Overcoming prokaryotic toxicity: a SUMO-fused secretory platform in Komagataella phaffii for high-yield production of phage holin Hol41

Abstract

Background

Phage holin Hol41 is a membrane protein with broad-spectrum antibacterial potential; however, its high hydrophobicity and prokaryotic expression toxicity severely limit its soluble expression and large-scale production.

Results

We constructed a synthetic biology-driven secretory expression platform for Hol41 in Komagataella phaffii (K. phaffii) by integrating small ubiquitin-related modifier (SUMO) fusion technology and the α-mating factor secretion pathway. This platform avoids prokaryotic toxicity by exploiting the structural differences between the cell membranes of K. phaffii and Escherichia coli. The SUMO tag combined with α-mating factor-mediated secretion enhances protein solubility, reduces toxicity, and inhibits Hol41 aggregation. After scale-up culture in a 2-L fermenter, the yield of SUMO-Hol41 reached 44.95 mg/L, with a yield coefficient (Y_p/s) of 0.00843 mg protein/mg dry cell weight (DCW). MIC and MBC assays revealed that SUMO-Hol41 exhibited selective antibacterial activity against Gram-positive bacteria among seven standard strains: the MIC values against Staphylococcus aureus ATCC 33,951 and Streptococcus pneumoniae ATCC 49,619 were both 128 µg/mL (MBC: 256 µg/mL), and 64 µg/mL (MBC: 128 µg/mL) for Enterococcus faecalis ATCC 29,212. Among Gram-negative bacteria, only Salmonella enteritidis S4 was susceptible (MIC: 128 µg/mL, MBC: 256 µg/mL). Moreover, the purified protein showed dose- and time-dependent antibacterial activity, whose inhibitory effect increased with elevated dosage and prolonged incubation time.

Conclusions

This study represents the first successful high-yield functional expression of phage holin Hol41 in K. phaffii. The established K. phaffii-SUMO fusion-secretion strategy effectively overcomes challenges of prokaryotic toxicity and poor solubility, providing a robust platform for developing novel antibacterial agents against drug-resistant bacteria.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12934-026-02983-2.

Introduction

The global antimicrobial resistance (AMR) crisis underscores an urgent need for novel antibacterial agents with distinct mechanisms of action. Bacteriophage-encoded holins, a class of transmembrane proteins capable of forming pores in bacterial membranes and inducing osmotic lysis, represent promising antimicrobial candidates owing to their membrane-targeting mechanism and low potential for resistance development [1, 2].

Holin function is closely associated with conserved structural features. Although holin genes exhibit low sequence similarity across different sources, the spatial organization and charge distribution of their transmembrane domains (TMDs) reflect evolutionary functional conservation [3]. Typical holins contain at least one hydrophobic TMD and a C-terminal domain enriched with charged residues [4].Based on the number and topology of TMDs, holins are classified into three categories: Class I, with three TMDs adopting an N-out/C-in orientation (Salmonella phage SMP); Class II, with two TMDs in an N-in/C-in orientation (phage φ21 S2168); and Class III, with a single TMD and an N-in/C-out orientation (T4 phage holin) [5–8]. Following membrane insertion, holin molecules oligomerize to form pores—a structural process that determines pore size and ion selectivity [9, 10]. The highly charged C-terminal domain not only serves as a site for anti-holin regulation but also precisely controls the timing of pore activation [11]. In phage genomes, holin genes are typically located adjacent to endolysin genes, forming a coordinated “lysis module” that mediates efficient host cell lysis [12].

Despite their therapeutic potential, the industrial application of holins remains constrained by challenges in achieving efficient soluble production. Owing to the high hydrophobicity and inherent membrane toxicity in prokaryotic hosts, conventional prokaryotic expression systems often yield low expression levels, exhibit poor expressibility, and produce poorly soluble products, posing major obstacles to scalable manufacturing. Previous studies have confirmed that induction of holin expression in prokaryotic systems leads to rapid inhibition of bacterial cell growth [4, 5]. For instance, expression of the DLP12 prophage holin HolinGFP in Escherichia coli(E. coli) AG1 strains strongly inhibited host growth (OD₆₀₀ stagnated at 0.754 within 5 h post-induction), and the protein frequently formed insoluble aggregates during purification due to its pronounced hydrophobicity [13]. Current strategies to improve solubility include fusion with chaperone tags (SUMO, GST) to assist folding and stability, or the use of detergents (DDM, Triton X-100) for solubilization [5]. Although these methods have facilitated the topological study of holins, they have not adequately addressed the issue of low expression yield [14]. Existing holin production processes generally suffer from low expression levels and require laborious purification steps—including ultrasonication, detergent extraction, and multiple chromatography steps—which hinder large-scale application [15]. To address these challenges, this study developed a SUMO-fused secretory expression platform in Komagataella phaffii (K. phaffii), integrating a tripartite synergistic strategy: First, differences in membrane lipid composition between eukaryotic and prokaryotic cells were exploited to mitigate toxicity. The membrane of K. phaffii is rich in ergosterol and long-chain unsaturated fatty acids, which enhance membrane rigidity and modify the lipid microenvironment, thereby inhibiting efficient insertion of SUMO-Hol41 into the membrane [16–19]. Coupled with the tightly regulated methanol-inducible AOX1 promoter, this system allows precise spatiotemporal control of toxic protein expression [20]. Second, SUMO fusion can significantly improve the solubility of target proteins and mediate efficient secretory expression in the host. In this study, an 11 kDa small ubiquitin-like modifier (SUMO) tag was fused to the N-terminus of Hol41. By virtue of its hydrophilic surface and chaperone-like activity, the SUMO tag neutralizes hydrophobic interactions among transmembrane domains, inhibits protein aggregation, and thereby promotes the correct folding of Hol41 and enhances its secretory efficiency [21–23]. The antimicrobial peptide CBF was expressed as a SUMO-fusion protein in Bacillus subtilis WB800N, and 3 mg/L of endotoxin-free bioactive product was obtained after purification, which confirmed the feasibility of the SUMO tag mediating efficient secretory expression in this host [24] Third, a secretion mechanism enables spatial compartmentalization. Directed by the α-mating factor signal peptide (encoded on the pPIC9K vector), the SUMO-Hol41 fusion protein is secreted extracellularly via the endoplasmic reticulum–Golgi (ER–Golgi) pathway. This physically separates the toxic protein from the host’s internal membrane systems, leading to substantially higher cell viability compared to that in E. coli. Furthermore, the secreted fusion protein present in the supernatant can be directly purified using Ni-NTA affinity chromatography, greatly simplifying downstream processing. The combination of these strategies, together with the high-cell-density fermentation capacity of K. phaffii, enabled high-yield soluble production of Hol41. Functional assays confirmed that the purified SUMO-Hol41 fusion protein retains potent antibacterial activity. Further analysis of its antibacterial spectrum and mechanism of action underscored the biological efficacy of this strategy.

This platform successfully achieves high-yield soluble production of Hol41, where the yield of SUMO-Hol41 reached 44.95 mg/L, with a Y_p/s of 0.00843 mg protein/mg DCW, overcoming the key limitations of prokaryotic expression. Its innovative aspects include: circumvention of membrane toxicity through eukaryotic lipid differences, enhancement of solubility via SUMO chaperone activity, isolation of toxicity through secretory expression, and scalability through high-density fermentation and the strong AOX1 promoter. This approach offers a universal technical framework for the soluble production of hydrophobic transmembrane proteins (such as pore-forming toxins and ion channels) and supports the development of novel antimicrobials against drug-resistant pathogens. Future modular optimizations—such as signal peptide engineering and screening of SUMO variants—could further broaden the applicability of this platform.

Materials and methods

Strains, plasmids, and culture conditions

Bacterial strains, phages and plasmids used in this study are listed in Table 1. The SUMO tag was the codon-optimized mature Smt3 sequence derived from Saccharomyces cerevisiae (GenBank accession No.: NM_001181430). The plasmid containing this sequence was synthesized by Tsingke Biotechnology Co, Ltd. (Tianjin, China). Bacterial strains were cultured in LB medium at 37 °C; recombinant E. coli was induced at 28 °C. K. phaffii GS115 and engineered strains were grown at 28 °C with shaking at 220 rpm in BMGY (1% glycerol) and induced in BMMY (1% methanol).

Table 1.

Bacterial strains, phages and plasmids used in this study

Bacterial strains, phage, plasmids, primers	Characteristics, function, or sequence	Source
Strains
Salmonella enteritidis S4	Host for phage SP_4	Our laboratory collection
Pseudomonas aeruginosa ATCC 15,692	Antibacterial spectrum determination
E. coli ATCC 25,922	Antibacterial spectrum determination
Klebsiella pneumoniae 911	Antibacterial spectrum determination
Staphylococcus aureus ATCC 33,951	Antibacterial spectrum determination
Streptococcus pneumoniae ATCC 49,619	Antibacterial spectrum determination
Enterococcus faecalis ATCC 29,212	Antibacterial spectrum determination
E. coli DH5α	For vector plasmid cloning	Shanghai Weidi Biotechnology Co., Ltd.
E. coli BL21(DE3)	For protein expression
K. phaffii GS115	For protein expression
Phage
Phage SP_4	Isolated from Salmonella enteritidis S4;Source of the Hol41 gene	Our laboratory collection
Plasmids
pET28a(+)	Prokaryotic expression; Vector plasmid	Tianjin Tsingke Biotechnology Co., Ltd.
pET28a(+)-SUMO	Prokaryotic expression; Vector plasmid
pPIC9K	Eukaryotic expression; Vector plasmid
pPIC9K-SUMO	Eukaryotic expression; Vector plasmid

Amino acid sequence analysis of Hol41

Open reading frames (ORFs) of phage SP_4 (OR001929) were predicted using GeneMark.hmm v4.28, and a putative holin gene (Hol41) adjacent to the endolysin gene (ORF40) was identified. Transmembrane domains (TMDs) were predicted with TMHMM Server v.2.0 [25]. Homology searches were performed using BLASTP against the NCBI nr database. The molecular weight and theoretical isoelectric point (pI) of Hol41 were calculated using ExPASy ProtParam [26]. Hydrophobicity analysis was conducted via ProtScale [27]. The tertiary structure of Hol41 was modeled using SWISS-MODEL with A0A718MXZ2.1 (class I holin, Salmonella typhimurium SL1344) as the template. The structure of SUMO-Hol41 was predicted by AlphaFold 3, with A0A0L8VUM0.1 as the SUMO domain template [28, 29]. 148–151

Construction of the Hol41 expression system

The gene cassette structure of the recombinant plasmids was designed as follows: for the SUMO-fused plasmids (pET28a(+)-SUMO-Hol41 and pPIC9K-SUMO-Hol41), the structure was vector-borne SUMO tag → Hol41 gene → 6×His tag; for the non-fused plasmids (pET28a(+)-Hol41 and pPIC9K-Hol41), the structure was vector-borne Hol41 gene → 6×His tag. The Hol41 gene was amplified via PCR using phage SP_4 genomic DNA as the template. The upstream primer used was Hol41-F (containing a BamHI restriction enzyme site; sequence: 5′-CCGGAATTCATGCTCAACCCACAAATCG − 3′), and the downstream primer was Hol41-R (containing a NotI restriction enzyme site; sequence: 5′-ATTTGCGGCCGCGTGCTGATGTTTCGGTT − 3′).

The PCR product and target vectors (pET28a(+)-SUMO, pPIC9K-SUMO, pET28a(+), pPIC9K) were double-digested with BamH I (Thermo Fisher Scientific Inc., USA; #FD0054) and Not I (Thermo Fisher Scientific Inc., USA; #FD0595), then ligated using T4 DNA ligase (Thermo Fisher Scientific Inc., USA; #EL0011) to generate recombinant plasmids (pET28a(+)-SUMO-Hol41, pET28a(+)-Hol41, pPIC9K-SUMO-Hol41, pPIC9K-Hol41). After transformation into E. coli DH5α, positive clones were screened by colony PCR and verified by DNA sequencing with using universal T7/AOXⅠ primers.

For the E. coli expression system, sequence-verified pET28a(+)-SUMO-Hol41 and pET28a(+)-Hol41 were transformed into E. coli BL21(DE3) competent cells by the heat shock method. Transformants were screened on LB agar supplemented with 50 µg/mL kanamycin, generating the recombinant strains BL21(DE3)/pET28a(+)-SUMO-Hol41 and BL21(DE3)/pET28a(+)-Hol41.

For the K. phaffii expression system, sequence-verified pPIC9K-SUMO-Hol41 and pPIC9K-Hol41 were linearized with SacI. Then 10 µg of linearized plasmid was mixed with 100 µL K. phaffii GS115 electrocompetent cells (Shanghai Weidi Biotechnology Co., Ltd., China, #PE1001) and transformed by electroporation (1.5 kV, 25 µF, 200 Ω). Transformants were selected on YPD agar plates supplemented with 500 µg/mL G418 (Geneticin, Seville Biotech Co., Ltd., China; #GC301011) at 28 °C for 3–5 days, yielding the engineered strains GS115/pPIC9K-SUMO-Hol41 and GS115/pPIC9K-Hol41.

Effect of Hol41 expression on host cell growth

To evaluate the effect of Hol41 expression on host growth, engineered E. coli and K. phaffii strains were tested as follows:

E. coli BL21(DE3)/pET28a(+)-SUMO-Hol41 (SUMO-fused), BL21(DE3)/pET28a(+)-Hol41 (non-fused), and the empty plasmid control BL21(DE3)/pET28a(+) were inoculated into 100 mL LB medium (50 µg/mL kanamycin) and cultured at 37 °C, 180 rpm to OD₆₀₀ = 0.5. Expression was induced with 0.8 mM IPTG, and OD₆₀₀ was measured every 30 min for 8 h post-induction to monitor growth.

K. phaffii GS115/pPIC9K-SUMO-Hol41 (SUMO-fused), GS115/pPIC9K-Hol41 (non-fused), and the empty plasmid control GS115/pPIC9K were inoculated into 100 mL BMGY (1% glycerol) and cultured at 28 °C, 220 rpm to OD₆₀₀ = 2.0. Cells were harvested by centrifugation (2500 × g, 5 min, 4 °C), washed twice with sterile saline, and resuspended in 100 mL BMMY (1% methanol, v/v) to OD₆₀₀ = 1.0. Cultures were incubated for 72 h at 28 °C, 220 rpm (1% methanol supplemented every 24 h; optimal induction conditions are shown in Suppl. Fig. S1), and OD₆₀₀ was measured every 4 h to assess growth.

Effect of SUMO fusion on the solubility of Hol41

To evaluate the effect of the SUMO tag on Hol41 solubility, K. phaffii GS115/pPIC9K-SUMO-Hol41 (SUMO-fused) and GS115/pPIC9K-Hol41 (non-fused) were induced at 28 °C with shaking at 220 rpm. After induction, one portion of the non-fused culture was directly centrifuged (10,000×g, 5 min, 4 °C) to collect the supernatant, while another portion was treated with 1% (v/v) Triton X-100 at 4 °C for 1 h prior to centrifugation [30]. The SUMO-fused culture was centrifuged under identical conditions. Western blotting was performed as follows: an aliquot of the supernatant containing the target protein was loaded per lane, separated by 12% SDS-PAGE, and transferred onto a 0.22 μm PVDF membrane via semi-dry transfer (100 mA, 90 min). The membrane was blocked with TBST containing 5% non-fat milk, then incubated overnight at 4 °C with primary mouse anti-His antibody (1:5000,Thermo Fisher Scientific, Cat. No. MA1-21315). After washing with TBST, the membrane was incubated with HRP-conjugated goat anti-mouse secondary antibody (1:10000,Thermo Fisher Scientific, Cat. No. MA1-10371) for 1 h at room temperature. Signals were detected by enhanced chemiluminescence (ECL) using a ChemiDoc imaging system, to clarify the influence of SUMO fusion on Hol41 solubility.

Expression and purification of SUMO-Hol41 fusion protein in K. phaffii

SUMO-Hol41 fusion protein was expressed in a 2 L fermenter (#MF-101, Shenzhen Siemen Technology Co., Ltd., China) with three biological replicates. Engineered K. phaffii GS115/pPIC9K-SUMO-Hol41 was inoculated into 250 mL BMGY (1% glycerol, w/v) in a 500-mL flask and cultured at 28 °C, 220 rpm for 24 h to prepare seed culture. Cells were harvested by centrifugation at 2500×g, 4 °C for 5 min, washed twice with sterile saline to remove residual glycerol, then resuspended in 2 L BMMY medium in a 5-L fermenter. Fermentation was performed at 28 °C, 300 rpm, 2.0 vvm aeration, and pH 6.0 (adjusted with 25% v/v ammonia water). Methanol concentration was maintained at 0.5% (v/v) for the first 6 h, then increased to 1.0% (v/v) for a further 72 h with real-time monitoring and supplementation (detailed parameters in Supplementary Table S1). After induction, the culture was centrifuged at 10,000×g, 4 °C for 5 min, and the supernatant was loaded onto a 5 mL Ni‑NTA column (Beijing Solarbio Sci. & Tech. Co., Ltd., Beijing, China; Cat. #P2012). The column was equilibrated with 5 column volumes (CV) of buffer (20 mM Tris‑HCl, 500 mM NaCl, pH 8.0) containing 20 mM imidazole, and target protein was eluted with the same buffer supplemented with 200 mM imidazole. Protein concentration was determined with a BCA Kit (Beijing Solarbio Sci. & Tech. Co., Ltd., Beijing, China; #PC0020). The fitting formula of the BSA standard curve is as follows:

where Y is absorbance at 562 nm (A₅₆₂) and X is BSA concentration (mg/mL). Dry cell weight (DCW) was measured in triplicate by gravimetry: 10 mL fermentation broth was centrifuged at 8000×g, 4 °C for 10 min; the pellet was washed twice, dried to constant weight at 70 °C in a pre-weighed crucible, and DCW was calculated. Total purified protein mass and yield per liter of fermentation broth were determined from eluate volume, protein concentration, and total fermentation volume. The identity of purified SUMO-Hol41 was confirmed by SDS-PAGE and Western blotting. The protein yield coefficient Y_P/S (mg protein per mg DCW) was calculated as:

where m_protein is the mass of purified protein (mg), and m_DCW is the corresponding dry cell weight (mg).

Effect of SUMO tag cleavage on the solubility of Hol41

To evaluate the effect of SUMO tag removal on Hol41 solubility, purified SUMO-Hol41 was digested with SUMO protease (Thermo Fisher Scientific Inc., USA; #12588018). The 200 µL reaction mixture contained 20 µL of 1× SUMO protease buffer, 10 U of SUMO protease, and 20 µg of SUMO-Hol41, with the final volume adjusted using double-distilled water (ddH₂O). After incubation at 30 °C for 6 h, the mixture was centrifuged at 10,000 × g for 1 min at 4 °C. Aliquots of 20 µL from the supernatant and precipitate were each mixed with 5× protein loading buffer. The cleavage products were analyzed by Western blot to assess changes in Hol41 solubility.

Determination of MIC, MBC and dose- and time-dependent antibacterial activity of SUMO-Hol41 against tested bacteria

The MIC and MBC of SUMO-Hol41 against seven common pathogenic bacteria were determined by broth microdilution following CLSI M07-Ed12 guidelines. Strains included Gram-positive (Staphylococcus aureus ATCC 33951, Streptococcus pneumoniae ATCC 49619, Enterococcus faecalis ATCC 29212) and Gram-negative (Pseudomonas aeruginosa ATCC 15692, Escherichia coli ATCC 25922, Salmonella enteritidis S4, Klebsiella pneumoniae 911) bacteria; ATCC strains ensured reproducibility, and clinical isolates reflected clinical efficacy to comprehensively evaluate its antibacterial spectrum.All strains were cultured to logarithmic phase (OD₆₀₀ = 0.6–0.8), adjusted to 0.5 McFarland standard (1–2 × 10⁸ CFU/mL) with sterile saline, and 1000-fold diluted to 5 × 10⁵ CFU/mL. SUMO-Hol41 stock solution was diluted to 1024 µg/mL with sterile saline, filter-sterilized through a 0.22 μm membrane, and stored at 4℃. MIC and MBC were determined via broth microdilution: MH broth and protein stock were added to the first column of a 96-well plate, followed by two-fold serial dilutions to 0.125–512 µg/mL. Except for blank control (only MH broth), 100 µL bacterial suspension was added to each well, with parallel bacterial growth control and ciprofloxacin positive control (Sigma-Aldrich, USA; #17850). The plate was incubated at 35℃ for 16 h; MIC was the lowest concentration without visible growth. Bacterial cultures from MIC and higher concentrations were spread on LB agar, and MBC was the lowest concentration with ≤ 5 CFU. No inhibition at the highest concentration was recorded as > ND. For dose- and time-dependent activity, strains were co-incubated with concentrations ≥ MIC for 0, 2, 4, 8, 12, 16 h; OD₆₀₀ was measured, and inhibition rate was calculated by the following formula. All experiments had three biological replicates. Formula for inhibition rate:

Where OD_control is the OD₆₀₀ value of the bacterial growth control (without SUMO-Hol41) and OD_treated is the OD₆₀₀ value of the sample treated with SUMO-Hol41.

Statistical analysis

All experiments were performed in triplicate. Data are expressed as mean ± standard deviation (SD). Statistical analysis was conducted using GraphPad Prism 6.0: differences between two groups were analyzed by Student’s t-test, and multiple comparisons were evaluated by one-way ANOVA. A P value ≤ 0.05 was considered statistically significant.

Results

Bioinformatic characteristics of Hol41 protein

ORF prediction of the phage SP_4 genome revealed that ORF41 (GenBank accession No.: WKK67931.1), located upstream of the endolysin gene (ORF40), contains three transmembrane domains (TMDs) (Fig. S2). ORF41 spans 291 bp, and its amino acid sequence shows high homology with known holins (Table 2). Based on the number of TMDs and sequence similarity, it was designated Hol41, a class I holin of phage SP_4. ProtParam analysis showed that Hol41 has a molecular weight of 10.7 kDa, a theoretical isoelectric point of 9.95, and an instability index of 41.42 (conformationally unstable). Its C-terminal hydrophilic region is enriched in positively charged arginine (Arg)/lysine (Lys) residues, accounting for approximately 32% of the 30 C-terminal amino acids. Hydrophobicity analysis yielded values ranging from − 4.5 to 4.5, displaying the typical characteristics of holins with distinct hydrophobic and hydrophilic regions. The predicted tertiary structures of both Hol41 and SUMO-Hol41 contain three α-helices (Fig. S3).

Table 2.

Amino acid sequence similarity comparison between Hol41 and known holin proteins

Protein	Accession number	Source	Identity (%)
Hol41	WKK67931.1	S. enteritidis phage SP_4	100.00
Putative holin	YP_010747087.1	Salmonella phage vB_SenS-EnJE6	97.92
Holin	YP_009280145.1	Salmonella phage MA12	96.84
Holin	YP_005098117.1	Salmonella phage SE2	95.83
Holin	YP_009009974.1	Salmonella phage vB_SenS-Ent2	95.79
Putative holin	YP_010746920.1	Salmonella phage JD01	94.79
Holin	YP_001110822.1	Salmonella phage SETP3	94.74
Putative holin	YP_010745833.1	Salmonella phage horsemountain	94.74

Construction of the Hol41 expression system

PCR amplification of the Hol41 gene yielded a 291 bp product on agarose gel electrophoresis, which was consistent with the expected size (Fig. 1A). After constructing the recombinant plasmids pET28a(+)-SUMO-Hol41, pET28a(+)-Hol41, pPIC9K-SUMO-Hol41, and pPIC9K-Hol41 and transforming them into the corresponding expression strains, colony PCR specifically amplified the 291 bp target fragment of Hol41 (Fig. 1B). DNA sequencing confirmed 100% identity between the cloned Hol41 gene and the expected sequence (Fig. S4), demonstrating that all recombinant expression systems were successfully established and harbored the target recombinant plasmids (Fig. 1C, D, E, F).

Fig. 1.

Construction and validation of recombinant plasmids for Hol41 and SUMO-Hol41 expression. A Agarose gel electrophoresis of PCR-amplified Hol41 gene. M: DNA marke. B Restriction digestion validation of recombinant plasmids. M: DNA marker. C Map of prokaryotic expression plasmid pET28a(+)-Hol41. Key elements include: T7 promoter, lac operator, ribosome binding site (RBS), C-terminal 6×His tag, T7 terminator, and KanR (kanamycin resistance gene). D Map of prokaryotic expression plasmid pET28a(+)-SUMO-Hol41. The plasmid carries an N-terminal SUMO fusion tag upstream of Hol41, with other core elements identical to pET28a(+)-Hol41. E Map of eukaryotic expression plasmid pPIC9K-Hol41. Key elements include: AOX1 promoter, α-mating factor (α-MF) signal peptide for secretion, C-terminal 6×His tag, AOX1 terminator, and AmpR (ampicillin resistance gene). F Map of eukaryotic expression plasmid pPIC9K-SUMO-Hol41. The plasmid carries an N-terminal SUMO fusion tag upstream of Hol41, with other core elements identical to pPIC9K-Hol41

Effect of SUMO-Hol41 fusion protein expression on the growth of different host strains

For the prokaryotic host E. coli, following IPTG induction, the OD₆₀₀ values of both BL21(DE3)/pET28a(+)-SUMO-Hol41 and BL21(DE3)/pET28a(+)-Hol41 were statistically significantly lower than those of their respective uninduced controls. Meanwhile, the empty vector control E. coli BL21(DE3)/pET28a(+) showed similar growth profiles between induced and uninduced groups with no obvious growth attenuation, further supporting that the observed growth inhibition was specifically caused by Hol41 or SUMO-Hol41 expression (Fig. 2A, B). This result indicates that the expression of either the SUMO-Hol41 fusion protein or the Hol41 protein exerted an inhibitory effect on the growth of the E. coli host cells.

Fig. 2.

Growth Curves of Recombinant E. coli and K. pastoris Strains After Induction. A E. coli BL21(DE3)/pET28a(+)-SUMO-Hol41; B E. coli BL21(DE3)/pET28a(+)-Hol41; C K. phaffii GS115/pPIC9K-SUMO-Hol41; DK. phaffii GS115/pPIC9K-Hol41. Data represent mean ± SD (n = 3)

In contrast, for the eukaryotic host K. phaffii, following methanol induction, the OD₆₀₀ values of GS115/pPIC9K-SUMO-Hol41 and GS115/pPIC9K-Hol41 exhibited only a slight reduction compared to their uninduced controls, with no statistically significant growth inhibition observed. Correspondingly, the empty vector control K. phaffii GS115/pPIC9K displayed comparable growth between induced and uninduced cultures without growth decline, which further corroborated the above conclusion (Fig. 2C, D). Notably, the OD₆₀₀ of the respective cultures continued to increase throughout the induction period, demonstrating that the expression of the SUMO-Hol41 fusion protein or the Hol41 protein did not cause marked suppression of K. phaffii host growth.

Effect of the SUMO tag on the soluble expression of Hol41 in K. phaffii

The soluble expression of the target protein in recombinant K. phaffii strains after induction was analyzed by Western blotting. For the non-fused expression strain GS115/pPIC9K-Hol41, no target Hol41 protein was detected in the culture supernatant after induction. However, after treatment of the fermentation broth with 1% Triton X-100, multiple Hol41-specific bands of different molecular weights appeared in the supernatant, confirming that non-fused Hol41 existed as insoluble intracellular aggregates(Fig. 3A).

Fig. 3.

Verification of SUMO-Hol41 fusion protein and Hol41. A Western blot analysis of non-fused Hol41: Lane 1, induced culture supernatant; Lane 2, induced supernatant treated with 1% Triton X-100. B Western blot analysis of SUMO-Hol41 fusion protein: Lane 3, uninduced culture supernatant; Lane 4, induced culture supernatant. C SDS-PAGE analysis of purified SUMO-Hol41 fusion protein. E Western blot analysis of SUMO-Hol41 fusion protein cleaved by SUMO protease: Lane 5, supernatant of SUMO-Hol41 sample treated with SUMO protease; Lane 6, precipitate of SUMO-Hol41 sample treated with SUMO protease. A 10–180 kDa protein marker was used for all SDS-PAGE and Western blot assays, labeled as Lane M in all panels

In contrast, the SUMO-fused expression strain GS115/pPIC9K-SUMO-Hol41 displayed a clear specific band of approximately 21 kDa in the post-induction supernatant, which was consistent with the predicted molecular weight of SUMO-Hol41 (Fig. 3B). These results demonstrated that the SUMO tag effectively prevented Hol41 aggregation and thus mediated its soluble secretory expression in K. phaffii.

Scalable production and purification of the SUMO-Hol41 fusion protein in K. phaffii

After fed-batch fermentation for recombinant protein production, the His-tagged SUMO-Hol41 fusion protein was purified from the fermentation supernatant using a His-tagged protein purification kit. For three independent fermentation batches, the concentration of the purified SUMO-Hol41 fusion protein (10 mL/batch) was determined as 9.24 mg/mL, 9.01 mg/mL, and 8.71 mg/mL, with a mean concentration of 8.99 ± 0.27 mg/mL and a CV of 3%. The DCW of K. phaffii at the end of fermentation was determined in three parallel replicates, yielding values of 5.5 g/L, 5.4 g/L, and 5.1 g/L with an average of 5.33 g/L. Based on the parameters that 2 L of fermentation broth yielded 10 mL of purified eluate with a mean concentration of 8.99 mg/mL, the total mass of purified SUMO-Hol41 was calculated to be 89.9 mg, corresponding to a purification yield of 44.95 mg per liter of fermentation broth, and the yield coefficient (Y_p/s) was 0.00843 mg protein/mg DCW. SDS-PAGE and Western blot analyses revealed a specific band for the purified protein in the range of 15–25 kDa, which was consistent with the expected molecular weight of the SUMO-Hol41 fusion protein (~ 21 kDa) (Fig. 3C, D).

Effect of SUMO tag cleavage on the solubility of Hol41

SUMO-Hol41 was treated with SUMO-specific protease to remove the SUMO tag. After centrifugation to separate soluble supernatant and insoluble precipitate, Western blotting of both fractions showed only faint free Hol41 bands in the supernatant, whereas multiple prominent bands were observed in the precipitate. Notably, these precipitated bands did not match the predicted molecular weight of monomeric Hol41 (~ 11 kDa, post-tag cleavage) (Fig. 3E). These results indicate that free Hol41, released by SUMO protease cleavage, rapidly formed high-molecular-weight insoluble multimers, which were mainly in the precipitate. This confirms the SUMO tag is essential for maintaining Hol41 solubility in the K. phaffii expression system.

Determination of MIC, MBC and dose- and time-dependent antibacterial activity of SUMO-Hol41 against tested bacteria

The in vitro antibacterial activity of SUMO-Hol41 was evaluated by determining the minimum inhibitory concentration (MIC) and minimum bactericidal concentration (MBC) against seven standard bacterial strains, with ciprofloxacin as the positive control (Table 3). SUMO-Hol41 exhibited moderate and selective antibacterial activity against Gram-positive bacteria: the MIC values against Staphylococcus aureus ATCC 33,951 and Streptococcus pneumoniae ATCC 49,619 were both 128 µg/mL (MBC: 256 µg/mL), while stronger activity was observed against Enterococcus faecalis ATCC 29,212 (MIC: 64 µg/mL, MBC: 128 µg/mL). SUMO-Hol41 showed weak inhibitory effects on most Gram-negative bacteria. No bacteriostatic or bactericidal activity was detected againstPseudomonas aeruginosa ATCC 15,692, Escherichia coli ATCC 25,922, or Klebsiella pneumoniae 911 at the maximum tested concentration. Only Salmonella enteritidis S4 was susceptible to SUMO-Hol41, with an MIC of 128 µg/mL and MBC of 256 µg/mL. As a positive control, ciprofloxacin displayed potent broad-spectrum antibacterial activity against all tested strains (MIC: 0.125–0.5 µg/mL, MBC: 0.25–2 µg/mL), verifying the reliability of the experimental system.

Table 3.

MIC and MBC of SUMO-Hol41 and ciprofloxacin against tested bacterial strains

Strains	SUMO-Hol41 MIC (µg/mL)	SUMO-Hol41 MBC (µg/mL)	Ciprofloxacin MIC (µg/mL)	Ciprofloxacin MBC (µg/mL)
Gram-positive bacteria
Staphylococcus aureus ATCC33951	128	256	0.25	0.5
Streptococcus pneumoniae ATCC49619	128	256	0.25	1
Enterococcus faecalis ATCC29212	64	128	0.5	2
Gram-negative bacteria
P. Aeruginosa ATCC15692	ND	ND	0.5	2
Escherichia coli ATCC25922	ND	ND	0.125	0.25
Salmonella enteritidis S4	128	256	0.125	0.5
Klebsiella pneumoniae 911	ND	ND	0.25	0.5

“ND” indicates that no bacteriostatic activity was detected at the maximum tested concentration

Purified SUMO-Hol41 exerted significant dose- and time-dependent antibacterial effects on the tested strains (Fig. 4). The inhibition rates of the three Gram-positive strains (Enterococcus faecalis ATCC 29212, Streptococcus pneumoniae ATCC 49619, and Staphylococcus aureus ATCC 33951) increased markedly with elevated protein concentrations. At the same concentration, the inhibition rate rose rapidly within 2–8 h and plateaued thereafter. Enterococcus faecalis ATCC 29,212 was the most sensitive strain, with an inhibition rate of 93.5% at 128 µg/mL for 8 h and nearly 99.9% at 512 µg/mL for 16 h. Streptococcus pneumoniae ATCC 49,619 and Staphylococcus aureus ATCC 33,951 showed moderate sensitivity, with inhibition rates exceeding 90% at 128 µg/mL for 8 h and over 99.5% at 512 µg/mL for 16 h. In contrast, Salmonella enteritidis S4, the only susceptible Gram-negative strain in this study, was significantly less sensitive to SUMO-Hol41 than Gram-positive bacteria. Although its inhibition rate increased with elevated protein concentrations and prolonged incubation time, the overall increase was slow. The inhibition rate was much lower than that of Gram-positive bacteria under identical conditions, with distinctly different antibacterial kinetic characteristics.

Fig. 4.

Dose- and time-dependent antibacterial activity of SUMO-Hol41 against the tested strains. A Dose- and time-dependent activity against Staphylococcus aureus ATCC 33,951; B Dose- and time-dependent activity against Streptococcus pneumoniae ATCC 49,619; C Dose- and time-dependent activity against Enterococcus faecalis ATCC 29,212; D Dose- and time-dependent activity against Salmonella enteritidis S4

Discussion

The escalating global crisis of antimicrobial resistance (AMR) has created an urgent demand for novel antibacterial agents with unique mechanisms of action. Traditional antibiotics mostly exert antibacterial effects by inhibiting bacterial cell wall synthesis, protein translation, or nucleic acid replication, and their long-term overuse has led to the widespread emergence of multidrug-resistant bacteria. In contrast, perforin kills bacteria mainly by forming transmembrane pores to destroy the integrity of bacterial cell membranes, with a mechanism of action completely different from that of most antibiotics. More importantly, such membrane-disrupting antibacterial proteins are less prone to induce bacterial resistance due to their unique mode of action [31–33]. Therefore, perforin can serve as an effective alternative or complement to conventional antibiotics, providing a new strategy for the control of drug-resistant bacterial infections.

In this study, a SUMO-fused secretory expression system was successfully constructed in K. phaffii, enabling functional, scalable and high-yield production of the holin Hol41, which provides a valuable paradigm for toxic and highly hydrophobic transmembrane proteins [34]. A striking difference in host tolerance to Hol41 was observed between prokaryotic and eukaryotic systems. Expression of either SUMO-Hol41 or non-fused Hol41 in E. coli BL21(DE3) strongly inhibited host growth, with OD₆₀₀ plateauing within 8 h, as Hol41 transmembrane domains (TMDs) insert into the cytoplasmic membrane, oligomerize into pores and cause cell lysis [13, 35]. Hydrophobic interactions and charge distribution in TMDs drive pore assembly in all holin families, making prokaryotic systems unsuitable for Hol41 production [5, 36]. In contrast, K. phaffii GS115 showed no obvious growth inhibition during 72 h of methanol induction, with steadily increasing OD₆₀₀. This tolerance arises from distinct membrane characteristics: K. phaffii membrane is rich in ergosterol (> 30% of total lipids) and long-chain unsaturated fatty acids, which prevent TMD insertion and pore formation [37–39]. It has been documented that ergosterol-rich membranes are resistant to disruption by hydrophobic peptides and toxins [40]. Moreover, secretion mediated by the α-mating factor signal peptide reduces intracellular Hol41 accumulation and avoids non-specific interactions with organelle membranes [41, 42].

Compared with Class II and III holins, the Class I holin Hol41 (containing 3 TMDs) has higher overall hydrophobicity, leading to a prominent solubility challenge. Our results clearly confirm that N-terminal fusion of the SUMO protein is the core strategy to solve this problem [43]. The SUMO tag has remarkable advantages: its small molecular weight (11 kDa) effectively reduces the metabolic burden on host cells; it outperforms maltose-binding protein (MBP) and glutathione S-transferase (GST) in improving the soluble expression level and total yield of difficult-to-express proteins; furthermore, specific SUMO protease enables efficient scarless cleavage, avoiding potential interference of tag residues with the structure and function of the target protein [44, 45]. The hydrophilic surface of SUMO neutralizes hydrophobic interactions between Hol41 TMDs, and its chaperone-like activity promotes correct protein folding; exposed polar amino acids on SUMO (e.g., aspartic acid, glutamic acid, lysine) shield the hydrophobic regions of Hol41 and inhibit intermolecular aggregation [22, 44]. In addition, the flexible hinge region of SUMO reduces misfolding of Hol41 and further improves its solubility.

Western blot analysis showed that SUMO-Hol41 was efficiently secreted into the culture supernatant of K. phaffii in a soluble form, whereas non-fused Hol41 mainly existed as insoluble oligomers. The multiple bands observed in the immunoblot of non-fused Hol41 were speculated to be oligomers with different molecular weights formed by polymerization, which is consistent with studies on other holins [14, 46–48]. For example, multiple bands corresponding to oligomers were detected in the Western blot analysis of the holin from Mycobacterio phage D29, confirming that it forms aggregates via hydrophobic interactions between TMDs [49].

The SUMO tag was retained rather than cleaved by SUMO protease in functional assays, based on two key findings: first, the functional domain of holins is a positively charged region at the C-terminus, which mediates binding to the bacterial cell membrane, and fusion of the SUMO tag at the N-terminus of Hol41 theoretically does not interfere with its C-terminal function [11, 50]. second, free Hol41 undergoes immediate and irreversible aggregation after enzymatic removal of the SUMO tag. Retention of the SUMO tag maintains the solubility, biological function and yield of Hol41, while eliminating additional cleavage and purification steps. The α-mating factor signal peptide integrated in the p PIC9K vector directs efficient secretion of SUMO-Hol41 via the endoplasmic reticulum (ER)-Golgi pathway, which brings two major advantages for large-scale production of Hol41: first, it reduces intracellular toxicity of Hol41 and enables high-cell-density fermentation of P. pastoris, which cannot be achieved in E. coli due to growth arrest induced by holin expression [4, 51, 52]. second, it simplifies downstream purification, as SUMO-Hol41 secreted into the culture supernatant can be captured in one step by Ni-NTA affinity chromatography via the 6×His tag, greatly reducing production costs and promoting industrial translation.

In this study, efficient production of soluble SUMO-Hol41 was achieved in a 2 L fermenter, with a purified yield of 44.95 mg per liter of fermentation broth and a yield coefficient (Y_p/s) was 0.00843 mg protein per mg DCW. The batch-to-batch CV was only 3%, further confirming the high reproducibility of this process. This yield is significantly higher than that of holins reported in prokaryotic expression systems; for instance, the yield of the holin Hol-4086 expressed in E. coli is extremely low, allowing only antibacterial activity assays using crude lysates [32]. The efficient large-scale production of SUMO-Hol41 benefits from the optimized methanol feeding strategy and strict control of fermentation parameters (pH 6.0, dissolved oxygen 20%), highlighting the excellent compatibility of this expression system with industrial bioprocesses [53, 54].

Functional characterization confirmed that SUMO-Hol41 retained potent antibacterial activity, with particularly significant inhibitory effects on Gram-positive pathogens such as Staphylococcus aureus ATCC 33,951, Streptococcus pneumoniae ATCC 49,619, and Enterococcus faecalis ATCC 29,212. Its antibacterial activity against tested strains showed obvious dose- and time-dependence. Notably, SUMO-Hol41 exhibited weak antibacterial activity against Gram-negative bacteria, with only moderate inhibition of its native host Salmonella enteritidis S4. This antibacterial selectivity is consistent with reports on other holins; for example, the holin HolASP mainly lyses its native host Pseudomonas aeruginosa L64 and has negligible effects on other Gram-negative species [55]. The underlying mechanism lies in the unique outer membrane (OM) barrier of Gram-negative bacteria: lipopolysaccharide (LPS) enriched in the OM forms a dense permeability barrier [56]. The negatively charged core oligosaccharides of LPS and their tightly packed structure repel or physically block SUMO-Hol41 from accessing the cell membrane, preventing the formation of functional pores in the inner membrane. This is in line with previous findings: antibacterial agents targeting the inner membrane (e.g., holins) usually show limited activity against intact Gram-negative bacteria unless the OM barrier is disrupted (e.g., via EDTA-mediated divalent cation chelation or combined use with phage-encoded outer membrane-degrading enzymes) [57]. In addition, SUMO-Hol41 inhibited a variety of pathogens, and its antibacterial spectrum is consistent with recent reports: the holin Hol6A3 derived from phage PD6A3 can lyse 83 multidrug-resistant Acinetobacter baumannii strains and also act on E. faecalis, S. aureus and Klebsiella pneumoniae, with a much broader lytic range than its parental phage PD6A3, which only targets A. baumannii [35]. This broad-spectrum characteristic greatly enhances the application potential of SUMO-Hol41 as a broad-spectrum antibacterial agent, but further modifications (e.g., fusion with outer membrane-penetrating peptides) are still required to improve its activity against Gram-negative bacteria.

Despite the progress achieved in this study, several unsolved issues point to directions for future research. First, it is necessary to systematically evaluate the in vivo stability and pharmacokinetic characteristics of SUMO-Hol41 in animal models: similar to other peptide antibacterial agents, SUMO-Hol41 may be easily degraded by serum proteases or rapidly cleared by the reticuloendothelial system, thus limiting its therapeutic efficacy. Strategies such as PEGylation and nanocarrier encapsulation (e.g., liposomes/exosomes) have been proven to enhance the in vivo stability of membrane-active peptides and prolong their circulating half-life, representing potential directions for molecular optimization of SUMO-Hol41 [58, 59]. Second, improving the activity of SUMO-Hol41 against Gram-negative bacteria is crucial. A feasible strategy is the combined use with its cognate endolysin, mimicking the natural holin-endolysin synergistic lysis module of phages [60]. Third, modular optimization of the expression system can further increase yield and expression efficiency, such as optimizing bioreactor fermentation conditions, screening alternative signal peptides with higher secretion efficiency, and constructing SUMO variants resistant to intracellular cleavage by endogenous proteases in K. phaffii [61–63].

Conclusion

In this study, a SUMO-fused secretory expression system was constructed in K. phaffii, which successfully overcame the two core problems of cytotoxicity and poor solubility associated with phage holin Hol41 in prokaryotic expression systems. Consequently, efficient production of soluble SUMO-Hol41 was achieved in a 2-L fermenter, with a purified yield of 44.95 mg/L and a yield coefficient Yₚ/ₛ = 0.00843 mg/mg DCW. The purified SUMO-Hol41 exhibited favorable antibacterial activity against a variety of pathogenic bacteria.

Author contributions

Zhichuang Huo: Writing–original draft; Software; Methodology; Formal analysis. Fangfang Wang: Data curation. Yanze He: Supervision. Zhihui Chang: Investigation. Jinzhe Kang: Supervision. Shasha Feng: Investigation. Cuihong Zhong: Supervision; Methodology; Software. Yuxiang Shi: Supervision; Validation;. Yongying Zhang: Supervision; Validation; Writing–review & editing.

Funding

This work was supported by the Modern Agricultural Industry Technology System (Egg and Poultry Innovation Team) of Hebei Province (Phase III), Project No.: HBCT2024260205.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.