Molecular engineering of Komagataella phaffii for venom toxin production Authors: Francielle Almeida Cordeiro, Karla de Castro Figueiredo Bordon, Henrique Ranieri Covali-Pontes, Isabela Gobbo Ferreira, Gisele Adriano Wiezel, Isadora Lino Mendes, Malson Neilson de Lucena, Eliane Candiani Arantes Published: 2026 ABSTRACT Animal venoms constitute a rich source of bioactive peptides and proteins with high target specificity, representing valuable scaffolds for therapeutic development. However, the biotechnological exploitation of venom-derived toxins is limited by challenges in achieving efficient, scalable, and reproducible production. Native venom extraction is constrained by low yields and biological variability, making recombinant platforms essential. Yet, most venom toxins are cysteine-rich peptides with complex disulfide bond architectures and stringent structure–function relationships, posing significant challenges to heterologous expression. Inefficient folding, proteolysis, and secretion bottlenecks frequently compromise functional yield. Among microbial hosts, Komagataella phaffii has emerged as a robust system combining eukaryotic protein processing with high cell-density fermentation and cost-effective cultivation. Its oxidative secretory pathway, strong and regulatable promoters, and suitability for strain engineering make it particularly attractive for producing disulfide-rich toxins. This review provides a critical analysis of recombinant venom toxin production in K. phaffii, focusing on molecular and bioprocess determinants of expression performance. We discuss post-translational modifications, yields, and bioactivity, as well as promoter selection and secretion signal optimization. By integrating data across toxin families, we identify recurring technical bottlenecks and highlight engineering approaches to enhance venom biomanufacturing within microbial biotechnology frameworks. Introduction Toxins from both microbial and animal sources (e.g., spiders, snakes, and scorpions) have been increasingly applied in the research and production of vaccines and therapeutic agents, including anticancer, antimicrobial, and antiviral drugs (Bordon et al. 2020; Shilova et al. 2021). Nevertheless, acquiring these compounds is exceedingly labor-intensive due to the low yield of isolated toxins and the substantial amount of venom required for testing and biotechnological applications. Heterologous expression represents a significant advancement in this field, enabling the production of these peptides and proteins on a larger scale (Amorim, Cordeiro, Pinheiro-Júnior, Boldrini-França, & Arantes 2018a, b; Wulff et al. 2019). Chemical synthesis also provides a cost-effective and flexible approach to creating small peptides with specific modifications. However, larger molecules or those with complex structures and the incorporation of post-translational modifications can be labor-intensive and technically demanding (Ma et al. 2023). Among the available expression platforms, Komagataella phaffii (formerly named Pichia pastoris) has emerged as one of the most efficient systems for producing recombinant toxins, particularly due to its high yield and eukaryotic folding capabilities (De Brabander et al. 2023; Mastropietro et al. 2021). Considering this point, this review consolidates and analyzes the use of heterologous expression in K. phaffii for animal toxin production. It emphasizes the advantages, such as high yields, post-translational modifications, efficient secretion, and scalability, that make it preferable over systems like Escherichia coli for complex and disulfide-rich molecules. This review also compiles successful examples of toxin production from snakes, scorpions, spiders, and other animals; evaluates key challenges, including hyperglycosylation and proteolytic degradation; and suggests strategies to address them, including new strategies of engineering. Advantages of the K. phaffii expression system K. phaffii has become a widely adopted platform for recombinant protein production due to its balanced combination of high productivity, eukaryotic processing capacity, and industrial scalability. However, its advantages must be interpreted within the context of the specific biochemical and structural requirements of the target protein, particularly for complex toxins where folding constraints, disulfide bond patterns, and post-translational modifications (PTMs) are critical determinants of functionality (Swichkow et al. 2022). One of the main advantages of K. phaffii is its ability to achieve high expression levels, often surpassing prokaryotic systems such as E. coli, particularly for proteins rich in disulfide bonds or secreted proteins (Karbalaei et al. 2020). For example, recombinant toxins expressed in yeast show superior performance compared to E. coli, with higher yields and better functional recovery, often eliminating the need for refolding. For instance, neurotoxin production in yeast exceeded 9 mg/L, whereas E. coli typically produced 1–2 mg/L and required renaturation steps (Borrego et al. 2022). This advantage is especially relevant for venom-derived peptides, where misfolding directly compromises bioactivity. Furthermore, advances in strain engineering and bioprocess optimization have significantly expanded the productive capacity of this system, with fed-batch fermentations reaching milligram-per-liter yields and, in optimized cases, even grams per liter (Barone et al. 2023; Carneiro et al. 2022; Lv & Cai 2025). However, these high yields are not universal and strongly depend on protein-specific factors, including secretion efficiency and folding complexity, highlighting that reported yields should be interpreted critically rather than directly compared between studies. In contrast to prokaryotic hosts, K. phaffii provides a eukaryotic intracellular environment capable of performing essential PTMs, including disulfide bond formation, signal peptide processing, and N- and O-linked glycosylation (Tsuda & Nonaka 2024). Its endoplasmic reticulum supports oxidative folding through chaperones such as protein disulfide isomerase, which is particularly advantageous for cysteine-rich toxins. Importantly, glycosylation in K. phaffii is less hypermannosylated than in Saccharomyces cerevisiae and can be further humanized through glycoengineering strategies, enabling the production of proteins with improved stability and reduced immunogenicity (Barone et al. 2023; Hardianto et al. 2023; Karbalaei et al. 2020). However, glycosylation remains a double-edged feature: While beneficial for some proteins, it may introduce heterogeneity or alter activity in toxins that are not naturally glycosylated, requiring careful design or strain selection. K. phaffii is also highly suitable for large-scale production. It reaches high cell densities in chemically defined media, facilitating reproducibility and regulatory compliance while avoiding animal-derived components. Its Crabtree-negative metabolism enables efficient carbon utilization without ethanol accumulation, which is advantageous in high-density aerobic fermentations (Ata et al. 2021). Furthermore, tightly regulated promoters, such as the inducible AOX1, allow for controlled gene expression, although methanol-based induction introduces operational and safety considerations that have driven the development of methanol-free systems in recent years (Karbalaei et al. 2020; Pan et al. 2022). In addition to the technical limitations associated with low yields and labor-intensive extraction, venom collection from wild animals or livestock also raises important ethical and environmental concerns. In this context, recombinant expression systems such as K. phaffii, which can be cultivated in fully chemically defined media without animal-derived components, represent a more sustainable and ethically responsible alternative. Another relevant advantage is its efficient secretory pathway, which enables recombinant proteins to be targeted to the extracellular medium using signaling peptides, such as the α-factor leader. This simplifies downstream processing, reducing contamination by host cell proteins and eliminating cell disruption steps (Barone et al. 2023). The system also benefits from a strong safety and regulatory profile, being classified as Generally Recognized As Safe (GRAS) and lacking endotoxin production, which is critical for pharmaceutical applications (Janpan et al. 2024; Tsuda & Nonaka 2024; Zhong et al. 2024). Stable genomic integration via homologous recombination ensures consistent expression in large-scale fermentations, supported by a wide array of genetic tools, including diverse promoters, selectable markers, and modular cloning platforms (Carneiro et al. 2022; Janpan et al. 2024). However, the efficient expression of certain toxins still requires the development of customized vectors and strain optimization, reinforcing the need for integrated metabolic and synthetic biology approaches. Finally, the versatility of K. phaffii is evidenced by its wide range of applications, with thousands of expressed proteins and dozens of commercial products extending the biopharmaceutical, industrial, and agricultural sectors (Lv & Cai 2025). Its successful use in the production of venom-derived molecules underscores its suitability for structurally complex peptides. Despite this, its performance is highly context-dependent, and its optimal use increasingly depends on rational engineering strategies rather than just intrinsic system properties. Recombinant toxins produced in the K. phaffii system The advantages of K. phaffii have enabled its application in the production of venom-derived toxins from diverse animal sources. However, expression outcomes remain highly dependent on toxin-specific features, particularly structural complexity and post-translational requirements. While K. phaffii can produce functional peptides and proteins across different toxin classes, variations in yield, folding efficiency, and processing highlight that its performance is strongly protein-dependent. This section summarizes representative examples of recombinant toxin production in K. phaffii, highlighting key challenges and determinants of expression performance. Animal toxins produced in K. phaffii to date are summarized in Table 1. Snake venom Snake venoms have gained considerable attention in recent years due to their complex and diverse protein compositions, which hold significant pharmacological and biotechnological potential (Bordon et al. 2020; Ferreira et al. 2023; Pinheiro-Junior et al. 2021). K. phaffii expression strategy is favored for its potential to achieve higher yields and reduce costs (Ata et al. 2021; Eskandari et al. 2023). However, the effectiveness of this approach is contingent upon the successful folding and functional expression of the proteins, which remains a critical bottleneck in the field. Echistatin, a disintegrin derived from the venom of Echis carinatus, serves as a pertinent example. Although it has been successfully expressed in K. phaffii with functional activity, it raises important questions about the reliability of such expression systems in maintaining biological activity. The study highlights that the C-terminal residues play a crucial role in its interaction with integrin β3 (Chen et al. 2012). This raises further concerns about whether the yeast system adequately mimics the post-translational modifications that may be critical for the in vivo functionality of such proteins. Thus, for the expression of echistatin in K. phaffii, the correct folding is critical to ensure its biological activity. In this study, it was successfully expressed in the X33 strain using the pPICZαA vector and exhibited functional activity, including inhibition of platelet aggregation. It was expressed with functional activities and in the correct folding, demonstrating that lysine at position 45 (K45) in the C-terminal tail plays an important role in its interaction with the αIIbβ3 integrin (Chen et al. 2012). Similarly, the recombinant expression of rCollinein-1, a snake venom serine proteinase, in the KM71H K. phaffii strain using the pPICZA vector demonstrated high stability across different pH and temperature conditions, as well as coagulant activity. Its esterase activity was inhibited in the presence of zinc (Zn2+) and copper (Cu2+), indicating that it was expressed with functional integrity on the evaluated parameters. This successful approach yields insights into the future applications of SVSPs (Boldrini-França et al. 2020, 2015). Despite successful recombinant expression, a major challenge in producing proteinases in K. phaffii is their glycosylation pattern, which may differ from that of native enzymes and often results in hyperglycosylation, potentially affecting their functional activity. This aspect will be further demonstrated and discussed below. Still on snake enzymes, batroxobin is a thrombin-like enzyme found in the venom of Bothrops moojeni [syn. Bothrops atrox moojeni]. Its recombinant form produced in K. phaffii has been investigated for its hemostatic potential (You et al. 2004). In addition to demonstrating stability across a range of pH and temperature conditions, in vivo toxicological studies revealed a favorable safety profile in animal models, supporting its therapeutic potential (Choi et al. 2018). Clinical evaluation further indicated dose-dependent defibrinogenating activity with acceptable tolerability in humans. Batroxobin has long been used clinically in its native form (Reptilase/Defibrase), particularly in diagnostic assays and as a defibrinogenating agent. Despite these promising outcomes, recombinant batroxobin may not fully recapitulate the functional complexity of its native counterpart and may exhibit distinct activities (You et al. 2004). Differences in post-translational modifications, especially glycosylation patterns in yeast-based expression systems, can influence enzymatic activity, stability, and immunogenicity. Despite these limitations, its recombinant form has been explored in human studies, demonstrating that it is well tolerated and exhibits a dose-dependent defibrinogenating effect in healthy adult subjects. (Choi et al. 2018). Jarastatin (rJast), a disintegrin from Bothrops jararaca venom, has also been successfully expressed recombinantly. Functional assays demonstrated preserved biological activity, including inhibition of platelet aggregation and cell adhesion (Succar et al. 2022). Nevertheless, disintegrins are typically small, cysteine-rich proteins that depend on precise disulfide bond formation for activity, making their recombinant production particularly challenging. While K. phaffii can support disulfide bond formation, variations in folding efficiency and potential heterogeneity in the final product may impact reproducibility and scalability. Similarly, rJararacin was expressed with relatively high yields (~ 40 mg/L) and retained biological activity (David et al. 2023). Although these results highlight the robustness of yeast expression systems, it is worth noting that yield alone does not necessarily correlate with functional equivalence. Subtle structural differences, including glycosylation and folding variants, may alter receptor binding affinity or in vivo pharmacokinetics, which are not always thoroughly assessed in early-stage studies. Regarding elapid snake toxins, the expression of α-cobratoxin from Naja kaouthia across multiple systems, including E. coli and K. phaffii, underscores a recurring bottleneck in the field (Damsbo et al. 2024). None of the systems tested was able to fully reproduce the native toxin, highlighting the intrinsic difficulty of expressing structurally constrained toxins with multiple disulfide bonds. However, the His6-tagged α-cobratoxin expressed in K. phaffii exhibited a secondary structure comparable to that of the variant, as indicated by circular dichroism spectra, and similar binding properties to the α7 subunit of the nicotinic acetylcholine receptor (Damsbo et al. 2024). Although several snake venom proteins have been expressed in E. coli and, to a lesser extent, in mammalian cell systems, these platforms present significant limitations, including improper folding, lack of post-translational modifications, or high production costs. Consequently, K. phaffii has emerged as a widely used alternative for recombinant expression; however, despite its clear advantages, several critical challenges remain, particularly discrepancies in post-translational modifications, folding efficiency, and structural fidelity compared to native toxins, as well as issues related to scalability and reproducibility. Addressing these limitations will be essential for advancing these molecules from experimental systems to clinically viable therapeutics. Scorpion venom Scorpion toxins expressed in K. phaffii represent the most abundant group among recombinant venom-derived molecules. In contrast to snake toxins, which are predominantly enzymatic and often glycosylated, scorpion toxins are mainly neurotoxins, and their recombinant production shows substantial variability in yield, proper processing, and correct folding. In total, 19 are peptides, and 2 are recombinant enzymes (Table 1). Among the strains employed for heterologous expression, X33 was notable for producing eleven recombinant toxins, including the recombinant enzymes rLqhIT2, rMgTx, and rBjαlT/GNA, with yields exceeding 1.4 g/L of rLqhIT2 (Li et al. 2025a, b; Li & Xia 2018b; Naseem et al. 2021). Table 1Examples of toxins produced in Komagataella phaffiiAnimalClassName of toxinSpeciesPTMsVectorStrainYieldPurification method/MWReferenceSnakesDisintegrinsEchistatinEchis carinatus4 disulfide bondspPICZαAX33Flasks: 2–7 mg/LNi2+ chelating chromatography and reversed-phase C18 HPLC6.5 kDa(Chen et al. 2012)15N-labeled echistatinEchis carinatus4 disulfide bondspPICZαAX33Flasks: 1–3 mg/mLNi2+ chelating chromatography and reversed-phase C18 HPLC7.6 kDa(Chen et al. 2012)Jarastatin (rJast)Bothrops jararaca6 disulfide bondspPIC9GS115Flasks: 40 mg/LMolecular exclusion chromatography7.73 kDa(Succar et al. 2022)rJararacinBothrops jararacaNDpPIC9GS115Flasks: 40 mg/LSuperose 12 10/300 GL7.7 kDa(David et al. 2023)EnzymesSVSPsaCollinein-1Crotalus durissus collilineatus6 disulfide bonds/N-glycosylation sitepPICZαAKM71HFlasks: 56 mg/LIon exchange and reversed-phase chromatography C428.8 kDa(Boldrini-França et al. 2015)BatroxobinBothrops atrox moojeni6 disulfide bonds/2 glycosylation sitespPIC9GS115Bioreactor: 7 mg/LPhenyl-Sepharose column/heparin–Sepharose affinity chromatography33 kDa(You et al. 2004)3FTXbα-CobratoxinNaja kaouthia5 disulfide bondspPICZαAKM71HFlasks: 3 mg/LHIS-Select® Nickel Affinity Gel resin8.8 kDa(Damsbo et al. 2024)ScorpionsNeurotoxinsAgTx2Leiurus quinquestriatus3 disulfide bondspPICZαAX-33Flasks: 14–18 mg/LNi2+ chelating chromatographyRP-HPLC5.19 kDa(Anangi et al. 2012)MgTxCentruroides margaritatus3 disulfide bondspPICZαAX-33Flasks: 12–15 mg/LNi2+ chelating chromatographyRP-HPLC5.27 kDa(Anangi et al. 2012)AnTx2Anuroctonus phaiodactylus4 disulfide bondspPICZαAX-33Flasks: 46 mg/LNi2+ chelating chromatographyRP-HPLC4.1 kDa(Borrego et al. 2022)Ts6Tityus serrulatus4 disulfide bondspPICZαAX-33Flasks: 7.5 mg/LNi2+ chelating chromatographyRP-HPLC4.5 kDa(Borrego et al. 2022)Vm24Vaejovis mexicanus4 disulfide bondspPICZαAX-33Flasks: 1.6 mg/LNi2+ chelating chromatographyRP-HPLC3.8 kDa(Borrego et al. 2022)TrMgTxCentruroides margaritatus3 disulfide bondspPICZαAX-33Bioreactor: 36 mg/LNi2+ chelating chromatographyRP-HPLC4.1 kDa(Naseem et al. 2021)BmαTX14Mesobuthus martensii4 disulfide bondspPIC9KGS115Flasks: 0.9 mg/mLMetal chelate affinity chromatography–gel filtration chromatography8.2 kDa(Wang et al. 2006)Insect-specific neurotoxinsrLqhIT2Leiurus quinquestriatus4 disulfide bondspPICZαAX-33Bioreactor: 1.4 g/LNi2+ chelatingDEAE Sepharose FF8 kDa(Li et al. 2025a, b)rAalTAndroctonus australis4 disulfide bondspPICZαAX-33Biorreactor: 240 mg/LNi+-NTA columnCM Sepharose FF8.6 kDa(Li & Xia 2018a)rAalTAndroctonus australis4 disulfide bondspPIC9KKM71Flasks: 10 mg/LUltracentrifuge filter8 kDa(Li & Xia 2018b)rBjαITButhotus judaicus4 disulfide bondspPICZαAX-33Bioreactor: 340 mg/LNi2+ chelatingCM Sepharose FF8 kDa(Li & Xia 2018a)rButaITMesobuthus tamulus4 disulfide bondspGAPZαBSMD1168HBench-top bioreactor: < 1 mg/LPhenyl-Sepharose10 kDa(Pham Trung et al. 2006)rButaIT/GNAMesobuthus tamulus4 disulfide bondspGAPZαBSMD1168HBench-top bioreactor: 25–35 mg/LCation-exchange chromatography16.5 kDa(Pham Trung et al. 2006)NDBPscrBmK AngM1Buthus martensii4 disulfide bondspPIC9KGS115Bioreactor: 500–1200 mg/LNi2+ chelating8 kDa(J.-L. Yang et al.)AEPButhus martensiiNDpPIC9KSMD1168HFlasks: 0.9 mg/mLNi2+ chelating8.3 kDa(Z. Wang et al. 2009)rTs8Tityus serrulatus3 disulfide bondspPICZαAKM71HNDNi2+ chelatingRP-HPLC ~ 8.1 kDa(Cordeiro et al. 2022)SD 4-kDa scorpion defensinLeiurus quinquestriatus3 disulfide bondspPICZαAX-33Flasks: 0.92 mg/mLNi2+ chelating chromatography5.1 kDa(Niu et al. 2013)CPMesobuthus martensii3 disulfide bondspPICZαAX-33Flasks: 271 mg/L11.6 kDa(Niu et al. 2008)EnzymesrTsHyal-1Tityus serrulatus5 disulfide bonds/5 N-glycosylation sitespPICZαAKM71HFlasks: 0.266 mg/mLCM-cellulose-5249.5 kDa(Amorim et al. 2018a, b)rPLA2(−5)Scorpio maurus4 disulfide bondspGAPZαBSMD-1168HFlasks: 18 mg/LUltracentrifuge filterGel filtration G7514 kDa(Krayem et al. 2022)SpidersrTx4 (6–1)Phoneutria nigriventer5 disulfide bonds/O-glycosylation sitepPICZαAX33Flasks: 5 mg/LBioreactor: 45 mg/LNi+-nitriloacetic acid affinity chromatography, carboxymethyl (CM) Sepharose ion-exchange chromatography7.6 kDa(Li & Xia 2019)Neurotoxinsd-Amaurobitoxin-PI1aPireneitega luctuosus4 disulfide bonds/N-glycosylation sitepGAPZαBSMD1168HBench-top bioreactor: 26 mg/LNi2+ chelating chromatography14 kDa(S. Yang et al. 2014)ω-Hexatoxina-Hv1aHadronyche versuta3 disulfide bondspGAPZαBSMD1168HBioreactor: 50 mg/LHydrophobic interaction and gel filtration chromatography20 kDa(Fitches et al. 2012)SnailNeurotoxinLt28.1Conus litteratus5 disulfide bondspPIC9GS115Flasks: 2 mg/LNi2+ chelating chromatographyGel filtration5.9 kDa(Lu et al. 2017)BeeEnzymeHyaApis mellifera4 disulfide bonds/2 glycosylation sitespPIC9GS115 (his4)Flasks: 10 mg/LCation exchangeSephadex 5050 kDa(Reitinger et al. 2008)LizardPeptideExendin-4Heloderma suspectumNDpPIC9GS115Bioreactor: 26 mg/LIon exchange, hydrophobic interaction chromatography (HIC), and size-exclusion chromatography4.3 kDa(Zhou et al. 2008)WaspEnzymeVest2aVespa tropica2 disulfide bonds/5 glycosylation sitesGoldenPiCS systemCBS 2612Bioreactor: 96.45 mg/LNi2+ chelating chromatography and reversed-phase59.35 kDa(Janpan et al. 2024)PTMs post-translational modificationsaSVSPs—snake venom serine proteinasesb3FTX—three finger toxinscNDBPs—non-disulfide-bridged peptide Table 1Examples of toxins produced in Komagataella phaffiiAnimalClassName of toxinSpeciesPTMsVectorStrainYieldPurification method/MWReferenceSnakesDisintegrinsEchistatinEchis carinatus4 disulfide bondspPICZαAX33Flasks: 2–7 mg/LNi2+ chelating chromatography and reversed-phase C18 HPLC6.5 kDa(Chen et al. 2012)15N-labeled echistatinEchis carinatus4 disulfide bondspPICZαAX33Flasks: 1–3 mg/mLNi2+ chelating chromatography and reversed-phase C18 HPLC7.6 kDa(Chen et al. 2012)Jarastatin (rJast)Bothrops jararaca6 disulfide bondspPIC9GS115Flasks: 40 mg/LMolecular exclusion chromatography7.73 kDa(Succar et al. 2022)rJararacinBothrops jararacaNDpPIC9GS115Flasks: 40 mg/LSuperose 12 10/300 GL7.7 kDa(David et al. 2023)EnzymesSVSPsaCollinein-1Crotalus durissus collilineatus6 disulfide bonds/N-glycosylation sitepPICZαAKM71HFlasks: 56 mg/LIon exchange and reversed-phase chromatography C428.8 kDa(Boldrini-França et al. 2015)BatroxobinBothrops atrox moojeni6 disulfide bonds/2 glycosylation sitespPIC9GS115Bioreactor: 7 mg/LPhenyl-Sepharose column/heparin–Sepharose affinity chromatography33 kDa(You et al. 2004)3FTXbα-CobratoxinNaja kaouthia5 disulfide bondspPICZαAKM71HFlasks: 3 mg/LHIS-Select® Nickel Affinity Gel resin8.8 kDa(Damsbo et al. 2024)ScorpionsNeurotoxinsAgTx2Leiurus quinquestriatus3 disulfide bondspPICZαAX-33Flasks: 14–18 mg/LNi2+ chelating chromatographyRP-HPLC5.19 kDa(Anangi et al. 2012)MgTxCentruroides margaritatus3 disulfide bondspPICZαAX-33Flasks: 12–15 mg/LNi2+ chelating chromatographyRP-HPLC5.27 kDa(Anangi et al. 2012)AnTx2Anuroctonus phaiodactylus4 disulfide bondspPICZαAX-33Flasks: 46 mg/LNi2+ chelating chromatographyRP-HPLC4.1 kDa(Borrego et al. 2022)Ts6Tityus serrulatus4 disulfide bondspPICZαAX-33Flasks: 7.5 mg/LNi2+ chelating chromatographyRP-HPLC4.5 kDa(Borrego et al. 2022)Vm24Vaejovis mexicanus4 disulfide bondspPICZαAX-33Flasks: 1.6 mg/LNi2+ chelating chromatographyRP-HPLC3.8 kDa(Borrego et al. 2022)TrMgTxCentruroides margaritatus3 disulfide bondspPICZαAX-33Bioreactor: 36 mg/LNi2+ chelating chromatographyRP-HPLC4.1 kDa(Naseem et al. 2021)BmαTX14Mesobuthus martensii4 disulfide bondspPIC9KGS115Flasks: 0.9 mg/mLMetal chelate affinity chromatography–gel filtration chromatography8.2 kDa(Wang et al. 2006)Insect-specific neurotoxinsrLqhIT2Leiurus quinquestriatus4 disulfide bondspPICZαAX-33Bioreactor: 1.4 g/LNi2+ chelatingDEAE Sepharose FF8 kDa(Li et al. 2025a, b)rAalTAndroctonus australis4 disulfide bondspPICZαAX-33Biorreactor: 240 mg/LNi+-NTA columnCM Sepharose FF8.6 kDa(Li & Xia 2018a)rAalTAndroctonus australis4 disulfide bondspPIC9KKM71Flasks: 10 mg/LUltracentrifuge filter8 kDa(Li & Xia 2018b)rBjαITButhotus judaicus4 disulfide bondspPICZαAX-33Bioreactor: 340 mg/LNi2+ chelatingCM Sepharose FF8 kDa(Li & Xia 2018a)rButaITMesobuthus tamulus4 disulfide bondspGAPZαBSMD1168HBench-top bioreactor: < 1 mg/LPhenyl-Sepharose10 kDa(Pham Trung et al. 2006)rButaIT/GNAMesobuthus tamulus4 disulfide bondspGAPZαBSMD1168HBench-top bioreactor: 25–35 mg/LCation-exchange chromatography16.5 kDa(Pham Trung et al. 2006)NDBPscrBmK AngM1Buthus martensii4 disulfide bondspPIC9KGS115Bioreactor: 500–1200 mg/LNi2+ chelating8 kDa(J.-L. Yang et al.)AEPButhus martensiiNDpPIC9KSMD1168HFlasks: 0.9 mg/mLNi2+ chelating8.3 kDa(Z. Wang et al. 2009)rTs8Tityus serrulatus3 disulfide bondspPICZαAKM71HNDNi2+ chelatingRP-HPLC ~ 8.1 kDa(Cordeiro et al. 2022)SD 4-kDa scorpion defensinLeiurus quinquestriatus3 disulfide bondspPICZαAX-33Flasks: 0.92 mg/mLNi2+ chelating chromatography5.1 kDa(Niu et al. 2013)CPMesobuthus martensii3 disulfide bondspPICZαAX-33Flasks: 271 mg/L11.6 kDa(Niu et al. 2008)EnzymesrTsHyal-1Tityus serrulatus5 disulfide bonds/5 N-glycosylation sitespPICZαAKM71HFlasks: 0.266 mg/mLCM-cellulose-5249.5 kDa(Amorim et al. 2018a, b)rPLA2(−5)Scorpio maurus4 disulfide bondspGAPZαBSMD-1168HFlasks: 18 mg/LUltracentrifuge filterGel filtration G7514 kDa(Krayem et al. 2022)SpidersrTx4 (6–1)Phoneutria nigriventer5 disulfide bonds/O-glycosylation sitepPICZαAX33Flasks: 5 mg/LBioreactor: 45 mg/LNi+-nitriloacetic acid affinity chromatography, carboxymethyl (CM) Sepharose ion-exchange chromatography7.6 kDa(Li & Xia 2019)Neurotoxinsd-Amaurobitoxin-PI1aPireneitega luctuosus4 disulfide bonds/N-glycosylation sitepGAPZαBSMD1168HBench-top bioreactor: 26 mg/LNi2+ chelating chromatography14 kDa(S. Yang et al. 2014)ω-Hexatoxina-Hv1aHadronyche versuta3 disulfide bondspGAPZαBSMD1168HBioreactor: 50 mg/LHydrophobic interaction and gel filtration chromatography20 kDa(Fitches et al. 2012)SnailNeurotoxinLt28.1Conus litteratus5 disulfide bondspPIC9GS115Flasks: 2 mg/LNi2+ chelating chromatographyGel filtration5.9 kDa(Lu et al. 2017)BeeEnzymeHyaApis mellifera4 disulfide bonds/2 glycosylation sitespPIC9GS115 (his4)Flasks: 10 mg/LCation exchangeSephadex 5050 kDa(Reitinger et al. 2008)LizardPeptideExendin-4Heloderma suspectumNDpPIC9GS115Bioreactor: 26 mg/LIon exchange, hydrophobic interaction chromatography (HIC), and size-exclusion chromatography4.3 kDa(Zhou et al. 2008)WaspEnzymeVest2aVespa tropica2 disulfide bonds/5 glycosylation sitesGoldenPiCS systemCBS 2612Bioreactor: 96.45 mg/LNi2+ chelating chromatography and reversed-phase59.35 kDa(Janpan et al. 2024)PTMs post-translational modificationsaSVSPs—snake venom serine proteinasesb3FTX—three finger toxinscNDBPs—non-disulfide-bridged peptide Examples of toxins produced in Komagataella phaffii Ni2+ chelating chromatography and reversed-phase C18 HPLC 6.5 kDa Ni2+ chelating chromatography and reversed-phase C18 HPLC 7.6 kDa Molecular exclusion chromatography 7.73 kDa Superose 12 10/300 GL 7.7 kDa Enzymes SVSPsa Ion exchange and reversed-phase chromatography C4 28.8 kDa Phenyl-Sepharose column/heparin–Sepharose affinity chromatography 33 kDa HIS-Select® Nickel Affinity Gel resin 8.8 kDa Ni2+ chelating chromatography RP-HPLC 5.19 kDa Ni2+ chelating chromatography RP-HPLC 5.27 kDa Ni2+ chelating chromatography RP-HPLC 4.1 kDa Ni2+ chelating chromatography RP-HPLC 4.5 kDa Ni2+ chelating chromatography RP-HPLC 3.8 kDa Ni2+ chelating chromatography RP-HPLC 4.1 kDa Metal chelate affinity chromatography–gel filtration chromatography 8.2 kDa Ni2+ chelating DEAE Sepharose FF 8 kDa Ni+-NTA column CM Sepharose FF 8.6 kDa Ultracentrifuge filter 8 kDa Ni2+ chelating CM Sepharose FF 8 kDa Phenyl-Sepharose 10 kDa Cation-exchange chromatography 16.5 kDa Ni2+ chelating 8 kDa Ni2+ chelating 8.3 kDa Ni2+ chelating RP-HPLC ~ 8.1 kDa Ni2+ chelating chromatography 5.1 kDa CM-cellulose-52 49.5 kDa Ultracentrifuge filter Gel filtration G75 14 kDa Flasks: 5 mg/L Bioreactor: 45 mg/L Ni+-nitriloacetic acid affinity chromatography, carboxymethyl (CM) Sepharose ion-exchange chromatography 7.6 kDa Ni2+ chelating chromatography 14 kDa Hydrophobic interaction and gel filtration chromatography 20 kDa Ni2+ chelating chromatography Gel filtration 5.9 kDa Cation exchange Sephadex 50 50 kDa Ion exchange, hydrophobic interaction chromatography (HIC), and size-exclusion chromatography 4.3 kDa Ni2+ chelating chromatography and reversed-phase 59.35 kDa PTMs post-translational modifications aSVSPs—snake venom serine proteinases b3FTX—three finger toxins cNDBPs—non-disulfide-bridged peptide Recombinant rMgTx (Naseem et al. 2021), Vm24, AnTx, Ts6 (Borrego et al. 2022), and AgTx2 (Anangi et al. 2012) are well-studied scorpion peptides. NMR and electrophysiological studies confirm that they retain their native structure and function. Anangi et al. (2012) proved K. phaffii was a superior host (12–15 mg/L) for MgTx compared to E. coli. rMgTx yields (36 mg/L purified) are about ten times higher than those from E. coli. However, optimal expression depends on maintaining strict pH (especially at 6.0) and methanol levels, as deviations can drastically reduce yield (Naseem et al. 2021). K. phaffii also successfully produced soluble forms of the toxins Vm24 at 1.6 mg/L, AnTx at 46 mg/L, and Ts6 at 7.5 mg/L, whereas E. coli attempts resulted in misfolding and insoluble inclusion bodies. Despite this, recombinant Ts6 had a 53-fold lower affinity (Kd) than the native, possibly due to missing post-translational modifications or cellular differences. Additionally, Vm24 yield stays low compared to other toxins (Borrego et al. 2022). AgTx2 yields in K. phaffii range from 14 to 18 mg/L, which is six to eight times higher than in E. coli, and the recombinant peptide remains functional. However, using enzymes like enterokinases for tag removal adds steps and can cut yield by 25–40%. AgTx2 may also be broken down by yeast proteases under stress (Anangi et al. 2012). BmaTX14 (~ 8.2 kDa), an α-NaTx, is a potential tool for studying Na+ channels and drug development (K. Wang et al. 2006). The soluble rBmaTX14 yield in K. phaffii is 100–120 mg/L, roughly ten times greater than in E. coli, and it does not need in vitro refolding or disulfide bond correction. The recombinant toxin is highly bioactive, reducing Na+ current by 80% in mouse trigeminal neurons. (Wang et al. 2006). A key application of scorpion toxins is the development of eco-friendly biopesticides targeting specific insect pests. Toxins like AaIT, LqhIT2, ButaIT, and BjαIT target insect sodium channels (Li & Xia 2018b). In K. phaffii, expression yields vary significantly depending on the system, reaching up to 240 mg/L with optimized strain/vector combinations and increased gene copy number. In contrast, E. coli produces low yields of non-native, non-toxic proteins, mainly due to its inability to correctly form the multiple disulfide bonds required for proper folding and activity of these toxins. LqhIT2 reported a yield of 1.4 g/L in high-cell-density fermentation, one of the highest for any scorpion insect neurotoxin and 20 times higher than shake-flask levels. This optimized K. phaffii yield is thousands of times higher than the early E. coli attempts. It successfully retains oral insecticidal activity, a trait missing in most other neurotoxins. However, the protein is highly unstable and undergoes substantial degradation by yeast proteases after 72 h of induction (H. Li et al. 2025a, b; Li & Xia 2008). Fusion with GNA has been widely used to overcome the lack of oral activity of scorpion toxins such as ButaIT and BjαIT, enabling their transport across the insect gut into the hemolymph. In K. phaffii, these fusion proteins also improve yields, with ButaIT/GNA reaching 25–35 mg/L (vs. < 1 mg/L for ButaIT alone) and BjαIT achieving up to 340 mg/L in fed-batch systems, while maintaining structural similarity to the native toxin. Additionally, fusion broadens the spectrum of activity and enables significant oral toxicity. However, this strategy involves trade-offs. Fusion proteins may show reduced specific toxicity compared to the native toxins and can suffer from instability and proteolytic degradation during yeast expression, as observed for BjαIT/GNA. Moreover, fusion orientation affects stability, with certain configurations being more prone to cleavage. The product’s stability depends on fusion orientation: Linking the toxin to GNA’s C-terminus (GNA/ButaIT) proved more stable than N-terminal linkage (ButaIT/GNA), which was prone to C-terminal cleavage (Li et al. 2024; Pham Trung et al. 2006). A recombinant antiepilepsy peptide (AEP) derived from Olivierus martensii (syn. Buthus martensii, Mesobuthus martensii) was used in pull-down experiments with rat hippocampal proteins. It was found to interact with SNAP-25 and the NMDA receptor, providing the first experimental evidence of its molecular mechanism in seizure control (Z. Wang et al. 2009). The small size of the single peptide (8.3 kDa) complicates direct expression without specialized construct design, like tandem repeats. Incorporating this tandem-repeat gene construct into K. phaffii improved stability and enabled the production of sufficient protein (0.9 mg/mL) for target-interaction research. Proteolytic degradation is a recurring limitation in the heterologous expression of scorpion venom components in K. phaffii. For example, the Ts8 toxin, in both native and recombinant forms, was shown to inhibit yeast growth, suggesting a potential antifungal activity that may also contribute to its degradation during expression (Cordeiro et al. 2022). Although this system enables the production of sufficient material for functional characterization, often not feasible with native venom, recombinant toxins frequently undergo proteolysis, generating multiple fragments rather than the full-length molecule. Additives such as casein hydrolysate only partially mitigate this effect (Amorim, Cordeiro, et al. 2018a, b). A 4-kDa scorpion defensin (SD) was expressed at high levels and showed broad-spectrum antibacterial activity against both Gram-positive and Gram-negative bacteria. Similarly, synthetic antimicrobial peptide constructs (CPs) achieved high yields (up to 271 mg/L) and retained strong bioactivity. However, these systems are also affected by processing limitations, including incomplete cleavage by KEX2 and increased susceptibility to proteolytic degradation during extended induction periods (Niu et al. 2013). While most studies focus on peptides, scorpion venoms also contain enzymatic components that play key roles in venom spreading and tissue damage. These include hyaluronidases and phospholipases, which are more structurally complex and often require post-translational modifications. The expression of rTsHyal-1 in K. phaffii demonstrated that yeast systems can produce active, glycosylated enzymes that are difficult to obtain in bacterial hosts (Amorim, Boldrini-França, et al. 2018) . Likewise, recombinant phospholipase A2 (rPLA2(−5)) was produced in a properly folded and active form, with yields significantly higher than those obtained in E. coli. However, enzymatic proteins remain particularly sensitive to proteolytic degradation, requiring careful control of fermentation time to preserve product integrity (Krayem & Gargouri 2020). Overall, comparative studies highlight the advantages of K. phaffii over bacterial systems, particularly for disulfide-rich peptides and structurally complex proteins. While E. coli often results in low yields and inactive inclusion bodies, K. phaffii enables higher production levels and proper folding, frequently exceeding 9 mg/L for neurotoxins and achieving substantial improvements in yield for several recombinant toxins and enzymes. The highest yields are often achieved through optimization techniques such as fed-batch fermentation or gene dosage control, primarily using K. phaffii for its secretion and disulfide bond formation. Other venomous taxa The venom of spiders, marine mollusks, bees, and lizards is often described as a valuable source of bioactive molecules for recombinant production in K. phaffii (Fig. 1 and Table 1). However, available studies suggest that, although the system is clearly capable, successful expression is often contingent on various factors and is not a trivial process. In the case of recombinant spider toxins, only a limited number of well-characterized examples are available, which already points to underlying technical challenges. For example, the expression of the Phoneutria nigriventer neurotoxin rTx4(6-1) in strain X33 demonstrates how much yield depends on process scale, increasing from ~ 5 mg/L in flasks to ~ 45 mg/L in bioreactors (Li & Xia 2019), indicating that upstream optimization plays a central role in achieving viable production levels. Similarly, the recombinant d-amaurobitoxin-PI1a peptide from Pireneitega luctuosus was produced in the protease-deficient SMD1168H strain, yielding ~ 26 mg/L (Yang et al. 2014), reinforcing the idea that host engineering can be an important strategy for improving stability and recovery. Notably, both rTx4(6-1) and d-amaurobitoxin-PI1a retained their insecticidal activity after expression and purification, highlighting a key strength of K. phaffii: its ability to produce biologically active toxins with preserved functional properties.Fig. 1Representative overview of venom toxin classes from different animal sources and their recombinant expression in K. phaffii. The diagram illustrates major toxin families, including neurotoxins, non-disulfide-bridged peptides (NDBPs), disintegrins, snake venom serine proteases (SVSPs), three-finger toxins (3FTx), phospholipases A₂ (PLA₂), and hyaluronidases, derived from reptiles, arachnids, insects, and scorpions. These toxins are heterologously expressed using common K. phaffii expression vectors (pPICZαA, pPIC9K, pPIC9, and pGAPZαB) and host strains (X-33, GS115, KM71H, and SMD1168H). Recombinant venom-derived molecules have potential applications in biomedicine and biotechnology, including insulin-releasing activity, pain and antiepileptic treatment, and insecticide development Representative overview of venom toxin classes from different animal sources and their recombinant expression in K. phaffii. The diagram illustrates major toxin families, including neurotoxins, non-disulfide-bridged peptides (NDBPs), disintegrins, snake venom serine proteases (SVSPs), three-finger toxins (3FTx), phospholipases A₂ (PLA₂), and hyaluronidases, derived from reptiles, arachnids, insects, and scorpions. These toxins are heterologously expressed using common K. phaffii expression vectors (pPICZαA, pPIC9K, pPIC9, and pGAPZαB) and host strains (X-33, GS115, KM71H, and SMD1168H). Recombinant venom-derived molecules have potential applications in biomedicine and biotechnology, including insulin-releasing activity, pain and antiepileptic treatment, and insecticide development Based on these examples, the expression of Hadronyche versuta ω-hexatoxin-Hv1a (Fitches et al. 2012) provides a broad overview of the system, illustrating both its potential and its limitations. On the one hand, the relatively high yield (~ 50 mg/L) and recovery of an active molecule reinforce that K. phaffii can successfully support the production of structurally complex neurotoxins rich in disulfide bonds. On the other hand, achieving this result required several non-trivial adaptations, including expression as a fusion protein to enhance secretion and stability, as well as the use of a protease-deficient strain to reduce degradation. These requirements indicate that, although the platform is versatile and capable of producing functional proteins, its performance depends heavily on tailored molecular and process optimization. Concerning the production of recombinant toxins from marine mollusks, even fewer examples are available, suggesting that these molecules present additional challenges for expression in K. phaffii. The conotoxin Lt28.1 from Conus litteratus was expressed in the GS115 strain with a relatively low yield (~ 2 mg/L) (Lu et al. 2017), highlighting the intrinsic difficulty of producing conopeptides, which are typically small and structurally constrained. Despite this limitation, the recombinant toxin remained active, reinforcing that K. phaffii can support the generation of biologically functional peptides even under suboptimal yield conditions. Based on this, (Zheng et al. 2022) proposed a GFP-based screening strategy to optimize conopeptide expression, demonstrating that folding efficiency and activity can be significantly improved through systematic and scalable approaches. Taken together, these findings suggest that, although basal expression may be limited, the platform offers sufficient flexibility to be enhanced through targeted optimization strategies. In addition to these observations, studies on toxins from bees, wasps, and lizards further illustrate both the strengths and limitations of K. phaffii. Honeybee hyaluronidase Hya was expressed in the GS115 strain with a yield of ~ 10 mg/L and biochemical properties comparable to those of the native enzyme (Reitinger et al. 2008), while exendin-4, derived from the saliva of the Gila monster (Heloderma suspectum), reached ~ 26 mg/L in a bioreactor and retained its hypoglycemic activity (Zhou et al. 2008), highlighting the system’s ability to produce functionally active molecules across different classes of toxins. Most notably, Vespa tropica hyaluronidase VesT2a achieved a high yield (96.45 mg/L) in a bioreactor, surpassing E. coli-based expression, which required refolding from inclusion bodies (Janpan et al. 2024) and reinforcing the advantages of a eukaryotic host for complex proteins. However, this molecule exhibited an increase in molecular weight due to yeast-specific glycosylation, highlighting a significant limitation, as such post-translational modification may affect pharmacological properties, immunogenicity, and comparability with native toxins. Overall, these studies indicate that K. phaffii represents a versatile and powerful platform, but one whose performance is highly context-dependent, requiring a balance between its capacity for functional expression and the need for careful control of post-translational processing. Major challenges in animal toxin production Glycosylation It is well known that glycosylation is a common eukaryotic post-translational modification that alters diverse physicochemical characteristics (e.g., molecular weight, polarity, volume, and thermal stability) as well as the folding, proteolytic stability, or enzymatic activity of a vast number of proteins. Even though glycan moieties are diverse (Andrade-Silva et al. 2018; Gerwig et al. 2013; Hykollari et al. 2019; Schjoldager et al. 2020), they are composed mainly of N-acetylglucosamine (GlcNAc), galactose, and sialic acid residues in higher eukaryotes or mannose residues in yeast glycoproteins (Gerwig et al. 2013; Khlebodarova et al. 2024; Varki et al. 2022) (Fig. 2). These few monosaccharide residues give rise to different glycosylation types, and this review will focus on N- and O-glycosylation due to their prevalence in animal venoms.Fig. 2Schematic overview of usual protein-linked carbohydrates. A Common cores of eukaryotic N-linked carbohydrates. B Typical N-linked high-mannose in K. phaffii (Man8-14GlcNAc2) and Saccharomyces cerevisiae (Man>50GlcNAc2) glycoproteins. C Mucin-type O-glycosylation cores. The differences among cores 1 and 8, cores 2 and 9, and cores 3, 5, 6, and 7 rely on the glycosidic bond type and its position Schematic overview of usual protein-linked carbohydrates. A Common cores of eukaryotic N-linked carbohydrates. B Typical N-linked high-mannose in K. phaffii (Man8-14GlcNAc2) and Saccharomyces cerevisiae (Man>50GlcNAc2) glycoproteins. C Mucin-type O-glycosylation cores. The differences among cores 1 and 8, cores 2 and 9, and cores 3, 5, 6, and 7 rely on the glycosidic bond type and its position The glycan core might be attached to the protein structures via the side chain of Asn (N-glycosylation), Ser (O-glycosylation), Thr (O-glycosylation), or Lys (glycation), or via the protein amino-terminal group (glycation) (de Melo-Braga et al. 2022; Schjoldager et al. 2020). Different from glycosylation reactions, glycation is an enzyme-independent reaction and a less prevalent modification among venom components, being described in the wasp allergen Ag5 (de Melo-Braga et al. 2022; Uceda et al. 2024). The N-glycosylated venom components are usually glycosylated via their Asn residue in the triad N-X-S/T (X ≠ Pro) (Taherzadeh et al. 2019). They are extremely common in animal venom proteins and examples include hyaluronidase, metalloprotease, venom allergen 5, phospholipase A2, allergen phospholipase A1, scorpion neurotoxins (e.g., Ts15, Aah6), serine protease, L-amino acid oxidase, nerve growth factor, and dipeptidyl peptidase (Andrade-Silva et al. 2021; de Melo-Braga et al. 2022; de Souza et al. 2019; Earl et al. 2006; Verano-Braga et al. 2013). Concerning the O-glycosylation pattern, the glycan moiety is commonly attached to the protein via the side chain hydroxyl group of any Ser or Thr residue via an ester bond without a consensus sequence (Taherzadeh et al. 2019). In 1976, vespulakinin-1 and vespulakinin-2 from wasp venom were the first O-glycosylated venom components evidenced (Yoshida et al. 1976). Many years later, neurotoxins from Conus snail venoms were characterized (Gerwig et al. 2013), and other O-glycosylated venom components were rarely identified, such as those from ants (Touchard et al. 2018) and snakes (Degueldre et al. 2017). Furthermore, they suggested the presence of O-linked carbohydrates in all snake venom natriuretic peptides due to the structural similarity to the conopeptides (Degueldre et al. 2017). Although there is a core sequence among the protein-linked carbohydrates, their structures are a vast world to be explored, and in recent years, the toxinologists have drawn attention to the “venom glycome” field. For instance, Hykollari et al. (2019) determined that Apis mellifera venom is dominated by complex and hybrid N-glycans with high isomeric variance and antennal residues of fucose, galactose, and N-acetylgalactosamine (GalNAc). After venom treatment with PNGases, 85 isomers were recovered among the neutral N-glycans, and more than 40 anionic and zwitterionic structures were also identified. Concerning snake venoms, more than 60 structures were identified in eight Bothrops spp. venoms after PNGase F treatment, with most of the hybrid/complex type and only two of the high-mannose type. Each venom presents a distinct molecular signature, and the presence of different isomers, abundance, and branching increases venom complexity and might impact its biological activity (Andrade-Silva et al. 2018). The glycans from conopeptides have also been investigated (Gerwig et al. 2013). They are classified as mucin-type glycoconjugate O-glycans, in which a GalNAc is linked to the Ser or Thr residues, and are usually composed of GalNAc, GlcNAc, galactose (Gal), fucose (Fuc), and sialic acid. Contulakin-G, a kappaA-conotoxin SIVA from Conus striatus venom, is 10 × more active than its non-glycosylated form, and the glycan moiety might be implicated in modulating peptide affinity or its access speed to the ion channel (Craig et al. 1998; Gerwig et al. 2013). Additionally, diverse roles for the glycosylation of venom components have been reported, but a detailed review on this subject is out of this report. For the moment, we highlight interesting findings as (i) role in the variability of metalloproteinases’ enzymatic activity and their neutralization by antivenoms (Debono et al. 2020), (ii) possible immunomodulatory role in venom allergens such as PLA1 and PLA2 (de Souza et al. 2019), (iii) protein folding/solubility (de Melo-Braga et al. 2022; Robinson et al. 2021), (iv) enhancement of wasp hyaluronidase thermal stability (Rungsa et al. 2019), and (v) participation in enzymatic/biological activity (Brás-Costa et al. 2023). As mentioned earlier, different venom components have been successfully produced using the K. phaffii expression system. Although glycosylation is an advantage of this system for protein production, hyperglycosylation can also represent a challenge, as it is often undesirable. It occurs when usually 50–150 mannose (Man) residues are coupled to the N-linked core Man8GlcNAc2 (Cereghino & Cregg 2000). Although K. phaffii typically produces less intense glycosylation than other yeasts (Khlebodarova et al. 2024), a high sugar content has been an issue for the phospholipases A2 inhibitor (PLI). rCdtPLI2 is the recombinant CdtPLI2 beta-PLI from Crotalus durissus terrificus snake venom (Wiezel et al. 2024). Better inhibitory PLA2 activity was observed against CB-Cdc and CTx-Cdc only after its deglycosylation by alpha-mannosidase, possibly due to a hyperglycosylation produced by K. phaffii (Wiezel et al. 2024). The hyperglycosylation has also been reported when producing recombinant arthropod venom hyaluronidases, but there is no information on its effect on their activities, and the glycan absence gave rise to an insoluble enzyme (Amorim, Boldrini-França et al., 2018; Janpan et al. 2024; Rungsa et al. 2019). One option to overcome this would be to use modified strains such as Pichia GlycoSwitch®. Although it is an interesting alternative, it is not commercially viable or available in all laboratories (Ingram et al. 2021). The presence of glycans in natural non-glycosylated sites has been reported for the spider short-chain insecticidal neurotoxin Tx4(6-1) (Li & Xia 2019). This recombinant toxin is 46% (2415 Da) larger than its native form due to the presence of O-glycans, and it is neurotoxic against the silkworm larvae (Bombyx mori Linnaeus) (Figueiredo et al. 1995; Li & Xia 2019). However, there is no comparison between the native and recombinant forms to ensure that O-glycosylation does not interfere with their biological activity (Li & Xia 2019). It is noteworthy that the snake venom serine protease Collinein-1 has been successfully expressed in K. phaffii (Boldrini-Franca et al. 2019; Boldrini-França et al. 2015). Nevertheless, troubles with solubility and proteolytic degradation were only fixed when this same enzyme was expressed in a complex culture medium (BMMY) (Boldrini-Franca et al. 2019). Then, a new challenge has to be overcome: The simple addition of peptone and yeast extract in the culture medium might cause excessive stress to the K. phaffii cells, inducing the production of two rCollinein-1 populations due to carbohydrate heterogeneity and decreasing protein expression recovery (Boldrini-Franca et al. 2019). On the other hand, the recombinant form of batroxobin (rBatroxobin) is 4 kDa smaller than its native counterpart (You et al. 2004). While rBatroxobin presents only the yeast-type carbohydrate, its native form (nBatroxobin) may contain 40 different glycan structures (You et al. 2004). They present different chromogenic substrate specificities, but both are coagulant proteins with slightly higher fibrinogen conversion to fibrin activity and plasma clotting velocity for rBatroxobin. Missing statistical analysis impairs a more detailed comparison (You et al. 2004). Similarly, Santos-Filho et al. (2015) discussed that a different glycosylation pattern on rBaltMIP might be the reason for its reduced activity in comparison to the native PLI αBaltMIP. Proteolytic degradation One of the major drawbacks of heterologous protein expression in K. phaffii is the proteolytic degradation of secreted proteins, which can significantly impair yield and product stability (Sinha et al. 2005). Fungal vacuoles contain a diverse array of proteases, and their abundance strongly depends on nutritional status and on the release of toxic metabolites that trigger cell lysis and subsequent enzyme leakage. In Saccharomyces cerevisiae, several vacuolar proteases have been well characterized, including proteinase A (PrA), proteinase B (PrB), aminopeptidases, and carboxypeptidases (Van Den Hazel et al. 1996). Although the proteolytic repertoire of K. phaffii is less characterized, strains such as KM71 and GS115 are known to produce a subtilisin-like serine protease, alongside aminopeptidases and PrA/PrB counterparts, encompassing acidic, neutral, and alkaline proteases (Salamin et al. 2010). Proteolysis may occur during trafficking through the secretory pathway by resident proteases or extracellularly due to secretion or release of vacuolar proteases following membrane damage (Ahmad et al. 2014; Marsalek et al. 2019). Since many of these proteases are stored in vacuoles, several stressors can trigger their release, including nutrient limitation, carbon source shifts, suboptimal pH, and production of intracellular toxic compounds (Juárez-Montiel et al. 2022; Sinha et al. 2005). Methanol feeding is a well-known inducer of cellular stress and protease release. During the glycerol-based growth phase, extracellular proteases are rarely detected; however, during methanol induction, large amounts of reactive oxygen species (ROS), hydrogen peroxide, and formaldehyde are generated as byproducts of methanol metabolism. These metabolites activate autophagic pathways and enhance cell lysis in K. phaffii cultures (K. Li et al. 2025a, b; S. Wang et al. 2025; Y. Wang et al. 2024). Proteolytic systems in fungi can be broadly grouped into three major categories: (i) the ubiquitin–proteasome system, which rapidly degrades misfolded or damaged proteins under stress conditions and relies on ubiquitin conjugation for substrate recognition; (ii) vacuolar proteases, the most abundant group, released under nutrient depletion, such as PrA, PrB, carboxypeptidases S and Y; and (iii) proteases of the secretory pathway, including the Kex2 endoprotease and dipeptidyl aminopeptidases, which participate in protein processing during secretion (Cao & Xue 2021; Hecht et al. 2014; Staszczak 2021). Despite this, the extent of proteolysis in K. phaffii is highly dependent on the structural and sequence properties of the heterologous protein. Some recombinant proteins remain intact, whereas others are highly susceptible to protease attacks. Sensitivity depends primarily on the amino acid sequence and the three-dimensional conformation, which may expose protease-sensitive motifs to cleavage. Oxidatively damaged proteins can further enhance the proteolytic response (Cordeiro et al. 2022; Dalvie et al. 2022). In the expression of the toxin from Tityus serrulatus rTs8, for example, proteolysis was observed near hydrophobic residues such as Ile and Leu. This pattern suggests the involvement of a protease with cleavage specificity like thermolysin, a bacterial metalloprotease known to preferentially cleave at bulky hydrophobic or aromatic residues (Cordeiro et al. 2022). Several bioprocess parameters can be optimized to reduce proteolytic degradation in K. phaffii, including modulation of temperature, pH adjustment, addition of protease inhibitors, medium optimization, protein engineering approaches, and the use of protease-deficient strains (Carneiro et al. 2022; Velez-Suberbie et al. 2020; Z. Yang & Zhang 2018). Another strategy is especially important when expressing proteins naturally prone to structural instability or to cleavage at exposed hydrophobic motifs and will be described below. Engineering K. phaffii for efficient production of venom-derived molecules The production of toxins in heterologous systems, particularly in K. phaffii, is still relatively recent; consequently, few studies report novel molecular engineering strategies for their expression. Therefore, in the “Engineering K. phaffii for efficient production of venom-derived molecules” section, we discuss more recent approaches and how they could be applied to improve the production of these molecules. CRISPR-based genome engineering for targeted host optimization Synthetic biology has enabled the development of programmable gene regulation systems that expand the capabilities of microbial hosts beyond conventional expression platforms (Guo et al. 2026). Among these, CRISPR-based transcriptional activation (CRISPRa) has emerged as a versatile tool for modulating gene expression without altering genomic sequences. This approach typically employs a catalytically inactive Cas9 (dCas9) fused to transcriptional activation domains, which is guided by single guide RNAs (sgRNAs) to specific promoter regions, thereby enhancing transcription of target genes in a sequence-specific and programmable manner. Recently, the integration of CRISPRa with orthogonal transcriptional machinery, such as the T7 RNA polymerase (T7 RNAP), has enabled the construction of synthetic genetic circuits with increased modularity and controllable systems (Ruan et al. 2025). In this system, sgRNAs are transcribed under T7 promoters, often combined with tRNA-based elements to improve their maturation and stability. This design enables efficient multiplexing, allowing the simultaneous activation of multiple genes. Additionally, inducible systems, such as tetracycline-responsive elements, and engineered versions of T7 RNA polymerase provide further control over gene expression. As a result, these systems can be used to build synthetic genetic circuits that respond to specific environmental or chemical signals (Ruan et al. 2025). By coupling the T7 RNAP–CRISPRa system to stress-responsive pathways, such as the unfolded protein response (UPR), it becomes possible to dynamically regulate key factors like HAC1, chaperones, and components of the secretory machinery in response to production burden. This enables the creation of self-regulated expression systems that adjust cellular resources to improve enzyme stability, activity, and yield (Ruan et al. 2025). CRISPR/Cas9 was also used to generate the K. phaffii recipient strains VKPM Y-5013 and Y-5014 from a high-expression parental strain, enabling the construction of markerless producers. The engineered strains maintained growth performance and showed higher heterologous protein production compared to the commercial GS115 strain. Additionally, an efficient strategy for sequential genome integration allowed the development of multicopy strains with enhanced productivity (Tkachenko et al. 2023). This approach holds significant potential for optimizing heterologous protein production, particularly for proteins that require fine-tuned expression levels and coordinated regulation of cellular pathways involved in folding, processing, and secretion. The adoption of CRISPR-based multigene editing approaches can be particularly valuable for reducing proteolytic degradation and alleviating secretory pathway burden through the targeted disruption of endogenous proteases and competing secretome components. In parallel, systematic screening of signal peptides, combined with the co-expression of folding assistants such as protein disulfide isomerase (PDI), may enhance secretion efficiency and promote correct disulfide bond formation, which is critical for the stability and activity of many toxin proteins. Strain engineering and system-level optimization Although K. phaffii provides a robust basis to produce recombinant proteins, its performance for structurally complex venom toxins rich in disulfide bonds is not intrinsic but strongly dependent on global strain properties that influence protein folding, stability, and overall cellular homeostasis. In this context, recent advances have shifted the focus from gene-level optimization to host-system-level refinement, encompassing proteostasis, redox balance, and intracellular resource allocation (Barone et al. 2023; Lv, Cai 2025). A central strategy in this regard is the development of protease-deficient strains, which aim to reduce the intracellular and extracellular degradation of recombinant proteins (Barone et al. 2023; Karbalaei et al. 2020). Targeted disruption of proteolytic pathways, particularly vacuolar and cell wall–associated proteases, such as members of the yapsin (YPS) family, has been shown to improve protein stability and increase recoverable yield (Kerry-Williams et al. 1998). Deletion of key vacuolar proteases (PEP4 and PRB1), as implemented in SMD strains (e.g., SMD1163: pep4 prb1; SMD1168: pep4), reduces degradation and improves recovery of full-length, biologically active proteins (Barone et al. 2023; Karbalaei et al. 2020). This approach is especially relevant for venom-derived toxins, which often contain flexible or exposed regions highly susceptible to proteolytic cleavage, frequently resulting in loss of biological activity rather than just a reduction in yield. Similarly, yapsin proteases such as Yps1p, responsible for cleavage at paired basic residues (KR, RR), have been associated with the fragmentation of recombinant products, including wasp venom hyaluronidase (VesT2a), which justifies the use of yps1Δ strains (where the Yps1p gene was deleted) or protease inhibition strategies (Janpan et al. 2024). In addition to proteolysis, intracellular folding capability is an important determinant of functional expression, particularly for cysteine-rich toxins that depend on the precise formation of disulfide bonds. Enhancing the host proteostasis network by increasing the availability of folding catalysts and improving the processing capacity of the endoplasmic reticulum (ER) enhances secretion competence and product quality (Barone et al. 2023; Lv, Cai 2025). Overexpression of folding catalysts such as PDI, Kar2, and Ero1 has been shown to promote oxidative folding and stabilize disulfide-rich proteins within the ER (Barone et al. 2023). However, the benefits of such modifications are not universal and often depend on the structural complexity of the target protein, reinforcing the need for protein-specific optimization. Closely linked to folding efficiency, redox homeostasis in the ER plays a crucial role in disulfide bond formation and protein maturation (Kastberg et al. 2025; Lv, Cai 2025). Although the oxidative environment of K. phaffii is generally favorable, its capacity can become limiting under high expression conditions. Disulfide bond formation in the ER requires an oxidizing environment and enzymatic systems such as Ero1-PDI, which catalyze thiol-disulfide exchange reactions during co-translational protein translocation (Barone et al. 2023). Despite the clear mechanistic relevance, systematic strategies to modulate ER redox balance remain relatively unexplored, particularly in the context of venom toxin production. This represents a significant gap, as subtle imbalances in redox conditions can lead to misfolding, aggregation, or the formation of non-native disulfide bonds (Krayem et al. 2022; Kastberg et al. 2025). Reductive conditions impair disulfide bond formation and favor aggregation, whereas excessive oxidation can increase reactive oxygen species (ROS) and trigger proteolytic degradation (Kastberg et al. 2025). At a broader level, overall cellular properties, including metabolic state, stress response, and intracellular trafficking capacity, also influence recombinant protein production (Barone et al. 2023; Lv, Cai 2025). Alterations in these parameters can indirectly affect secretion efficiency, folding kinetics, and protein stability. Accumulation of overflow metabolites such as ethanol or acetate can disrupt redox balance and impair biosynthetic pathways required for folding and secretion (Steimann et al. 2024). Cellular stress responses triggered by high expression can exacerbate proteolysis or impair proper folding, highlighting the importance of maintaining cellular homeostasis along with productivity. ER overload activates the unfolded protein response (UPR) and ER-associated degradation (ERAD), which restore homeostasis but may reduce yield when persistently engaged (Barone et al. 2023; Lv, Cai 2025). Despite these advances, it is important to emphasize that most strain engineering strategies are still empirical and highly context-dependent, with limited predictive frameworks available. This limitation is particularly critical for venom toxins, where small perturbations in folding or processing can result in disproportionate losses in biological function. Furthermore, trade-offs between expression level, cellular stress, and product quality are frequently observed but rarely systematically quantified. High-cell-density fermentations, for instance, can increase productivity while simultaneously promoting cell lysis and release of vacuolar proteases, intensifying product degradation; mitigation strategies include pH control (e.g., pH 3.0–6.0) and nitrogen supplementation (e.g., yeast extract, casamino acids) (Janpan et al. 2024; Karbalaei et al. 2020). Taken together, these observations indicate that the successful production of venom-derived molecules in K. phaffii depends on fine-tuning host physiology at a systemic level, rather than isolated genetic modifications. The absence of standardized design principles remains a major obstacle, highlighting the need for integrative approaches capable of linking toxin-specific structural constraints to rational strain engineering strategies. A combined framework integrating protease-deficient backgrounds, redox and folding optimization (e.g., PDI/Ero1 overexpression), and controlled bioprocess conditions emerges as a coherent strategy to balance productivity, stability, and biological activity of complex venom-derived proteins. Vector and promoters engineering Vector engineering in K. phaffii is critical for optimizing the expression of bioactive molecules, including toxins and venom-derived peptides. Traditionally, conventional expression systems are primarily based on the AOX1 promoter, which is present in vectors widely used for the expression of recombinant toxins, such as pPICZαA, pPIC9, and pPIC9K. This promoter is strongly induced by methanol and enables high levels of protein expression, making it one of the key factors that established this system as a robust heterologous expression platform (Ahmad et al. 2014; Pan et al. 2022). However, the dependence on methanol represents a significant limitation. Methanol is flammable and toxic, requiring strict control conditions in bioreactors, in addition to increasing operational costs and hindering industrial scaling (Eissazadeh et al. 2017). Alternatively, methanol-free systems have been developed using constitutive promoters, such as the GAP promoter, found in vectors like pGAPZαB. This system allows for the continuous expression of the recombinant protein in carbon sources, such as glucose or glycerol, eliminating the need for methanol induction. Studies show that, although expression under the GAP promoter may, in some cases, be lower than that of the AOX1 system, volumetric productivity and process reproducibility may be higher due to greater culture stability and operational simplification (Bernat-Camps et al. 2023; Vijayakumar & Venkataraman 2024). Recently, modular engineering approaches have been introduced, such as the GoldenPiCS system. This system uses Golden Gate cloning to enable the rapid and standardized assembly of multiple genetic elements, including promoters, secretion signals, and genes of interest. One of the main advantages of GoldenPiCS is its flexibility, allowing the construction of both methanol-dependent and methanol-independent vectors, depending on the choice of promoter. In addition, this platform enables the generation of combinatorial libraries for expression optimization, which is particularly relevant for complex proteins, such as venom toxins, that often require fine-tuning of expression and processing. The GoldenPiCS system was successfully used to express Vest2a toxin (Janpan et al. 2024). Taken together, these advances demonstrate a clear shift from traditional AOX1 promoter-based systems toward safer and more versatile strategies, such as methanol-free systems and modular platforms. The choice between these systems should take into account not only the desired expression level but also factors such as safety, cost, scalability, and the complexity of the target protein. AI (artificial intelligence) in toxin design The development of new technological tools contributes to the improvement of methodologies applied to heterologous expression. Computational models based on machine learning have been developed with the aim of optimizing codon usage and, consequently, increasing the efficiency of protein expression in different organisms (Fallahpour et al. 2025; Fu et al. 2020). However, most currently available models are focused on optimization for E. coli, with a lack of studies dedicated to optimization in K. phaffii, highlighting the need for further research in this area. In addition, the use of deep learning tools, such as AlphaFold3 (Abramson et al. 2024), enables the prediction of three-dimensional structures with high accuracy, allowing in silico evaluation of the thermodynamic stability of toxin variants before experimental validation. This approach is important to ensure proper protein folding and to avoid degradation pathways, which often limit the yield of complex proteins rich in disulfide bonds in the expression system of K. phaffii (Puxbaum et al. 2015). Similarly, AI-based tools for cleavage site prediction, such as SignalP 6.0 (Teufel et al. 2022), can be used to design secretion leader sequences and to estimate the efficiency of signal peptidase cleavage. This computational approach helps ensure that the recombinant toxin is correctly translocated and processed, resulting in proper termini and reduced intracellular retention, thereby facilitating production scale-up and purification (Obst et al. 2017). Future perspectives The K. phaffii system is widely regarded as one of the most suitable platforms for producing complex, disulfide-bond-rich peptides, as it effectively overcomes several limitations of E. coli. Compared to bacterial systems, K. phaffii frequently produces higher expression levels, and the secretion of heterologous proteins into the culture medium, combined with the low background of endogenous secreted proteins, facilitates downstream purification. Despite challenges related to proteolytic degradation, post-translational modifications such as undesired glycosylation, and effects on host cell physiology that can compromise yield, purity, and biological activity, K. phaffii remains one of the most suitable systems for expressing properly folded, cysteine-rich peptides. To overcome these limitations, several complementary strategies can be employed. The use of protease-deficient or glycoengineered K. phaffii strains can significantly reduce proteolytic degradation and undesired glycosylation. Optimizing cultivation parameters, such as reducing methanol concentration, improving nutrient balance, and controlling induction time, helps minimize cellular stress and enhance protein stability (Cordeiro et al. 2022). In addition, fusion to protective carrier proteins followed by site-specific cleavage, as well as intracellular expression combined with oxidative folding enhancement, co-expression of molecular chaperones, or in vitro refolding approaches, can further improve folding efficiency and biological activity. Despite inherent challenges, K. phaffii remains one of the most robust and versatile platforms for cysteine-rich peptide production (Barone et al. 2023). Despite recent advances in synthetic biology and recombinant protein production in K. phaffii, several important gaps remain in the current literature, particularly in the context of complex or disulfide-rich proteins such as venom-derived toxins. First, there is currently no clear evidence supporting the application of artificial intelligence or machine learning approaches to optimize secretion pathways in K. phaffii, indicating that this remains an underexplored area with significant potential. Second, while promoter optimization has been investigated within the species, systematic cross-species comparative studies involving other yeast platforms are still lacking. Finally, although K. phaffii is widely used for heterologous protein expression, there is limited documentation regarding the successful production of venom-derived toxins, particularly those outside commonly studied enzyme classes. Together, these limitations highlight key opportunities for future research aimed at expanding the applicability of this host for the production of structurally complex and biotechnologically relevant proteins.