Fungal expression strategies for heterologous production of antimicrobial peptides

Abstract

Antimicrobial peptides (AMPs) are promising candidates for next-generation therapeutics due to their broad-spectrum activity and reduced propensity for resistance, making them valuable in medicine, agriculture, and biotechnology. However, traditional AMP production methods including isolation from natural sources and chemical synthesis are costly, inefficient, and environmentally unsustainable, particularly for longer or post-translationally modified peptides. While heterologous expression has emerged as a scalable and versatile alternative, its success depends strongly on host selection and tailored optimisation strategies. This review examines recent advances in fungal systems as platforms for AMP production. Fungal systems, and particularly yeasts such as Pichia pastoris, offer rapid growth, low-cost fermentation, secretion capacity, and the ability to perform key post-translational modifications (PTMs), making them leading hosts for recombinant AMPs. We outline strain choice and engineering strategies that enhance AMP yield and bioactivity, including promoter and codon optimisation, secretion signal choice, fusion partners, and the construction of tandem or chimeric AMPs. By integrating current methodologies and case studies, this review aims to guide future efforts toward efficient, scalable, and commercially viable AMP manufacturing in fungal hosts, positioning fungal biotechnology as a key enabler in the development of next-generation antimicrobial solutions.

Introduction

Antimicrobial resistance (AMR) is a crisis of immediate priority for humanity. The severity of this crisis, largely a result of heavy antibiotic use in livestock agriculture, is exemplified by the reservoirs of resistance that now exist even for drugs of last resort such as colistin [1]. Globally, AMR is now among the leading causes of death and in 2021 was associated with approximately 1.14 million deaths worldwide [2]. New antimicrobial drugs to which major pathogens are yet to develop resistance are needed to address this crisis. Antimicrobial peptides (AMPs) are a diverse group of bioactive molecules that have the potential to serve this need in new drug discovery [3].

AMPs are found across the three domains of life and exhibit activity against a broad spectrum of pathogens [4, 5]. Over the last few decades, modern techniques including metagenomic sequencing and genome mining have diversified the number of known AMPs and their respective biological sources [6]. Site-directed mutagenesis has enabled insight into structure–function relationships, and enabled rationally-designed libraries of synthetic AMP derivatives with enhanced antimicrobial activity, protease resistance, stability, toxicity profiles, and target specificity [7, 8]. As of January 1, 2026, more than 3379 natural, 2290 synthetic, and 373 predicted AMPs have been documented in the Antimicrobial Peptide Database (APD), which categorizes AMPs across seven classifications (Fig. 1) [9].

Fig. 1.

Overview of the seven different classification systems for AMPs used by the APD. Created in https://BioRender.com

AMPs are typically comprised of fewer than 100 amino acids, with masses of less than 5 kDa. Most are cationic, with hydrophobic residues that enable amphipathic conformations when interacting with membranes through AMP-membrane electrostatic interaction [10, 11]. The distinct composition of prokaryotic and eukaryotic membranes can allow AMPs to selectively target specific cells such as invading pathogens while sparing host cells [12]. Interactions between AMPs and target membranes typically lead to membrane permeabilization by pore formation. Further to directly killing pathogens, in higher organisms AMPs have also been shown to play an immunomodulatory role [13]. AMPs' unique mechanisms of action, rapid bactericidal activity, broad specificities, reduced likelihood of evolved resistance, high mobility, and high permeability make them promising alternatives to conventional antibiotics [14]. They are attractive therapeutic candidates for treating a range of diseases, with additional applications across agriculture and biotechnology applications as summarised in Fig. 2 [10, 15–23]. Despite their promise, the commercialization of AMPs is limited by costly production, which has necessitated the innovation of better production methods to make AMP production both commercially and environmentally viable [24].

Fig. 2.

Overview of AMP applications in healthcare, farming and biotechnology sectors. Including example AMPs employed or under investigation. Created in https://BioRender.com

Three methods exist for AMP production (Fig. 3). Direct purification from natural sources is typically expensive due to the resulting low quantities of partially purified products and potential for severe environmental consequences related to peptides produced from rare and non-farmable species. While some AMPs are naturally expressed constitutively, basal expression is generally low. For example, the extraction of apidaecin from Apis mellifera (western honeybee) requires 50,000 bees to achieve yields of only 1 g biologically active purified peptide [25]. Chemical synthesis methods can enable the production of both natural and synthetic peptides with high yield and purity and have resulted in the production of several bioactive AMPs [26, 27]. However, the cost and time requirements for chemical synthesis are infeasible for commercial-scale production and are challenging for AMPs with complex structures or longer sequences [28]. Biotechnological approaches such as heterologous expression can provide key functionality missed by chemical synthesis, and cheaper and more scalable production compared to direct purification from native hosts; with demonstrated high yields of active AMP. While host extraction of apidaecin was estimated to cost $45 USD per gram, heterologous expression can produce this for a mere $1 USD per gram [25], a preferable solution for both bees and financial bottom lines.

Fig. 3.

Comparison of the three methods for AMP production. Created in https://BioRender.com

Selecting an appropriate host for recombinant AMP expression depends on how well the host’s biological capabilities align with the peptide’s properties and intended application. A variety of biological systems—bacteria, yeasts, plants, and cell lines—have been used for AMP production (Fig. 4), with bacterial and yeast platforms being the host of choice for the majority of cases [29]. While bacterial systems have permitted high yield production due to their rapid growth rates, relatively inexpensive feedstocks, simple genetics, and comprehensive expression toolkits, they often experience issues with host toxicity and lack the cellular machinery to perform PTMs [30]. As these PTMs are required for the functionality of many eukaryotic AMPs, the resulting bacterially produced peptides can exhibit poor antibiotic activity. Fungal species, particularly yeasts, combine rapid growth and low nutritional requirements with the ability to perform essential PTMs. They are therefore a common next step when bacterial systems fail due to inadequate modification or toxicity. Their strong secretion capabilities also streamline purification for large-scale precision fermentation. As a result, yeast have become promising platforms for the expression of a broad range of AMPs.

Fig. 4.

Comparison of bacterial, yeast, plant, insect and mammalian hosts for heterologous AMP expression. Created in https://BioRender.com

This review will examine current strategies for AMP production in yeast and filamentous fungal hosts. It discusses key aspects of AMP expression and host optimization in yeast, including species selection and expression approaches such as codon and promoter optimization, secretion signal design, and protein fusion techniques, supported by examples reported in the literature.

Heterologous expression of AMPs in fungal hosts

Several species of yeast have been used for recombinant AMP expression, with Pichia pastoris (P. pastoris) being most commonly employed [31]. P. pastoris is the preferred choice due to its strong capacity for heterologous protein secretion and its lack of ethanol production under aerobic conditions, which support high cell densities and improved protein yields. P. pastoris has been engineered to optimise AMP production yield with strains listed in Table 1. While many AMP expression attempts have employed an unmodified X-33 strain (Table 2), AMPs are vulnerable to protease degradation and strains engineered to express lower protease activity can be beneficial for AMP yields and stability. Several protease deficient P. pastoris strains containing deletions in genes such as pep4 (proteinase A) and prb1 (encoding proteinase B) have been employed to produce various AMPs including cecropin D, cecropin B, apidaecin, and LL-37 [25, 32–34]. While these strains have shown improved yields for non-AMP proteins, only two studies have directly assessed the benefit of these deletions for AMP expression, and no increase in yield was observed [34–38]. Pichia has 104 predicted proteases across 50 families; thus, it is highly unlikely that these two deletions have comprehensively explored all of the broad proteolytic activities Pichia could display, presenting ample opportunity for further engineering [39].

Table 1.

Overview of P. pastoris engineered strains

Species	Strain	Genotype	Characteristics
P. pastoris	X-33	Mut⁺, his4 variant	Derived from CBS7435; AOX1 and AOX2 present, growth on methanol, high transformation efficiency due to more permeable cell wall
P. pastoris	CBS7435	WT (Mut⁺)	Wild-type parent strain (Y-11430), AOX1 and AOX2 present, robust growth on methanol
P. pastoris	GS115	Mut⁺, his4-	Histidine-deficient, AOX1 and AOX2 present, growth on methanol
P. pastoris	KM71	MutS	Histidine-deficient, AOX1 replaced by S. cerevisiae ARG4, slow/weak methanol utilization
P. pastoris	MC100-3	Mut⁻	Deficient in both AOX genes, no growth on methanol
P. pastoris	SMD1163	pep4-, prb1-	Protease-deficient background
P. pastoris	SMD1168	pep4-, his4-	Protease-deficient and histidine-deficient
P. pastoris	PPY12	his4, arg4 auxotrophic	Double auxotroph
P. pastoris	BSYBG11	MutS, aox1Δ1	Derived from CBS7435; AOX1-deficient, slow/weak methanol utilization

Mut⁺, methanol utilization proficient; Mutˢ, slow/weak methanol utilization; Mut⁻, methanol utilization deficient

Table 2.

Heterologous expression attempts of AMPs in yeast host P. pastoris since 2010

Peptide name	Source	Expression host	Vector name	Codon optimized	Yield	References
ABP-dHC-cecropin A	Hyphantria cunea (drury moths)	P. pastoris GS115	pPICZαA	No	21 mg/l	[110]
acidocin 4356 (ACD)	Lactobacillus acidophilus ATCC 4356	P. pastoris GS115	pPICZαA	Yes	ND	[113]
AP138L-arg26	Recombinant Pseudoplectania nigrella derived	P. pastoris X-33	pPICZαA	Yes	3.1 mg/ml	[114]
Apidaecin	Apis mellifera (Honeybee)	P. pastoris SMD1168H	pPICAP	Yes	418 mg/l	[25]
Apidaecin Ia	Apis mellifera (Honeybee)	P. pastoris	pPICZαA	No	700 mg/l	[68]
Bacteriocin E 50–52 (BacE50-52)	Enterococcus faecium NRRL B-3274	P. pastoris X-33	pPICZαA	Yes	ND	[42]
Bovine Lactoferricin (BLF)	Recombinant (Bos taurus-derived)	P. pastoris X-33	pPICZαA	Yes	193.9 mg/l	[35]
Cathelicidin-BF	Bungarus fasciatus (Many-banded Krait)	P. pastoris X-33	pPICZαA	Yes	500 mg/l	[115]
Cathelicidins (FBC)	Bos taurus (Bovine)	P. pastoris SMD1168	pGAPZαA	No	ND	[62]
Cecropin A-thanatin	Hyalophora cecropia/Podisus maculiventris	P. pastoris X-33	pPICZαA	Yes	ND	[111]
Cecropin D	Bombyx mori (domestic silkworm)	P. pastoris SMD1168	pGAPZαA	Yes	485.2 mg/l	[32]
Ceropin P4	Ascaris suum (pig large round worm)	P. pastoris X-33	pPICZB	No	NS	[107]
CGA-N12	Recombinant Homo sapiens (Human) derived	P. pastoris X-33	pPIC9K	Yes	30 mg/l	[116]
Dybowskin-2 CAMa	Rana dybowskii (Dybowsky's Frog)	P. pastoris X-33	pPICZαA	No	500 mg/l	[117]
Enterocin A (EntA)	Enterococcus faecium T136	P. pastoris X-33	pPICZαA	No	45.1 ng/mg	[118]
Enterocin A (EntA)	Enterococcus faecium (Common Enterococcus)	P. pastoris X-33	pPICZαA	Yes	180 mg/l	[119]
Enterocin L50A (EntL50A)	Enterococcus faecium L50	P. pastoris X-33	pPICZαA	No	228.5 ng/mg	[73]
Enterocin L50B (EntL50B)	Enterococcus faecium L50	P. pastoris X-33	pPICZαA	No	1240 ng/mg	[73]
Epidermicin NI01	Staphylococcus epidermidis	P. pastoris CBS7435/ BSBG11	pPpT4	Yes	100 mg/l	[101]
Fowlicidin-2 (Sc: α)	Gallus gallus domesticus (Chicken)	P. pastoris X-33	pPICZαA	Yes	85.6 mg/l	[82]
gcIFN-20H	Recombinant Ctenopharyngodon idella (grass carp) derived	P. pastoris GS115	pPIC9K	Yes	ND	[23]
Hepcidin (HepcD)	Camelus dromedarius (Dromedary Camel)	P. pastoris X-33	pPICZαA	No	2.8 mg/l	[69]
Hepcidin 25 (Hep25)	Homo sapiens (Human)	P. pastoris X-33	pPICZαA	No	1.9 mg/l	[71]
Hepcidin (LJ-hep2)	Lateolabrax japonicus (Japanese seabass)	P. pastoris GS115	pPIC9K	Yes	ND	[120]
Hispidalin	Benincasa hispida (Wax Gourd)	P. pastoris GS115	pPICZαA	No	98.6 µg/ml	[121]
Human β-defensin 2 (hBD-2)	Homo sapiens (Human)	P. pastoris X-33	pPICZαA	No	2mg/l	[122]
interleukin-2 (IL-2) & bovine cathelicidins (FBC)	Sus scrofa domesticus (Domestic Pig), Bos taurus (Bovine)	P. pastoris SMD1168	pGAPZαA	No	ND	[62]
L2	Streptococcus agalactiae	P. pastoris X-33	pPIC-SUMO	Yes	629 mg/l	[91]
Lactoferricin (Lfcin)	Camelus dromedarius (Dromedary Camel)	P. pastoris KM71	pPICZαA	Yes	ND	[123]
Lactoferricin (LfcinB)	Bos taurus (Bovine)	P. pastoris GS115	pPIC9K-Hi	Yes	NA	[124]
LG (LL-37 derived)	Homo sapiens (Human)	P. pastoris X-33	pPICZaA	Yes	4.3 mg/l	[103]
LI (LL37-Indolicidin)	Homo sapiens/Bos taurus	P. pastoris GS115	pPIC9K	Yes	5.9 mg/l	[88]
LL-37 Tα1	Synthetic (Homo sapiens-derived)	P. pastoris X-33	pPIC9K	Yes	20 mg/l	[93]
LL37	Homo sapiens (Human)	P. pastoris GS115	pPIC3.5K	No	446 g/l*	[67]
Lunasin	Glycine max (Soybean)	P. pastoris GS115	pPIC9K	Yes	0.2 mg/ml	[87]
Mytichitin A	Mytilus coruscus (Mussel)	P. pastoris GS115	pPICZαA	No	18.2 mg/l	[106]
Neutrophil peptide 1 (HNP1)	Homo sapiens (Human)	P. pastoris GS115	pPIC9K	No	33.8 mg/l	[46]
NFAP (Antifungal protein)	Neosartorya fischeri	P. pastoris KM71H	pPICZαA	Yes	11.3 mg/l	[125]
NFAP (Antifungal protein)	Neosartorya fischeri	P. pastoris KM71H	pPICZαA	No	5.958 mg/l	[126]
NZ17074	Arenicola marina (Lugworm)	P. pastoris X-33	pPICZαA	Yes	4.1 mg/l	[104]
NZ2114	Recombinant Pseudoplectania nigrella (Black Pseudoplectania) derived	P. pastoris X-33	pPICZαA	Yes	1309 mg/l	[127]
NZL	Recombinant Pseudoplectania nigrella (Black Pseudoplectania) derived	P. pastoris X-33	NA	NA	1505 mg/L	[65]
N-terminal lobe of human lactoferrin (hLF)	Homo sapiens (Human)	P. pastoris SMD1168H	pPIC9K	Yes	458 mg/l	[128]
OH-CATH30	Ophiophagus hannah (snake)	P. pastoris X-33	pPIC9K	NA	NA	[129]
PaDef	Persea americana (Avocado)	P. pastoris GS115	pPICZαA	Yes	32.8 mg/l	[97]
Parasin I-lysozyme (PI-hLY)	Siluriformes (Catfish), Homo sapiens (Human)	P. pastoris X-33	pPICZαA	Yes	86 mg/l	[79]
PAD102	Synthetic (rationally designed)	P. pastoris X-33	pGAPHA	NA	180 mg/l	[100]
Peptide K	Synthetic (rationally designed)	P. pastoris GS115	pPIC9K	Yes	6.67 mg/ml	[130]
PBD-2	Sus scrofa domesticus (Domestic Pig)	P. pastoris KM71	PIC9k	Yes	NA	[90]
Plectasin	Pseudoplectania nigrella (Black Pseudoplectania)	P. pastoris X-33	pPICZαA	Yes	879mg/l	[57]
Plectasin	Pseudoplectania nigrella (Black Pseudoplectania)	P. pastoris X-33	pPICZαA	Yes	537 µg/ml	[83]
Plectasin	Pseudoplectania nigrella (Black Pseudoplectania)	P. pastoris X-33	pPICZαA	No	143 µg/ml	[86]
Plectasin	Pseudoplectania nigrella (Black Pseudoplectania)	P. pastoris GS115	pPICZαA	Yes	426.3 mg/l	[61]
rPlectasin (Ple-AB)	Recombinant Pseudoplectania nigrella (Black Pseudoplectania) derived	P. pastoris X-33	pPICZαA	Yes	2.9 g/l	[131]
Porcine β-defensin 2 (pBD2)	Sus scrofa domesticus (Domestic Pig)	P. pastoris X-33	pPIC9K	Yes	383.7 mg/l	[56]
Porcine β-defensin 2 (pBD2)	Sus scrofa domesticus (Domestic Pig)	P. pastoris X-33	pPICZαA	Yes	383.7 mg/l	[132]
PMAP-37(F34-R)	Sus scrofa domesticus (Domestic Pig)	P. pastoris GS115	pPICZαA	Yes	ND	[133]
Protegrin-1 (rPG-1)	Sus scrofa domesticus (Domestic Pig)	P. pastoris	pPIC9K	Yes	104 mg/l	[58]
Protegrin-1 (rPG-1)	Sus scrofa domesticus (Domestic Pig)	P. pastoris GS115	pGAP4	Yes	NA	[134]
PR-FO	Sus scrofa domesticus (Domestic Pig)/ Gallus gallus domesticus (Chicken)	P. pastoris X-33	pPICZαA	Yes	259.6 mg/L	[108]
rAc-AMP2	Amaranthus caudatus (Amaranth)	P. pastoris GS115	pPICZαA	Yes	210 mg/l	[135]
rMiAMP1	Macadamia integrifolia (Macadamia Nut)	P. pastoris GS115	pPICZαA	Yes	220 mg/l	[135]
Sakacin A	Latilactobacillus sakei Lb706	P. pastoris X-33	pPICZαA	No	3.6 mg/l	[109]
Snakin-1 (SN-1)	Solanum tuberosum (Potato)	P. pastoris GS115	pPIC9	No	40 mg/l	[136]
Spgillcin177–189	Scylla paramamosain (Mud crab)	P. pastoris GS115	pPICZαA	Yes	126.1 mg/L	[59]
Tachyplesin I	Podisus (Spined soldier bug)	P. pastoris GS115	pGAPZαB	Yes	29.5 mg/l	[89]
Thanatin	Podisus maculiventris (Spined soldier bug)	P. pastoris KM71H	pPICZαA	Yes	NA	[137]
Thanatin	Podisus maculiventris (Spined soldier bug)	P. pastoris GS115	pPIC9K	Yes	20 mg/l	[63]
Thanatin	Podisus maculiventris (Spined soldier bug)	P. pastoris GS115	pGAP4	Yes	NA	[134]
rTEWP	Recombinant Caretta caretta (Sea turtle) derived	P. pastoris GS115	pPICZαA	Yes	0.128 mg/mL	[138]
Tilapia piscidin 4 (TP4)	Oreochromis niloticus (Nile tilapia)	P. pastoris X-33	pPICZαA	Yes	1.07 mg/g	[139]
Thrombocidin-1 (TC-1)	Homo sapiens (Human platelets)	P. pastoris KM71H	pPICZαA	Yes	190 µg/ml	[140]
Thymosin β4 (rTβ4)	Pinctada fucata (Akoya pearl oyster)	P. pastoris X-33	pPICZαA	Yes	104 µg/ml	[141]
Turgencin A	Synoicum turgens (sea squirt)	P. pastoris GS115	pPICZαA	Yes	11.23 µg/ml	[142]

Includes peptide source organism, expression host, expression vector, codon optimization approach and yield

ND, not determined; NA, not available

A number of other reports of AMP production employ alternative yeasts to P. pastoris including Saccharomyces cerevisiae (S. cerevisiae), Kluyveromyces lactis (K. lactis), Saccharomyces boulardii (S. boulardii), Hansenula polymorpha, and Arxula adeninivorans have exemplified varying degrees of success in terms of yield, bioactivity and scalability (Table 3). The appeal of these alternative yeasts includes unique glycosylation patterns and high lipid tolerance which may improve the AMP yield and stability, though it can be challenging to empirically anticipate which AMPs are best matched with specific strains [40]. S. cerevisiae was the first yeast to be used for recombinant protein production and has a deep history with well-characterized physiology and genetics, and a plethora of available of vectors, genetic tools, and strains. For AMP production, however, S. cerevisiae has shown limited success; there are only a few reports of AMPs expression, often with low yields as documented in Table 3. One challenge is that S. cerevisiae’s metabolism redirects carbon to ethanol production, limiting biomass accumulation and peptide yield. P. pastoris has consistently outperformed S. cerevisiae in terms of AMP yield. P. pastoris and S. cerevisiae additionally differ in their glycosylation machinery. Glycosylation plays a crucial role in the folding, stability, and function of many proteins and peptides. P. pastoris produces simpler, less immunogenic N- and O-glycans and has also been used for the development of GlycoSwitch technology, a protein production system comprising engineered strains capable of human-like glycosylation [34]. S. cerevisiae exhibits hyper mannosylated glycans that can be both immunogenic and challenging to modify. In one study exploring the expression of the primarily O-glycosylated penaeidins Pen-2 and Pen-3a from Penaeus vannamei (tropical shrimp) in S. cerevisiae, the substitution of the native dimannosyl group with O-glycans resulted in significantly reduced penaeidin activity from half of the tested strains, rendering the strain non-viable for production of these AMPs [35]. Alternative Saccharomyces have also been explored for in situ AMP production [41], providing an AMP delivery mechanism in conjunction with recombinant production. As the transient gut colonization and superior heat and acid tolerance of Saccharomyces boulardii permits better survival under gastrointestinal conditions as compared to S. cerevisiae, the AMP leucocin C from Leuconostoc carnosum was recombinantly produced in S. boulardii and exhibited antimicrobial activity against Listeria monocytogenes. (L. monocytogenes). Another yeast species, K. lactis, was compared to P. pastoris for expression of two Enterococcus-derived AMPs, entA and BacE50-52 [42]. While no specific rationale for the selection of K. lactis was discussed in the study, K. lactis has well-established safety record and long history of use in food biotechnology, where it has been used as a heterologous protein expression system for commercial production of the milk-clotting enzyme bovine chymosin as well as over 40 other proteins from diverse source organisms [43]. Compared to the native producer, Enterococcus faecium T136, P. pastoris showed a 21.4-fold entA production increase, with 235-fold greater antimicrobial activity in after 12 h incubation in YPD medium at 30° [42]. K. lactis, however, only showed a meagre 1.2-fold increase in production, with reduced antimicrobial activity. While employment of alternative yeast strains has shown sporadic success, empirical host screening ideally via comparative expression trials is often the only way find an optimal yeast host for AMP production. One advantage of yeast is that high throughput (HTP) automated heat shock transformation workflows are available for some species (Fig. 5), allowing for rapid screening of diverse panels of production strains. Access to diverse and well-characterised strain collections is valuable as natural and engineered variation between strains can strongly influence secretion efficiency, stress tolerance, and protease activity, which could accelerate the development of efficient AMP production platforms.

Table 3.

Heterologous expression attempts of AMPs in alternative yeast hosts since 2010

Peptide	Source	Expression host	Vector name	Codon optimized	Yield	References
Albusin B	Ruminococcus albus 7	Saccharomyces cerevisiae DBY 747	pYEX-S1	No	1 mg/g	[143]
Bacteriocin E 50–52 (BacE50-52)	Enterococcus faecium NRRL B-32746	Kluyveromyces lactis GG799	pKLAC2	Yes	ND	[42]
Cecropin P1	Sus scrofa domesticus (Domestic Pig)	Saccharomyces cerevisiae INVSc1	pYES2/CT	Yes	7.8 mg/l	[74]
Crustin	Portunus trituberculatus (Blue crab)	Saccharomyces cerevisiae S-78	pVT102U	–	ND	[144]
Enterocin A (EntA)	Enterococcus faecium T136	Kluyveromyces lactis GG799	pKLAC2	No	10.5 ng/mg	[70]
Enterocin A (EntA)	Enterococcus faecium T136	Hansenula polymorpha KL8-1	s pBTEA	No	4.8 ng/mg	[70]
Enterocin A (EntA)	Enterococcus faecium T136	Arxula adeninivorans G1212	pBYEA	No	ND	[70]
Enterocin A (EntA)	Enterococcus faecium T136	Kluyveromyces lactis GG799	pKLAC2	Yes	ND	[42]
Enterocin P-Sakacin A (EntA-SakA)	Latilactobacillus sakei Lb706	Kluyveromyces lactis GG799	pKLAC2	No	110.4 mg/l	[109]
Leucocin C	Leuconostoc carnosum 4010	Saccharomyces boulardii CNCM I-745	pSF-TEFI-TPI1-Blast	Yes	NA	[41]
Mundticin ST42A (MunX)	Enterococcus mundtii CUGF08	Saccharomyces cerevisiae Y294	pMR	Yes	20 mg/l	[76]
Pediocin (PA-1)	Pediococcus acidilactici	Saccharomyces cerevisiae Y294	YEp352	No	ND	[145]
Plantaricin 423 (PlaX)	Lactobacillus plantarum	Saccharomyces cerevisiae Y294	pMR	Yes	18.4 mg/l	[76]
Plectasin	Pseudoplectania nigrella	Saccharomyces cerevisiae Y294	pMT2548	No	NA	[48]
Sakacin A	Latilactobacillus sakei Lb706	Kluyveromyces lactis GG799	pKLAC2	No	22.9 mg/l	[109]

Includes peptide source organism, expression host, expression vector, codon optimization approach and yield

ND, not determined; NA, not available

Fig. 5.

Workflow for the heterologous expression of AMPs in yeast hosts. Created in https://BioRender.com

Other fungal systems such as filamentous fungi have been explored for their supposed high potential for heterologous protein expression. Notable filamentous fungi cell factories include Trichoderma reesei, Aspergillus spp., and Penicillium spp. Filamentous fungi exhibit a saprophytic lifestyle which has driven the evolution of highly sophisticated protein secretion systems capable of mediating various PTMs and translocating large quantities of hydrolytic enzymes outside of the cell. Despite these hypothetical advantages, there have been few reports of recombinant AMP expression in filamentous fungi, and none yet approach the yields or functionality achieved with yeast production (Table 4). Filamentous fungi have larger, more complex genomes, complicated fermentation requirements, express large quantities of AMP-destroying proteases, and have much slower development timelines due to less-developed protocols and expression toolkits [44], leading to a significant lag behind yeast for heterologous protein production. While the hypothetical advantages of filamentous fungi could in theory provide benefits over yeast platforms, significant resources and toolkits will be required before any advantages can be reaped for recombinant AMP production.

Table 4.

Homologous or heterologous expression attempts of AMPs in filamentous fungal hosts since 2010

Peptide name	Source	Expression host	Vector name	Codon optimized	Yield					References
Antifungal protein (afpB)	Penicillium chrysogenum	Penicillium digitatum CECT	pSK275	No	20 mg/l					[45]
Cecropin B (CB)	Aspergillus awamori	Cordyceps militaris	pCB130-CB	Yes	2.9 mg/g					[46]
Magainin II-cecropin B (Mag II-CB)	Aspergillus awamori	Cordyceps militaris	pCB130-CB	Yes	3.9 mg	1168H	pPICAP	Yes	418 mg/l	[25, 46]
NFAP	Neosartorya fischeri	Penicillium chrysogenum Δpaf mutant strain	pSK275	Yes	80 mg/l					[47]
PAF	Penicillium chrysogenum	Penicillium chrysogenum Δpaf mutant strain	pBHt2	Yes	3 mg/l					[47]
Plectasin	Pseudoplectania nigrella	Aspergillus oryzae 55@@BECh2	pMT2188,	No	-					[48]
Plectasin	Pseudoplectania nigrella	Aspergillus niger MBin118.30	pMT2188,	No	-					[48]

Includes source organism, expression host, expression vector, codon optimisation approach and yield

ND, not determined; NA, not available

Expression strategies for increasing AMP yield

Several factors influence recombinant AMP production including transcriptional, translational, and post-translational factors, the methodology of each of which can be optimized to increase peptide yields (Fig. 6).

Fig. 6.

Points of optimization for heterologous expression of AMPs in yeast hosts. Created in https://BioRender.com

Increasing gene transcription via expression loci strategies

To perform heterologous expression, a gene cassette must first be introduced to the cells; this can either be maintained on a plasmid (episomal, separate to the genome), or integrated directly into the genome via homologous recombination. The overwhelming majority of reported AMP expression systems have utilised genome integration of the expression cassette, either via random (ectopic) integration or integration into a defined locus. Targeted integration minimises the probability of undesired off target effects such as the unintentional disruption of growth or metabolism related genes. Where a strain is well characterised, highly transcriptionally-active genomic loci can be selected; in P. pastoris and S. cerevisiae, the vectors pPICZαA and pPIC9K are commonly used to integrate genes into the widely used and transcriptionally active AOX1 locus. Ectopic integration but can result in high-expressing clones when the cassette lands in a highly active part of the genome or a high number of copies are integrated [49]. Random ectopic integration can also result in perturbed growth or poorly expressing strains, where integrations disrupt core functions, display low copy number, or are in poorly-transcribed genomic regions. As such, ectopic integration requires more screening to determine the best-producing strain and makes the approach more laborious and particularly challenging for low yielding AMPs that lack robust quantification assays.

Increasing gene transcription via copy number strategies

By increasing gene copy number, stronger transcription can be achieved, leading to increased translation and thus more product [50–52]. To achieve high copy number integrations, a number of strategies can be employed. For example, transformation with vectors harbouring multiple copies of expression cassettes, targeting multi-copy genes or repetitive regions such as the rDNA, can lead to higher numbers of gene copy numbers [53]. Dose-dependent antibiotic resistance markers can also be used; in some cases, expressing more copies of an antibiotic resistance gene confers resistance to higher concentrations of the antibiotic (e.g. amdS, bleoR). This property can be hijacked to also select for extra copies of an AMP expression construct when integrated alongside the antibiotic resistance marker [54]. Several studies have optimized copy number to increase AMP expression [55–57]. In one example P. pastoris X-33 transformants harbouring 1–14 copies of the expression recombinant protegrin-1 (PG-1) expression cassette were analysed for expression [58]. This linearly increased yields in strains with up to six copies, however higher copy number clones (> 11) did not improve yield, indicating an upper limit on benefits reaped from multi-copy integrations. This correlation also translated to peptide activity levels. Interestingly two clones, both with six copies, expressed the AMP at different levels, suggesting the involvement of additional factors. In another study investigating plectasin expression in Pichia pastoris X-33, 1-, 2-, 4- and 8-copy transformants yielded 296, 409, 516 and 879 mg/L respectively, showing a positive correlation between copy number and yield [57]. In a study expressing Spgillcin177–189 in P. pastoris, strains with higher integration copy numbers showed markedly increased expression, with the optimal 4‑copy integrant producing recombinant Spgillcin177–189 at 126.1 mg/l—a 2.75‑fold increase over the single‑copy strain [59]. However, beyond this copy‑number threshold, no further improvement was observed. While multi-copy gene insertions can be effective in yield optimisation, increasing copy number beyond a certain threshold can lead to diminishing returns or even decreased yield. This can be due to metabolic burden, toxic levels of protein accumulation, promoter saturation or post-translational bottlenecks such as ER folding capacity, membrane translocation, signal sequence processing and secretion efficiency limitations [60]. An alternative way to increase AMP production is to link multiple peptide coding units in a single transcript using self-processing 2A sequences. This approach allows ribosomal skipping during translation, permitting the generation of multiple discrete peptide products from a single mRNA without the need for in vitro cleavage unlike a tandem repeat fusion approach – which is discussed under Post translational strategies for increasing AMP yield. This strategy was applied for the expression of plectasin in P. pastoris by constructing a four-copy cassette with interspaced 2a sequences [61]. 183.2 mg/L of plectasin with 60.8% purity was produced, a 2.3-fold increase compared to the one-copy producing strain. This was further optimized to 426.3 mg/L total protein by upscaling to a 5-litre fermenter.

A similar 2A-based design was used for the co-expression of pig IL-2 and a fusion bovine cathelicidin gene (FBC) from the same vector in P. pastoris SMD1168 [62]. This study showed that 2A-mediated co-expression could support simultaneous production of functionally relevant antimicrobial and immunomodulatory proteins while specific peptide yield were not reported, retained antimicrobial activity and immune response modulation was demonstrated in vitro and in vivo.

Increasing gene transcription via promoter choice

Promoter strength is an important factor to balance for AMP expression, with strong transcription initiation desirable to yield high quantities of AMP. Promoter libraries with a range of strengths allows tailored expression of AMPs. Inducible promoters can control the timing of transcription, allowing a high cell density to be achieved before initiating peptide expression and mitigating toxicity. Expression inducers can be prohibitively expensive at scale and increase process complexity. Constitutive promoters are much more practical at industrial scale but can result in lower yields due to the increased metabolic burden and reduced growth from continuous high-level expression [35, 63, 64]. As shown in Table 5, the inducible AOX1 promoter is most commonly employed in literature for AMP expression in P. pastoris due to its tight expression control and relatively cheap inducible regulation. The highest reported yield was of NZL, a variant of plectacin expressed in P. pastoris X-33, which yielded 1505 mg/l in a 5 l fermenter [65]. Inducer dosage requires careful optimization. Insufficient methanol fails to drive efficient transcription while excessive levels are toxic to P. pastoris due to formaldehyde and hydrogen peroxide accumulation [66]. Although less frequently employed, constitutive promotes like GAP (glyceraldehyde-3-phosphate dehydrogenase) have been used in P. pastoris for the expression of AMPs [32], and have also shown high titres, e.g. Cecropin D expressed in P. pastoris SMD1168 yielded 485.24 mg/l. Both inducible and constitutive promoters can yield high titres of AMP; the best choice will depend on the AMP and the expression host, particularly in relation to the risk of toxicity of the AMPs on the host strain.

Table 5.

Summary of promoters used for AMP expression attempts in yeast and filamentous fungal hosts

Promoter	Type	Frequency of use in literature	Example peptides	Typical host(s)	References
AOX1	Induced	Employed very regularly	LL-37, Pediocin (PA-1), Hepcidin (HepcD), Apidaecin Ia, α-defensin 5, Enterocin P, Hepcidin 25	P. pastoris K. lactis	[32, 56, 67–71]
GAP	Constitutive	Infrequent reports	Cecropin D, α-defensin 5, Thanatin rThan	P. pastoris	[32, 63, 72]
PGAL1	Induced	Infrequent reports	Enterocin L50A (EntL50A), Enterocin L50B (EntL50B)	S. cerevisiae	[73]
T7	Induced	Infrequent reports	Cecropin P1	S. cerevisiae	[74]
MFa1	Induced	Infrequent reports	Drosomycin	S. cerevisiae	[75]
paf	Constitutive	Infrequent reports	NFAP, PAF	P. chrysogenum ∆paf mutant strain	[47]
TEF1	Constitutive	Infrequent reports	Mundticin ST42A (MunX), Plantaricin 423 (PlaX)	S. cerevisiae Y294	[76]
AHD1	Constitutive	Infrequent reports	Crustin	S. cerevisiae S-78	[77]

Increasing protein translation via codon modification

Codon optimization aims to improve translation efficiency by harmonizing codons with those preferred by the host. This is particularly important when expressing in an evolutionarily-distant host, e.g. mammalian or bacterial AMPs expressed in yeast. Whilst widely employed, there are few studies specifically evaluating the impact of codon optimization on AMP expression. One study expressed codon optimized and native forms of plantaricin 423 (pla423) and mundticin ST4SA (munST4SA) in S. cerevisiae [76]. For optimized versions, yields were very similar between both AMPs, however the authors did not report the yields for non-optimized variants, so the impact of codon optimization on peptide yield is unclear. Interestingly, the authors did compare the variants' activities, with munST4SA codon optimization leading to an eight-fold increase in activity, whilst codon optimization of pla423 showed no significant improvement in activity. The authors suggest this might be due to the extent to which the two peptides were codon optimized (expressed through the codon bias index (CBI)). The CBI was higher for optimized munST4SA than for optimized pla423, meaning the codons on average were more optimal for munST4SA. From these indirect inferences it is difficult to draw concrete conclusions on whether codon optimization tangibly impacts AMP yield or activity. Controlled comparisons are scarce, likely because other bottlenecks such as toxicity, proteolysis, and secretion mask its effects, leading researchers to apply codon optimization routinely as best practice rather than a test variable. Codon optimization remains fundamental and recommended for heterologous expression in distantly related hosts like yeast.

Post translational strategies for increasing AMP yield

Protein fusions are widely used in yeast systems to enhance stability, secretion, solubility, folding, PTMs, and purification of target proteins and have been widely employed for AMP expression (Table 6). The fused domains may be secretion signals, affinity tags, carrier proteins, or hybrid peptide dimers as depicted in Fig. 7. These domains can be connected via linker sequences which vary in flexibility, length, and amino acid composition to meet specific functional requirements [78]. Linkers are utilised to increase the stability of the connection between the peptide and the fusion domain and can aid in correct folding. Cleavable linkers contain specific cleavage sites which allow for liberation of the AMP from the carrier protein via a chemical or enzymatic digest. Multiple cleavage sites have been employed for AMP fusion protein separation including enterokinase, formic acid, and TEV [68, 79, 80]. However, cleavage increases downstream processing costs and could limit scalability and economic feasibility.

Table 6.

Fusion tags and carrier protein approaches used for heterologous AMP expression attempts in yeast hosts

Element	Type	Regularity of employment	Example peptides	Typical host(s)	References
SUMO / SUMO3	Solubility tag	Regularly employed	Enterocin HF (EntHF), Enterocin CRL35 (CRL35), LG (LL-37 derived), NZ17074, L2	P. pastoris	[89, 102–104]
His-tag (6 × His)	Affinity purification tag	Regularly employed	β-defensin-1 (sBD-1), Scygonadin (scy), Mytichitin A, PaDef, Hepcidin (HepcD), Pediocin (PA-1), LL-37 Tα1, Mdcec, Hepcidin 25 (Hep25)	P. pastoris	[69, 71, 92, 93, 95, 98, 105, 106]
GST	Solubility/affinity tag	Occasionally employed	Cecropin D	P. pastoris@@SMD1168	[32]
FLAG	Detection/affinity tag	Occasionally employed	pBD-2	P. pastoris@@KM71	[90]
MYC-tag	Detection/affinity tag	Occasionally employed	Cecropin P4, β-defensin-1 (sBD-1)	P. pastoris	[95, 107]
Ole18	Purification	Rare	PAD102	P. pastoris	[100]
Human Serum Albumin	Solubility Enhancer carrier protein	Occasional	Apidaecin Ia	P. pastoris	[68]
iLOV	Transformant library screening	Rare	NI01	P. pastoris BSYBG11	[101]
EGFP-Polyhedrin	Expression Enhancer carrier protein	Rare	Neutrophil peptide 1 (HNP1)	P. pastoris GS115	[46]
PEP1/PEP2	Charge neutraliser carrier protein	Rare	Bovine Lactoferricin (BLF)	P. pastoris X-33	[35]

Fig. 7.

Overview of fusion peptide approaches employed for heterologous AMP expression in yeast hosts. Created in https://BioRender.com

Secretion of peptides can be advantageous for several reasons, including simplified purification, reduced host cell toxicity, proper folding, peptide maturation, and PTM annotation. Secretion can be achieved by tagging AMPs with a secretion peptide, typically a host-native N-terminal sequence that targets the AMP for secretion. The α-mating factor MFα1 signal peptide, a 13-residue pheromone from S. cerevisiae α-cells that directs secretion— is widely used for AMP secretion via the sec-dependent pathway and has been used for nearly all P. pastoris AMP expression studies [81]. Secretion tags are typically cleaved upon secretion; for MFα1s this is performed by the Kex2 and Ste13 proteases. While the α-mating factor is highly effective, there have been multiple reports of inefficient tag cleavage impacting AMP production due to either protease overload from high expression saturating Kex2/Ste13, suboptimal cleavage site sequences or pro-region misfolding with cationic AMPs [82–84]. Thus, other signal peptides have been explored for yeast AMP expression. In one study, nine were evaluated for bovine AMP Lfcin in P. pastoris X-33 [35]. The tags originated in a range of organisms, 0030 and SCW10 from P. pastoris, INU1 and UTH1 from Kluyveromyces marxianus, SP from Bos taurus, and MEL1, OST1, PHO11 and SUC2 from S. cerevisiae. All nine significantly reduced Lfcin production, with most strains failing to produce detectable levels of peptide. This suggested that both pre- and pro-sequences present on the MFα1 were required for proper Lfcin processing. Researchers then hybridized the MFα1 by substituting its pre-sequence with pre-sequences derived from the nine signal peptides. Hybrid signal peptides 0030-α, OST-α-, and UTH1-α exhibited higher Lfcin titres, with 0030-α showing the highest yield at 28.8 mg/l and outperforming unhybridized MFα1. Hybrid signal peptides can be an effective AMP yield booster, presumably maintaining the MFα1 pro-peptide and ensuring proper AMP maturation, with secretion boosted via a more active secretion signal. Signal peptides can also be used to target AMPs to an organelle, for example to the peroxisome using the C-terminal Ser-Lys-Leu (SKL) signal peptide. This can prevent undesirable modifications and proteolytic degradation and could mitigate host toxicity. For example, this approach was used to alleviate the host toxicity of AMP LL37 in P. pastoris GS115 [67]. The resulting expression strain showed no significant inhibition of cell growth after expression was induced and the peptide maintained antimicrobial activity. This underlines the potential benefits of peroxisomal targeting to alleviate host toxicity of AMPs and is a strategy worth considering.

Tandem repeat peptide fusions are an employed strategy to increase the stability, detectability and recovery of AMPs. Multiple copies of the AMP-coding sequence are concatenated in a head-to-tail configuration under a single promoter and terminator typically separated by protease-recognition sites. This arrangement generates a single polypeptide multimer that can be cleaved in vitro to release individual AMP monomers, thereby amplifying the yield of active peptide per transcript. Although this approach requires additional biochemical processing. This approach was used for the expression of tachyplesin I (TP-I) in P. pastoris, in which four tandem TP-I copies were separated by protease-cleavage sites. The secreted 74 amino acid peptide multimer was processed to release authentic TP-I monomers and yielded about 27.24–29.53 mg/l with retained antibacterial activity [85]. In some scenarios the multimeric peptide polymer has shown to retain antimicrobial effectivity or even enhance it without monomer liberation [86, 87] A tandemly arrayed plectasin gene (Ple-8) expressed it in P. pastoris using multicopy episomal plasmids [86]. The peptide multimer was efficiently secreted yielding 143 µg/ml with stability and antimicrobial activity comparable to that of monomeric plectasin.

Affinity tags such as Fusion Locus Antigen Gene (FLAG), MYC, Glutathione S-transferase (GST), and Small Ubiquitin-like Modifier Outcome (SUMO) have been used for peptide detection, to aid solubility, and for purification of AMPs in yeasts [88–91]. The SUMO and FLAG tags were directly compared for their ability to enhance expression of the synthetic peptide L2 in P. pastoris X-33. Fusion SUMO-L2 achieved a satisfactory yield of 629 mg/l, and the C-terminally SUMO tagged peptide was stable and did not adversely affect L2 conformation. In contrast, the FLAG tag seemed to compromise peptide stability and resulted in a ~ 50% lower yield [91]. The poly-histidine tag (His-tag) is routinely employed for yeast AMP expression to facilitate purification and detection [92–97]. His-tags consist of at least six histidine residues and typically do not affect peptide function when C- or N-terminally fused for various AMPs including hepcidin, pediocin and LL-37 [69, 93, 98]. Multiple studies have demonstrated His-tags can even improve functionality and yield [92, 95]. His-tagged sheep β-defensin-1 (msBD-1-T) expressed in P. pastoris GS115 maintained antimicrobial efficacy while improving purification yield from 35 to 80 mg/l [95]. Conversely, while a His-tag on corn defensin PDC1 expressed in P. pastoris GS115 did not impact peptide structure, the tag reduced antifungal efficacy against Fusarium graminearum [99]. This demonstrates the varied impacts different tags can have on different AMPs. As this is hard to predict, it is recommended that new AMP expressions test a range of tags to find the best fit. Carrier proteins fusing a heterologous protein to a highly secreted native protein have been proven to improve solubility, folding, and secretion. Further, using well-expressed native proteins can improve transcription by piggybacking on the already efficient transcription of the native gene. Efficient carrier proteins are characterized by factors such as size, isoelectric point, secretion efficiency, and hydrophobicity. While not common, there is precedent for carrier proteins in yeast AMP expression, usually employed to mitigate host toxicity (Table 6). For example, for AMP human neutrophil peptide 1 (HNP1), a carrier protein fusion approach was taken for expression in P. pastoris GS115 [46]. HNP1 was fused with polyhedrin from Bombyx mori (domestic silkworm) and enhanced GFP (eGFP) to indicate expression, enhance solubility, and reduce host toxicity. The fusion protein was successfully expressed and secreted at a yield of 33.8 mg/l. An enterokinase cleavage site was incorporated to allow for the release of the AMP following purification, and the recombinant HNP1 retained strong antimicrobial activity. In another study, the issues with host toxicity when expressing Lfcin in P. pastoris X-33 were resolved using a carrier protein approach [35]. Researchers hypothesized that the antibacterial efficacy of Lfcin was linked to its net positive charge, and fusing Lfcin with anionic antioxidant peptides PEP1 and PEP2 (derived from porcine myofibrillar protein) might therefore reduce inhibitory effects. Three fusions were expressed: PEP1-Lfcin, PEP2-lactoferricin, and PEP1/2-Lfcin. Of these, the PEP1/2-Lfcin fusion achieved the highest titre at 55.3 mg/l after six days of fermentation and successfully displayed reduced host toxicity. Human serum albumin (HSA) served as an effective carrier protein for the expression of AMP Apidaecin Ia when expressed in P. pastoris [68]. To enable purification, a His-tag and a TEV protease cleavage site were incorporated between apidaecin and HSA. This approach yielded over 418 mg/l of bioactive apidaecin with antimicrobial activity against E. coli. Additionally, a specialized lipid based fusion strategy that uses plant oleosin technology have been recently employed as an approach to improving AMP expression and downstream processing in P. pastoris [100]. The antifungal peptide PAF102 was fused to rice oleosin 18 (Ole18), which targets the fusion to lipid droplets allowing for high level accumulation which yielded 180 mg/L. The Ole18–PAF102 fusion enabled straightforward downstream extraction by flotation of lipid droplets on density gradients, and the recovered PAF102 retained full antifungal activity against pathogenic fungi. In contrast, the same oleosin‑fusion approach did not yield detectable accumulation of the alternative AMP Cecropin A, which authors speculate was due to peptide susceptibility to proteases. In a recent study, the flavin‑based fluorescent protein iLOV was employed as an N‑terminal fusion partner to the AMP NI01 in P. pastoris to enable fluorescence‑based identification of high‑yielding strains [101]. The iLOV–NI01 fusion was used to investigate the relationship between iLOV fluorescence and copy number, the effect of targeted versus random integration, and the impact of codon‑optimisation of the NI01 sequence. iLOV allowed for the selection of high‑expressing “jackpot” clones while incorporation of an enterokinase (rEK) cleavage site between iLOV and NI01 enabled recovery of the native AMP sequence, which retained antimicrobial activity. Mixed‑codon‑usage libraries paired with random integration delivered the highest‑expressing strains, with top clones showing up to a 30% increase in iLOV‑based expression over the benchmark strain. The system was extended to seven additional AMPs. Compared with GFP‑based reporters iLOV offers a smaller size (~ 13.5 kDa vs ~ 30 kDa) and a lower cellular burden, while still allowing visualization and purification and therefor might be a powerful tool for streamlining difficult‑to‑express AMPs.

Beyond tags and carrier proteins, AMPs can also be fused to themselves, or to other AMPs. These dimeric or chimeric AMPs have been shown to possess increased antimicrobial activity and improved target spectrum. Generating longer mRNAs and peptides can also result in more efficient transcription and translation and more stable peptides. AMP dimers fuse two heterologous (hybrid dimer) or homologous AMPs, usually with a linker. Chimeric AMPs, however, combine functional components from different AMPs typically comprising two or more distinct domains such as a targeting domain and a killing domain. These synthetic AMPs are usually rationally designed and chemically synthesized, but biological synthesis can be cheaper and achieve better functionality and scalability. A range of AMP dimers and chimeric AMPs have been successfully expressed in yeast (Table 7). Hybrid fusion dimer Lfcin-Lfa was successfully expressed in P. pastoris KM71 at 30 mg/l, the transformant supernatant showed improved bactericidal activity to either individual constituent subunit, meaning this case showed an acceptable titre of AMP with enhanced activity. In another case, a hybrid fusion dimer of human LL37 and indolicidin (LI) (called LIG), was expressed in P. pastoris with a flexible Gly–Ser–Gly (G–S–G) linker and N-terminal His and FLAG tags [88], yielding 5.9 mg/l. This maintained similar antimicrobial efficacy compared to the monomeric LI, demonstrating the use of linkers to separate the two fused AMPs can yield active AMP. A recent study optimized the length of a flexible linker sequence between a chimeric fusion of the AMP PR‑FO, composed of porcine PMAP‑36 and avian Fowlicidin‑2 variants, and a hyperstable four‑helix bundle domain (DAMP4), which functions as a solubility‑enhancing carrier in P. pastoris X‑33 [108]. Total recombinant protein yields were similar across the three linker variants, ranging from 248.5 to 259.6 mg/L, but the longer‑linker construct D2L promoted proper domain folding, higher α‑helical content in hydrophobic environments, and stronger antibacterial activity than both the shorter‑linker variants and the parent PR‑FO. The authors also developed a selective thermal‑precipitation‑based, chromatography‑free purification process for D2L, which yielded 79.02 mg/L of recombinant peptide with 85.19% purity, demonstrating that rationally designed linker‑separated DMAP4–PR‑FO dimers can combine enhanced activity with scalable, chromatography‑free purification for industrial‑scale applications. There have been cases where chimeras also lose activity [109]. The hybrid chimera PBD-2::PoIFNgamma, consisting of porcine beta-defensin-2 (PBD-2) and porcine interferon-gamma (PoIFNgamma) was successfully expressed in P. pastoris X-33, but did not exhibit any antibacterial activity [83]. Significant antiviral activity was observed for PoIFNgamma alone, suggesting that improper folding of the fusion protein may have impaired the antimicrobial function of PBD-2-PoIFNgamma. Interestingly, a chimeric AMP was able to provide novel functionality; PG-1 is naturally activated via cleavage with a neutrophil elastase into the cathelin and PG-1 domains. This was exploited for controlled activation of the AMP in the intestines, where the AMPs activity was desired. The researchers replaced the native neutrophil elastase cleavage site with an enterokinase cleavage site. This meant PG-1 would remain inactive until cleaved by enterokinase in the intestinal duodenum, reducing off target effects and avoiding host toxicity. The approach yielded 104 μg/ml of recombinant PG-1 after 24 h in fermentation culture medium and produced an AMP with the novel function of intestinally-targeted activation. This reduces the likelihood of adverse side effects, negative impacts on commensal gut microflora, and demonstrates the creative ways chimeric AMPs can be applied to make smarter drugs. These examples illustrate that diverse fusion-protein strategies can be deployed to enhance AMP solubility, stability, purification or an optimisation screening approach although the success of each approach is strongly AMP-dependent and must be evaluated empirically.

Table 7.

Tandem repeat and hybrid dimer/chimer approaches used for heterologous AMP expression in yeast hosts

Type	Regularity of employment	Example peptides	Typical host(s)	References
Tandem repeat	Few reports	Lunasin, Plectasin, TP1, LL37	P. pastoris	[61, 67, 87, 89]
Hybrid dimer	Multiple reports	ABP-dHC-cecropin A, LFA-LFC, hLYZ-TP I, EntA-EnHF, EntA-EntCRL55, Mag II-CB, PI-hLY, EntP-SakA, LFA-LFC-6His	P.pastoris, K. lactis, C. militaris	[79, 94, 102, 109–112]
Hybrid chimer	Multiple reports	LI (LL37:Indolicidin), CA::MA, EntA::EntCRL55, EntA::EntHR, EntP::EntHR, Cecropin A::thanatin hybrid, DAMP4-PR::FO	P. pastoris	[56, 92, 102, 108, 111]

Conclusion

With their unique modes of action and inherent customizability, AMPs hold significant promise for addressing the global challenge of antimicrobial resistance. However, traditional production methods—such as direct extraction from natural sources and chemical synthesis—are often inefficient, costly, and environmentally unsustainable. In contrast, biological production via heterologous expression offers a more scalable and sustainable alternative. This review highlights the importance of tailoring expression systems to the specific characteristics of individual AMPs, with factors such as sequence complexity, PTM requirements, and intended application guiding host selection and optimization.

Bacterial systems are well suited for simple AMPs that do not require PTMs, whereas yeasts, particularly P. pastoris, are more appropriate for complex peptides or when mitigating host toxicity is necessary. Yeasts offer several advantages, including efficient secretion, scalability, and the ability to perform many eukaryotic PTMs, making them highly attractive for industrial-scale AMP production. Numerous engineering strategies can further enhance production and functionality, including codon optimization, multicopy gene integration, fusion protein design, tandem repeats, and chimeric constructs. Signal peptide engineering can improve secretion efficiency, while protease gene deletions may reduce peptide degradation. As synthetic biology and bioengineering continue to advance, yeast-based systems are poised to enable the development of next-generation AMPs with improved activity, reduced off-target effects, and novel functionalities.

Despite these advances, several knowledge gaps and technical bottlenecks remain. Systematic, quantitative comparisons of core expression parameters—such as codon optimization strategies, gene copy number, promoter architecture, and strain selection—are still limited, making it difficult to establish robust, generalizable design principles. Future work should prioritize controlled, head-to-head comparisons to generate AMP-specific optimization frameworks. Protease-mediated degradation remains a major challenge in yeast, particularly given the continued reliance on strains such as X-33; evaluating protease-deficient strains or implementing rational peptide design strategies to enhance stability could help address this issue. While fusion protein approaches can improve stability and secretion, they often introduce additional downstream processing complexity that may hinder scalability. Accordingly, strategies such as self-cleaving tags, tag-free expression systems, and chromatography-free purification workflows should be explored to streamline production. Additionally, current expression systems are heavily dominated by a narrow set of AOX1-based vectors, limiting flexibility and optimization potential. The integration of high-throughput screening approaches—particularly fluorescence-based systems—offers a powerful avenue to rapidly evaluate large libraries of constructs, signal peptides, and strains. These platforms enable efficient identification of high-producing variants and facilitate systematic optimization of expression parameters at scale, accelerating the development of robust production systems.

By addressing these challenges, yeast expression platforms are well positioned to deliver next-generation AMPs that are highly active, specific, and industrially scalable, enabling transformative applications spanning healthcare, biotechnology, and agriculture.

Author contributions

AD conceived, designed, and prepared the initial manuscript draft and figures, and revised all manuscript drafts. SMMP revised and edited manuscript drafts. SAK conceived and supervised the initial draft development, and critically reviewed and revised all manuscript drafts. All authors read, revised and approved the final manuscript.

Funding

This work was supported by the New Zealand Ag Emissions Centre Science Program.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.