 16 mg/l was achieved that was higher than the levels produced by E. coli [56]. In addition, Xia et al. showed production of 30 mg/l of recombinant angiogenin in this yeast [57]. Furthermore, the productivity of 111 mg/l for human adiponectin [58], 8.1 g/l for recombinant xylanase [59] and 260 mg/l of anti-HIV antibody [60] have been reported in P. pastoris.

Pichia pastoris Strains

Pichia pastoris expression strains are commonly derived from NRRL-Y 11430 strain. This parental strain, as an auxotrophic mutant, is deficient in one or few genes that allow selection of transformants after transformation [61]. AOX1 and AOX2 genes encode alcohol oxidase as the first enzyme in the methanol oxidation pathway. Based on methanol utilization as the only carbon/energy source and depending on deletion of one or two of these genes, there are three types of P. pastoris strains [62] that indicated as Mut+, Muts and Mut [63]. Mut+ phenotype includes variants containing both functional forms of AOX genes, Muts phenotype composed of variants lacking AOX1 gene and Mut phenotype includes variants that both AOX1 and AOX2 genes are deleted. In a commercially accessible strain, MC100-3 (his4 arg4 aox1Δ: SARG4 aox2Δ: Phis4), both AOX1 and AOX2 genes have been deleted from the genome; however, the strain retained the capacity to induce high levels of protein production by AOX1 promoter on vector. This strain demonstrates the first example of separating growth and protein expression phases during the same cultivation, hence, co-feeding with an another carbon source, e.g. glucose, in addition to a certain concentration of methanol that acts only as an inducer (without metabolization), is essential [51].

Auxotroph strains with one auxotrophy include JC254 (ura3), GS115 (his4) and GS190 (arg4); two auxotrophies include GS200 (arg4 his4), and Jc227 (ade1 arg4); three auxotrophies comprise Jc300 (ade1 arg4 his4) and four auxotrophies include Jc308 (ade1 arg4 his4 ura3). Furthermore, there are three types of protease-deficient mutants, including SMD1163 (his4 pep4 prb1), SMD1165 (his4prb1) and SMD1168 (his4pep4) [62] (Table 2).

Table 2 Common methylotroph yeast strains: P. pastoris and H. polymorpha

At present, CRISPR/Cas9 technology is extensively used for targeted genome modifications of P. pastoris through several mechanisms such multiplex gene deletions and targeted integration of homologous DNA cassettes. In this regard, Weninger et al. employed an optimized CRISPR/Cas9 system for genome engineering in P. pastoris with high-targeting efficiency [69].

Further to Crispr, several other systems have also been used for genome engineering, for instance, Gibson assembly is a quick cloning system in which PCR-amplified fragments are used for cloning [70]. In this respect, Vogl et al. initially utilized this modular method that led to an increase of 31-fold in the culture supernatant [71]. Newly, an efficient cloning system with high flexibility, called as GoldenPiCS, has been introduced for engineering of P. pastoris [4, 70] using multigene constructs. This system is capable of integration of up to eight expression cassettes in one plasmid, consisting integration loci, appropriate promoters, and terminators [72]. Golden gate system could be used for optimizing expression and secretion of recombinant proteins by changing the promoter and secretion signal in the genetic construct [70].

Hansenula polymorpha

Hansenula polymorpha belongs to a number of few yeast species, such as P. pastoris, that are able to use methanol as the only carbon and energy source [68]. This methylotrophic yeast has become a relatively popular host for heterologous protein production due to the accessibility of the multicopy integration system and the powerful inducible promoters. The thermotolerant feature of H. polymorpha, that allows ethanolic fermentation at high temperature is one of the advantages of this yeast over the traditional yeast S. cerevisiae [73]. At high temperatures (48–50 °C), this yeast naturally ferments xylose to ethanol [74]. Unusual physiological properties of H. polymorpha, including resistance to heat, oxidative stress and heavy metals make this yeast attractive for numerous biotechnological applications [75, 76]. H. polymorpha is a suitable host for the production of human-like glycoproteins, so that most of the N-linked glycans expressed in this expression host showed reduced hypermannosylation without hyper-immunogenic terminal α-1,3-linked mannose residues [77, 78]. Market available products from H. polymorpha include recombinant hepatitis B vaccine, insulin [79], interferon alpha-2a, hirudin, phytase, hexose oxidase, and lipase. In 1999, Mayer and colleagues reached a yield of 13.5 g/l for phytase enzyme in this expression system. Permanent integration of vectors into the H. polymorpha genome provides stable expression. Due to resistance to high temperature (48–50 °C) [74], production of enzymes and other proteins intended for crystallography at high temperatures is highly recommended in this yeast [79, 80].

Hansenula polymorpha Strains

Hansenula polymorpha, has three main strains with unknown relationships, various features, and independent sources. Examples of H. polymorpha strains with application in biotechnology and basic research are CBS4732 (CCY38-22-2; ATCC34438, NRRL-Y-5445), DL-1 (NRRL-Y-7560; ATCC26012) and NCYC495 (CBS1976; ATAA14754, NRLLY-1798) [68]. These strains have independent sources: CBS4732 was primarily derived from soil water, NCYC495 from the intestine of insects, and DL-1 from orange juice concentrates [81]. Strain DL1 has been used in studies related to peroxisome function and biogenesis, methanol metabolism, microbial cell factory, and metabolic engineering for ethanol production at high temperature [82] (Table 2). Recent works concentrated on H. polymorpha as the most thermotolerant yeast strain identified up to the present, with maximal fermentation and high-temperature growth [83, 84]. H. polymorpha is capable of fermenting glycerol, glucose, xylose, and cellobiose to ethanol [85, 86] but, ethanol productivity from xylose is very low using the wild-type strains [87]. It has recently been shown that the CAT8 gene has a critical role in the regulation of xylose alcoholic fermentation in H. polymorpha. Ruchala et al. (2017) studied CAT8 knock-out strain and CAT8 overexpressed mutant for characterization of the role of CAT8 homolog in the native xylose-fermenting H. polymorpha. Finally, it has found that ethanol production from xylose is increased in cat8Δ mutant but decreased in CAT8 overexpressed strain. Moreover, glucose alcoholic fermentation remained unchanged in both CAT8 knock-out strain and CAT8 overexpressed mutant [88].

Expression Vectors and Selection Markers

An appropriate vector should have a targeting sequence for homologous recombination/integration, a multiple cloning cite for foreign gene insertion, a promoter element, a selection marker for transformation and a secretory signal for secretion of foreign proteins [6, 89]. An attractive feature of the yeast vectors is that they can replicate in the host cell by autonomous replication (episomal vectors). Both vector systems (integrative and episomal vectors) have been utilized for expression of recombinant proteins. Integrated vectors exhibit a very high stability, that is a significant demand for scalable production processes, but their copy number is often low. Conversely, episomal vectors provide higher copy numbers of the expression cassette but will be unstable in the absence of selection. In spite this limitation, episomal vectors provide advantages such as simple transformation protocols [90]. Obviously, utilization of a shuttle vector simplifies the transformation of yeast. Shuttle vectors are able to propagate in two different host species. The gene of interest is inserted into the shuttle vector and introduced into the E. coli for cloning of gene into the expression vector. After selection of transformants, the construct is amplified and transformed into the yeast host [91]. Presence of strong promoters is additional advantages of the yeast hosts that provide high level expression of the gene of interest inexpensively compared to other eukaryotic expression systems [65].

Vectors applied for transformation of appropriate S. cerevisiae host strains are hybrids of bacterial and yeast-derived sequences. The bacterial fragment carries elements indispensable for plasmid proliferation in E. coli, such as selection marker and Ori. The yeast-derived fragment contains the components for selection of yeast transformants such as genes for oritidine 5′-decarboxylase (URA3) or β-isopropylmalate dehydrogenase (LEU2) [92]. There are various types of S. cerevisiae vectors, but most of the expression systems rely on plasmids originated from YEp vectors. They commonly exist in the cells in 30 copies or more. They reproduce autonomously, as they contain elements of the 2 μm yeast plasmid. Nevertheless, the use of YEp vectors can lead to unstable strains that are common case for batch variations in the production process [93]. YIp vectors contain an Ori, an E. coli Ori, needed for plasmid replication. They are integrated into the chromosomes via homologous recombination. These vectors are very stable but persist in low copy numbers only [91]. In a strategic method to overcome this drawback, the vector may be targeted to the rDNA cluster, existing as about 150 tandem repeats. In this engineered recombinant strain, the stably integrated construct is present in high copy numbers. Copy numbers could also be increased by using a weak promoter for expression of the selection marker. Such imperfect promoters can be generated by elimination of 5′-flanking sequences as performed in the case of the LEU2-D promoter [94]. Common vectors used in S. cerevisiae are shown in Table 3.

Table 3 Yeast expression vectors and selective markers

Similar to other yeast systems, vectors applied for transformation of Y. lipolytica are shuttle vectors. Two main kinds of expression vectors are available for Y. lipolytica; episomal replicative vectors do not appear to meet stringent industrial necessities and demands [105]. The integrative vectors deliver the heterologous DNA into Y. lipolytica genome mostly by homologous recombination which its frequency is increased by the linearization of the plasmids before transformation [46]. Integration of monocopy vectors in Y. lipolytica genome provides numerous advantages over other yeast expression hosts such as P. pastoris. A stringent targeting of the monocopy integration into the genome extremely improves the efficiency of transformation. The most common vectors for expression in Y. lipolytica are shown in Table 3 [79].

Yarrowia lipolytica, in contrast to S. cerevisiae, predominantly uses the nonhomologous end-joining for repairing double-strand DNA breaks. The homologous integration of exogenous DNA into Y. lipolytica genome with suitable rates can occur only with large (0.5–1 kb) homologous flanking regions. To resolve this problem, a Y. lipolytica strain with KU70 gene (responsible for DSB repair in the NHEJ pathway) deletion has recently developed. In KU70 deficient Y. lipolytica Strains, frequency of homologous recombination of URA3 marker at the ADE2 locus was 43%, even with 50 bp flanking sequence [106].

Manipulation of gene expression by different kinds of plasmids can increase expression levels and plasmid copy numbers. Nevertheless, for the strong industrial workhorse Y. lipolytica, only a low copy CEN plasmid is accessible. Liu et al. have reported the engineering of CEN plasmid for Y. lipolytica by combining various promoters upstream of the centromeric region to regulate its function and increase its host range. Finally, they showed 80% increase in the copy number and gene expression that offering a starting point for more potent plasmids in Y. lipolytica [107].

Pichia pastoris expression vectors that are commercially available involve the following elements: a multiple cloning site (MCS); the alcohol oxidase promoter (AOX1); a secretion signal sequence (SIG) [includes SUC2 (invertase) and PHO5 (acid phosphatase)], the pGKL killer protein and alpha-MF (S. cerevisiae α-mating factor); a transcription termination site (TT); Ampr (used for selection by ampicillin); HIS4 (marker for selection using hydroxyhistidinase); and ColB1, a replication element for plasmid proliferation in E. coli [24].

The most common vectors for recombinant expression in P. pastoris are given in Table 3. Figure 2 shows the map of pPinkα-HC and pPICZα-E vectors containing general elements. Different P. pastoris host strains and expression vectors have presented. Additional detailed data on vectors and strains may be found in Higgins and Cregg [96, 108]. Moreover, there are several vector DNA sequences at the Invitrogen website (http://www.invitrogen.com) [109].

Fig. 2
figure 2

Diagram of pPinkα-HC and pPICZα-E vectors

Nowadays, to raise expression of the target protein, various systems have been developed. Cre/lox system, as a site-specific recombination technology is available for genetic manipulation of different yeast strains including P. pastoris. This technology has been designed for integration of multiple genes with single selection marker. Li et al. used Cre/lox in a novel expression vector for multicopy expression in P. pastoris. For this, 2–16 copies of appA were constructed with a single selection marker. In this system, the amount of appA gene expression was dependent on gene dosage. The yield of appA protein raised to 4.45-folds when 12 copies of appA were integrated [110].

In H. polymorpha, plasmids that have been established for industrial use in CBS4732-based strains contain pFPMT121 (for the phytase production) and a derivative of pMPT121 (for production of the anti-coagulant hirudin) [111]. The multiple integration of plasmids, AMIpLD1, AMIpSU1, and AMpL1, are based on auxotrophy complementation and have been utilized in eliciting multiple plasmid copy numbers in DL-1-based recombinant hosts. Transformants with low, moderate, or high copy integrated plasmids might be selected when a mutant strain is transformed by these vectors [112]. Using different concentrations of antibiotics such as G418 and hygromycin B, it is possible to select transformants with copy numbers from 1 to 50 [79].

Gateway-compatible destination vectors provide fast and efficient transfer of sequences into a variety of yeast expression systems via site-specific recombination. Saraya et al. recently constructed new modular Gateway-compatible vectors for engineering of H. polymorpha genome, modules for gene deletion/expression (promoters, terminator), and for producing hybrid genes encoding fusion proteins with a green fluorescent protein [113].

Selection markers are necessary elements to ensure plasmid persistence through generations. Common selection markers include antibiotic resistance genes, such as chloramphenicol, G418, hygromycin, zeocin and auxotrophic selection markers including HIS4, LEU2, LYS2, and URA3. The most popular markers in various yeast expression systems are listed in Table 3 [6].

Promoters for Protein Production in Various Yeast Strains

Well-characterized inducible or constitutive promoters with powerful transcriptional activity are usually utilized for overexpression of heterologous proteins [114]. It has been shown that in secretory protein expression, overexpression may result in lower protein yield due to protein aggregation in the endoplasmic reticulum. Therefore, some of the promoters with variable transcriptional activity can be beneficial to achieve optimal secretory expression [10]. Great attempts have been made to develop constitutive promoters with a broad range of transcriptional activities [99]. The conversion of a constitutive promoter to a set of synthetic promoters with a tailor-made regulatory profile has recently been reported [115]. Inducible promoters provide the advantage of control of gene expression levels in response to the presence of particular inducer or repressor [99]. The well-known and powerful promoters that have been used for high-level expression of foreign genes in S. cerevisiae include GAL1, GAL10, JUB1, SNR52, MET17, TDH3, TPI1, ENO1, and PDC1 [100, 116].

Most of S. cerevisiae expression vectors allow the expression of only a single gene. However, it is advantageous to provide an expression of more than one gene per plasmid, for instance, in the case of the expression of whole metabolic pathways. Partow et al. constructed two novel expression vectors that contain two promoters, PGK1 and PTEF1, in an opposite direction to each other. Using this new vector, simultaneous expression of two dissimilar genes from a single vector in S. cerevisiae is provided. Upon the fusion, PPGK1 and PTEF1 activity do not significantly alter compared to the single promoter vector [114]. Dual promoter systems are also promising for improvement of recombinant protein expression in S. cerevisiae [117]. Vickers et al. reported the construction of synthetic bidirectional promoters by combination of the inducible HXT7 (high affinity hexose transporter) or GAL10 promoters for expression of gene cassettes. They found that PTEF1, PHXT7, and PGAL10 direct high level expression with galactose-containing media, but PTEF1 and PHXT7 show better results in continuous fermentations [101].

Two novel inducible promoters in Y. lipolytica, PEYK1 and PEYD1, with practical applications in metabolic engineering and synthetic biology has recently been developed [118]. These promoters are induced by erythrulose and/or by erythritol and repressed by glucose and glycerol [118, 119].

LEU2 promoter is a recombinant promoter (is called hp4d), which is independent from environmental situations, including pH, carbon, nitrogen, and peptone sources and can strongly expresses proteins in various environments. Two other strong and inducible promoters used in Y. lipolytica, include PFBA1 and PFBA1IN(PFBA1+FBA1 intron). Available promoters for the protein production in various host strains of yeasts have listed in the Table 4. Two very strong and adjustable hybrid promoters have recently been designed by adding different copy numbers of an upstream activating sequence (UAS), inserted upstream of an endogenous promoter. These types of recombinant promoters that have been used in Y. lipolytica include UAS1Bn-Leum (a hybrid promoter of PXPR2UAS1B element and PLEU2) and UAS1B8/16-TEF (a hybrid promoter of the PXPR2UAS1B element and PTEF1). These promoters were significantly stronger than all known endogenous Y. lipolytica promoters [46].

Table 4 Common promoters used for protein production in various yeast host strains

Different types of promoters are available for P. pastoris among them glyceraldehyde-3-phosphate dehydrogenase (PGAP) and alcohol oxidase 1(PAOX1) are common constitutive and methanol-dependent inducible promoters, respectively [132]. In addition to AOX1 promoter and GAP promoter, several other promoters have been used in P. pastoris. These include the sodium-coupled phosphate symporter promoter (PPHO89) and the thiamine biosynthesis gene promoter (PTHI11), as inducible promoters and PGCW14, PG1, and PG6 that are considered as constitutive promoters (Table 4) [24, 133, 134]. Recently, Ahmad et al. (2014) introduced a list of promoters for P. pastoris [24]. These promoters suffer from some drawbacks. For example, safety concerns of methanol usage that frequently leads to serious problem in fermentation, especially in large-scale process. Recent improvements in P. pastoris promoters have concentrated on decreasing the required volume of methanol or even avoid its utilization [135]. Synthetic AOX1 promoter variants have recently been reported in which the expression level is regulated by using glucose or glycerol as the only substrate based on principles of repression/derepression [132, 135, 136]. Recently, Wang et al. established a methanol-free AOX1 promoter expression system as a novel glucose–glycerol-shift-induced AOX1 promoter for production of insulin precursor. Compared to induction using methanol, the novel mutant strain produced 77% GFP amounts of wild type strain in glycerol-containing media [137]. Development of dual promoter systems, based on AOX1 and GAP promoters, are promising approaches to improve the expression of heterologous proteins. These promoter systems have been described as the combinatorial usage of two promoters that control transcription of the identical gene. Compared to single promoter-based expression systems, protein expression rate in P. pastoris has been increased via double promoter systems up to sevenfold [138,139,140,141,142].

Foreign gene expression in H. polymorpha is generally carried out under control of a strong and regulatable promoter derived from either methanol oxidase (MOX) [143] or fumarate dehydrogenase (FMD) genes. H. polymorpha methanol oxidase promoter is induced in low concentrations or lack of glucose or in the presence of glycerol as the only source of carbon. This ability is unique among methylotrophic yeasts [68]. Other regulatory and less prevalent inducible promoters are derived from genes encoding for nitrate pathway, including YNT1, YNI1, and YNR1, which are induced by nitrate and inhibited by ammonium [144]. Constitutive promoters of H. polymorpha include PPEX14, PPEX11, and PTEF1 [113]. It has recently been shown that the constitutive PMA1 promoter is able to regulate the expression of two recombinant proteins, human serum albumin (HSA) and glucose oxidase. Using the PMA1 promoter, the expression yield of glucose oxidase and human serum albumin was reached to 185 mg/l and 460 mg/l in high cell density fed batch cultures, respectively. In addition, a productivity of 500 mg/l glucose oxidase and 280 mg/l human serum albumin have been achieved using MOX promoter. These results indicated that HpPMA1 provides a constitutive expression of heterologous proteins and avoids the consumption of methanol as an explosive and toxic hazard. In addition, protein productivity was comparable to MOX promoter [145].

CRISPR/Cas9 System for Yeast Genome Engineering: Advances and Applications

In recent years, CRISPR-Cas9 genome editing technology has been applied for yeast genome engineering (Fig. 3). The popularity of the CRISPR/Cas9 system compared to other conventional methods is related to its outstanding features including multiplexing capability, the efficient introduction of site-specific mutation, flexibility and retargeting efficiency [69]. Most of the studies describing CRISPR/Cas9 genome editing in yeasts have concentrated on S. cerevisiae [146]. The use of installed synthetic Cas9 target sequences into the genome of yeast has recently been reported that faciliate multiplexing using a single sgRNA, precise control on integration of the sequence of interest in to the specific regions of the genome and elimination of off-target effects [147]. The CRISPR/Cas9 system has a lot of potentials in Y. lipolytica and gene deletion studies have shown appropriate efficiency. In addition, several vectors have designed for multiplex gene targeting in Y. lipolytica [148]. In recent years, the expression of different Cas9 genes and gRNA molecules were extensively analyzed in P. pastoris. Weninger et al. recently developed CRISPR/Cas9 strategy for P. pastoris and showed targeting efficiency of almost 100%. They used this optimized technology for gene disruption studies, to introduce deletions of multiple genes and to investigate the integration of site-specific homologous DNA. CRISPR/Cas9 strategy allows fast and marker-free genome engineering in P. pastoris providing unique metabolic and strain engineering applications [69]. It has been reported that expression of HH/HDV-ribozyme-flanked gRNA transcript and Cas9 gene has led to single-gene nonsense mutations with up to 90% efficiency. Although two genes were targeted, nonsense mutations in both ORFs were identified with 69% efficiency. Despite donor template with 1-kbp homologous arms, the efficiency of integration was only 2%. This suggested that NHEJ was the main method of double strand break repair [146]. In another study, the CRISPR-Cas9 system was developed for application in H. polymorpha, using the tRNALeu promoter. For correct sgRNA maturation, this system utilizes the endogenous tRNA processing system. The efficiency of disruption was reached up to 71% that simplified future of metabolic engineering methods [149].

Fig. 3
figure 3

Genome editing using CRISPR-Cas9 technology. The Cas9-sgRNA complex locates the complementary sequence by scanning genomic DNA. After binding, the endonuclease domains create double strand break [150] Upstream of the PAM sequence. DSB is then repaired either by homologous recombination (HR) or by nonhomologous end-joining (NHEJ). Homologous recombination provides precise genome editing at the target site, but nonhomologous end-joining resulting in small insertions or deletions (indels)

Optimization of Recombinant Protein Expression in Yeast Cells

Host strain engineering to enhance the efficiency of heterologous protein production is costly and time consuming. Another economic and easier strategy is the optimization of culture conditions, using the design of experimental strategies. Variations in culture conditions, including pH, oxygen density, temperature, aeration, and induction techniques can extensively affect the cultures yields of yeast [151]. Following parts will describe a brief overview of the optimization procedures used in common yeast expression systems.

The optimum conditions for protein expression in P. pastoris is varied based on host strain and expressed proteins. Wang et al. reported optimization of Apostichopus japonicus (Aj) lysozyme expression by various medium conditions, including different pHs, temperatures, oxygen densities, and methanol levels. The optimum protein yield was achieved in 1% methanol density, pH 5 and induction time of 120 h [152]. Some studies, have shown that degradation of target proteins is diminished in low temperatures. Li et al. studied the effect of different temperatures (23, 25, 27, and 30) on protein productivity and reported that low temperatures resulted in the enhancement of protein yield and reduction of protease activity [109].

Cell growth, protein production, and stability are highly affected by optimal pH. Jahic et al. reported that reduction of pH from 5 to 4 in P. pastoris bioreactor cultures resulting in 50% increase of CBM CALB (cellulose-binding module) productivity. In the induction phase, pH medium about three leads to the efficient protection of IGF-1 (insulin like growth factor) product from proteolysis [153]. Although methanol is utilized for induction of AOX1 promoter, in Mut+ and Muts strains, the methanol concentration of more than 4 g/l hampers cellular growth [154]. A significant alternative method includes utilization of glycerol as a cosubstrate [155, 156]. It has been shown that AOX1 promoter is not repressed by moderately low levels of glycerol [120, 157]. Three feeding phases is required for keeping the methanol concentration below the toxic limit in the fed-batch cultures. In the initial phase, the modified strain is cultured in a medium containing glycerol as the carbon source for enhancing biomass. In the second phase (as fed-batch phase), the low amount of glycerol is used to increase cell density without growth inhibition. In the third phase, induction is initiated by adding methanol to the fermentation environment [158].

Although the rate of dissolved oxygen is one of the most influencing factors for cell growth and recombinant protein expression, the extent of the effect has been variable in case of different proteins. Trentmann et al. have been carried out production of a single-chain antibody fragment in P. pastoris under a low levels of dissolved oxygen [159]. They indicated that the production was not affected by oxygen reduction. An identical method was used by Hellwig et al. for production of heterologous scFv antibody fragment [160]. In these researches, the dissolved oxygen was approximately zero during the induction phase, since the oxygen uptake rate and the maximum oxygen transfer rate were equal. Previous studies have shown that higher levels of dissolved oxygen may increase the product titer. Jazini et al. studied the effects of various dissolved oxygen concentrations on protein expression during the induction phase [161]. They perceive that diverse dissolved oxygen rates did not vary the product-specific yield [162]. In Y. lipolytica, it is critical to select an optimized oleic acid feeding and production of protein in the bioreactor to increase the productivity. In this yeast, the optimum inducer concentration and selection of the appropriate medium is vital to raise the production of recombinant protein. Gasmi et al. in 2011 conducted a research about the optimization of hunIFNa2b production. In this analysis, the effects of different feeding strategy and pre-induction cell density were evaluated in protein production. In this research, the effects of simultaneous feeding of oleic acid and glucose, different methods of oleic acid addition to induce POX2 promoter and stability of cell density was investigated. As a result of optimization, huIFN a2b protein production was reached to 200 mg/l [163]. In H. polymorpha, fermentation is based on FMD or MOX promoters using glucose or glycerol and will continue to the second phase with the limitation of carbon source. Protein expression optimization in this yeast is related to medium condition (pH, aeration and dissolved oxygen density), carbon source (glucose and methanol) and higher levels of expression in the engineered strains [123].

N-glycan Engineering in Yeast Expression Systems

Due to the important effects of glycosylation on the biological characteristics of glycoproteins, the pattern of glycosylation should carefully be controlled through development processes [164, 165]. It has been revealed that the glycosylation pattern is significantly different between human and yeast species (Fig. 4a). In this regard, extensive efforts have been made to compensate glycan heterogeneity and develop glycoengineering strategies for the production of efficient therapeutic glycoproteins [166]. Recently glycoengineered yeasts have been developed with the capability to generate glycoproteins with human-like N-glycans structures [167, 168]. In a recent study, a triple mutant S. cerevisiae strain has been created by deletion of ALG3, OCH1, and MNN1 genes [169]. The Δalg3 Δoch1 Δmnn1 triple mutant strain produced Man5GlcNAc2 intermediate of human N-glycosylation without any growth defects. This modified strain can be utilized as the first strain to produce an expression system based on the yeast for the production of therapeutic glycoprotein [170]. Construction of engineered Y. lipolytica strains generating glycoproteins homogeneously glycosylated with either Man8GlcNAc2 or Man5GlcNAc2 has been recently reported. Resultant strains, YlOch1p and YlMnn9p, do not have yeast-specific Golgi α-1,6 mannosyltransferases and made homogeneous Man8GlcNAc2 N-glycans. Using expression of a fungal α-1,2- mannosidase, this strain was more modified for producing homogeneous Man5GlcNAc2 residues [171]. The same research groups produced a Δoch1 Δmpo1 double mutant strain without yeast-specific mannosyl phosphorylation and hypermannosylation [172]. Recently, this Δoch1 Δmpo1 double-mutant strain was more engineered via a fungal α-1,2- mannosidase. These novel host strains will promote Y. lipolytica as an expression system capable of producing humanized N-linked oligosaccharides suitable for therapeutic uses [46]. To humanize yeast N-glycosylation pathways in P. pastoris, the first step is the deletion of the OCH1 gene due to its important role in catalyzing the initial reaction of hypermannosylation. Disruption of the OCH1 gene from P. pastoris resulting in a novel strain for the production of heterologous proteins with homogeneous shorter glycans [173]. Compared to a wild type strain, the mannoses number in glycans of engineered strain decreased from ten to eight. In addition, removal of the ALG3 with numerous modifications led to human-like Man3GlcNAc2 glycoform (Fig. 4 b) [168]. Unlike S. cerevisiae, H. polymorpha is less prone to hyperglycosylation due to the fact that the glycosylation pathway does not add alpha-1,3-linked residues [174]. Furthermore, significant attempts have been made for optimization of human-like glycosylation [78].

Fig. 4
figure 4

An overview of engineering the N-linked glycosylation pathways in yeasts. a N- linked glycosylation in S. cerevisiae and human. Synthesis of highly mannosylated N-glycan in S. cerevisiae containing OCH1 gene. The Golgi apparatus is the place of N-Glycan processing that includes sequential attachment of mannose residues by mannosyltransferases. In this case, the final compound is very mannosylated with various numbers of mannose up to almost 50. b (1) Synthesis of human-complex type N-glycan in P. pastoris, S. cerevisiae and H. polymorpha via removal of OCH1 gene. (2) Synthesis of human-complex type N-glycan in P. pastoris, S. cerevisiae and H. polymorpha via removal of OCH1 and ALG3 genes. (3) Synthesis of Human complex-type N-glycan in P. pastoris, S. cerevisiae and H. polymorpha via removal of ALG3 and ALG11 genes

Products on—or on the Way to—the Market

Saccharomyces cerevisiae, as the first completely sequenced eukaryotic genome, [7], has been developed to produce a large variety of heterologous proteins for the last three decades. The first licensed vaccine against hepatitis B [175] was generated in recombinant S. cerevisiae. Until 2010, FDA and EMEA approved recombinant therapeutic proteins from microbial eukaryotic organisms are almost exclusively manufactured by S. cerevisiae [176]. Products on the market from S. cerevisiae, include hepatitis B surface antigen, insulin, hirudin, urate oxidase, granulocyte, glucagons, platelet-derived growth factor and macrophage colony stimulating factor (GM-CSF) (Table 5) [6, 25].

Table 5 Products on the market in various yeast host strains

American Food and Drug Administration organization has approved many processes based on Y. lipolytica as generally regarded as safe (GRAS) [46]. Y. lipolytica is an outstanding host for secretory production of several proteins such as two proteases, many lipase and phosphatase, an RNase and an esterase [188]. Under appropriate induction conditions, this yeast is able to secrete high levels (1–2 g/l) of alkaline proteases to the culture medium [189].

Until now, several recombinant proteins comprising industrial enzymes and biopharmaceuticals have been produced in P. pastoris that have been approved and marketed (presented on the web page: http://www.pichia.com) [190]. Products under development include Novozymes (a highly active antimicrobial agent), Elastase inhibitor against Cystic Fibrosis and the plectasin peptide derivative of NZ2114 (an antimicrobial peptide) [191, 192]. In addition to approved pharmaceutical products, some of the recombinant cytokines and growth factors produced in Pichia, including human stem cell factor (SCF), human HSA-IFN-Alpha 2b and ovine IFN-τ, murine TNF-α have been marketed for use in research (Table 5) [24].

Hansenula polymorpha is a safe expression system which has been established for heterologous protein production at a commercial scale [193, 194]. Numerous H. polymorpha-derived proteins, like insulin and hepatitis B vaccines, have been approved. These products have been marketed and utilized in some countries [79, 195]. A heterologous hexose oxidase is used for baking (Table 5) [196].

Conclusion and Future Views

Over the years, a variety of yeast strains have been successfully utilized for production of therapeutic proteins. Many genetic elements of yeast expression system containing a broad variety of vectors and host strains are well developed. However, further optimization of protein expression and secretion dependent on the product of interest is needed [6]. Marker-less integration and rapid deletion of various genes as the well-established tools for strain engineering will offer a potent set of engineered expression hosts in the future [24]. Glycosylation has a significant effect on protein half-life and function; hence it must efficiently be controlled. Many efforts have been made for producing therapeutic glycoproteins with human-like glycosylation pattern in certain host strains such as P. pastoris [52]. In addition, yeasts are applied as a strong tool in high-throughput analyses for drug screening and functional genomics. This expression system provides very efficient homologous recombination feature, hence is extensively used as an engineering tool to overcome the problems in cloning larger DNA pieces [197, 198], opening the doors to new eras in molecular biology and biotechnology.